SOCIAL JUSTICE AND SOCIOLOGY 1 Social Justice and ... · roots in a sociology committed to social justice, to cultivate and extend the long-standing “countersystem” approach to

SOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGY 11111

[T]oday the contradictions of American civi-lization are tremendous. Freedom of politi-cal discussion is difficult; elections are notfree and fair. . . . The greatest power in theland is not thought or ethics, but wealth. . . .Present profit is valued higher than futureneed. . . . I know the United States. It is mycountry and the land of my fathers. It is stilla land of magnificent possibilities. It is stillthe home of noble souls and generouspeople. But it is selling its birthright. It isbetraying its mighty destiny. (Pp. 418–19)

Today the social contradictions of Ameri-can and global civilizations are still im-mense. Many prominent voices tell us that itis the best of times; other voices insist that itis the worst of times. Consider how theapologists for modern capitalism now cel-ebrate the “ free market” and the global capi-talistic economy. Some of these analystseven see modern capitalism as the last andbest economic system, as the “ end of his-tory” (Fukuyama 1992). In contrast, from

Direct correspondence to Joe R. Feagin, De-partment of Sociology, Box 117330, Universityof Florida, Gainesville, FL 32611, ([email protected]). I would like to thank the numerous col-leagues who made helpful comments on variousdrafts of this presidential address. Among thesewere Hernán Vera, Sidney Willhelm, BerniceMcNair Barnett, Gideon Sjoberg, Anne Rawls,Mary Jo Deegan, Michael R. Hill, PatriciaLengermann, Jill Niebrugge-Brantley, TonyOrum, William A. Smith, Ben Agger, KarenPyke, and Leslie Houts.

American Sociological Review, 2001, Vol. 66 (February:1–20) 1

Joe R. Feagin

University of Florida

The world’s peoples face daunting challenges in thetwenty-first century. While apologists herald the globaliza-tion of capitalism, many people on our planet experiencerecurring economic exploitation, immiseration, and envi-ronmental crises linked to capitalism’s spread. Across theglobe social movements continue to raise the issues ofsocial justice and democracy. Given the new century’sserious challenges, sociologists need to rediscover their

roots in a sociology committed to social justice, to cultivate and extend the long-standing “countersystem” approach to research, to encourage greater self-reflectionin sociological analysis, and to re-emphasize the importance of the teaching of soci-ology. Finally, more sociologists should examine the big social questions of thiscentury, including the issues of economic exploitation, social oppression, and thelooming environmental crises. And, clearly, more sociologists should engage in thestudy of alternative social futures, including those of more just and egalitarian soci-eties. Sociologists need to think deeply and imaginatively about sustainable socialfutures and to aid in building better human societies.

e stand today at the beginning ofa challenging new century. Like

ASA Presidents before me, I am consciousof the honor and the responsibility that thisaddress carries with it, and I feel a specialobligation to speak about the role of sociol-ogy and sociologists in the twenty-first cen-tury. As we look forward, let me quote W. E.B. Du Bois, a pathbreaking U.S. sociologist.In his last autobiographical statement, DuBois (1968) wrote:

W

Social Justice and Sociology:

Agendas for the

Twenty-First Century

22222 AMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEW

the late 1930s to the 1950s many influentialeconomists and public leaders were commit-ted to government intervention (Keynesian-ism) as the way to counter the negative ef-fects of capitalist markets in the UnitedStates and other countries—effects clearlyseen in the Great Depression of the 1930s.The view that a capitalistic market aloneshould be allowed to make major social andeconomic decisions would then have beenmet with incredulity or derision (George1999; also see Block 1990). Half a centuryago, Karl Polanyi ([1944] 1957), a prescienteconomic historian, critically reviewed thehistory of the free-market idea: “ To allowthe market mechanism to be sole director ofthe fate of human beings and their naturalenvironment, indeed, even of the amountand use of purchasing power, would resultin the demolition of society” (p. 73).

Since the 1960s, conservative businessgroups have pressed upon the world’s politi-cal leaders, and upon the public generally,the idea of a self-regulating market mecha-nism, thereby organizing a successfulcounter-attack against Keynesian ideas(Steinfels 1979). These new apologists forcapitalism have heralded the beneficial as-pects of a globalizing capitalism and haveexported the free-market model in an eco-nomic proselytizing project of grand scope.Free marketeers have persuaded many peopleacross the globe that class conflict is in de-cline and that capitalism and its new tech-nologies will bring prosperity to all coun-tries. Similarly, other influential supportersof the status quo have argued optimisticallythat major forms of social oppression, suchas racial and gender oppression, are also insharp decline in Western societies.

THE DOWNSIDE OF A

CAPITALISTIC WORLD

Nonetheless, many people in the UnitedStates and across the globe insist that this isnot the best of times. Karl Marx long agounderscored the point that modern capital-ism creates bad economic times that encom-pass both social injustice and inequality.Looking at the present day, I will briefly de-scribe a few examples of the troubling con-ditions currently being created or aggravatedby modern capitalism:

Many of the World’s People Still

Live in Misery

First, while it may be the best of times forthose at the top of the global economy, it isnot so for the majority of the world’speoples. The pro-capitalist polices of manynational governments and international orga-nizations have fostered a substantial transferof wealth from the world’s poor and work-ing classes to the world’s rich and affluentsocial classes. Social injustice in the form ofmajor, and sometimes increasing, inequali-ties in income and wealth can be observedacross the globe. Thus, in the United Statesincome inequality has reached a record levelfor the period during which such data havebeen collected: The top one-fifth of house-holds now has nearly half the income; thebottom one-fifth has less than 4 percent.Moreover, the top 1 percent of U.S. house-holds holds more in wealth than the bottom95 percent, and the wealthy have doubledtheir share since 1970. Moreover, moreAmericans live in poverty than a decade ago.As of the late 1980s, 31.5 million peoplelived at or below the officially defined pov-erty level, while in 1999 the figure had in-creased to 34.5 million (Collins, Hartman,and Sklar 1999; Oxfam 1999). In recent de-cades the number of millionaires and billion-aires has grown dramatically. Yet many or-dinary workers have seen their real wagesdecline—even while the costs of housing,transportation, and medical care have in-creased significantly in real terms.

Of the 6 billion people on earth, a largeproportion live in or near poverty and desti-tution, with 1.2 billion living on less than onedollar a day. The numbers living in povertyare increasing in areas of South Asia, Africa,and Latin America. Today one-fifth of theworld’s people, those in the developed coun-tries, garner 86 percent of the world’s grossdomestic product, with the bottom fifth gar-nering just one percent. In recent years theworld’s richest 200 people, as a group, havedoubled their wealth, to more than 1 trilliondollars for the year 2000 (Oxfam 1999).While there has been much boasting abouteconomic growth among those pushing glo-bal capitalism, between 1980 and the late1990s most of the world’s countries saw sus-tained annual growth rates of less than 3 per-


cent per capita, and 59 countries actually ex-perienced economic declines (Toward Free-dom 1999). Moreover, in most countriesgreat income and wealth inequalities createmajor related injustices, including sharp dif-ferentials in hunger, housing, life satisfac-tion, life expectancy, and political power.

Viewed from a long-term perspective, thehigh levels of wealth and income inequality,and the increase in that inequality, signal yetanother critical point in human history wherethere is a major foregrounding of social jus-tice issues.

Working Families Are Exploited and

Marginalized

Second, global capitalism may bring the bestof times for corporate executives and thewell-off, yet for many of the world’s peopleit brings recurring economic disruption, ex-ploitation, marginalization, and immis-eration. The international scene is increas-ingly dominated by highly bureaucratizedmultinational corporations, which often op-erate independently of nation states. Work-ing for their own economic interests, thesetransnational corporations routinely “ de-velop” their markets—and destroy and dis-card regions, countries, peoples, cultures,and natural environments. For example,transnational corporations now control muchof the world’s agricultural system. In devel-oping countries small farmers are shovedaside by large agribusiness corporations orare pressured to produce crops for an inter-national market controlled by big trans-national corporations—thereby reducing theproduction of essential foodstuffs for localpopulations (Sjoberg 1996:287).

Today there are an estimated 1 billion un-employed or underemployed workers aroundthe world, with 50 million unemployed inthe European countries alone. Hundreds ofmillions, including many millions of chil-dren, work in onerous or dangerous work-places. Some 30 million people die fromhunger annually in a world whose large ag-ricultural enterprises produce more thanenough food for every person (Ramonet1999). The real effects of expanding capital-ism for a large proportion of the planet’s in-habitants are not only greater inequality butalso job restructuring, unsafe working con-

ditions, low wages, underemployment or un-employment, loss of land, and forced migra-tion. Ordinary working people and theirfamilies—in most nationality, racial, andethnic groups across the globe—face signifi-cant negative social impacts from an encir-cling capitalism.

Capitalism Imposes Huge

Environmental Costs

Third, the global capitalistic economy gen-erates profits at the huge cost of increasingenvironmental degradation. Since the 1970s,the levels of some greenhouse gases (e.g.,carbon dioxide) in the earth’s atmospherehave grown significantly because of the in-creasing use of fossil fuels, widespread de-forestation, and industrial pollution. Globalwarming, which results from this increase ingreenhouse gases, is melting polar ice packs,increasing coastal flooding, generating se-vere weather, creating droughts and reshap-ing agriculture, and facilitating the spread ofdisease. In addition, as a result of human ac-tions, the earth’s ozone layer is severely de-pleted in some areas. This alone results in arange of negative effects, including in-creases in skin cancer incidence and majorthreats to essential species, such as phy-toplankton in the oceans (M. Bell 1998;Hawken, Lovins, and Lovins 1999).

A lack of sufficient water and poor waterquality are large-scale problems in manycountries. Half the world’s wetlands andnearly half the forests have been destroyedin just the last century. The destruction offorests is killing off many plant species, in-cluding some supplying the oxygen webreathe. The consequences of these environ-mental changes will be the greatest for theworld’s poorest countries, many of whichare in areas where the increasing heat of glo-bal warming is already having a serious im-pact on water availability, soil erosion, de-struction of forests, agriculture, and thespread of disease (Sachs 1999).

Today, some environmental experts are se-riously discussing the possibility that mostof the planet’s plant and animal species willbe gone by the twenty-second century. JaredDiamond, a leading physical scientist, hasreviewed the evidence and concludes thatmovement toward an environmental catas-


trophe is accelerating. The only question, inhis view, is whether it is likely to “ strike ourchildren or our grandchildren, and whetherwe choose to adopt now the many obviouscountermeasures” (Diamond 1992:362). Andthere are yet other related problems facinghumanity, such as those arising out of thenew technologies associated with world-wide, capitalist-led economic development.

Global Capitalism Reinforces

Other Injustice and Inequality

Fourth, in addition to the economic and en-vironmental inequalities generated or aggra-vated by contemporary capitalism, otherforms of social injustice and inequality re-main central to the United States and othersocieties. I only have space here to notebriefly such major societal realities as racialand ethnic oppression, patriarchy, homopho-bia, bureaucratic authoritarianism, violenceagainst children, and discrimination againstthe aged and the disabled. These persistingforms of discrimination and oppression gen-erally have their own independent social dy-namics, yet they too are often reinforced orexacerbated by the processes of moderncapitalism.

WHAT KIND OF A WORLD

DO WE WANT?

The world’s majority now lives, or soon willlive, in difficult economic and environmen-tal times. By the end of the twenty-first cen-tury, it is likely that there will be sustainedand inexorable pressures to replace the so-cial institutions associated with corporatecapitalism and its supporting governments.Why? Because the latter will not have pro-vided humanity with just and sustainable so-cieties. Such pressures are already buildingin the form of grassroots social movementsin many countries.

A few of the world’s premier capitalistsalready see the handwriting on the wall.The billionaire investor George Soros(1998), for instance, has come to the con-clusion that free markets do not lead tohealthy societies:

Markets reduce everything, including humanbeings (labor) and nature (land), to com-

modities. We can have a market economy,but we cannot have a market society. In ad-dition to markets, society needs institutionsto serve such social goals as political free-dom and social justice. (P. 24)

As Soros sees it, without a more egalitarianglobal society, capitalism cannot survive.

In a recent interview, Paul Hawken (Haw-ken and Korten 1999), an environmentallyoriented critic of modern capitalism, has re-counted the story of a business consultantwho conducted a workshop with middlemanagers in a large corporation that makes,among other things, toxic chemicals such aspesticides. Early in the workshop the execu-tives discussed and rejected the idea that cre-ating social justice and resource equity is es-sential to the long-term sustainability of asociety such as the United States. Later,these managers broke into five groups andsought to design a self-contained spaceshipthat would leave earth and return a centurylater with its occupants being “alive, happy,and healthy” (Hawken and Korten 1999).The executives then voted on which group’shypothetical spaceship design would bestmeet these objectives.

The winning design was comprehensive:It included insects so no toxic pesticideswere allowed on board. Recognizing the im-portance of photosynthesis, the winninggroup decided that weeds were necessary fora healthy ecosystem, so conventional herbi-cides were not allowed. The food systemwas also to be free of toxic chemicals. Thesemanagers “also decided that as a crew, theyneeded lots of singers, dancers, artists, andstorytellers, because the CDs and videoswould get old and boring fast, and engineersalone did not a village make.” In addition,when the managers were asked if it was rea-sonable to allow just one-fifth of those onboard to control four-fifths of the ship’s es-sential resources, they vigorously rejectedthe idea “as unworkable, unjust, and unfair”(Hawken and Korten 1999).

Note that this example spotlights the criti-cally important ideas of human and environ-mental interdependence and of social justice.Even these corporate managers, when hypo-thetically placing themselves in the closedsystem of a spaceship, rejected environmen-tal degradation, a boring monoculture, andmajor resource inequalities.


As I see it, social justice requires re-source equity, fairness, and respect for di-versity, as well as the eradication of exist-ing forms of social oppression. Social jus-tice entails a redistribution of resourcesfrom those who have unjustly gained themto those who justly deserve them, and italso means creating and ensuring the pro-cesses of truly democratic participation indecision-making. A common view in West-ern political theory is that, while “ thepeople” have a right to self-rule, they del-egate this right to their representatives—tothe government leaders who supposedly actin the public interest and under the guid-ance of impartial laws (Young 1990:91–92).However, there is no impartial legal and po-litical system in countries like the UnitedStates, for in such hierarchically arrangedsocieties those at the top create and main-tain over time a socio-legal framework andpolitical structure that strongly support theirgroup interests. It seems clear that only adecisive redistribution of resources and de-cision making power can ensure social jus-tice and authentic democracy.

The spaceship example explicitly recog-nizes the interdependence of human beingsand other living species. For some decadesnow central ideas in physics and biologyhave stressed the interconnectedness of whatwere once thought to be discrete phenomena.Thus, the “gaia theory” in biology suggests,according to Lovelock (1987), that

. . . the entire range of living matter onEarth, from whales to viruses, and from oaksto algae, could be regarded as constituting asingle living entity, capable of manipulatingthe Earth’ s atmosphere to suit its overallneeds and endowed with faculties and pow-ers far beyond those of its constituent parts.(P. 9)

This is more than a metaphorical descrip-tion, for in fact we live on a planet that, weare increasingly realizing, is truly interwo-ven. All of earth’s aspects—from biosphere,to soils and oceans, to atmosphere—are seenas parts of one interconnected living systemwith important cybernetic features. Thus,environmental irresponsibility in one place,such as the excessive burning of fossil fuelsin the United States, contributes to negativeeffects elsewhere, such as to global warm-ing in Australia.

Perhaps there are clues in the gaia theoryfor a broader sociological framework forviewing the development of human societ-ies. We human beings are not just part of aninterconnected biosphere, but are also linkedin an increasingly integrated and global webof structured social relationships. This com-plex “ sociosphere” consists of some 6 billionpeople living in many families and commu-nities in numerous nation states. Nationstates and their internal organizations arelinked across an international web. Indeed,we human beings have long been more in-terconnected than we might think. Accord-ing to current archaeological assessments,we all descended from ancestors who mi-grated out of Africa some millennia in thepast. Today, most human beings speak re-lated languages; about half the world’speople speak an Indo-European language. Inrecent decades the expansion of telecommu-nication technologies has placed morepeople in potential or actual contact with oneanother than ever before. For the first timein human history, these technologies are rap-idly creating one integrated body of human-ity (Sahtouris 1996).

Yet, this increasingly interconnectedsociosphere remains highly stratified: Greatbenefits accrue to those classes dominant ininternational capitalism. Today most of theglobe’s political and business leaders, aswell as many of its academic experts, havecome to accept capitalism as the more orless inevitable economic system for allcountries. However, at the same time, grow-ing numbers of people are recognizing that,because of globalizing capitalism, the earthis facing a massive environmental crisis,one that has the potential to destroy the ba-sic conditions for human societies within acentury or two. Issues of ecological de-struction—as well as broader issues of so-cial inequality and injustice—are beingforced to the forefront not by corporate ex-ecutives but by some 30,000 people’sgroups and movements around the globe.These include environmental groups, indig-enous movements, labor movements,health-policy groups, feminist groups, anti-racist organizations, and anti-corporategroups (Klein 2000). Such groups agree onmany critical environmental and political-economic goals.


Indeed, many people in other regions ofthe world seem to be ahead of us in theUnited States in their understanding of thedamage done by the unbridled operations ofmultinational corporations. These groups arepressing for meaningful international decla-rations and treaties, such as the variousUnited Nations declarations on the environ-ment and human rights. In the United Statesawareness of the negative impact of global-izing capitalism is now substantial and maybe growing. A 1999 U.S. poll found that justover half the respondents said they weresympathetic with the concerns of activistswho had aggressively protested a recentWorld Trade Organization summit in Seattle(Business Week 1999). In many places in theUnited States today there is growing opposi-tion to the economic and environmental de-cisions of those executives heading trans-national corporations.

Unquestionably, social justice appears asa recurring concern around the globe. Forthat reason alone, we sociologists must vig-orously engage issues of social justice or be-come largely irrelevant to the present andfuture course of human history.

A LONG TRADITION: SOCIOLOGY

AND SOCIAL JUSTICE

Given impending national and internationalcrises, sociology appears to be the right dis-cipline for the time. Sociology is a broad in-terdisciplinary field that draws on ideas fromother social sciences, the humanities, and thephysical sciences. Our intellectual and meth-odological pluralism, as well as our diver-sity of practitioners, are major virtues. Suchrichness gives sociology a particularly goodposition as a science to examine the com-plexities and crises of a socially intercon-nected world. Those sciences with diverseviewpoints and constructive conflicts overideas and issues have often been the mostintellectually healthy. As P. H. Collins(1998) has put it, “Sociology’s unique sociallocation as a contested space of knowledgeconstruction allows us to think through newways of doing science” (p. 10; also seeBurawoy 1998).

Views of sociology’s goals have long re-flected a dialectical tension between a com-mitment to remedy social injustice and the

desire to be accepted as a fully legitimatediscipline in the larger society, especially bypowerful elites. The lead article in the July1895 issue of the American Journal of Soci-ology, written by Albion Small, founder ofthe first graduate sociology department (atthe University of Chicago), listed among themajor interests of the journal editors theanalysis of “plans for social amelioration”(Small 1895:14). A decade later, Small pre-sented a paper at the American SociologicalSociety’s first meeting in which he arguedvigorously that social research was not anend in itself but should serve to improve so-ciety (Friedrichs 1970:73). Small was notalone in this commitment. In the first decadeor two of U.S. sociology, leading scholarsadvocated the pursuit of knowledge for itsown sake and the assessment of that knowl-edge in relation to its current usefulness tosociety.

Moreover, from the beginning there hasbeen a robust “ countersystem” traditionwithin U.S. sociology—a tradition whoseparticipants have intentionally undertakenresearch aimed at significantly reducing oreliminating societal injustice. The counter-system approach is one in which social sci-entists step outside mainstream thought pat-terns to critique existing society (Sjobergand Cain 1971). From the perspective of thisresearch tradition, social scientists have alltoo often accepted the status quo as theirstandard. It is noteworthy too that muchcountersystem analysis develops ideas aboutalternative social systems. For instance, anyserious exploration of the countersystem tra-dition must acknowledge the past and cur-rent influence of Marx’s critical analysis ofcapitalism, which included ideas about analternative social system. Marx’s counter-system analysis has, directly or indirectly,influenced many social scientists, includingseveral of the sociologists to whom I nowturn.

In the late nineteenth and early twentiethcenturies, a number of white women, blackmen, and black women sociologists—as wellas a few white male sociologists—did muchinnovative sociological research and at thesame time took strong informed positions inregard to ending the oppression of women,black Americans, the poor, and immigrants.Among the now forgotten women and black


male sociologists were Jane Addams, Flo-rence Kelley, Emily Greene Balch, Ida B.Wells-Barnett, Charlotte Perkins Gilman,and W. E. B. Du Bois. All were practicingsociologists, and all developed important so-ciological ideas and research projects. Mostwere members of the American SociologicalSociety (Deegan 1987).

Jane Addams was a key founder of U.S.sociology. Head resident of Chicago’s pio-neering Hull-House complex, she was an ac-tive sociologist and charter member of theAmerican Sociological Society. She inter-acted professionally with other leading soci-ologists and intellectuals. During the 1890sand later, there was great intellectual fer-ment at Hull-House. Not only were unionleaders, socialists, and other social reform-ers welcomed there, but a few major malesocial theorists, such as John Dewey andGeorge Herbert Mead, regularly interactedwith the women sociologists there (Deegan1988:5). Addams was one of the first U.S.sociologists to deal conceptually and empiri-cally with the problems of the burgeoningcities, and she was advanced in her socio-logical analysis of justice and democracy.She viewed democracy as entailing morethan fairness and legal equality:

We are brought to a conception of Democ-racy not merely as a sentiment which desiresthe well-being of all men, nor yet as a creedwhich believes in the essential dignity andequality of all men, but as that which affordsa rule of living as well as a test of faith.(Addams 1902:6)

In her view ordinary Americans had to par-ticipate actively in major decisions affectingtheir lives for there to be real democracy.

Addams and the numerous women (and afew men) sociologists working at Hull-House not only accented a cooperative anddemocratic model of society but also usedtheir sociological research and analysis toground their efforts for tenement reform,child-labor legislation, public health pro-grams, feminism, and anti-war goals (Dee-gan 1988). They worked in immigrant andother poor urban communities and sought tobuild a grassroots base for social change.Moreover, working in collaboration, they didthe first empirical field research in U.S. so-ciology. Like Harriet Martineau earlier in thenineteenth century (see below), Addams and

her colleagues accented a new sociologicaltradition that developed empirical data in or-der to better deal with issues of both socialtheory and public policy. Their 1895 book,Hull-House Maps and Papers (Residents ofHull-House [1895] (1970), reported on thesociodemographic mapping of Chicago’s ur-ban areas well before that statistical ap-proach became important for the Universityof Chicago’s male sociologists. Interestingly,these sociodemographic data were used tohelp local residents understand their commu-nity patterns, not just to provide data forpublications in academic journals. More-over, one indication of the disciplinary im-pact of these early women sociologists isthat between 1895 and 1935 they publishedmore than 50 articles in what was then theleading sociology journal, the AmericanJournal of Sociology (Deegan 1988:47).

In 1896 W. E. B. Du Bois became an as-sistant in sociology at the University ofPennsylvania. Du Bois was hired to do astudy of black Philadelphians using, as henoted, the “best available methods of socio-logical research” (Du Bois [1899] 1973:2).His book, The Philadelphia Negro ([1899]1973), was the first empirical study of ablack community to be reported in sociologi-cal depth and at book length. Therein DuBois not only analyzed sociological data onpatterns of life in the black community (in-cluding racial discrimination) but also as-sessed what he viewed as the immorality ofdiscrimination. The last part of this path-breaking book includes a study of domesticworkers by Du Bois’s white colleague IsabelEaton, a former Hull-House sociologist. Theresearch collaboration of these early blackand white sociologists is also part of the nowforgotten history of sociology. Moreover, inspite of Du Bois’s stellar qualifications—major sociological research, a HarvardPh.D., and work with leading European so-cial scientists—no white-run sociology de-partment offered him a regular position.Over time, Du Bois would make very impor-tant contributions to the sociological studyof community, family, social problems, andclass relations, as well as to the historicalstudy of slavery and Reconstruction.

We should recognize too that in this earlyperiod there were important black womensociologists, such as Ida B. Wells-Barnett


and Anna Julia Cooper, whose work has re-cently been rediscovered (Lemert and Bhan1998; Lengermann and Niebrugge-Brantley1998). Though neither was affiliated withacademic sociology, both were practicingsociologists and theorists of society. In theirwork they were among the earliest social sci-entists to analyze data on the conditions ofAfrican Americans and of women in U.S.society in terms of social “ subordination”and “ repression” (Cooper 1892; Wells-Barnett 1895).

By the 1920s and 1930s leading whitemale sociologists were downplaying or ig-noring the pioneering sociological work ofthe early countersystem sociologists. For ex-ample, the dominant introductory textbookof the interwar decades, Park and Burgess’s(1921) lengthy Introduction to the Science ofSociology, views sociology as an academicand abstract science. This text contains in its1,040 pages only a few bibliographical ref-erences to the work of Du Bois, but no dis-cussion of his research work, and only oneterse sentence on, and two bibliographicalreferences to, the work of Addams.

Park and other prominent sociologistswere increasingly critical of an activist soci-ology and were moving away from a con-cern with progressive applications of socialresearch toward a more “detached” sociol-ogy. Their work was increasingly linked tothe interests of certain corporate-capitalistelites, such as those represented by theRockefeller family foundations. While theyfrequently researched various types of urban“ disorganization,” usually in qualitativefield studies, they rarely analyzed deeply theharsher realities of social oppression—espe-cially gender, class, and racial oppression—in the development of cities. Park and sev-eral of his colleagues played a major role inshifting the emphasis from a sociology con-cerned with studying and eliminating serioussocietal problems to a more detached andacademic sociology concerned with “natu-ral” social forces—without the humanitarianattitude or interpretation of what Park some-times called the “ damned do-gooders”(Lengermann and Niebrugge-Brantley 1998:15–18; Raushenbush 1979:96).

Moreover, during the 1920s and 1930ssupport for a detached and instrumental-positivist sociology increased at major U.S.

universities. This approach is “ instrumental”in that it limits social research mainly tothose questions that certain research tech-niques will allow; it is “positivist” in that itcommits sociologists to “ rigorous” researchapproaches thought to be like those used inthe physical sciences (Bryant 1985:133). Apioneer in this approach was Franklin H.Giddings at Columbia University. In an early1900s’ American Journal of Sociology dis-cussion, Giddings (1909) argued, in stronglygendered language,

We need men not afraid to work; who willget busy with the adding machine and thelogarithms, and give us exact studies, suchas we get in the psychological laboratories,not to speak of the biological and physicallaboratories. Sociology can be made an ex-act, quantitative science, if we can get in-dustrious men interested in it. (P. 196, ital-ics in original)

By the 1920s the influential William F.Ogburn, who trained at Columbia Universityunder Giddings and was later hired at theUniversity of Chicago, aggressively arguedfor such a detached and quantitative researchapproach. In his 1929 presidential address tothe American Sociological Society he calledfor a sociology emphasizing statistical meth-ods and argued that sociologists should notbe involved as sociologists in improving so-ciety; instead they should focus on effi-ciently discovering knowledge about society.Whoever is in power, “ some sterling execu-tive,” might then apply this objective socio-logical research (Bannister 1992:188–90).Survey methods and statistical analyses weregradually becoming the emphasized and pre-ferred research strategies in mainstream so-ciology.

Over the next few decades, most main-stream sociologists, including those in lead-ing departments, did not research majorpublic events and issues, especially from acritical perspective. One study of 2,559 ar-ticles appearing in the American Sociologi-cal Review from 1936 to 1984 examinedmajor social and political events for five pe-riods within this time frame—events such asthe Great Depression and McCarthyism—and found that overall only 1 in 20 articlesdealt with the major events examined forthese periods (Wilner 1985). Moreover, fromthe 1920s to the 1940s remarkably few of the


leading U.S. sociologists researched, orspoke publicly and critically of, thegrowing fascist movements in the UnitedStates and Europe, some of which wouldsoon help generate a catastrophic war. Ap-parently, one reason for this neglect was theincreasing emphasis on a “ value-free,”“pure-science” approach to sociology (Ban-nister 1992:175–89).

Still, some important critics emerged.Writing in the early 1940s in an appendix tohis An American Dilemma, Gunnar Myrdal([1944] 1964) specifically criticized themove by Park and Ogburn toward a moredetached sociology:

The specific logical error is that of inferringfrom the facts that men can and should makeno effort to change the “natural” outcome ofthe specific forces observed. This is the olddo-nothing (laissez-faire) bias of “ realistic”social science. (P. 1052)

Anticipating later discussions and debates,Myrdal developed a critique of the new ac-cent on a “value-free” social science:

Scientific facts do not exist per se, waitingfor scientists to discover them. A scientificfact is a construction abstracted out of acomplex and interwoven reality by means ofarbitrary definitions and classifications. Theprocesses of selecting a problem and a basichypothesis, of limiting the scope of study,and of defining and classifying data relevantto such a setting of the problem, involve achoice on the part of the investigator.([1944] 1964:1057)

As Myrdal viewed the matter, value neu-trality in social science is impossible, for inmaking choices about how to assess and re-search society there is always something ofvalue at stake. While scientific conventionsprovide guidelines for choices, they neces-sarily involve value judgements, and no onecan avoid value judgments simply by focus-ing on just social “ facts.”

By the 1930s and 1940s the critical, coun-tersystem approaches of sociologists likeAddams and Du Bois were losing out to apolitically safe, academic, and distancingsociology. Sociology was increasingly be-coming a discipline whose college and uni-versity departments were dominated bywhite male sociologists and often linked toelite interests—including ties such as grants

from corporate foundations and governmentagencies. As Deegan (1988) has noted re-garding the dominant sociologists at the Uni-versity of Chicago,

These later men therefore condemned politi-cal action for sociologists, while the ideasof the elite, in fact, permeated theirwork. . . . Rather than condemn the exploi-tation and oppression of daily life, the laterChicago men described it. They justified itthrough their acceptance of it. (P. 304)

In the decades after World War II, manymainstream sociologists continued the movetoward the pure-science ideal and away fromthe concerns for social justice and the mak-ing of a better society. There was a great ex-pansion of federally funded research in thephysical sciences, and leading sociologistsworked aggressively to grasp a share of thenew federal money, often by stressing an in-strumental-positivist sociology that at-tempted to imitate those physical sciences.In the late 1950s some 15 prominent socialscientists, including leading sociologists,signed onto a statement, “National Supportfor Behavioral Science,” which pressed theU.S. government for funds for social sci-ence:

We assume the probability of a break-through in the control of the attitudes andbeliefs of human beings. . . . This could be aweapon of great power in Communist hands,unless comparable advances in the Westproduce effective counter-measures.(Quoted in Friedrichs 1970:88)

Contrary to their statements elsewhere aboutvalue neutrality, the political orientation ofthese and other influential social scientists ofthe time made transparent the centrality ofvalues that were then shaping social scienceresearch.

Also evident is the strong interest of lead-ing social scientists in state-funded research.These researchers were largely successful intheir efforts, and substantial bureaucracieshave developed to fund social science re-search under the auspices of the federal gov-ernment and private foundations. This gov-ernment and corporate underwriting of muchmainstream sociological research has fed theemphasis on a quantitatively oriented or in-strumental-positivist sociology and on soci-

1 01 01 01 01 0 AMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEWAMERICAN SOCIOLOGICAL REVIEW

ologists as research entrepreneurs. Not sur-prisingly, social scientists who have securedmajor funding from federal governmentagencies and large corporate foundationshave rarely done research that draws on thecountersystem tradition and is strongly criti-cal of established institutions in the corpo-rate or governmental realms. From the 1930sto the present, the accent on academic grant-getting, the heavy emphasis on certain typesof quantitatively-oriented research, and themovement away from the social justice con-cerns of earlier sociologists have been asso-ciated trends (see Cancian 1995).

A detached-science perspective has beeninfluential in many areas of sociology forsome decades now, but not without strongcountering perspectives (e.g., see Vaughan1993). Since the late 1960s there has been aperiodic resurgence of interest in an activistsociology, including an increased concernwith research on (and the eradication of) in-stitutional discrimination and other forms ofsocial oppression (e.g., see Omi and Winant1994). Significantly, the recent history ofsociology has been dialectical, with support-ers of the detached-science perspective of-ten being central, yet regularly challengedby those advocating a sociology committedto both excellent sociological research andsocial justice.

AGENDAS FOR SOCIOLOGY:

THE NEW CENTURY

Looking toward the next few decades, I seeimportant conceptual, empirical, policy, andactivist tasks for which the rich diversity ofcontemporary sociology can help prepare us.These tasks often relate to questions of so-cial justice. Indeed, one major reason thatsome subfields of sociology are periodicallyattacked by conservative, and often ill-in-formed, journalists and media commentatorsis that analyses of discrimination, domina-tion, and social justice are generally threat-ening to those who desire to maintain the sta-tus quo. Moreover, we should keep in mindthat sociologists have already had a broadimpact. Sociological ideas and research arefrequently used in public discourse by thosegrappling with societal problems, and soci-ology books are more widely reviewed (andperhaps even read) outside the discipline

than any other social science books excepthistory books (Bressler 1999:718).

Let us now consider a few of the socially-relevant agendas for the twenty-first centurythat can be inaugurated or accelerated by so-ciologists with many different research per-spectives and methods.

Bring Social Justice Back

to the Center

First, it is time for the discipline to fully re-cover and celebrate its historical roots in asociology committed to social justice in ide-als and practice. In recent decades no soci-ologist has published even one substantialarticle in a major sociology journal (e.g., theAmerican Journal of Sociology, AmericanSociological Review, and SociologicalTheory) on the sociological ideas of thewomen sociologists in the founding genera-tion (Lengermann and Niebrugge-Brantley2001). It is time for us sociologists to rem-edy this neglect and help to reclaim the im-portant ideas of those women sociologistsand sociologists of color who are among thefounders of our discipline.

A strong case can be made that the Britishsocial scientist Harriet Martineau (1802–1876) is the founder of empirical sociologyin the West. She was apparently the first so-cial scientist both to use the term sociologyand to do systematic sociological research inthe field (Hoecker-Drysdale 1992). Shehelped to invent a new sociological ap-proach that brought empirical data to bearon questions of social theory and publicpolicy. She wrote the first book on socio-logical research methods (Hill 1989), inwhich she argued—preceding Emile Durk-heim by half a century—that research on so-cial life is centrally about studying social“ things” accurately and should involve re-search on “ institutions and records, in whichthe action of a nation is embodied and per-petuated” (Martineau [1838] 1989:73).She was a contemporary of Auguste Comteand translated his major work on positivephilosophy (sociology) into English.Martineau’s first major sociological analysiswas based on observations from a field tripacross the United States—a multi-volumeset titled Society in America (1837). In thatwork she developed sociological insights as

SOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGYSOCIAL JUSTICE AND SOCIOLOGY 1 11 11 11 11 1

penetrating and original as those of her morecelebrated male counterpart, Alexis deTocqueville. Martineau was also a feministtheorist and anti-slavery activist, and wroteextensively and sociologically on social is-sues for the general public.

Contemporary sociologists should alsorecognize the importance of, and draw morefrom, the ideas of early U.S. sociologistslike Jane Addams and W. E. B. Du Bois. AsI noted previously, these pioneering U.S.sociologists offer solid role models in theirdual commitments to social-scientificknowledge and to social justice, equality,and democracy. They gave central attentionto the theoretical, empirical, and policy di-mensions of sociological research. Thework of the early women and AfricanAmerican sociologists, as well as that ofprogressive white men, may well point ustoward a new conceptual paradigm for soci-ology. Such a paradigm would accent thecentrality of differences, oppressions, andinequalities—as well as recurring move-ments for social justice—within societieslike the United States.

It is also time that we recognize these so-cial justice themes in the writings of someof the classical “ founding fathers” of sociol-ogy. For example, Emile Durkheim has of-ten been portrayed in relatively conservativeterms, as being principally concerned withsocial order and stability. Yet Durkheimwrote eloquently about the impetus for so-cial justice in societies. He argued that aforced division of labor, like that found in aclass-riven society, was pathological andmade organic solidarity impossible. Socialinequality, created by such social mecha-nisms as routine inheritance across genera-tions, compromises organic solidarity. ForDurkheim ([1893] 1933:384–88) organicsolidarity and social justice require theelimination of inequalities not generated byvariations in personal merit:

If one class of society is obliged, in order tolive, to take any price for its services, whileanother can abstain from such action thanksto the resources at its disposal which, how-ever, are not necessarily due to any socialsuperiority, the second has an unjust advan-tage over the first at law. . . . [The] task ofthe most advanced societies is, then, a workof justice. . . . [O]ur ideal is to make social

relations always more equitable, so as to as-sure the free development of all our sociallyuseful forces. (P. 387)

A successful movement to complex organicsocieties requires ever more social justice,

. . . and we can be sure that this need willbecome ever more exacting if, as every factpresages, the conditions dominating socialevolution remain the same. (P. 388)

Nurture the Countersystem

Approach

Second, contemporary sociologists need toenlarge and cultivate the long-standingcountersystem approach, not only in regardto investigating social inequality and injus-tice but also in regard to assessing alterna-tive social systems that may be more just.

Today, the sociology handbooks and ency-clopedias on my bookshelves have little tosay about the concept of social justice. Onesignificant task for social scientists is todocument empirically, and ever more thor-oughly, the character of major social injus-tices, both nationally and internationally. Wealso need more conceptual work that devel-ops and enriches the concepts of social jus-tice and equality. In my view, social justiceis not only a fundamental human right but isalso essential for a society to be sustainablein the long term. Even the corporate execu-tives in the aforementioned spaceship ex-ample developed some understanding thatjustice and equity are essential to the long-term sustainability of a social system.

As I have suggested above, social injus-tice can be examined not only in terms ofthe maldistribution of goods and services,but also in regard to the social relations re-sponsible for that maldistribution. These so-cial relations, which can range from cen-trally oppressive power relations to less cen-tral mechanisms of discrimination, deter-mine whether individuals, families, andother groups are excluded from society’s im-portant resources and decision making pro-cesses. They shape the development ofgroup and individual identities and the senseof personal dignity. In the end, social justiceentails a restructuring of the larger frame-works of social relations generally (Feaginand Vera forthcoming).


We sociologists have made a good starttoward understanding certain types of socialinjustice and inequality. Some of us havedone considerable work to document thecharacter and impact of class, racial, andgender subordination. In the United Statesand in Latin America some sociologists us-ing participatory-action-research strategieshave honed countersystem ideas and meth-ods and worked interactively with people atthe grassroots level seeking assessments of,and alternatives to, an onerous status quo(e.g., Fals-Borda 1960). The commitmenthere is to get out of the ivory tower and tohelp build a resource and power base forthe disenfranchised in their communities.The legitimacy of this type of sociologicalresearch must be enhanced. As one group ofparticipatory-action researchers has put it,“To map and analyze the dimensions of so-cial problems . . . is seen as scientific re-search. To discuss and describe alternativepractices and develop solutions is seen asmoving toward politics and advocacy—ar-eas that are perceived as a threat to the ob-jectivity of research” (Nyden et al. 1997;also see Stoecker 1996). Collaborative re-search between sociologists and communitygroups seeking solutions to serious localproblems of housing, work, education, pov-erty, discrimination, and environmental pol-lution should not be shoved aside, as itsometimes is, with cavalier comments aboutsociological “ do-goodism,” but should beplaced in the respected core of sociologicalresearch—where it was at the birth of U.S.sociology.1

In everyday practice all sociology is amoral activity, whether this is recognized ornot. In a society deeply pervaded and struc-tured by social oppressions, most sociologi-cal research will reflect these realities tosome degree, and attempts to deny these re-alities or their impact on research are mis-guided at best. All social science perspec-tives have an underlying view of what theworld ought to be. As Moore (1971) noted,

[Questions] that arouse human passions, es-pecially in a time of change, have had to dowith the forms of authority and justice, andthe purposes of human life. . . . It is impos-sible therefore to avoid taking some kind ofa moral position, not only in writing aboutpolitics but also in not writing about them.(P. 3)

A countersystem approach attempts to assessthe status quo from a viewpoint at leastsomewhat outside the frame of the existingsociety and/or nation state. In practice, so-cial scientists can accept the prevailing na-tion-state or bureaucratic-capitalistic moral-ity or they can resist this morality by mak-ing a commitment to social justice and hu-man rights. Contemporary countersystemapproaches often accent a broad humanrights framework in which each person isentitled to fair treatment and justice simplybecause they are human beings, not becausethey are members of a particular nation-state. Moreover, some social scientists (e.g.,Sjoberg 1996) have suggested that the Uni-ted Nation’s Universal Declaration of Hu-man Rights—with its strong array of social,political, and economic rights—may be agood starting place for developing a robusthuman rights framework for social scienceresearch.

We should seek a sociology that isgrounded in empirical and theoretical re-search and that hones a critical perspectiveless restricted by established institutions.Careful data collection, reasoned argument,and critical moral judgments are not incom-patible. The great sociologist of race andclass, Oliver C. Cox, underscored this point:

Clearly, the social scientist should be accu-rate and objective but not neutral; he [or she]should be passionately partisan in favor ofthe welfare of the people and against the in-terests of the few when they seem to sub-merge that welfare. (Cox 1948:xvi)

Numerous sociologists, from Jane Addamsand W. E. B. Du Bois, to Robert and HelenLynd and Gunnar Myrdal, to more contem-porary scholars as diverse as AlfredMcClung Lee, Jessie Bernard, James Black-well, Robert Bellah, and Orlando Fals-Bordahave accented the importance of bringingmoral discourse and research on “what is thegood society” into the center of sociological

1 Interestingly, one 1990s survey of 12,000Ph.D. sociologists revealed that over half spentat least 10 hours a week doing what they view as“ applied” research (Dotzler and Koppel1999:79).


debate and analysis. Even more, today weneed to look beyond the borders of the na-tion-state to address the possibility of aworld moral community.

Be More Self-Critical

Third, as part of an ongoing self-renewalprocess, I see the need for accelerated self-reflection in sociology. This is a task closelyrelated to my last point. The communities,colleges, universities, agencies, companies,and other settings in which we practice soci-ology are shaped in part by the oppressivesocial relations of the larger society. Weneed a liberating and emancipating sociol-ogy that takes risks to counter these oppres-sive social relations in our own bailiwicks.

As social scientists, we should regularlyexamine our research environments, includ-ing our metascientific underpinnings andcommitments. Critical social perspectives,such as those of feminists, gay/lesbian schol-ars, critical theorists, anti-racist scholars,and Marxist researchers, among others, havebeen resurgent since the 1960s. Scholars re-searching from these perspectives, as well assymbolic interactionists and ethnometh-odologists, have called for more internal re-flection in the social sciences. In one suchdisciplinary reflection, feminist sociologistsStacey and Thorne (1996:1–3) argue that,while anthropology and history have incor-porated feminist ideas better than sociology,the questioning of androcentric concepts andstructures is finally beginning to have abroader impact in sociology. In an earliercritical reflection, Dorothy Smith (1987) ar-gues that mainstream sociology has histori-cally been part of the dominant ideologicalapparatus, which focuses on issues primarilyof concern to men. Mainstream sociology’scentral themes are “organized by and articu-late the perspectives of men—not as indi-viduals . . . but as persons playing determi-nate parts in the social relations of this formof society . . .” (p. 56). Feminist sociologistshave pressured the discipline to view and re-search the social world from the perspectiveof women and thereby greatly expand itsfund of knowledge.

African American, Latino, Native Ameri-can, Asian American, gay/lesbian, and otherformerly excluded sociologists also have

pressed the discipline of sociology to viewand research society from their standpointsand thus to broaden sociological knowledge.For instance, in an introduction to the reprintof her pioneering book, The Death of WhiteSociology ([1973] 1998), Ladner notes nu-merous ways in which the presence of schol-ars of color, as well as women and gay/les-bian scholars, has forced issues of socialsubjugation to be considered seriously inboth the academy and the larger society.Similarly, racial-ethnic feminists haveforced the academy to consider seriouslymultiple statuses and the intersectionality ofoppressions (Baca Zinn and Dill 1994). Thegoal of all these scholars is not just to de-velop alternative funds of knowledge, butalso to push this knowledge in from the mar-gins, where it too often resides, toward thecentral trends and debates in sociology. In-side and outside the discipline, this accumu-lating knowledge can then become part ofthe process of eroding the historical relationsof social oppression.

Hopefully, more self-reflection among so-ciologists can also lead us and other socialscientists to destroy the insidious boundarieswe often draw around ourselves, such as theartificial dichotomy of quantitative versusqualitative research, the ranking of basicover applied research, and the valuing of re-search over teaching.

Recognize and Stress the

Importance of Teaching Sociology

Fourth, we need to recognize and accent theimportance of teaching sociology—espe-cially the kind of quality teaching that willprepare present and future generations forthe coming social, economic, technological,and environmental challenges. Indeed, manyof us were recruited into sociology by first-rate teachers. Our graduate programs need torecognize that most people who secure Ph.D.degrees in sociology do not become profes-sors in research universities, but rather be-come applied sociologists or faculty mem-bers with heavy teaching loads in a diversearray of public and private educational insti-tutions (see Eitzen, Baca Zinn, and Gold1999:57–60).

The majority of undergraduate and gradu-ate students in sociology are looking for


meaningful ways to contribute to making abetter society. Thus, it is disturbing to hearreports from some of these students at vari-ous colleges and universities that their pro-fessors are asserting that there is no room insociology for idealism and activism. Socialscientists who attempt to avoid social better-ment issues often defend themselves withphrases like, “ We are not out to save theworld.” C. Wright Mills (1958) once sug-gested,

Sometimes this is the disclaimer of a mod-est scholar; sometimes it is the cynical con-tempt of a specialist for all issues of largerconcern; sometimes it is the disillusionmentof youthful expectations; often it is the pos-ture of men who seek to borrow the prestigeof The Scientist, imagined as a pure and dis-embodied intellect. (P. 133)

As teachers of sociology, we should makeclear to the coming generations of sociolo-gists not only that there is plenty of roomfor idealism and activism in the field butthat these qualities might be required for hu-manity to survive the next century or so. Weneed to communicate the excitement andimportance of doing sociology. AlfredMcClung Lee (1978) was eloquent in thisregard:

The wonder and mysteries of human creativ-ity, love, and venturesomeness and thethreatening problems of human oppressionand of sheer persistence beckon and involvethose with the curiosity and courage to becalled sociologists. Only those who chooseto serve humanity rather than to get caughtup in the scramble for all the immediate re-wards of finance and status can know thepleasures and lasting rewards of such a pur-suit. (Pp. 16–17)

In my view, sociology students should beshown how the diversity of theories, meth-ods, debates, and practitioners in sociologyis generally healthy for the field and for so-ciety. We also should strive to help our stu-dents think critically about their social livesand about building a better society. WendellBell (1998) has underscored the importanceof showing social science students how toengage in debates about important issues,critically assess necessary moral judgments,and explore possible social futures for them-selves and their societies.

Study the Big Social Questions

Finally, contemporary sociologists need tospend much more effort studying the big so-cial questions of the twenty-first century. In-terestingly, Kai Erikson (1984:306; also seeWilner 1985) once suggested that a reviewof leading sociology journals over severaldecades would likely find that many decisiveevents had been ignored there by sociolo-gists. When social scientists become tooprofessionalized and too narrowly commit-ted to a discipline or area of study, researchissues tend to be defined from within theirdominant professional paradigm. They relyheavily on a narrow range of theories andmethods. Only those research topics and in-terpretations are accepted that do notthreaten the basis of the profession and itsestablished intellectual capital. However,technological and other knowledge develop-ments are now moving so fast that a socialscientist who is too narrowly trained or fo-cused may be incapable of making sense outof the ongoing currents of change.

In many U.S. colleges and universities theadministratively sanctioned goal of generat-ing grant money—often for its own sake—still distorts too much social science re-search in the direction of relatively minorsocial issues. This heavy focus on grantmoney reduces the amount of research onkey public issues and diminishes the poten-tial for colleges and universities to be arenasfor critical debate and discussion of those is-sues (Black 1999).

C. Wright Mills (1958) called for socialscientists to challenge dominant ideas:

If truly independent ideas are not even for-mulated, if we do not set forth alternatives,then we are foolishly trapped by the diffi-culties those now at the top have gotten usinto. (P. 137)

Sociologists need to formulate more originaland independent ideas, and to illuminate anddirectly and critically address recurring na-tional and global crises. We need to imple-ment Gans’s (1989) call for more sociolo-gists to become public intellectuals who willspeak critically, and from data, about majorsocietal issues. Especially in our journals,many social scientists need to break from theconventional style of research presentation


and jargonistic writing that targets a special-ized audience and move to a style accessibleto broad audiences and to an approach thataddresses the big social questions and theimplications of research for society. At thesame time, we should recognize that thereare numerous sociologists who write welland accessibly, yet often face the censorshipof ideas that are seen as too critical—an ex-perience still common in this society. Thus,we also should insist that the relevant pub-lishing outlets consider and publish impor-tant critical analyses of momentous socialissues, and not rule them out as “ too contro-versial” or as “only thought pieces” (Agger1989:220).

Yes, some sociologists do work on the bigand tough questions; yet we need many moreto ask major questions about such societaltrends as the huge and ongoing wealth trans-fers from the working classes to the rich, thesocial impact of environmental crises, theimpact of globalizing capitalism on localcommunities, and the human costs of racism,sexism, and other social oppressions.

One major research question requiringmuch attention relates to the internationalimpact of multinational capitalism and its“ free markets.” We hear much today aboutthe global capitalistic economy, but all toolittle social science research is examining itsdeep structure and broad range of humanconsequences. Half a century ago, in a fore-word to Polanyi’s book, The Great Transfor-mation, sociologist Robert M. MacIver([1944] 1957) noted that some research oncapitalistic markets already indicated thatformulas like “ world peace through worldtrade” were dangerous simplifications:

Neither a national nor an international sys-tem can depend on the automatic regulants.Balanced budgets and free enterprise andworld commerce and international clearing-houses . . . will not guarantee an interna-tional order. Society alone can guarantee it;international society must also be discov-ered. (P. xi)

Other major research questions deservingmore attention from sociologists center onthe character, costs, and future of contempo-rary racism. While some sociologists havepressed forward in researching the white-generated oppression targeting Americans ofcolor, more researchers should address the

ways in which racial oppression becomesdisguised or subtle in its character and prac-tice, the ideological defense of that oppres-sion, and the social costs for its targets andthe larger society.

We should also encourage similar socio-logical research on other major forms of so-cial oppression that pervade this and othercontemporary societies. In recent years so-ciologists and other social scientists haveundertaken significant empirical and theo-retical work on sexism, homophobia, age-ism, and discrimination against the disabled,yet today these areas cry out for much moreresearch and analysis.

In addition, more sociologists shouldstudy societal futures, including the alterna-tive social futures of just and egalitarian so-cieties. The United States spends severalmillion dollars annually on the scientificsearch for extraterrestrial life, yet very littleon examining the possible or likely socialfutures for terrestrial societies. Today, weshould encourage more sociologists andother social scientists to investigate societalprobabilities and possibilities, and assessthem for the general public. Social scientistscan extrapolate critically from understand-ings of the trends and possibilities alreadyapparent in various societal arenas, as wellas probe an array of societal alternativeswith imaginative research approaches.

Major societal transformations loomahead of us. There are, for example, the de-mographic changes well described by somesociologists, such as the graying of societ-ies. Such trends will likely be associatedwith other societal changes: Aging societies,for example, may have less interest in war,experience less street crime, and focus them-selves more on issues of health care, socialservices, and euthanasia. Another demo-graphic shift already underway is an increas-ing racial and ethnic diversity in some na-tional populations. According to some U.S.Census Bureau projections, in the year 2050the U.S. population will reach about 383million; just under half will be Americans ofcolor (Murdock 1995:33–47). And by the2050s, it is estimated, Americans of Euro-pean descent will become a statistical minor-ity. For the most part, in-depth analysis ofthe social significance of this demographictrend has been left to journalists or popular


commentators, most of whom have limitedsociological knowledge. There is ample op-portunity right now for sociological researchinto the possible or likely societal futuresassociated with trends such as these, particu-larly assessments from a countersystemframework accenting the goals of social jus-tice and multiracial democracy.

In addition, more sociologists should bedoing research on, and showing the publicthe social consequences of, the likely tech-nological advances in biomedicine, artificialintelligence, genetics, and telecommunica-tions. A central aspect of human societies isthe ability to collect, amass, and analyze in-formation. Today new developments in in-formation generation, storage, and applica-tion are emerging at an explosive rate. Forinstance, technological optimists predict thatover the next few decades the biomedicalrevolution will greatly extend the human lifespan and augment our mental and physicalcapacities dramatically. What are the socialconsequences of such striking biomedicaldevelopments for the world’s many peoples?A leading medical expert on immunology,Jerome Groopman (1999), has speculated onthe inequalities likely to emerge:

I don’ t see the wealthy western nations ral-lying to make major inroads into the devel-oping world, where infant mortality is stillhigh and life expectancy is much lower. Willyou have this very lopsided set of popula-tions, where people in America and WesternEurope are playing tennis and taking Viagraat 115, while in Zaire people are still dyingat 15 from HIV, malaria, tuberculosis, andEbola? (1999:n.p.)

Moreover, in a provocative article, “Whythe Future Doesn’ t Need Us?” Sun Micro-systems’ co-founder and chief scientist BillJoy (2000) has warned of a major techno-logical threat to human beings—the newtechnologies of robots and other human-en-gineered organisms. In Joy’s informed pre-diction, uncontrolled self-replication by ro-bots with artificial intelligence could pose aserious threat to human beings in the com-ing decades. A number of computer scien-tists have predicted that by the 2030s com-puters will be ever more human, “ con-scious,” and intelligent (Kurzweil 1999).They predict that computers will have ca-pacities a million times greater in the future

than at present, and that computerized robotswill be much “ smarter” than human beings.A generally cautious computer scientist, Joy(2000) does not see himself as writing sci-ence fiction, but as one who asks toughquestions about social futures: “ Given theincredible power of these new technologies,shouldn’ t we be asking how we can best co-exist with them? . . . [S]houldn’ t we proceedwith caution?” (n.p.).

Reviewing policy options, Joy (2000) sug-gests the almost unthinkable solution of hu-mans giving up entirely the development ofthis robotic technology because of its likelynegative consequences for human societies.Physical scientists like Joy are questioningthe modern faith in the benign character ofnew technologies. They are asking toughquestions about the failure of physical andsocial scientists, policymakers, and ordinarycitizens to be centrally concerned with thesocial consequences of technologies. Criti-cal assessments of possible or probable so-cial futures for technologically “advanced”civilizations are natural research and analyti-cal tasks for contemporary sociologists.

CONCLUSION

In an 1843 letter, the young Karl Marx sug-gested that critical social analysis should laybare the hidden societal realities. The goalmust be the “ reform of consciousness notthrough dogmas but by analyzing mysticalconsciousness obscure to itself, whether . . .in religious or political form” (Marx [1843]1975:209). Marx added that the task for in-volved social scientists, as for other citizensof the world, was the clarification of the“ struggles and wishes of the age” (p. 209).

For many millennia human beings havebeen tool-makers, yet in just a few decadeswe have created economies and technolo-gies—such as polluting industries, fossil-fuel consuming engines, and nuclear weap-ons—that may well threaten the survival ofour species and of our living planet itself. Itseems likely that the fate of our planet andits many species will be decided within thenext few generations by just one of its spe-cies. As moral beings, we need to ask insis-tently: What would alternatives to our self-destructive societies look like? And how dowe get there?


Much of humanity might agree on a newglobal social system that reduces injustice,is democratically accountable to all people,offers a decent standard of living for all, andoperates in a sustainable relation to earth’sother living systems (e.g., see Korten 1999;Sahtouris 1996). Determining whether this isthe case and how such a just global societymight be developed are enormous questionsthat sociologists—and other citizens of theworld—should be tackling.2 In a pioneeringbook, The Image of the Future (1973), FredPolak argued that we need a new generationof visionaries who can think clearly anddeeply about sustainable social futures:

Social scientist, intellectual, artist, leader,middleman of any breed, and the CommonMan (and Woman) to whom, after all, thiscentury belongs—each must ask himself [orherself], what is my vision of the future?And what am I going to do about it? (P. 305)

While social science analysis can help usto understand our ailing societal dreams anddecide what dreams to accept or reject, suchanalysis is beneficial only if it frees us todecide on a better future. Let me concludeby closely paraphrasing Polak (1973:305):Human beings have the ability to dream bet-ter futures than we have yet succeeded indreaming. We have the ability to createmuch better societies than we have yet suc-ceeded in creating.

Joe R. Feagin is Professor of Sociology at theUniversity of Florida. His research interests con-cern the development and structure of institution-alized discrimination, oppression, and exploita-tion in contemporary societies, as well as relatedresistance struggles and movements. He is cur-rently working on research examining the racialviews of white elites, the individual and socialcosts of racism, racial barriers in business sec-

tors, and the character of the discriminationfaced by women in U.S. society. His recent andforthcoming books include Racist America:Roots, Current Realities, and Future Reparations(Routledge 2001); White Racism: The Basics(with Hernán Vera and Pinar Batur, 2d ed.,Routledge 2001); and The First R: How ChildrenLearn Race and Racism (with Debra VanAusdale, Rowman and Littlefield 2001), and Lib-eration Sociology (with Hernán Vera, Westview,forthcoming).

REFERENCES

Addams, Jane. 1902. Democracy and Social Eth-ics. New York: Macmillan.

Agger, Ben. 1989. Reading Science: A Literary,Political, and Sociological Analysis. Dix Hills,NY: General Hall.

Baca Zinn, Maxine and Bonnie Thornton Dill,eds. 1994. Women of Color in U.S. Society.Philadelphia, PA: Temple University Press.

Bannister, Roger C. 1992. “ Principle, Politics,Profession: American Sociologists and Fas-cism.” Pp. 172–213 in Sociology Responds toFascism, edited by S. P. Turner and D. Kasler.London, England: Routledge.

Bell, Michael M. 1998. An Invitation to Environ-mental Sociology. Thousand Oaks: Pine Forge.

Bell, Wendell. 1998. “ Making People Respon-sible: The Possible, the Probable, and the Pref-erable.” American Behavioral Scientist 42:323–39.

Black, Timothy. 1999. “Going Public: How So-ciology Might Matter Again.” Sociological In-quiry 69:257–75.

Block, Fred. 1990. Postindustrial Possibilities: ACritique of Economic Discourse. Berkeley,CA: University of California Press.

Bressler, Marvin. 1999. “ Contemporary Sociol-ogy: A Quarter Century of Book Reviews.”Sociological Forum 14:707–20.

Bryant, Christopher G. A. 1985. Positivism inSocial Theory and Research. New York: St.Martin’ s.

Burawoy, Michael. 1998. “Critical Sociology: ADialogue between Two Sciences.” Contempo-rary Sociology 27:12–20.

Business Week. 1999. “Economic Growth: Hey,What About Us?” Business Week, December27. Retrieved December 22, 1999 (http://www.businessweek.com).

Cancian, Francesca M. 1995. “Truth and Good-ness: Does the Sociology of Inequality Pro-mote Social Betterment.” Sociological Per-spectives 38:339–56.

Collins, Chuck, Chris Hartman, and Holly Sklar.1999. Divided Decade: Economic Disparity atthe Century’s Turn. Boston, MA: United for a

2 For example, visualizing the path to a betterfuture for the world’ s poor is not difficult. The1997 Human Development Report of the UnitedNations indicated that for about 15 percent of theU.S. defense budget, or about $40 billion a year,the basic needs for health, nutrition, education,reproductive health, safe water, and sanitationcould be met for the entire population of theplanet. Another $40 billion would be enough tobring the poorest residents of the planet out ofextreme poverty (Williamson 2000).


Fair Economy. Retrieved December 1999(http://www.stw.org/my_html/Divided.html).

Collins, Patricia Hill. 1998. “On Book Exhibitsand New Complexities: Reflections on Sociol-ogy as Science.” Contemporary Sociology27:7–11.

Cooper, Anna Julia. 1892. A Voice from theSouth by a Black Woman from the South. Xe-nia, OH: Aldine.

Cox, Oliver C. 1948. Caste, Class, and Race: AStudy in Social Dynamics. New York:McGraw-Hill.

Deegan, Mary Jo. 1987. “An American Dream:The Historical Connections between Women,Humanism, and Sociology, 1890–1920,” Hu-manity and Society 11:353–65.

———. 1988. Jane Addams and the Men of theChicago School, 1892–1918. New Brunswick,NJ: Transaction Books.

Diamond, Jared. 1992. The Third Chimpanzee:The Evolution and Future of the Human Ani-mal. New York: HarperCollins.

Dotzler, Robert J. and Ross Koppel. 1999. “WhatSociologists Do and Where They Do It—TheNSF Survey on Sociologists’ s Work Activitiesand Workplaces.” Sociological Practice 1:71–83.

Du Bois, W. E. B. 1968. The Autobiography ofW. E. B. Du Bois: A Soliloquy on Viewing MyLife from the Last Decade of Its First Century.New York: International Publishers.

———. [1899] 1973. The Philadelphia Negro: ASocial Study. Together with a Special Reporton Domestic Service by Isabel Eaton. Reprint,Millwood, NY: Kraus-Thomson.

Durkheim, Emile. [1893] 1933. The Division ofLabor in Society. Translated by G. Simpson.New York: Free Press.

Eitzen, D. Stanley, Maxine Baca Zinn, and StevenJ. Gold. 1999. “ Integrating Professional Social-ization and Training for Sociology GraduateStudents.” The American Sociologist 30:56–63.

Erikson, Kai. 1984. “Sociology and Contempo-rary Events.” Pp. 303–10 in Conflict and Con-sensus: A Festscrift in Honor of Lewis A.Coser, edited by W. W. Powell and R.Robbins. New York: Free Press.

Fals-Borda, Orlando. 1960. Acción Comunal enUna Vereda Colombiana: Su Applicación, SusResultados y Su Interpretación (CommunalAction in a Columbian Area: Its Application,Its Results, and Its Interpretation). Bogotá, Co-lombia: Universidad Nacional de Colombia,Departamento de Sociología.

Feagin, Joe R. and Hernán Vera. Forthcoming.Liberation Sociology. Boulder, CO: Westview.

Friedrichs, Robert W. 1970. A Sociology of Soci-ology. New York: Free Press.

Fukuyama, Francis. 1992. The End of Historyand the Last Man. New York: Free Press.

Gans, Herbert J. 1989. “Sociology in America:The Discipline and the Public.” American So-ciological Review 54:1–16.

George, Susan. 1999. “A Short History of Neo-Liberalism.” Paper presented at the Confer-ence on Economic Sovereignty in a Globaliz-ing World, March 24, Bangkok, Thailand. Re-trieved December 21, 1999 (http://www.zmag.org).

Giddings, Franklin H. 1909. In discussion afterL. L. Bernard’ s “The Teaching of Sociologyin the United States.” American Journal of So-ciology 15:195–211.

Groopman, Jerome. 1999. “ Is There Life onMars?” Boston Magazine. Retrieved December12, 1999 (http://www.alternet.org/PublicArchive/Groopman1217.html).

Hawken, Paul and David Korten. 1999. “Corpo-rate Futures.” Interview published in Yes! AJournal of Positive Futures, Summer 1999.Retrieved December 12, 1999 (http://www.f u t u r e n e t . o r g / 1 0 c i t i e s o f e x u b e r a n c e /corporatefutures).

Hawken, Paul, Amory Lovins, and L. HunterLovins. 1999. Natural Capitalism: Creatingthe Next Industrial Revolution. Boston, MA:Little, Brown.

Hill, Michael R. 1989. “Empiricism and Reasonin Harriet Martineau’ s Sociology.” Pp. xv–lx,Introduction to reprinted edition of How to Ob-serve Morals and Manners, by H. Martineau.New Brunswick, NJ: Transaction.

Hoecker-Drysdale, Susan. 1992. HarrietMartineau: First Woman Sociologist. Oxford,England: Berg.

Joy, Bill. 2000. “Why the Future Doesn’ t NeedUs?” Wired, April. Retrieved April 2000(http://www.wired.com:80/wired/archive/8.04/joy.html).

Klein, Naomi 2000. No Space, No Choice, NoJobs, No Logo: Taking Aim at the Brand Bul-lies. New York: Picador.

Korten, David. 1999. The Post-Corporate World:Life after Capitalism. San Francisco, CA:Berrett-Koehler Publishers and KumarianPress.

Kurzweil, Ray. 1999. Age of Spiritual Machines:When Computers Exceed Human Intelligence.New York: Viking.

Ladner, Joyce A. [1973] 1998. “ Introduction tothe Black Classic Press Edition.” In The Deathof White Sociology: Essays on Race and Cul-ture, by J. A. Ladner. Reprint, Baltimore, MD:Black Classic Press.

Lee, Alfred McClung. 1978. Sociology forWhom? New York: Oxford University Press.

Lemert, Charles and Esme Bhan, eds. 1998. TheVoice of Anna Julia Cooper. Lanham, MD:Rowman and Littlefield.

Lengermann, Patricia Madoo and Jill Niebrugge-


Brantley. 1998. The Women Founders: Sociol-ogy and Social Theory, 1830–1930. New York:McGraw-Hill.

———. 2001. “The Meaning of ‘Things’ : Theoryand Method in Harriet Martineau’ s How toObserve Morals and Manners and EmileDurkheim’ s The Rules of SociologicalMethod.” Pp. 212–37 in Harriet Martineau:Theoretical and Methodological Perspectives,edited by M. R. Hill and S. Hoecker-Drysdale.New York: Garland.

Lovelock, James E. 1987. Gaia: A New Look atLife on Earth Oxford, England: Oxford Uni-versity Press.

MacIver, Robert M. [1944] 1957. “ Foreword.”Pp. ix–xxi in The Great Transformation: ThePolitical and Economic Origins of Our Time,by K. Polanyi. Boston, MA: Beacon.

Martineau, Harriet. 1837. Society in America. 3vols. New York: Saunders and Otley.

Martineau, Harriet. [1838] 1989. How to ObserveMorals and Manners. Reprint, New Bruns-wick, NJ: Transaction.

Marx, Karl. [1843] 1975. “ Letters from theFranco-German Yearbooks.” 1843 letter to A.Ruge, pp. 206–209 in Early Writings [of]Marx, introduced by L. Colletti, translated byR. Livingstone and G. Benton. London: Pen-guin Books.

Mills, C. Wright. 1958. The Causes of WorldWar III. New York: Simon and Schuster.

Moore, Barrington. 1971. Reflections on theCauses of Human Misery, and upon CertainProposals to Eliminate Them. Boston, MA:Beacon.

Murdock, Steve H. 1995. An America Chal-lenged: Population Change and the Future ofthe United States. Boulder, CO: Westview.

Myrdal, Gunnar. [1944] 1964. An American Di-lemma. Vol. 2. New York: McGraw-Hill.

Nyden, Philip, Anne Figert, Mark Shibley, andDarryl Burrows. 1997. “Conclusion: Collabo-ration Gives Hope and Voice in an Age of Dis-illusionment.” Pp. 240–42 in Building Commu-nity: Social Science in Action, edited by P.Nyden, A. Figert, M. Shibley, and D. Burrows.Thousand Oaks, CA: Pine Forge.

Omi, Michael and Howard Winant. 1994. RacialFormation in the United States: From the1960s to the 1990s. 2d ed. New York:Routledge.

Oxfam. 1999. “New Millennium—Two Futures.”Oxfam Policy Papers. Retrieved December 30,1999 (http://ww.oxfam.org.uk/policy/papers).

Park, Robert E. and Ernest W. Burgess. 1921. In-troduction to the Science of Sociology. Chi-cago, IL: University of Chicago Press.

Polak, Fred. 1973. The Image of the Future.Translated and abridged by E. Boulding. SanFrancisco, CA: Jossey-Bass.

Polanyi, Karl. [1944] 1957. The Great Transfor-mation: The Political and Economic Origins ofOur Time. Boston, MA: Beacon.

Ramonet, Ignacio. 1999. “ The Year 2000.” LaMonte Diplomatique, December 15. RetrievedD e c e m b e r 1 9 9 9 ( h t t p : / / w w w . m o n d e -diplomatique.fr/en/1999/12?c=01leader).

Raushenbush, Winifred. 1979. Robert E. Park:Biography of a Sociologist. Durham, NC:Duke University Press.

Residents of Hull-House. [1895] 1970. Hull-House Maps and Papers, by Residents of Hull-House, a Social Settlement. Reprint, NewYork: Arno.

Sachs, Jeffrey. 1999. “Apocalypse Soon.” Bos-ton Magazine. Retrieved December 21, 1999(http:/ /www.alternet.org/PublicArchive/Sachs1217.html).

Sahtouris, Elisabet. 1996. Earthdance: LivingSystems in Evolution. Alameda, CA: MetalogBooks.

Sjoberg, Gideon. 1996. “ The Human RightsChallenge to Communitarianism: Formal Or-ganizations and Race and Ethnicity.” Pp. 273–93 in Macro Socio-Economics: From Theoryto Activism, edited by D. Sciulli. Armonk, NY:M. E. Sharpe.

Sjoberg, Gideon and Leonard D. Cain. 1971.“Negative Values, Countersystem Models, andthe Analysis of Social Systems.” Pp. 212–29in Institutions and Social Exchange: The So-ciologies of Talcott Parsons and George C.Homans, edited by H. Turk and R. L. Simpson.Indianapolis, IN: Bobbs-Merrill.

Small, Albion. 1895. “ The Era of Sociology.”American Journal of Sociology 1:1–27.

Smith, Dorothy E. 1987. The Everyday World asProblematic: A Feminist Sociology. Boston,MA: Northeastern University Press.

Soros, George. 1998. “ Toward a Global OpenSociety.” Atlantic Monthly, January, pp. 20–24, 32.

Stacey, Judith and Barrie Thorne. 1996. “ Is So-ciology Still Missing Its Feminist Revolu-tion?” Perspectives: The ASA Theory SectionNewsletter 18:1–3.

Steinfels, Peter. The Neoconservatives: The MenWho Are Changing America’s Politics. NewYork: Touchstone.

Stoecker, Randy. 1996. “ Sociology and SocialAction: Introduction.” Sociological Imagina-tion 33: 3-17.

Toward Freedom. 1999. “ What a World! TenTroubling Faces from the 1999 UN Develop-ment Report.” Toward Freedom. RetrievedMarch 2, 2000 (http://www.undp.org/hdro/99.htm).

United Nations. 1997. Human Development Re-port. New York: Oxford University Press, forthe United Nations Development Programme.


Vaughan, Ted R. 1993. “The Crisis in Contem-porary American Sociology: A Critique of theDiscipline’ s Dominant Paradigm.” Pp. 10–53in A Critique of Contemporary American So-ciology, edited by T. R. Vaughan, G. Sjoberg,and L. T. Reynolds. New York: General Hall.

Wells-Barnett, Ida B. 1895. A Red Record. Chi-cago, IL: Donohue and Henneberry.

Williamson, Thad. 2000. “Global Economic In-

equality.” Dollars and Sense, January–Febru-ary 2000. Retrieved March 2000 (atwww.igc.org/dollars/2000/0100econin.html).

Wilner, Patricia. 1985. “The Main Drift of Soci-ology between 1936 and 1984.” Journal of theHistory of Sociology 5:1–20.

Young, Iris Marion. 1990. Justice and the Poli-tics of Difference. Princeton, NJ: PrincetonUniversity Press.

Past Presidents of the American Sociological Association

Lester F. WardWilliam G. Sumner

Franklin H. GiddingsAlbion W. SmallEdward A. Ross

George E. VincentGeorge E. HowardCharles H. Cooley

Frank W. BlackmarJames Q. DealeyEdward C. Hayes

James P. LichtenbergUlysses G. WeatherlyCharles A. Ellwood

Robert E. ParkJohn L. Gillin

William I. ThomasJohn M. Gillette

William F. OgburnHoward W. OdumEmory S. BogardusLuther L. Bernard

Edward B. ReuterErnest W. BurgessF. Stuart Chapin

Henry P. FairchildEllsworth Faris

Frank H. HankinsEdwin H. SutherlandRobert M. MacIver

Stuart A. QueenDwight Sanderson

George A. LundbergRupert B. VanceKimball YoungCarl C. TaylorLouis Wirth

E. Franklin FrazierTalcott Parsons

Leonard S. Cottrell Jr.Robert C. Angell

Dorothy Swaine ThomasSamuel A. StoufferFlorian Znaniecki

Donald Young

Herbert BlumerRobert K. Merton

Robin M. Williams Jr.Kingsley DavisHoward Becker

Robert E. L. FarisPaul F. LazarsfeldEverett C. HughesGeorge C. Homans

Pitirim SorokinWilbert E. MooreCharles P. LoomisPhilip M. HauserArnold M. RoseRalph H. TurnerReinhard Bendix

William H. SewellWilliam J. Goode

Mirra KomarovskyPeter M. Blau

Lewis A. CoserAlfred McClung Lee

J. Milton YingerAmos H. Hawley

Hubert M. Blalock Jr.Peter H. Rossi

William Foote WhyteErving GoffmanAlice S. Rossi

James F. Short Jr.Kai T. Erikson

Matilda White RileyMelvin L. KohnHerbert J. Gans

Joan HuberWilliam Julius Wilson

Stanley LiebersonJames S. Coleman

Seymour Martin LipsetWilliam A. Gamson

Amitai EtzioniMaureen Hallinan

Neil J. SmelserJill Quadagno

Alejandro PortesJoe R. Feagin

RADIO AND SOUTHERN TEXTILE WORKER INSURGENCYRADIO AND SOUTHERN TEXTILE WORKER INSURGENCYRADIO AND SOUTHERN TEXTILE WORKER INSURGENCYRADIO AND SOUTHERN TEXTILE WORKER INSURGENCYRADIO AND SOUTHERN TEXTILE WORKER INSURGENCY 2 12 12 12 12 1


Media and Mobilization:

The Case of Radio and Southern

Textile Worker Insurgency,

1929 to 1934

Collective action rests, in part, on group identity and political opportunity. Just howgroup identity is manifested and perceptions of political opportunity are altered,however, remain unclear, particularly in the case of a geographically dispersedpopulation. An often overlooked mechanism is media technology. This article ana-lyzes an important yet underexamined instance of worker mobilization in the UnitedStates: the southern textile strike campaigns of 1929 to 1934 during which morethan 400,000 workers walked off their jobs. Using historical data on textile manu-facturing concentration and strike activity, FCC data on radio station foundings,and analyses of political content and song lyrics, the authors show that the geo-graphic proximity of radio stations to the “textile belt” and the messages airedshaped workers’ sense of collective experience and political opportunity: Walk-outsand strike spillover across mill towns resulted. The implications of the analyses forsocial movement theory generally, and for the understanding of how media canenable or constrain collective struggle, are discussed.

sponsored violence (Griffith 1988; Hall et al.1987; McLaurin 1971).

A pivotal moment in U.S. labor history,the movement’s eventual defeat has had con-sequences for labor practices, organizing ef-forts, economic development, and persistentpoverty and inequality in the U.S. South upto the current day (Roscigno and Kimble1995). Yet, little sociological attention hasbeen devoted to this instance of southernworker unrest. Indeed, much of the researchon worker insurgency overlooks the South ortreats it as a union-resistant region (Corn-field and Leners 1989). How is it that nearly

etween 1929 and 1934, the U.S.South experienced a truly remarkable

moment in labor history. Estimates suggestthat approximately half a million southerntextile mill workers walked off the job dur-ing this period, culminating in the GeneralTextile Workers Strike of 1934. Interest-ingly, this mobilization occurred with littleorganization by labor unions and in the faceof coercive paternalistic practices and state-

B

Direct all correspondence to Vincent Ros-cigno, 300 Bricker Hall, 190 N. Oval Mall, TheOhio State University, Columbus, Ohio, 43210([email protected]). A version of this paperwas presented at the meetings of the AmericanSociological Association, Washington, D.C., Au-gust 2000. The authors thank Jack Bloom,Martha Crowley, Rick Della Fave, Cathy Evans,Randy Hodson, Craig Jenkins, Elizabeth Kamin-ski, Jeffrey Leiter, Susan Roscigno, David Snow,Verta Taylor, Donald Tomaskovic-Devey, MariekeVanWilligen, Al Wall, Karen Weissman, and theASR Editors and anonymous reviewers for theirhelpful comments on early drafts. We also thank

Vincent J. Roscigno

The Ohio State UniversityWilliam F. Danaher

College of Charleston

Richard Schraeder, Director of the University ofNorth Carolina’ s Southern Folk Life ArchivesCollection, and numerous archivists at the Fed-eral Communications Commission, the Univer-sity of South Carolina, and the Walter P. ReutherLibrary at Wayne State University for their helpduring the data collection phase. This researchwas supported by grants from the College of So-cial and Behavioral Sciences at the Ohio StateUniversity and the College of Charleston.


half a million workers, most of whom weregeographically isolated in rural mill towns,collectively mobilized in the face of localelite repression? Were collective identity andpolitical opportunity—prerequisites to col-lective action according to social movementtheorists—achieved and, if so, through whatmechanism? What networking resource wasat the disposal of workers that fostered strikespillover from one mill town to the next?

We extend the literature on collective be-havior and social movements, labor insur-gency, and class consciousness by address-ing these questions. We first embed ouranalysis in the historical specifics, and thenin collective identity (e.g., Melluci 1985;Taylor and Whittier 1992) and political op-portunity (e.g., Jenkins 1985; McAdam1983) frameworks of collective action. Weargue that insights from these literatures canbe effectively integrated by focusing on aunique and key historical event—the adventof radio.

BACKGROUND:

SOUTHERN TEXTILE WORKER

INSURGENCY, 1929 TO 1934

In 1921, southern cotton-producing statesproduced 54 percent of the nation’s totalyardage of woven cotton goods. This yieldincreased to 67 percent by 1927, partly theresult of the relocation of textile manufactur-ing operations from the North to the South.This regional shift occurred because cheaplabor was abundant in the South and unionactivity was virtually nonexistent: SouthernChambers of Commerce focused on thesefacts when enticing northern mill owners tomove south. Indeed, wages in southern millswere approximately one-third of those in theNorth, even after controlling for the cost ofliving. In addition, southern mill workersworked longer hours (Yellen 1936).

Workers typically lived in villages underthe control of mill owners. Whole familieslabored together for the sake of subsistence,yet housing, food, and medical care re-mained substandard in many instances (Gell-horn 1933; Hall et al. 1987). Some mill own-ers, employing paternalistic policies to sta-bilize their workforces, offered company-sponsored social programs, housing, medi-cal services, credit at the company store, and

religious services to workers. These com-pany programs failed, however, to offset lowwages and instead came to be seen as an im-portant mechanism of labor control and co-ercion (Leiter, Schulman, and Zingraff1991). Exorbitant interest rates were chargedat mill stores, ministers and doctors were onthe company payroll, and workers who werenot performing to the company standard orwho got out of line risked losing their homes(Cornfield and Leners 1989; Griffith 1988;McLaurin 1971; Pope 1942). Given suchconditions, worker resistance eventuallyemerged.

Strikes broke out in large numbers in1929; the main grievances were workingconditions, wages, and hours. On March 12,1929, 500 women walked out of the inspec-tion department at the American GanzstoffCorporation in Elizabethton, Tennessee. Thefollowing day, 3,000 more workers walkedout demanding higher pay. Later that week,2,000 workers walked out of the neighbor-ing Bemberg plant in Gastonia, North Caro-lina making the same demands. Strikes notimmediately related to those in Gastonia andElizabethton occurred soon afterward inSouth Carolina: In late March, 800 workerswalked out at Ware Shoals ManufacturingCompany, and 1,250 workers walked out ofthe New England Southern plant in Pelzer.Within three weeks, 8,000 workers hadwalked out of 15 plants in the Piedmont areaof South Carolina. Strikes followed shortlythereafter in the North Carolina towns ofPineville, Forest City, Lexington, BessemerCity, Draper, and Charlotte (Hall et al. 1987;Yellen 1936).

Local newspapers tended to be connectedto traditional economic interests, such astextiles, and thus took a vehement and ag-gressive stance against this early wave ofworker protest. Indeed, mill owners and lo-cal newspapers often worked hand in handto sway public opinion away from strikersby appealing to anti-Communist sentiments,despite the fact that few workers had suchaffiliations. Most, in fact, were simply pro-testing unfair conditions rather than defend-ing or fighting for a broader ideologicalstance (Salmond 1995). Perhaps most well-known in this regard were the editorials pub-lished in the Gastonia Daily Gazette entitled“ A Deep Laid Scheme” and “ Red


Russianism Lifts its Gory Hands Right Herein Gastonia,” both of which were publishedduring the infamous Loray Mill Strike of1929. Such editorials and the red-scarerhetoric they espoused, which continuedthrough the mid-1930s, created public angertoward strikers, caused workers to questiontheir own national and religious loyalties,and had long-term consequences for south-ern attitudes toward organized labor (Bill-ings 1990; Nolan and Jones 1976; Salmond1995; Simon 1998).1 Yet poor conditionspersisted into the 1930s, and strikes againemerged.

By June of 1933, newly elected PresidentFranklin Delano Roosevelt signed a bill in-tended to alleviate the plight of overworkedmillhands—the National Industrial Recov-ery Act (NIRA). This bill seemingly gavemill workers the right to push for decenthours and working conditions through col-lective bargaining. Section 7a of the TextileCode called for a minimum wage, a 40-hourwork week, and prohibited child labor. Thiseffort was part of the newly formed NationalRecovery Administration (NRA). Both

Roosevelt and the head of the NRA, HughS. Johnson, were opposed to strikes as ameans of solving disputes between workersand mill owners. Instead, they favored con-trolling work hours and child labor in an ef-fort to limit production, drive up profits formill owners, and improve economic condi-tions for workers through the “ trickle-down”of profits (Hall et al. 1987; Hodges 1986).

Prices, sales, and employment increased tothe highest level in five years by late sum-mer, 1933, but by fall this prosperity soured.The Depression reached its worst period inthe winter of 1933–1934, and mill owners,while seemingly supportive of the coopera-tive message in Section 7a, began to practiceold strategies of oppression or in some casesinstituted new ones. The “ stretch-out,” forinstance, was used to circumvent laws limit-ing working hours. This was the workers’term for the cumulative changes that “ setthem tending machines ‘by the acre,’ filledevery pore in the working day, and robbedthem of control over the pace and method ofproduction” (Hall et al. 1987: 211). Spinners,mostly women, were often stretched from 24to 48 looms, and then from 48 to 96, “with-out a commensurate increase in pay, oftenwith no increase whatsoever, or even an ac-tual decrease” (Yellen 1936:299).2 Thus,workers found themselves working as muchin the new eight-hour shift as they had inshifts lasting two to four hours longer. Fur-ther, by enabling industry to curtail produc-tion when mills were producing sufficientproduct through “ short time,” the NIRA Tex-tile Code inadvertently led to a surplus ofgoods and higher rates of unemployment(Tullos 1989). In short, mill owners saw thelaws enacted under the NIRA as bothersomebut easy to manipulate (Wood 1986).

Unions in the coal mining industry jumpedat the chance to take advantage of the NIRA.Yet, in the textile industry, the United Tex-tile Workers (UTW) did not institute aunionization drive (Tippett 1931). Indeed, it

1 Other print communications, such as pam-phlets, labor-oriented newspapers, and under-ground newspapers did little to foster thesestrikes. In a few cases, National Textile WorkersUnion (NTWU) pamphlets were distributed, butonly after the strike had already begun; and theyreduced strike support among workers because oftheir emphasis on promoting racial and genderequality in wages. Similarly, certain strikes, suchas those in Gastonia in 1929, received coveragein the Communist Daily Worker, despite little ifany striker affiliation with the Communist Party.This coverage, rather than fostering greatermovement cohesion or participation, was used bylocal papers to criticize strikers and had the ef-fect of reducing support (Salmond 1995). Oneexception to this generally negative impact ofprint media was the Augusta Labor Review, asmall pro-labor paper in the somewhat isolatedHorse Creek Valley of South Carolina. Thisnewspaper seems to have had a positive impacton strikes in that area, perhaps because itsfounder and editor, Paul W. Fuller, a MethodistEpiscopal minister, integrated Christian and pa-triotic discourse and symbolism into the presen-tation of worker grievances. Thus, the news-paper’ s message and related worker actions wereless easily interpreted or attacked as being Com-munist-oriented by the local power structure(Simon 1998).

2 A 1929 South Carolina House of Representa-tives committee found that in one South Carolinamill, a force of five men, each paid $23 per week,was reduced to three men at $20.23 per week; atanother mill, a weaver who had operated 24looms at $18.91 per week was stretched to 100looms for $24 a week (Yellen 1936).


had fewer than 10 paid organizers in theSouth at that time and represented only asmall fraction of the entire mill work force(Hodges 1986). According to historical ac-counts, this was mostly due to a lack of or-ganizational resources and a vast, hard-to-cover rural area. Nonetheless, southern millworkers walked off their jobs, formed localunions, and organized against unfair laborpractices. According to Hall et. al (1987):

[The UTW] launched no Piedmont organiz-ing campaign. Agents did not throng to thesouthern field. Yet within less than a monthafter passage of the act, union locals had re-portedly sprung to life in 75 percent ofSouth Carolina’ s mills. From an estimated40,000 in September 1933, UTW Member-ship leaped to 279,000 by August 1934. Tothe shock of labor leaders, government offi-cials, and businessmen alike, southern work-ers began “ organizing just as fast as wecan.” (P. 304)

Strike activity intensified throughout theSouth in 1934. On February 12, 1934, astrike broke out at K. S. Tanners StonecutterMills in Spindale, North Carolina. Fivemonths later, on July 14, a strike occurred inGuntersville, Alabama, and wildcat strikessoon rolled across that state involving20,000 workers. On Labor Day, many work-ers in North Carolina and South Carolina,states that did not observe the holiday, re-fused to come to work. Newspapers reported400,000 workers on strike by September14—the largest strike in American history.What prompted workers across dispersedmill villages to strike despite the lack ofclear-cut union support or organization?

COLLECTIVE IDENTITY AND

POLITICAL OPPORTUNITY

FRAMEWORKS

Collective identity and political opportunityperspectives offer a starting point for ex-plaining how collective behavior was mani-fested in the case of southern textiles. Whileboth perspectives deal with the developmentof collective action and the preconditions forinsurgency, their foci differ.

Collective identity theorists emphasizeideological, normative, and cultural pro-cesses that induce individual participation incollective action and ensure social solidarity,

even in the face of harsh countermobil-ization. These researchers also argue that al-ternative belief structures provide movementparticipants with a structure of nonmaterialrewards, not necessarily tied to movementsuccess (Epstein 1990; J. Gamson 1995; W.Gamson 1992a; Melucci 1985; Taylor andWhittier 1992). An alternative belief struc-ture and collective identity have been impor-tant in a variety of struggles, including thosepromoting racial justice (Morris 1984; Nagel1994; Stotik, Shriver, and Cable 1994),women’s rights (Mathews 1982; Meyer andWhittier 1994; Taylor and Whittier 1992),and class-based politics (Fantasia 1988;Hodson, Ziegler, and Bump 1987). Fantasia(1988) makes this focus explicit in relationto workingclass politics and highlights theimportance of “cultures of solidarity,” defin-ing them as “cultural formations that arise inconflict, creating and sustaining solidarity inopposition to the dominant structure” (p. 19).

Discussions of collective identity resonatewith classical and contemporary theoreticalideas pertaining to “ class consciousness”and when it may emerge. Indeed, “ the mostimportant blank spots in the theory of classconcern the processes whereby ‘ economicclasses’ become ‘ social classes’” (Giddens1982:157). Mann (1973) conceives of classconsciousness as a complex process, occur-ring in stages, that is often curbed by domi-nant ideologies, class ambiguities, conces-sions by elites, or outright defeat. Mann’sstages include (1) class identity, wherebyone defines oneself as working class; (2)class opposition, whereby one perceivescapitalists and their agents as opponents; (3)class totality, whereby class identity and op-position define the total of one’s social situ-ation and society as a whole; and (4) con-ceiving of an alternative. During this finalstage, Mann continues, an “explosive poten-tial” may emerge and create either a “con-flict consciousness,” which aims to alleviatethe immediate problem, or a more “ revolu-tionary consciousness,” wherein the neededchange involves overall systemic reorgani-zation (also see Giddens 1982).3 Given the

3 Many factors can influence the progressionof class consciousness through the stages de-scribed by Mann (1973), including mobility clo-sure, the division of labor within economic en-


correct progression, the delegitimization ofexisting ideology, and the existence of an al-ternative interpretational frame, class con-sciousness will emerge (Della Fave 1980,1986; Oliver and Johnston 1999).

In contrast to identity theorists and thosedealing explicitly with class consciousness,political opportunity theorists focus on thepolitical context in which groups are embed-ded and the shifting levels of opportunitythat emerge across time and place. The like-lihood of mobilization and the degree of le-verage exerted by insurgents, it is argued,will be heightened in situations in whichelites are divided in their defense of the ex-isting order (Gamson and Meyer 1992; Jen-kins 1985; Jenkins and Perrow 1977;McAdam 1982; Pichardo 1995; Tilly 1976).When elites are coordinated, in contrast, thereproduction of dominant relations is morelikely, as is countermobilization againstthose engaging in insurgent action (Lach-mann 1990; Tomaskovic-Devey and Ros-cigno 1996). McAdam (1983) emphasizessuch countermobilization in his analysis ofelite response to tactics implemented bycivil rights activists during the 1960s. Bar-kan (1984) and James (1988) highlight therole of other actors in the civil rights strug-gle—namely the southern racial state, whichconstrained movement participants, and thefederal government, which eventually inter-vened on behalf of participants.

Griffin, Wallace, and Rubin (1986) andMontgomery (1987) stress themes of eliteresponse to labor organization in their analy-ses of capitalist countermobilization duringthe 1930s and 1940s. Coercion and controlthrough paternalism proved effective as apreventive strategy (Leiter et al. 1991; Mc-Laurin 1971). More obvious were efforts ofcapital to divide workers racially, to curtailworking-class mobilization with subversiveactivities and violence, and to control labororganization and labor practices through ma-nipulation of the state and state policy (seeC. Brown and Boswell 1995; Brueggemanand Boswell 1998; Kimeldorf 1999;Roscigno and Kimble 1995; Wood 1986).

For collective identity theorists then, thecentral task is to explain how interpretationis altered, collective identity manifested,and solidarity maintained. For political op-portunity theorists, the focus is on the de-gree of elite unity, elite countermobil-ization, and the extent to which these di-mensions of political opportunity enable orconstrain the collective expression of griev-ances in a given historical context. Eachperspective, however, has problems whenapplied to Southern textile worker mobili-zation. How would collective identitytheory explain the manifestation of solidar-ity across this geographically dispersed tex-tile mill population? What was the mecha-nism through which structural political op-portunity, if it existed, translated into andshaped political perceptions and the degreeof efficacy among mill workers? We believethat media technology offers a bridge be-tween these two perspectives.

MEDIA AND MOBILIZATION

How were mill worker identity and sense ofpolitical opportunity manifested in the 1920sand 1930s despite the geographically dis-persed nature of mill towns? This questionis integral to those interested in the diffusionof collective action (Oliver 1989; Olzak1992; Rogers 1995; Soule 1997; Soule andZylan 1997). Such spatial “ spillover” re-quires some form of network structurethrough which information is communicatedand shared (Fantasia 1988; Morris 1984;Oberschall 1989). Assuming nonparticipantshave the same structural relation to the net-work as social movement participants, non-participants become potential adopters(Myers 1998; Strang and Meyer 1993).

While information networks may includefamily, friendship, or transportation ties, themedia may be particularly important for in-formation flow across geographically dis-persed populations (W. Gamson 1995;Kahan 1999; Oberschall 1989; Spilerman1970). Myers (1998, 2000), in his analysesof racial rioting, 1964–1971, characterizesthis potential influence as a concentric areaaround the network origin defined by therange of the medium’s distribution, ratherthan as lines connecting individuals. This isan important theoretical extension of previ-

terprises, authority relationships, patterns of re-source distribution, geographic dispersion, andpatterns of institutional power (see Blau 1977;Dahrendorf 1959; Giddens 1982; Parkin 1979).


ous perspectives because it offers a potentialmechanism through which group conscious-ness and perceptions of opportunity may bealtered across geographic space.

To assert the media’s influence requiresspecifying the structural and instrumentalways in which it can shape collective actionacross a dispersed population. On the struc-tural end, the introduction of new media mayprovide opportunities not directly associatedwith collective action, but which alter the le-verage and/or autonomy of subgroups. Thisappears to be true in the case of radio stationfoundings in the South, which had the unin-tended consequence of creating a relativelyautonomous community of musicians, manyof whom were ex-mill workers, who traveledfrom mill town to mill town and radio sta-tion to radio station. This group alone, wesuggest, represented an important conduitfor information flow among mill towns.4 In-deed, indirect network ties may be as impor-tant to social movement diffusion as directlinks (Soule 1997).

Media, including the radio, can be moredirectly influential when it shapes prospec-tive movement participants’ perceptions ofpolitical opportunity. It is here—in drawinga distinction between political opportunity ata structural level and perceived political op-portunity among potential insurgents, andspecifying the mechanism(s) through whichperceived opportunity may be altered—thatpolitical opportunity theory has been limited(Kurzman 1996; McAdam 1982; Tarrow1988). By disseminating information geo-graphically, media can mold the politicalperceptions of a dispersed population(Kahan 1999). This was the case with radioand its establishment in the U.S. South. Forthe first time in U.S. history, a presidentspoke over this medium to southern workersin the format of “ fireside chats,” duringwhich a national political commitment to theplight of workers and workers’ right to col-lectively organize was communicated de-spite local elite repression (R. Brown 1998;Hall et al. 1987).

Media can also be instrumental by alter-ing workers’ sense of collective experienceand solidarity. Historically, one of the mostobvious means through which group identityhas been manifested and shared is throughlanguage generally, and music specifically(Eyerman and Jamison 1998). Language andvocabularies of motive, of which music lyr-ics are no exception, are important facts insocial action not reducible to individual so-cial psychology. Rather, verbalization,through speech or song, is always conversa-tional and dynamic, often political, and po-tentially consciousness-altering (Flacks1999; W. Gamson 1992b; Goffman 1981;Lichterman 1999; Mills 1939, 1940). Assuch, “ the language of situations as givenmust be considered a valuable portion of thedata to be interpreted and related to theirconditions” (Mills 1940:913).

Although consistent with classical the-ory’s interest in culture and more recent ef-forts to develop social movement theory’semphasis on cultural processes (Melucci1985; Taylor and Whittier 1992), it is no-table that so few analyses systematicallyconsider music as a component of the col-lective action repertoire or as a form of dis-course through which collective identity isfostered and movement solidarity isachieved. In a study of American left-wingmusic, Denisoff (1972) distinguishes be-tween songs that are rhetorical, highlightingdiscontent, and songs that aimed at recruit-ment and solidarity maintenance during ac-tive, collective protest (also see Flacks 1999;McLaurin and Peterson 1992). Eyerman andJamison (1998) concur and suggest that thearticulation of identity through music is cen-tral to movement formation. Indeed, musicnot only adds an authentic air to the plea forsocial action because of its emotional appeal(Pratt 1990), but it also builds and reinforcesidentity and group commitment throughritual and the act of singing collectively(Flacks 1999).5 In the southern case, the folk

4 This resonates with Gurlach’ s (1999) discus-sion of “ traveling evangelists.” By carrying mes-sages, spreading ideology, and building personalrelationships across the network, they play a rolein social movement links and spatial diffusion.

5 Drawing from symbolic interactionism,Flacks (1999) suggests that singing is a form ofrole playing, requiring one to take the identity ar-ticulated in the song, at least momentarily. Thisprocess may be further reinforced by collectivesinging, a symbolic gesture whereby participantsdemonstrate membership in, and commitment to,the group.


tradition of storytelling through music has along and important history (Malone 1979).6

Thus, we expect to find that music, and itsdissemination via radio and ex-mill workermusicians, was an influential part of the so-cial movement repertoire for southern textileworkers.

DATA

We draw our data from a number of sources.Data on radio station foundings in the Southprior to 1935 were gathered from the ar-chives of the Federal Communication Com-mission (FCC). These records provide theday, month, and year of each radio stationfounding, along with the radio station name,ownership, and the city in which it operated.Coupling FCC data with data on textile millconcentration, derived from Clark’s Direc-tory of Southern Textile Mills (1929) andDavidson’s Textile Blue Book (1935), anddata on strike activity drawn from varioussources,7 allows us to address the most im-portant empirical question—was radio spa-tially proximate enough to textile mills tohave played a part in the insurgency that oc-curred? This data is supplemented with his-torical evidence on radio ownership amongmill workers.

Along with establishing the spatial linkbetween mill concentration, radio stationfoundings, and strike activity is the need tospecify and analyze how radio was influen-tial. Here, we rely on historical data pertain-ing to the impact of politically orientedbroadcasts, archival and interview data onex-mill worker musicians during the periodin question, and on content analyses of mu-sic lyrics. Political data are drawn from his-torical accounts and the archives of the NewDeal Network Library and Franklin D.Roosevelt Library. We gathered informationon musicians and songs dealing with south-ern textile mills, textile mill town life, andtextile worker insurgency from 1929 to 1934from the Archie Green Papers of the South-ern Folk Life Collection at the University ofNorth Carolina Archives and a variety ofother sources.8 The resulting collection of 35songs represents the most comprehensivecompilation of Southern textile songs ofwhich we are aware.9

ANALYTIC STRATEGY

AND RESULTS

We first examine radio station foundings inthe South and the proximity of these stations

6 Of course, music has been important for otherlabor insurgencies in the United States, back tothe classical industrial ballads by Joe Hill. Un-fortunately, and despite the existence of collec-tions of songs of insurgency across a variety ofindustries (see Greenway 1956; Hille 1948;Lieberman 1989; Lomax 1960; Lomax, Guthrie,and Seeger 1967), the link between music, socialprocesses, and mobilization has, with few excep-tions (e.g., McLaurin and Peterson 1992), re-ceived little systematic attention among sociolo-gists.

7 Complete strike data for the U.S. South dur-ing the early 1930s is difficult to find, as no sys-tematic records for the region were kept. We did,however, compile a relatively extensive list ofstrike activity drawing from Hall et al. (1987),Nolan and Jones (1976), Salmond (1995), Simon(1998), and Yellen (1936). While our list and theresulting map of strike activity that follow un-doubtedly capture the general concentration ofstrike events and the largest, most pronouncedstrikes in the region between 1929 and 1934, wesuspect that some small, less visible strikes maybe missing from our data.

8 These sources include, but are not limited to,American Folksongs of Protest (Greenway1953), The People’s Song Book (Hille 1948),Folk Songs of North America (Lomax 1960),Hard Hitting Songs for Hard-Hit People (Lomaxet al. 1967), American Industrial Ballads (Seeger1992), Babies in the Mill (Dixon 1998), and in-terviews undertaken by the authors.

9 We limited our sample of songs to those ema-nating from the South that were recorded, sung,played, or transcribed prior to 1935. Ideally, wewould also have radio “play lists” for the timeperiod in question. Unfortunately, given the new-ness of radio at the time, such records were notkept. What we do know, given the historical lit-erature and interviews we undertook with surviv-ing mill musicians and mill workers, is that manyof the musicians who wrote and sang mill-relatedsongs traveled from radio station to radio stationto play, sang some of these songs during live ra-dio shows, and spent significant time with thoseresiding in the mill towns (Malone 1968; Wig-gins 1987). This suggests that the impact of ra-dio on collective identity may have been directthrough the playing of mill songs, or indirectthrough the creation of this autonomous group of


to textile mills and mill towns where themost pronounced strikes of 1929–1934 oc-curred. Spatial patterns are analyzed by geo-graphical mapping. These visual representa-tions show whether radio was a viable mech-anism through which perceptions of politi-cal opportunity could have been altered andcollective identity manifested. Supplemen-tary quantitative analyses of the populationof Southern mill towns (N = 542), whether astrike occurred, and whether there was a ra-dio station within the city limits or near tothe city, help us further establish the link be-tween radio stations and actual strike events.

The analysis then focuses on transmissioncontent, directly or indirectly via radio, andits implications for perceptions of politicalopportunity and collective identity. For anal-yses of political content, we outline a majorshift in political structure and opportunityduring the time period in question—that is,Roosevelt’s New Deal. We then draw fromarchival sources to describe the role radioplayed in communicating this new context tosouthern mill workers, magnifying their per-ceptions of opportunity and offering legiti-macy to their experiences and claims of in-justice at the hands of mill owners.

Consistent with Hodson’s (1999) recom-mendation for systematically analyzing qual-itative content data and converting it intoquantitative and descriptive summary statis-tics, we created a coding scheme for analyz-ing music lyrics. Both authors coded each ofthe 35 songs along various dimensions, re-ported in Tables 2, 3, and 4. Inter-coder reli-ability was approximately 91 percent. Incases of disagreement, we went back to theoriginal source to rectify the difference. Ouranalysis of music lyrics employs a dichotomysimilar to Denisoff ’s (1972), distinguishingbetween songs focusing on discontent andcollective experience (N = 21) and songsdealing with protest (N = 14).10 For songs of

discontent, we report frequency breakdownsof the problems workers faced and the inter-pretation of causes. For songs of protest, wealso use frequency breakdowns but focus onthe primary intent of the song and the type ofelite countermobilization described, if any.We supplement these summary statisticsthroughout with discussion of the musiciansand illustrative lyrics.

Radio Station Foundings, Proximity

to Textile Mills, and Strike Activity

Radio quickly found its way into the U.S.South in 1922, when on February 3, the firstlicense was granted to WGH in Montgom-ery, Alabama. Within one month, stationswere founded in Charlotte, Memphis, At-lanta, Charleston, Richmond, and Mor-ganton. Interest in this new medium was in-tense, to say the least, as 43 operating li-censes were granted to various stationsacross the South by the end of that year.

Early ownership patterns in the South mir-rored those in the country as a whole, withheavy reliance on department stores, insur-ance companies, universities, amateurs, andmajor electronics manufacturers (e.g., Gen-eral Electric, Westinghouse, etc.) who hadlinks to the newly emerging recording indus-try (e.g., RCA/Victor) (Garafalo 1997). Ourexamination of FCC archival records indi-cates that traditional industries in the South(e.g., agriculture and textiles) played little orno role, while colleges, music companies,battery companies, and relatively new indus-tries, such as automobile and insurance deal-ers, did. Take, for example, station WBT inCharlotte and station WRBU in Gastonia,North Carolina. Although located near tex-tile mills, mill owners apparently had no in-volvement or control in either station. Rather,WBT was owned by C. C. Coddington, anentrepreneur and Buick dealer, while WRBUwas owned by A. J. Kirby Music Company.

Two important political issues emergedduring the early years of radio. The first hadto do with the nature of broadcasts, and de-bates over public versus commercial inter-ests. Despite considerable discussion, littlefederal intervention occurred until the midto late 1930s, when the Communications Actof 1934 created the FCC. Even then, how-ever, regulation and oversight lagged behind

musicians who drew some of their songs fromworkers and shared these songs across communi-ties.

10 Music probably serves a different functionin each context. Prior to mobilization, musicshapes collective experience, group-building, andinterpretive understanding. Thereafter, the goal issolidarity maintenance and description of poten-tial countermobilization in the face of active pro-test (Flacks 1999).


the new, developing technology (Bittner1982; Garafalo 1997). Thus, stations wererelatively free to broadcast what theywanted. Most stations depended on the pro-gramming of NBC and CBS, while allocat-ing between one and three hours a day toprogramming and live shows that catered tothe specific interests of local populations(Summers 1958). In the South, for instance,“depending on the location, one could usu-ally begin the day or spend the noon hourlistening to a program of hillbilly, Cajun,blues, or gospel music” (Malone 1979:71).

The other issue had to do with transmis-sion range. This concern was addressed bythe Radio Act of 1927, which limited thepower of most stations in order to reduce in-terference among stations sharing the samefrequency (Hogan [1930] 1971). Althoughinitially a problem in the South, given itsfew stations and the inability of broadcaststo reach rural populations, this was partiallyremedied by the increase in foundings in thelate 1920s and into the 1930s. Figure 1 dis-plays the increase in number of stations dur-ing this time period. The increase between1922 and 1930 was significant, and largeSouthern cities like Atlanta, Memphis, andNashville boasted as many as five stations.Small stations were also established, andthey found a niche in rural towns with re-spectable population concentrations (Hall etal. 1987). By 1931, foundings tapered off

because of increased competition and mar-ket saturation.

Radio station transmissions in the largeSouthern cities were not strong enough toreach the high textile concentration countiesof western North Carolina and northwesternSouth Carolina, where worker mobilizationoccurred between 1929 and 1934. Thus, ifradio was central to information flow, and toperceptions of political opportunity and col-lective identity for mill workers specifically,some stations needed to have been estab-lished in areas geographically proximate tomill concentration. Map 1 overlays the con-centration of textile mill manufacturing withcity-specific radio station foundings prior to1935 in an effort to assess whether this wasthe case.

Map 1 shows a clear “ radio belt” cuttingthrough the center of the densest textile con-centration—an area where the most pro-nounced strike activity and the largest strikesof the late 1920s and mid 1930s occurred(Map 2). This link between radio stationfoundings and strike activity is further con-firmed in associational analyses. Table 1shows that strikes were significantly morelikely to occur in southern mill towns thathad a radio station. Because transmissionsextended beyond the limits of a particularcity (i.e., 25 to 40 miles, on average), wealso examine the impact of having a radiostation near the city—that is, not in the city,

1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934

140

130

120

110

100

90

80

70

60

50

40

30

20

Nu

mb

er o

f R

adio

Sta

tio

ns

Year

Figure 1. Number of Radio Stations in the Southern States of Alabama, Georgia, North Carolina,South Carolina, Tennessee, and Virginia, 1922 to 1934

Source: FCC Archives.


Spr

ing

Hill

Ral

eigh

Ash

evill

e

Cha

rlotte

Gre

ensb

oro

Wilm

ingt

on

Hen

ders

onW

inst

on-

Sal

emD

urha

m

Mem

phis

Nas

hvill

e

Law

renc

ebur

g Col

dwat

er

Tulla

hom

a

Col

umbi

a

Kno

xvill

e

Cha

ttano

oga

Spr

ingf

ield

Uni

on C

ity

Atla

nta

Dec

atur

Col

lege

Par

k

Mac

on

Sav

anna

h

Gai

nesv

ille

Tho

mas

ville

Col

umbu

s

Tocc

oa

Ath

ens

Tifto

n

Aug

usta

Cha

rlest

on

Ora

ngeb

urg

Gre

envi

lleC

lem

son

Spa

rtan

burg

Mon

tgom

ery

Birm

ingh

am

Mob

ile

Aub

urn

Gad

sden

Talla

dega

Ric

hmon

d

Bla

cksb

urg

Mon

roe

Por

tsm

outh

Roa

noke

Arli

ngto

n

Mou

ntV

erno

n

Pet

ersb

urg

Virg

inia

Bea

ch

Em

ory

Dan

ville

Lync

hbur

g

Har

risbu

rg

Gas

toni

a

Nor

folk

Col

umbi

a

Nu

mb

er o

f S

pin

dle

s in

Co

un

ty

Less

than

50,

000

Bet

wee

n 50

,000

and

200

,000

Mor

e th

an 2

00,0

00

Map

1. T

exti

le M

anu

fact

uri

ng

Com

mu

nit

ies

and

Rad

io S

tati

on F

oun

din

gs i

n t

he

U.S

. Sou

th P

rior

to

1935

Sour

ces:

Cla

rk’s

Dir

ecto

ry o

f So

uthe

rn T

exti

le M

ills

(19

29),

Dav

idso

n’s

Tex

tile

Blu

e B

ook

(193

5), a

nd F

CC

Arc

hive

s.

Den

otes

cit

y w

ith

at l

east

one

rad

io s

tati

on f

ound

ing.


Bel

ton

Pie

dmon

t

Eliz

abet

hton

Bel

mon

t

Gra

nite

Fal

ls

Dur

ham

She

lby

Gas

toni

a Pin

evill

e

Mar

ion

For

est C

ity

Lexi

ngto

n

Dra

per

Cha

rlotte

Hen

ders

on

Hig

h P

oint

Bur

lingt

on

And

erso

n

Bat

h

Gre

enw

ood

Hon

ea P

ath

War

e S

hoal

s

Pel

zer

Gre

envi

lleW

oodr

uff

Sen

eca

Col

umbi

a

Abb

evill

eC

alho

un F

alls

Che

ster

Gaf

fney

Roc

k H

illU

nionYo

rk

Birm

ingh

am

Gun

ters

ville

New

nan

Trio

n Col

umbu

s

Dan

ville

Cen

tral

Whi

tmire

Arc

adia

Aik

en

Roc

king

ham

Aug

usta

Sar

gent

Laur

inbu

rg

Gre

er

Wav

erly

Sel

ma

Mob

ile

Hun

tsvi

lleF

lore

nce

Gad

sden

Ann

isto

n

Con

cord

Gre

ensb

oro

Por

terd

ale

Ced

arto

wn

Cow

pens

Jasp

er

Alb

ertv

ille

John

son

City

Har

riman

Aus

tell

Spi

ndal

eK

anna

polis

Bes

sem

er C

ity

Map

2. T

exti

le M

ill

Str

ike

Act

ivit

y ac

ross

th

e U

.S. S

outh

, 192

9 to

193

4

Sour

ces:

Hal

l et

al.

(19

87),

Nol

an a

nd J

ones

(19

76),

Sal

lmon

d (1

995)

, Sim

on (

1998

), a

nd Y

elle

n (1

936)

.

Den

otes

cit

y w

here

str

ike

occu

rred

.


but in the county or in an adjoining county.This association, albeit weaker, is also sig-nificant.11 It is clear, given these patterns,that many southern mill workers lived andworked within the concentric rings of radiotransmission and that this may have shapedthe insurgency that unfolded.

But did workers have access to radios?Radio ownership among mill workers wassurprisingly high given their low economicstatus. Hampton’s (1935) analysis of leisure-time activities among 122 mill workersacross three mill villages during this periodsuggests that up to 70 percent had radios intheir homes. Furthermore, when asked torank 46 leisure-time activities, listening tomusic on the radio was ranked highest. “Theradio is kept going all the time there ain’ t nostatic” (woman, quoted in Hampton 1935:61). While music programs were most popu-

lar, others opted for “preaching and talks onthe government” (elderly man, quoted inHampton 1935:61).

Radio and Political Opportunity

Media may influence collective action by al-tering actors’ perceptions of political oppor-tunity. Pratt (1990) notes that the rapid dif-fusion of information to more peoplethrough media was intensified with the ad-vent of radio. The South was no exception,particularly when it came to political infor-mation. For the first time, southern workersfelt as if they had a direct line to the presi-dent, as Franklin Roosevelt took to the air-waves and entered their homes. Rooseveltused the new medium of radio to move be-yond traditional means of political discourseand to “ reach over and around the networksof state and local party structures and politi-cal personalities” (Kahan 1999:185). Roose-velt, in fact, created “a new political con-text” through radio, utilizing its directnessand its potential to circumvent local powerbases.

With respect to the rights and grievancesof workers, Roosevelt used weekly “ firesideradio chats” to signify his support. The thirdof these broadcasts, which was titled “On thePurposes and Foundations of the RecoveryProgram” and aired on July 24, 1933, dealtexplicitly with the need for industrial reformand better working conditions. This leftworkers, including those in the South whotraditionally felt isolated from national basesof power, with the impression that theycould count on “ the intervention of the fed-eral government as a lever against localelites and guarantor of workers’ rights” (Hallet al. 1987:292). Roosevelt also urged work-ers to write him—and they did, in unprec-edented numbers (Sussman 1963).

Many southerners took part in the write-incampaign to the president (McElvaine 1983;Sussman 1963). Those who felt excludedfrom the political process, most notablywomen, now felt empowered to share theirgrievances, discussed the need and desire toorganize collectively and, through their de-tailed letters, encouraged the powerful toconsider the mill workers’ plight. Many ofthese letters also made clear that southernworkers believed there would be federal in-

11 We also ran logistic regression models pre-dicting the likelihood of a strike that includedboth radio indicators along with controls for thestate and textile mill density in the city. Thesefindings (available from the authors upon re-quest) are consistent with the analyses reported.They suggest that a strike was greater than threetimes more likely to occur if the mill town had aradio station, and between one and one-half andtwo times more likely if there was a radio stationin the county or in an adjoining county.

Table 1. Association between StrikeOccurrence and the GeographicLocation of Radio Stations Relative toSouthern Mill Towns, 1929 to 1934

Proximity of Radio Strike Occurred

Station to Town No Yes

In the Town No 456 50

Yes 22 14

χ2 (d.f. = 1) 27.2***

Near the Towna

No 263 24

Yes 215 40

χ2 (d.f. = 1) 7.0**

Note: Number of mill towns = 542.a Station not located in the town, but in the county

or in an adjoining county.**p < .01 ***p < .001 (one-tailed tests)


tervention (Hall et al. 1987). The followingletter, sent to the president during the 1934strike, describes the consequences of thewalkout and includes a “personal” appeal toF.D.R. to get involved.

Dear President Roosevelt,

I hope you can spare the time for a fewwords from a cotton mill family, out of workand almost out of heart and in just a shortwhile out of a house in which to live. Youknow of course that the realators are puttingthe people out when they cannot pay the rentpromptly. and how are we to pay the rent solong as the mills refuse us work, merely be-cause we had the nerve to ask or “demand,”better working conditions.

I realize and appreciate the aid and foodwhich the government is giving to the poorpeople out of work Thanks to you.

but is it even partly right for us to bethrown out of our homes, when we have nochance whatever of paying, so long as thebig corporations refuse of work. I for one amvery disheartened and disappointed guessmy notice to move will come next.

what are we to do. Wont you try to helpus wont you appeal, “ for us all,” to the realestate people and the factories

hoping you’ ll excuse this, but I’ ve al-ways thought of F.D.R. as my personalfriend.

C.L.F. (Columbus, Ga.)

(Henry Morgenthau Jr. Collection 1934,emphasis and punctuation as in original)

The impact of Roosevelt’s radio transmis-sions on the consciousness of southern tex-tile workers was witnessed firsthand in 1933by Martha Gellhorn, a reporter hired by Fed-eral Emergency Relief Administration direc-tor Harry Hopkins to investigate social andeconomic conditions in the South. In her re-port to Hopkins on the conditions in GastonCounty, North Carolina, Gellhorn describespoor health conditions along with unfair millowner practices, while also noting:

All during this trip I have been thinking tomyself about that curious phrase “ red men-ace,” and wondering where said menace hiditself. Every house I visited—mill worker orunemployed—had a picture of the President.These ranged from newspaper clippings (indestitute homes) to large colored prints,

framed in gilt cardboard. The portrait holdsthe place of honour over the mantel; I canonly compare this to the Italian peasant’ sMadonna. And the feeling of these peoplefor the president is one of the most remark-able emotional phenomena I have ever met.He is at once God and their intimate friend;he knows them all by name, knows theirlittle town and mill, their little lives andproblems. And, though everything else fails,he is there, and will not let them down.(Gellhorn 1933)

Workers also spoke directly with Gellhornabout Roosevelt, sharing their confidencethat the president was on their side.

You heard him talk over the radio, ain’ tyou? He’ s the only president who ever saidanything about the forgotten man. We knowhe’ s going to stand by us.

He’ s a man of his word and he promised us;we aren’ t worrying as long as we got him.

The president won’ t let these awful condi-tions go on.

The president wanted the Code [NIRA, Sec-tion 7a, of the Textile Code]. The presidentknows why we struck.

(Gellhorn 1933)

The initial local political autonomy of ra-dio also enabled local organizers to gain ac-cess to the southern airwaves. At the out-break of the massive strike of 1934, for in-stance, UTW vice-president FrancesGorman

. . . took his cue from the rising generationof Millhands. He went on the radio, gaininghours of air time at no expense. He encour-aged “ flying squadrons” of cars and trucksto speed through the countryside—and theydid, closing mills so rapidly that “ tabulatorsalmost lost check.” (Hall et al. 1987:329)

Consequently, the airways became an arenain which political battles over the right ofworkers to collectively organize were foughtwhen George Sloan, head of the Cotton Tex-tile Institute, went on the airwaves to expressthe position of mill owners (Hall et al. 1987).

Use of the airwaves by local organizers anda progressive president intent on addressingwork-related issues was important for millworker perceptions of political opportunity.


The initial autonomy of radio stations,mostly a function of ownership patterns anda lack of local elite control, allowed the ex-pression of grievances by local organizersand the use of radio as a networking resource.Roosevelt’s “ fireside chats” communicated anational political commitment to these work-ers along with an explicit recognition of theirright to organize. Write-in campaigns bysouthern mill workers showed that they rec-ognized a political opening and were listen-ing. What remained unclear was the degreeto which federal intervention on their behalfwould actually occur.

Music, Collective Experience,

and Discontent

Beyond explicitly political broadcasts, radiohad a direct and indirect impact on the cul-tural life of mill towns by creating a nichefor musicians, many of whom were ex-millworkers. These musicians knew mill life,and the songs they wrote showed they werestill wedded to their mill experiences. Theyalso made it a point to visit mill towns whentraveling from radio station to radio stationto perform (Malone 1968; Wiggins 1987).Ex-mill workers, like Charlie Poole and his

North Carolina Ramblers, for instance, wereable to play music full-time after landing arecording contract in 1925. Poole’s band wasextremely popular throughout the mill townsof the Southeast, playing popular tunes andsongs that spoke to the lives of mill work-ers. In fact, the bulk of the music that millworkers listened to reflected not only millwork but the whole existence of mill life(Hall et al. 1987).

The Dixon Brothers, former mill workers,recorded “Weave Room Blues,” “Weaver’sLife,” “Factory Girl,” and “Hard Times inHere,” while also performing traditionalsouthern folk tunes and mill-related songslive on the airwaves (Photograph 1). Lesserknown singers also recorded cotton-millsongs. Frank Welling and John McGee, alsoknown as the Martin Brothers, recorded“ The Marion Massacre” for Paramount,while David McCarn recorded “Cotton MillColic” and “Serves Them Fine” for Victor,the latter song chiding textile workers forleaving their mountain homes for the cottonmills (Malone 1968; Peterson 1992).

Record sales dropped during the Depres-sion, and live broadcasts of music becameeven more important for these musicians.Sometimes, during live “barn dance” broad-

Photograph 1. The Dixon Brothers Performing Live on the Radio in Charlotte, North Carolina

Source: John Edwards Memorial Collection, Southern Folklife Collection, University of North CarolinaArchives.


casts, like J. W. Fincher ’s Crazy WatersCrystals Saturday Night Jamboree broadcaston WBT in Charlotte, these musicians andgroups would play together and share mill-related songs (Green 1963a; Malone 1968).Their sharing of popular tunes and their abil-ity to travel “paved the way for some of thefirst hillbilly bands to earn their livings per-forming at a Spartanburg high school audi-torium one night, at a Gastonia mill recre-ation center the next evening, and on a Char-lotte radio station the following morning”(Tullos 1989:2).

Woody Dewey, a radio archivist associ-ated with the John Rivers Communications

Museum at the College of Charleston, inCharleston, South Carolina, remembershearing musicians such as Bill and CharlieMonroe, Bill Carlisle, and others sing songsof interest to mill workers. He also recallslistening to specific broadcasts in the early1930s as a 10-year-old while visiting hisuncle:

They listened to a program each day fromWIS, I think that’ s right, in Columbia. Any-way, they listened to a program called theAristocratic Pigs. . . . Fisher Hendley wasthe head pig, I suppose [laughs]. His signa-ture song was “Weave Room Blues,” and heplayed the banjo, and he would play it

Table 2. Frequency Distribution of Textile Workers’ Concerns as Expressed in Songs about theTextile Worker Experience

Family Well-Being Worker Well-Being

Family Children Future and Low Physical MentalSong Title Subsistence Working Children Wages Well-Being Well-Being

The Big Fat Boss and the × Worker

Brown Lung Blues × ×Cotton Mill Blues ×Cotton Mill Colic × ×Cotton Mill Girl ×Cotton Mill Man × ×Factory Girl ×Hard Times Cotton Mill Girls ×Hard Times in Here ×Hard Times in the Mill ×Let Them Wear Their × × Watches Fine

Ma and Pa ×Mill Mother’ s Lament × ×No More Shall I Work in × the Factory

Ol’ Man Craft

Rich Man/Poor Man ×Serves Then Fine ×Shirt Factory Blues

Weave Room Blues × ×Weaver’ s Life ×Winnsboro Cotton Mill Blues ×

Total 5 3 2 7 6 2


[Weave Room Blues] every once in a while.He came to Lancaster with the AristocraticPigs . . . of course, we all had to go up andsee them at the high school there in Lan-caster. We all enjoyed it. Back then, yaknow, there weren’ t no TV. . . ya didn’ tknow what they looked like and in order tosee them and get a glimpse of what theylooked like, you go out when they playedthose personal appearances. They did thattwo or three times a week all around thestate. (Dewey 1999)

Homer “Pappy” Sherrill, himself a legend-ary bluegrass fiddle player who began play-ing on Gastonia, North Carolina, radio sta-tion WSOC in 1929 as a 13-year-old, simi-larly recalls hearing the Dixon Brothers play“Weave Room Blues” and other mill-relatedsongs over the airwaves:

I’ve heard’em play it. They were on WBT[Charlotte] when we were up there. Theywere connected to Fisher Hendley. It was inthe early thirties. [Starts singing WeaveRoom Blues during interview]. “ Elevencents of cotton, 40 cents of meat, how in theworld can a poor man eat, I got them lone-some weave room blues. (Sherrill 1999)

Thus, radio transmission was directly in-fluential for the cultural life of mill towns.More indirectly, radio created a network ofex-mill workers (i.e., musicians) who trav-eled between towns, drawing from and con-tributing to the cultural life and experiencesof those still working in the mills. The mu-sic and the emergence of radio “ put mill-hands across the region in touch with eachother, allowing those who missed the travel-ing musicians’ performances to hear and en-joy the same music,” thus fostering a strongsense of group identity (Hall et al. 1987:261). Mill owners, on the other hand, peri-odically saw the emerging music and its con-sciousness-altering potential as a threat. Forexample, in Danville, Virginia in 1930, localauthorities and mill owners attempted to for-bid workers from singing Dave McCarn’srecently released “ Cotton Mill Colic”(Rorrer 1982).

Songs pertaining to mill life often em-ployed a collective sense of experience, us-ing the words “ we,” “ us,” and “ our,” andcommunicated anxieties specific to the ex-periences of most mill workers. Table 2 (on

p. 35) shows these songs (N = 21), catego-rized by the concerns they address, specifi-cally family well-being and/or the workerher or himself. This table illustrates the rich-ness of these songs, the multiplicity of con-cerns they touch on, and the fact that the is-sues addressed could vary by verse. Recog-nizing the overlap in themes highlights thecomplexity of these songs and their lyrics.

Ten of the 21 songs dealing with millwork generally relate concerns for family,while fifteen, or seventy-one percent, de-note problems faced by workers. Amongconcerns for family well-being, children areoften mentioned, something that undoubt-edly evoked a broader concern and univer-sal appeal among listeners.12 Five of thesedeal with family subsistence—the ability toprovide for one’s family’s basic needs. Thisconcern is most obvious in some of the lyr-ics to a song aptly titled Mill Mother’sLament:

And when we draw our money,

Our grocery bills to pay,

Not a cent to spend for clothing,

Not a cent to lay away.

And on that very evening

Our little son will say:

“ I need some shoes, mother,

And so does sister May.”

How it grieves the heart of a mother,

You every one must know;

But we can’ t buy for our children,

Our wages are too low.

Another song, Cotton Mill Man, reflects thegrieving heart of a mill worker and his fearthat his son may also end up working in themill:

12 The following quoted lyrics for Mill Moth-ers Lament, Let Them Wear Their Watches Fine,Weave Room Blues, Big Fat Boss Man and theWorker, and The Marion Strike were obtainedfrom Greenway (1953). Lyrics for Cotton MillMan, Winsboro Cotton Mill Blues, Union Spe-cial, Here We Rest, and On a Summer Eve weregathered from Green (1963b).


I watched my woman cry when

our baby daughter died.

I couldn’ t make her understand

why the doctor never came,

The lack of money was to blame.

I cussed the day that I became a

cottonmill man.

Lord, don’ t let my son grow up

to be a sweaty cottonmill man.

While these lyrics reflect the collectiveexperiences of mill life and their conse-quences for family sustenance and the well-being of children, over half of these songs(15) describe the toll mill work takes on theworker her or himself. Seven of these songsdeal with the low economic return for theamount of work put in. In four verses of LetThem Wear Their Watches Fine, this griev-ance is coupled with a discussion of the so-cial status consequences of mill work.

We work from week end to week end,

And never miss a day,

And when that awful pay day comes

We draw our little pay.

On pay day night we go back home

And set down in a chair.

The merchant knocks upon our door

He’s come to get his share. . . .

Those fancy folks that dress so fine

And spend their money free,

They don’ t have time for a factory hand

That dresses like you and me.

As we go walking down the street

All dressed in lint and strings,

They call us fools and factory trash

And other low down things.

Many of the 21 songs dealing with thegeneral experiences of southern mill work-ers in the 1920s and 1930s also specify thecause(s) of the problems they face. Table 3shows the coding according to the cause(s)

specified—the work process and/or the neg-ative impact of human agents (i.e., bosses,managers, and/or scabs). Given our focus onthe manifestation of collective identity, classconsciousness, and insurgency through mu-sic, the distinction between cause and effectis important. If lyrics do not address a cause,then consciousness relating to where griev-ances should be aimed will remain unclear.This interpretational link between cause andeffect is indeed crucial if social movementdiscourse and framing processes are to beeffective (W. Gamson 1995; Snow and Ben-ford 1992; Taylor and Whittier 1995). It isnoteworthy that a cause is specified in morethan three-quarters of these songs. Sixteen ofthe 21 associate discontent with the workprocess, while the same number specify ahuman culprit.

Lyrics pointing to the work process fallinto three principal categories: general workconditions, the length of the work day, andthe introduction of scientific management.Twelve of the 21 songs place the blame formill worker problems on general work con-ditions, including the speed, cleanliness, andnoise associated with mill work. One verseof Weave Room Blues provides a vivid im-age of these conditions:

Slam out, break out, makeouts

by the score,

Cloth all rolled back and

piled up on the floor.

The bats are running into strings,

they’ re hanging to your shoes,

I’m simply dying with them

weave room blues.

Notably, much of the worker complaint isdirected specifically at employers and man-agers rather than being seen as a conse-quence of mill work. Vallas’s (1987) analy-sis of the labor process and the social andorganizational aspects of work suggests thata focus on managers and owners, rather thanthe labor process generally, has strongerramifications for class consciousness (Bill-ings 1990; Della Fave 1980). W. Gamson(1995) concurs and suggests that an injus-tice frame will be more effective at recruit-ing and mobilizing if the target is a concrete


actor, preferably a person or corporation pre-sented as malicious, selfish, or greedy. TheBig Fat Boss and the Worker—a songpenned by Ella May Wiggins, a mill workerand local organizer who was killed in anambush during the Gastonia uprisings—con-veys such specificity and resulting polariza-tion by straightforwardly attributing workerproblems to mill owner exploitation. Twoverses in particular stand out:

The boss man hates the workers,

the workers hates the boss.

The boss man rides in a big fine car

and the workers has to walk.

The boss man sleeps on a big fine bed

and dreams of his silver and gold.

The workers sleeps in an old straw bed

and shivers from the cold.

Some of this exploitation, as workers rec-ognized, took the form of paternalistic prac-tices, as related in the following two versesof Cotton Mill Man:

The company taught us all the rules

on how to work the spinning spools,

So the boss’s son could drive a

big black sedan.

Table 3. Frequency Distribution of Causes of Textile Workers’ Concerns as Expressed in Songsabout the Textile Worker Experience

Work Process Bosses, Managers, Scabs

General Work Length of Scientific Exploitation Managerial PresenceSong Title Conditions Work Day Management by Owner Control of Scabs

The Big Fat Boss and the × Worker

Brown Lung Blues ×Cotton Mill Blues × ×Cotton Mill Colic × ×Cotton Mill Girl × ×Cotton Mill Man × ×Factory Girl × ×Hard Times Cotton Mill Girls ×Hard Times in Here × ×Hard Times in the Mill ×Let Them Wear Their × Watches Fine

Ma and Pa ×Mill Mother’ s Lament ×No More Shall I Work in × × the Factory

Ol’ Man Craft × ×Rich Man/Poor Man ×Serves Then Fine ×Shirt Factory Blues × ×Weave Room Blues × ×Weaver’ s Life ×Winnsboro Cotton Mill Blues × ×

Total 12 3 1 8 7 1


The company owned the houses and the

company owned the grammar school,

You’ ll never see an educated

cottonmill man.

They figure you don’ t need to learn

anything but how to earn

The money that you paid upon demand

To the general store they owned or else

they’d take away your home

And give it to some other homeless

cottonmill man.

Managers also are blamed for problemsexperienced by workers. In this verse fromWinnsboro Cotton Mill Blues, the cause ofworker duress is clearly managerial over-sight and greed:

Old man Sargent sitting at the desk,

The damned old fool won’ t give us no rest.

He’d take the nickels off a dead man’s eyes,

To buy a Coca Cola and a Eskimo pie.

Clearly, songs of mill worker experienceand discontent have a general appeal—anappeal that transcends the specifics of a par-ticular mill town and that reverberates withthe day-to-day realities of mill life in theSouth. Not only do these songs appeal tocollective understanding and concerns relat-ing to family subsistence, the well-being andfuture of children, and specific problems af-fecting workers, but they also provide aframework through which such concerns areinterpreted in a causal fashion. This is cru-cial if the framing aspect of social movementculture is to invoke focused collective action(W. Gamson 1995; Snow and Benford 1992;Snow et al. 1986; Taylor and Whittier 1995).Put simply, songs afforded workers a frame-work through which the similarity of theirplight became increasingly obvious; theyalso shifted accountability away from theworkers and toward the labor process and itsbeneficiaries.

Songs of Protest

Music is influential not only in its impact oncollective experience and group identity but

also because it serves as a unifying forceduring mobilization. As Morris (1984) notesin his analysis of the southern civil rightsmovement, this is particularly importantwhen countermobilization is fierce (Denisoff1972; Flacks 1999). Symbolism, ritual, anddiscourse through music are crucial formaintaining solidarity among participantsand for shaping the consciousness of nonpar-ticipants so that they become sympathetic to,or are actually recruited into, the movement.

In the case of southern textiles, songs thatemerged out of earlier strikes became songsof unification across mill towns during laterstrikes (Denisoff 1971; Huber 1998; Malone1979). The Gastonia Strike of 1929 standsout as one of the first labor conflicts in theSouth to create a repertoire of protest songsoutlining the plight of workers as well astouching on strike issues and elite counter-mobilization strategies. The importance ofsong during an active protest is evident inPhotograph 2. It shows the Four TobaccoTags (foreground), a popular recording andradio group during the era, playing and sing-ing at a 1934 strike near Austell, Georgia,while striking workers danced in order toblock the doorways to the Clark Threadmill.

Table 4 classifies these 14 textile protestsongs by their primary intent. Unlike thesongs of general mill worker experience de-scribed previously, these songs are more di-rect in their goal and thus were easier tocode. Most (9) have as their main aim soli-darity maintenance and recruitment during astrike period. Like the general songs of milllife summarized previously, language islargely inclusive, referring to “we,” “our,”and “ fellow workers.” Some lyrics, such asthese from Here We Rest, attempt to main-tain solidarity in the face of strike-breakingby scabs:

We are standing on guard

Both night and day,

We are doing our best

To keep scabs away.

We are 1200 strong

And the strike still is on,

And the scabs still are standing

But they won’ t scab for long.


Some lyrics appeal to the worker’s broadsense of commitment to his/her fellow work-ers, as exemplified in this verse from On aSummer Eve:

If we love our brothers as we

all should do,

We’ ll join this union help fight it through.

We all know the boss don’ t care

if we live or die,

He’d rather see us hang on the

gallows high.

Table 4 also reports the type of counter-mobilization described, if any. Interestingly,13 of the 14 songs describe some form ofcountermobilization, and in more than halfof these, the countermobilization entails co-ercion of employees and employer manipu-lation of the legal-political system. Elitecountermobilization is obvious in the lyricsto The Marion Strike. In this well-knownsong used in subsequent strikes, the per-former provides the listener with a detailedaccount of both corporate influence over thelocal police and the interactive nature of thestruggle that occurred:

The sheriff came down there to the factory,

And brought all of his men along,

And he says to the mill strikers,

“Now boys, you all know this is wrong.”

“But sheriff, we just can’ t work for nothing,

For we’ve got a family to feed,

And they’ve got to pay us more money

To buy food and clothes that we need.

You’ve heard of the stretchout system,

A-goin through the country today,

They put us on two men’s jobs,

And just give us half enough pay.

You know we helped give you your office,

And we helped to give you your pay,

And you want us to work for nothing,

That’s why we are down here today.”

So one word just brought on another,

And the bullets they started to flying,

And after the battle was over,

Six men lay on the ground a-dying.

Photograph 2. Musicians Furnish Music to Dancing Strikers Blocking the Entrance of the ClarkThreadmill Near Austell, Georgia, 1934

Source: Walter P. Reuther Library, Wayne State University.


Note that despite the coercion and legal-po-litical manipulation described, these songsof protest have an optimistic underpinningthroughout—that positive change will hap-pen, albeit through struggle. The last verseof Union Special couples worker optimismwith the recognition that powerful forcesstand in their way:

The city officials are against us,

And the big men of the town.

But they will see in the future

That their playhouse will come down.

These illustrative lyrics demonstrate thecomplex nature and substance of these pro-test songs. They were an important tool inthe protest repertoire, appealing to workers’sense of commitment and similarity of ex-perience, to their own economic well-beingand future, and to social justice, broadly de-fined. Thus, these songs offered a valuabletool with which to recruit and maintain soli-darity. Further, they offered the listener,

whether protest participant or not, a justifi-cation of the mobilization that was unfold-ing by specifying corporate abuses of eco-nomic, legal, and political power. The shar-ing of these themes, directly and indirectlyvia radio, for the first time offered southernmill workers sense of unity and solidaritythat extended beyond the particulars of theirspecific mill town.

CONCLUSIONS

The establishment of radio stations in theSouth and the high concentration textilemanufacturing areas of North Carolina andSouth Carolina was influential for the wor-ker insurgency that unfolded in the 1920sand 1930s. Local organizers used the air-waves to express grievances and to coordi-nate action. Presidential “ fireside chats” al-tered southern mill workers’ perceptions ofpolitical opportunity, leading to the beliefthat national political power was now ontheir side and that the federal governmentwould intervene to counter local elite repres-

Table 4. Frequency Distribution of Characteristics of Textile Workers Protest Songs

Primary Intent

Description Hostility Countermobilization

Maintenance of Strike toward Employee Incarceration HiringSong Title of Solidarity Activity Scabs Coercion of Leaders of Scabs

All Around the Jailhouse × ×Ballad of the Blue Bell Jail × ×Chief Aderholt × ×Come On You Scabs If You × × Want to Hear

Here We Rest × ×Let Me Sleep In Your Tent × Tonight, Beal

The Marion Strike × ×The Marion Massacre × ×On A Summer Eve × ×Roane County Strike at Harriman × ×The Speakers Didn’ t Mind × ×Song Ballet by Ella May × ×Up in Old Loray Waiting for × × a Trial

Union Special × ×

Total 9 4 1 7 5 1


sion. A new collective identity and move-ment solidarity was formed through musicplayed on the radio and spread by local mu-sicians traveling from mill town to milltown. Unfortunately for mill workers, littlefederal intervention occurred, and the strikecampaigns were largely defeated throughstate-sponsored violence and legal-politicalcontrol. Southern businessmen also helpedlobby for, and eventually pushed through,the Taft-Hartley Act, which banned theclosed shop, allowed states to instigate“ right to work” laws, and resulted in a dras-tic reduction in worker petitions for unionelections (Minchin 1997; Roscigno andKimble 1995; Wood 1996).

Our explication of radio’s role in this in-stance of mobilization extends on collectiveidentity and political opportunity frame-works. Collective identity theory, as wehave noted, is useful for establishing theimportance of interpretational, identity, andsolidarity processes for subordinate-groupchallenge, in a manner consistent with tra-ditional emphases on class consciousness.What is often lacking, though, is elabora-tion on how media can play a fundamentalpart in these processes and, indeed, shapethe collective experience and feelings of“ groupness” across a geographically dis-persed population. Furthermore, little sys-tematic attention among collective identitytheorists has been devoted to examining therole of music as an important element of thesocial movement repertoire—a role that notonly provides a basis through which collec-tive identity may be realized but also one inwhich an interpretational frame of causeand effect is offered to the listener. In thecase of mill-related music, this conscious-ness-altering potential was evident duringthe 1920s and 1930s and persisted even intothe mid-1960s when, for example, a record-ing of “Cotton Mill Man” was considered“ too provocative” to be played on many ra-dio stations in southern mill towns(Peterson 1992). Following previous workon culture and framing processes (W.Gamson 1995; Snow and Benford 1992;Snow et al. 1986; Taylor and Whittier1995), we believe that music and the inter-pretational frame it can provide is funda-mental if collective identity is to be trans-formed into coordinated collective action.

We do not believe that the importance ofmusic is limited to textile workers or to oneparticular era of worker unrest. In fact, wefound during our data collection a signifi-cant amount of material highlighting the im-portance of music for a number of historicalstruggles pertaining to class, race, and gen-der, and across a number of industries andgeographic locations. Many movementshave had a well-developed repertoire ofsongs, used before and during collective pro-test. Women’s workplace issues, for in-stance, have received attention in songs per-taining to the coal mining industry, amongothers, throughout this century. Music isclearly an important part of the AfricanAmerican experience, from slave gospels, toblues lyrics, to jazz, to contemporary soci-etal critiques embedded in rap music. Songalso has been central to other working classmovements—including striking longshore-man, lumberers, steel workers, and automo-bile workers—and for as long as we can tell(e.g., see Greenway 1953; Lomax, Guthrieand Seeger 1967; Pratt 1990). What is lack-ing, despite archival collections and somehistorical accounts, are systematic analysesand substantive sociological interpretationsof these lyrics—if and when they are impor-tant and how they are tied to stratificationprocesses and efforts to remedy inequality.13

Political opportunity theory has specifiedthe importance of leverage and its histori-cally contingent nature. However, an expli-cation of the mechanism(s) through whichpotential movement participants’ percep-tions of opportunity may be altered is oftenlacking. Media, whether in an earlier histori-cal era or the contemporary day, are impor-tant in this process (Brown 1998; Calhoun1998; Kahan 1999). It is also the case that

13 In general, music allows oppositional cultureto exist, persist, and possibly spread, not onlyduring pivotal moments when the political struc-ture is ideal but also during relatively calm peri-ods with little active or visible protest. Such op-positional culture can foster discontent and theseeds of protest, and may become more explicitand goal-directed at key historical moments(Denisoff 1972). It is also important to note thatmusic can have a conservative influence by re-signing the listener to his or her plight instead ofencouraging action aimed at changing one’ s situ-ation (Peterson 1992).


structural political opportunity must be dis-entangled from perceptions of opportunity,particularly when the analytic focus is onmovement success and the forces that per-suade or dissuade social movement partici-pation (Kurzman 1996; McAdam 1982; Tar-row 1988). Where there is a disjuncture be-tween the two, insurgency may result, butwill probably be curtailed by countermobil-ization. Southern mill workers’ perceptionsof political opportunity were altered via ra-dio, creating a belief that F.D.R. was on theirside, that they had (federal) justification fortheir actions, and that the federal govern-ment would intervene when the pivotal mo-ment came. Little intervention, however, ac-tually occurred.

Our focus on media technology helpsbridge the divide between collective identityand political opportunity perspectives by ad-dressing the question of how processes rel-evant to social movement formation aremanifested across space. However, the studyof media and social movement dynamicsmust be supplemented further by theory thatexplicitly incorporates aspects of both iden-tity and opportunity into a single framework(Meyer and Staggenborg 1996; Oliver andJohnston 1999). One of the most promisingcontemporary lines of work that undertakesthis task is that dealing with social move-ment culture.

Social movement culture, rather than anambiguous construct, is an influential andclearly defined component of the socialmovement dynamic comprising normativeguidelines and practices that create and re-inforce (1) a sense of group identity, (2) analternative interpretational frame of causeand effect, and (3) a sense of political effi-cacy (W. Gamson 1995; Taylor and Whittier1995). Such a conceptual frame is well-suited for analyzing the influential nature ofmusic in the social movement repertoire, andwe suspect that extending the focus to otherforms of creative, linguistic, and/or perfor-mance expression conducive to conscious-ness-raising, group-building, and solidaritymaintenance, would be worthwhile.14 In our

view, the melding of these two foci—socialmovement culture and the media—offers re-searchers the most useful set of theoreticaltools for understanding the complex and dy-namic character of historical, contemporary,and future movement formation across space(J. Gamson 1998; W. Gamson 1995; Gamsonand Wolfsfeld 1993; Taylor 1996).

Our analyses also contribute to the grow-ing body of research on media, communica-tion technology, and community (Calhoun1998; Cerulo and Ruane 1998; Purcell1997). While much of this work focuses onthe contemporary era, specifically on televi-sion and the Internet, many similar themesemerge. Do media and new communicationtechnologies enhance social integration? Dothey produce collective identity? Does suchcollective identity facilitate group action?What tends to be overlooked in this litera-ture, however, is the historical context inwhich new information media unfold and theconsequences for social groups. “We need toset our discussions of electronic media in abit deeper historical context—not just oftechnology but of the spatial organization ofpower and movements challenging thatpower” (Calhoun 1998:375).

While our analyses highlight the power ofradio to transform consciousness and to in-stigate challenges to existing structures ofinequality, we also acknowledge radio’slimitations. Our analyses reveal lessons thatmay be applicable to understanding the po-tential influence of television or the Interneton collective action. These newer technolo-gies enhance collective experience throughthe maintenance of dispersed networks, theencouragement of “ socio-spatial” enclaves,and the facilitation of group activities (Cal-houn 1998; Cerulo and Ruane 1998). Likeradio, however, the influence of newer me-dia and communication technologies onthese social processes may vary, dependingon the level of political autonomy and thedegree to which the information transmittedappeals to the unique experiences of indi-viduals and specific social groups (W. Gam-son et al. 1992).

In the case of radio, the autonomy of localstations was curbed by the FCC in the mid-14 Like Pattillo-McCoy (1998), who analyzes

church culture and action in the black commu-nity, we find Swidler’ s (1986) discussion of “cul-ture as a tool kit” from which actors can draw to

be a useful way of conceiving of the cultural rep-ertoire available to social movement participants.


to late 1930s when small owners and univer-sities, advocating radio in the “public inter-est” and greater flexibility in what was aired,lost out to “ commercial interests.” Morestringent guidelines for broadcasts that couldbe interpreted as “political” or “propaganda”were put in place, and radio operators whoviolated the new policies risked losing theiroperating licenses (Federal Radio Commis-sion 1929; McChesney 1993). The successof the recording industry and its links to thebig corporate broadcast networks also na-tionalized music played over the airwaves,leaving little room or market for music deal-ing with worker grievances and the concernsof local populations (Cantril and Allport[1935] 1971; Malone 1979). Thus, while thedecline in local radio station autonomy andtransmission specificity was in part a func-tion of institutional bureaucratic tendencies,it was also a political process—one whosetrajectory leaned toward, although was notcompletely determined by, corporate-politi-cal hegemony.

These issues of limited autonomy and anoverly-generalized appeal apply moststraightforwardly to television, suggesting alimited impact on collective experience andcollective action.15 However, the internet isa medium of multidirectional informationflow—and it exists in a multinational con-text. Thus, it affords its users freedom frompolitical controls and specificity of group in-terest, at least at the present time, and it willtherefore probably have strong conse-quences for group-specific identities and thecoordination of collective action. Like tex-tile workers and radio in the 1920s and1930s, however, such relations are tenuousat best, as they require “connection mainte-nance” (Cerulo and Ruane 1998)—some-thing that is difficult to nurture and preserveover time and on a wide geographic andsocio-political scale. Furthermore, like ra-dio, the potential impact of the Internet mayvery well be curbed by regulations and po-litical oversight, particularly if its use runscounter to dominant ideological positionsand stratification structures.

15 The impact of media is often conservativeand supportive of the status quo. For elaborationof the conservative versus critical potential ofmedia, see W. Gamson (1992b, 1995).

Vincent J. Roscigno is Assistant Professor ofSociology at the Ohio State University. In addi-tion to his research in collective behavior and la-bor insurgency, he has written on contemporarylabor organization and its historical context,race and class stratification in the U.S. South,labor market opportunity and adolescent well-being, and educational inequality and its repro-duction. Some of this work has been publishedrecently in the American Sociological Review,Social Problems, Social Forces, and Sociology ofEducation. With William Danaher, he is extend-ing his research on southern mill workers by de-lineating the impact of speeches by FranklinDelano Roosevelt on the consciousness of millworkers, and is also undertaking case studies ofseveral mill towns to further specify the pro-cesses outlined in this article.

William F. Danaher is Assistant Professor of So-ciology at the College of Charleston. He has pub-lished recently on group identity, stratification,and music. Aside from these interests, he has alsopublished work on poverty and migration. He iscurrently working with Vincent Roscigno on abook manuscript that focuses on group identity,music, and networks in the forging of mobiliza-tion efforts among southern textile workers. Heis also investigating race and gender inequalitydiscourse in relation to blues and rap music.

REFERENCES

Barkan, Steven E. 1984. “Legal Control of theSouthern Civil Rights Movement.” AmericanSociological Review 49:552–65.

Billings, Dwight B. 1990. “Religion as Opposi-tion: A Gramscian Analysis.” American Jour-nal of Sociology 96:1–31.

Bittner, John. R. 1982. Broadcast Law and Regu-lation. Englewood Cliffs, NJ: Prentice-Hall.

Blau, Peter. 1977. Inequality and Heterogeneity:A Primitive Theory of Social Structure. NewYork: Free Press.

Brown, Cliff and Terry Boswell. 1995. “Strike-breaking Solidarity: The Great Steel Strike of1919.” American Journal of Sociology 100:1479–1519.

Brown, Robert J. 1998. Manipulating the Ether:The Power of Broadcast Radio in ThirtiesAmerica. Jefferson, NC: McFarland and Com-pany.

Brueggeman, John and Terry Boswell. 1998.“ Realizing Solidarity: Sources of InterracialUnionism During the Great Depression.” Workand Occupations 25:436–82.

Calhoun, Craig. 1998. “Community without Pro-pinquity Revisited: Communications Technol-ogy and the Transformation of the Urban Pub-lic Sphere.” Sociological Inquiry 68:373–97.


Cantril Hadley and Gordon W. Allport. [1935]1971. The Psychology of Radio. New York:Arno.

Cerulo, Karen A. and Janet M. Ruane. 1998.“Coming Together: New Taxonomies for theAnalysis of Social Relations.” Sociological In-quiry 68:398–425.

Clark’s Directory of Southern Textile Mills.1929. Charlotte, NC: Clark Publishing Com-pany.

Cornfield, Dan and Mark V. Leners. 1989.“ Unionization in the Rural South: RegionalPatterns of Industrialization and the Process ofUnion Organizing.” Research in Rural Sociol-ogy and Development 4:137–52.

Dahrendorf, Ralf. 1959. Class and Class Conflictin Industrial Society. Stanford, CA: StanfordUniversity Press.

Davidson’s Textile Blue Book, July 1934–July1935. 1935. New York: Davidson.

Della Fave, L. Richard. 1980. “The Meek ShallNot Inherit the Earth.” American SociologicalReview 45:955–71.

———. 1986. “ Toward and Explication of theLegitimation Process.” Social Forces 65:476–500.

Denisoff, R. Serge. 1971. Great Day Coming:Folk Music and the American Left. Urbana, IL:University of Illinois Press.

———. 1972. Sing a Song of Social Significance.Bowling Green, KY: Bowling Green Univer-sity Popular Press.

Dewey, Woody. 1999. Interview by William F.Danaher. Tape Recording. Charleston, SC, Oc-tober 16.

Dixon, Dorsey. 1998. Babies in the Mill. Soundrecording. Oakland, CA: HighTone Records/HMG.

Epstein, Barbara. 1990. “ Rethinking SocialMovement Theory.” Socialist Review 20:35–66.

Eyerman, Ron and Andrew Jamison. 1998. Mu-sic and Social Movements: Mobilizing Tradi-tions in the Twentieth Century. Cambridge,England: Cambridge University Press.

Fantasia, Rick. 1988. Cultures of Solidarity:Consciousness, Action, and ContemporaryAmerican Workers. Berkeley, CA: Universityof California Press.

Federal Radio Commission. 1929. Third AnnualReport of the Federal Radio Commission to theCongress of the United States Covering thePeriod from October 1, 1928, to November 1,1929. Washington, DC: U.S. GovernmentPrinting Office.

Flacks, Richard. 1999. “ Culture and SocialMovements: Exploring the Power of Song.”Paper presented at the annual meeting of theAmerican Sociological Association, August,Chicago, IL.

Gamson, Joshua. 1995. “ Must Identity Move-ments Self-Destruct—A Queer Dilemma.” So-cial Problems 42:390–407.

———. 1998. Freaks Talk Back: Tabloid TalkShows and Sexual Nonconformity. Chicago,IL: University of Chicago Press.

Gamson, William A. 1992a. “ The Social Psy-chology of Collective Action.” Pp. 53–76 inFrontiers in Social Movement Theory, editedby A. Morris and C. McClurg. New Haven,CT: Yale University Press.

———. 1992b. Talking Politics. New York:Cambridge University Press.

———. 1995. “Constructing Social Protest.” Pp.85–106 in Social Movements and Culture, ed-ited by H. Johnston and B. Klandermans. Min-neapolis, MN: University of Minnesota Press.

Gamson, William A., David Croteau, W. Hoynes,and T. Sasson. 1992. “Media Images and theSocial Construction of Reality.” Annual Re-view of Sociology 18:373–93.

Gamson, William A. and David S. Meyer. 1992.“The Framing of Political Opportunity.” Paperpresented at the annual meeting of the Ameri-can Sociological Association, August, Pitts-burgh, PA.

Gamson, William A. and G. Wolfsfeld. 1993.“ Movements and Media as Interacting Sys-tems.” Annals of the American Academy of Po-litical and Social Science 528:114–25.

Garofalo, Reebee. 1997. Rockin’ Out: PopularMusic in the USA. Boston, MA: Allyn and Ba-con.

Gellhorn, Martha. 1933. Letter to Harry Hopkins,Director of the Federal Emergency Relief Ad-ministration, November 11, 1934. RetrievedDecember 20, 1998 (http://newdeal.feri.org/hopkins/hop08.htm).

Giddens, Anthony. 1982. “ Class Structurationand Class Consciousness.” Pp. 157–74 inClasses, Power, and Conflict: Classical andContemporary Debates, edited by A. Giddensand D. Held. Berkeley, CA: University of Cali-fornia Press.

Goffman, Erving. 1981. Forms of Talk. Philadel-phia: University of Pennsylvania Press.

Green, Archie. 1963a. Correspondence, Inter-view, and Biographical Material on DorseyDixon. Archie Green Papers, Files 795–825,Southern Folklife Collection, Library of theUniversity of North Carolina, Chapel Hill, NC.

———. 1963b. Textile Mill Songs (Subseries4.23). File 1277 of the Archie Green Papers,Southern Folklife Collection, Library of theUniversity of North Carolina, Chapel Hill, NC.

Greenway, John. 1953. American Folksongs ofProtest. Philadelphia, PA: University of Penn-sylvania Press.

Griffin, Larry, Michael Wallace, and Beth Rubin.1986. “Capitalist Resistance to the Organiza-


tion of Labor Before the New Deal: Why?How? Success?” American Sociological Re-view 51:147–67.

Griffith, Barbara S. 1988. The Crisis of Ameri-can Labor: Operation Dixie and the Defeat ofthe CIO. Philadelphia, PA: Temple UniversityPress.

Gurlach, Luther P. 1999. “The Structure of So-cial Movements: Environmental Activism andIts Opponents.” Pp 85–97 in Waves of Protest:Social Movements Since the Sixties, edited byJ. Freeman and V. Johnson. New York: Row-man and Littlefield.

Hall, Jacquelyn Dowd, James Leloudis, RobertKorstad, Mary Murphy, Lu Ann Jones, andChristopher B. Daly. 1987. Like a Family: TheMaking of a Southern Cotton Mill World.Chapel Hill, NC: University of North CarolinaPress.

Hampton, Francis. 1935. “New Leisure: How IsIt Spent? A Study of What One HundredTwenty-Two Textile Workers of Leaksville,Spray, and Draper are Doing with the NewLeisure Created by the N.R.A., as Applied toCertain Types of Activities.” Masters thesis,Department of Sociology, University of NorthCarolina, Chapel Hill, NC.

Henry Morgenthau Jr. Collection. 1934. Speechon Radio: Reaction Letters A thru H. Box 259.Franklin D. Roosevelt Library Archives,Franklin D. Roosevelt Library, Hyde Park,NY.

Hille, Waldemar. 1948. The People’s Song Book.New York: Boni and Gaer.

Hodges, James A. 1986. New Deal Labor Policyand the Southern Cotton Textile Industry,1933–1941. Knoxville, TN: University of Ten-nessee Press.

Hodson, Randy. 1999. Analyzing DocumentaryAccounts. Newbury Park, CA: Sage.

Hodson, Randy, Deborah Ziegler, and BarbaraBump. 1987. “Who Crosses the Picket Line?An Analysis of the CWA Strike of 1983.” La-bor Studies Journal 12:19–37.

Hogan, John W. L. [1930] 1971. “ How RadioWorks.” Pp. 267–74 in Radio and Its Future,edited by M. Codel. New York: Arno.

Huber, Patrick. 1998. “Battle Songs of the South-ern Class Struggle: Songs of the Gastonia Tex-tile Strike of 1929.” Southern Cultures 4:109–125.

James, David R. 1988. “The Transformation ofthe Southern Racial State: Class and Race De-terminants of Local-State Structures.” Ameri-can Sociological Review 53:191–208.

Jenkins, J. Craig. 1985. The Politics of Insur-gency. New York: Columbia University Press.

Jenkins, J. Craig and Charles Perrow. 1977. “ In-surgency of the Powerless: Farm WorkerMovements (1946–1972).” American Socio-

logical Review 42:249–68.Kahan, Michael. 1999. Media as Politics:

Theory, Behavior and Change in America. Up-per Saddle River, NJ: Prentice-Hall.

Kimeldorf, Howard. 1999. Battling for AmericanLabor: Wobblies, Craft Workers, and the Mak-ing of the Union Movement. Berkeley, CA:University of California Press.

Kurzman, Charles. 1996. “Structural Opportunityand Perceived Opportunity in Social Move-ment Theory: The Iranian Revolution of1979.” American Sociological Review 61:153–70.

Lachmann, Richard. 1990. “ Class Formationwithout Class Struggle: An Elite ConflictTheory of the Transition to Capitalism.” Amer-ican Sociological Review 55:398–414.

Leiter, Jeffrey, Michael Schulman, and RhondaZingraff. 1991. Hanging by a Thread: SocialChange in Southern Textiles. New York: ILRPress.

Lichterman, Paul. 1999. “Talking Identity in thePublic Sphere: Broad Visions and SmallSpaces in Sexual Identity Politics.” Theory andSociety 28:101–41.

Lieberman, Robbie. 1989. My Song Is My Wea-pon: People’s Songs, American Communism,and the Politics of Culture, 1930–1950. Ur-bana, IL: University of Illinois Press.

Lomax, Alan. 1960. The Folk Songs of NorthAmerica. New York: Doubleday.

Lomax, Alan, Woody Guthrie, and Pete Seeger.1967. Hard Hitting Songs for Hard-Hit Peo-ple. New York: Oak Publications.

Malone, Bill C. 1968. Country Music USA: AFifty Year History. Austin, TX: University ofTexas Press.

———. 1979. Southern Music, American MusicLexington, KY: University of Kentucky Press.

Mann, Michael. 1973. Consciousness and Actionamong the Western Working Class. London,England: Macmillan.

Mathews, Jane D. 1982. “The New Feminism andthe Dynamics of Social Change.” Pp. 397–425in Women’s America: Refocusing the Past, ed-ited by L. K. Kerber and J. De Hart Mathews.New York: Oxford University Press.

McAdam, Doug. 1982. Political Process and theDevelopment of Black Insurgency, 1930–1970.Chicago, IL: University of Chicago Press.

———. 1983. “Tactical Innovation and the Paceof Insurgency.” American Sociological Review48:735–54.

McChesney, Robert W. 1993. Telecommunica-tions, Mass Media, and Democracy: The Battlefor the Control of Broadcasting, 1928–1935.New York: Oxford University Press.

McElvaine, Robert S. 1983. Down and Out in theGreat Depression: Letters from the “Forgot-ten Man.” Chapel Hill, NC: University of


North Carolina Press.McLaurin, Meltin A. 1971. Paternalism and Pro-

test: Southern Cotton Mill Workers and Orga-nized Labor, 1875–1905. Westport, CT:Greenwood.

McLaurin, Meltin A. and Richard A. Peterson.1992. You Wrote My Life: Lyrical Themes inCountry Music. Philadelphia, PA: Gordon andBreach.

Melucci, Alberto. 1985. “ The Symbolic Chal-lenge of Contemporary Movements.” SocialResearch 52:781–816.

Meyer, David S. and Nancy E Whittier. 1994.“ Social Movement Spillover.” Social Prob-lems 41:277–98.

Meyer, David S. and Suzanne Staggenborg.1996. “ Movements, Countermovements, andthe Structure of Political Opportunity.” Ameri-can Journal of Sociology 101:1628–60.

Mills, C. Wright. 1939. “Language, Logic, andCulture.” American Sociological Review 4:670–80.

———. 1940. “Situated Actions and Vocabular-ies of Motive.” American Sociological Review5:904–13.

Minchin, Timothy J. 1997. What Do We Need aUnion For? The TWUA in the South, 1945–1955. Chapel Hill, NC: University of NorthCarolina Press.

Montgomery, David. 1987. The Fall of the Houseof Labor. New York: Cambridge UniversityPress.

Morris, Aldon D. 1984. Origins of the CivilRights Movement: Black Communities Orga-nizing for Change. New York: Free Press.

Myers, Daniel J. 1998. “Collective Violence andHeterogenous Diffusion Processes: U.S. RacialRioting from 1964 to 1971.” Working PaperSeries, vol. 1, no. 1. American SociologicalAssociation Section on Collective Behaviorand Social Movements. Retrieved November1, 1999 (http://www.nd.edu/~dmyers/cbsm/vol1/myers1.pdf).

———. 2000. “The Diffusion of Collective Vio-lence: Infectiousness, Susceptibility, and MassMedia Networks.” American Journal of Soci-ology 106: 173–208.

Nagel, Joane. 1994. “ Constructing Ethnicity:Creating and Recreating Ethnic Identity andCulture.” Social Problems 41:152–76.

Nolan, Dennis R. and Donald E. Jones. 1976.“ Textile Unionism in the Piedmont, 1901–1932.” Pp. 48–69 in Essays in Southern LaborHistory: Selected Papers, Southern Labor His-tory Conference, 1976, edited by G. M. Finkand M. E. Reed. Westport, CT: Greenwood.

Oberschall, Anthony. 1989. “ The 1960 Sit-Ins:Protest Diffusion and Movement Take-Off.”Research in Social Movements, Conflict, andChange 11:31–53.

Oliver, Pamela E. 1989. “ Bringing the CrowdBack In: The Nonorganizational Elements ofSocial Movements.” Research in Social Move-ments, Conflict and Change 11:1–30.

Oliver, Pamela E. and Hank Johnston. 1999.“What a Good Idea! Frames and Ideologies inSocial Movement Research.” Paper presentedat the Annual Meetings of the American So-ciological Association, Chicago, August.

Olzak, Susan. 1992. The Dynamics of EthnicCompetition and Conflict. Stanford, CA: Stan-ford University Press.

Parkin, Frank. 1979. The Marxist Theory ofClass: A Bourgeois Critique. London, En-gland: Tavistock.

Pattillo-McCoy, Mary. 1998. “Church Culture asa Strategy of Action in the Black Community.”American Sociological Review 63:767–84.

Peterson, Richard A. 1992. “Class Unconscious-ness” Pp. 35–62 in You Wrote My Life: Lyri-cal Themes in Country Music, edited by M. A.McLaurin and R. A. Peterson. Philadelphia,PA: Gordon and Breach.

Pichardo, Nelson A. 1995. “The Power Elite andElite-Driven Countermovements: The Associ-ated Farmers of California During the 1930s.”Sociological Forum 10:21–49.

Pope, Liston. 1942. Millhands and Preachers: AStudy of Gastonia. New Haven, CT: Yale Uni-versity Press.

Pratt, Ray. 1990. Rhythm and Resistance. NewYork: Praeger.

Purcell, Kristen. 1997. “Towards a Communica-tion Dialectic: Embedded Technology and theEnhancement of Place.” Sociological Inquiry67:101–12.

Rogers, Everett M. 1995. Diffusion of Innova-tions. New York: Free Press.

Rorrer, Kinney. 1982. Rambling Blues: The Lifeand Songs of Charlie Poole. London, England:Old Time Music.

Roscigno, Vincent J. and M. Keith Kimble. 1995.“Elite Power, Race, and the Persistence of LowUnionization in the South.” Work and Occu-pations 22:271–300.

Salmond, John A. 1995. Gastonia 1929: TheStory of the Loray Mill Strike. Chapel Hill,NC: University of North Carolina Press.

Seeger, Pete. 1992. American Industrial Ballads.Sound recording. Washington, DC: Smith-sonian Folkways Recordings.

Sherrill, Homer “ Pappy.” 1999. Interview byWilliam F. Danaher. Tape Recording. Charles-ton, SC, December 5.

Simon, Bryant. 1998. The Fabric of Defeat: ThePolitics of South Carolina Millhands, 1910–1948. Chapel Hill, NC: University of NorthCarolina Press.

Snow, David A. and Robert D. Benford. 1992.“ Master Frames and Cycles of Protest.” Pp


133–155 in Aldon Morris and Carol Mueller,eds., Frontiers of Social Movement Theory.New Haven, CT: Yale.

Snow, David A., E. Burke Rochford, Steven K.Worden, and Robert D. Benford. 1986. “FrameAlignment Process, Micromobilization, andMovement Participation.” American Sociologi-cal Review 51:464–81.

Soule, Sarah A. 1997. “The Student DivestmentMovement in the United States and TacticalDiffusion: The Shantytown Protest.” SocialForces 75:855–83

Soule, Sarah A. and Yvonne Zylan. 1997. “Run-away Train? The Diffusion of State-Level Re-form in ADC/AFDC Eligibility Requirements,1950–1967.” American Journal of Sociology103:733–62.

Spilerman, Seymour. 1970. “The Causes of Ra-cial Disturbances: A Comparison of Alterna-tive Explanations.” American Sociological Re-view 3:627–49.

Stotik, Jeffrey, Thoman E. Shriver, and SherryCable. 1994. “ Social Control and MovementOutcome: The Case of AIM.” Sociological Fo-cus 27:53–66.

Strang, David and Meyer, John W. 1993. “ Insti-tutional Conditions for Diffusion.” Theory andSociety 22:487–511.

Summers, Harrison B. 1958. A Thirty-Year His-tory of Programs Carried on National RadioNetworks in the United States, 1926–1956.New York: Arno Press.

Sussman, Leila A. 1963. Dear FDR: A Study ofPolitical Letter-Writing. Totowa, NJ: Bedminster.

Swidler, Ann. 1986. “Culture in Action: Symbolsand Strategies.” American Sociological Review51:273–86.

Tarrow, Sidney. 1988. “National Politics and Col-lective Action: Recent Theory and Research inWestern Europe and the United States.” AnnualReview of Sociology 14:421–40.

Taylor, Verta. 1996. Rock-A-By Baby: Feminism,Self Help, and Postpartum Depression. NewYork: Routledge.

Taylor, Verta and Nancy E. Whittier. 1992. “Col-lective Identity in Social Movement Commu-nities: Lesbian Feminist Mobilization.” Pp.104–29 in Frontiers in Social Movement The-ory, edited by A. Morris and C. McClurg. NewHaven, CT: Yale University Press.

———. 1995. “Analytical Approaches to SocialMovement Culture: The Culture of the Wo-men’ s Movement.” Pp.163–86 in Social Move-ments and Culture, edited by H. Johnston andB. Klandermans. Minneapolis, MN: Universityof Minnesota Press.

Tilly, Charles. 1976. From Mobilization to Revo-lution. Englewood Ciffs, NJ: Prentice-Hall.

Tippett, Thomas. 1931. When Southern LaborStirs. Vol. 4, The Strike at Marion. New York:Jonathan Cape and Harrison Smith.

Tomaskovic-Devey, Donald and Vincent J. Ros-cigno. 1996. “Racial Economic Subordinationand White Gain in the U.S. South.” AmericanSociological Review 61:565–89.

Tullos, Allen. 1989. Habits of Industry: WhiteCulture and the Transformation of the Caro-lina Piedmont. Chapel Hill, NC: University ofNorth Carolina Press.

Vallas, Stephen P. 1987. “The Labor Process asa Source of Class Consciousness: A CriticalExamination.” Sociological Forum 2:237–56.

Wiggins, Gene. 1987. Fiddlin’ Georgia Crazy:Fiddlin’ John Carson, His Real World, and theWorld of His Songs. Chicago, IL: Universityof Illinois Press.

Wood, Philip. 1986. Southern Capitalism: ThePolitical Economy of North Carolina, 1880–1980. Durham, NC: Duke University Press.

Yellen, Samuel. 1936. “ The Southern TextileStrikes and Gastonia.” Pp. 292–326 in Ameri-can Labor Struggles. New York: S.A. Russell.

HOW MOVEMENTS WINHOW MOVEMENTS WINHOW MOVEMENTS WINHOW MOVEMENTS WINHOW MOVEMENTS WIN 4 94 94 94 94 9


How Movements Win:

Gendered Opportunity Structures

and U.S. Women’s Suffrage

Movements, 1866 to 1919

State women’s suffrage movements are investigated to illuminate the circumstancesin which social movements bring about political change. In 29 states, suffragistswere able to win significant voting rights prior to passage of the Nineteenth Amend-ment. In addition to resource mobilization, cultural framing, and political opportu-nity structures, the authors theorize that gendered opportunities also fostered thesuccesses of the movements. An event history analysis provides evidence that gen-dered opportunity structures helped to bring about the political successes of thesuffragists. Results suggest the need for a broader understanding of opportunitystructure than one rooted simply in formal political opportunities.

(Larson 1965), but in New York, whichpassed full suffrage in 1917, by the 1910s, aquarter of a million suffragists participatedin various state suffrage organizations(McDonald 1987:150).

We examine the circumstances that ledstates to adopt full, primary, or presidentialsuffrage prior to ratification of the Nine-teenth Amendment. To do this, we investi-gate the U.S. state suffrage movements andthe contexts in which they mobilized. Withfew exceptions (Gamson 1975), sociologistsonly recently have begun to consider whysome movements succeed and others fail intheir attempts to bring about social change(Guigni 1998). As McAdam, McCarthy, andZald’s (1988) extensive review of the move-ments literature reveals, most movementstudies have focused on movement emer-gence and development. But recently, a fewpolicy studies have shed light on movementpolitical outcomes (Amenta and Poulsen1996; McCammon 1995; Quadagno 1992).Their theoretical focus, though, is on the in-terests of state actors and policy change, andthus these studies typically offer only pass-

n 1869 Wyoming became the first state(then a territory) to grant women suffrage.

By 1920, when the Nineteenth Amendmentwas ratified, 15 states had granted womenfull suffrage, 2 southern states had givenwomen the right to vote in primary elections,and 13 states had awarded women the rightto vote for president. During these years,suffragists in all states mobilized in thestruggle for voting rights. In Wyoming ahandful of women were active in 1869

I

Direct all correspondence to Holly McCammon,Department of Sociology, Vanderbilt University,Nashville, TN 37235 ([email protected]). We thank Paul Burstein, GlennDeane, Larry Griffin, Patricia McManus, KentRedding, and Pam Walters along with membersof the Political Economy Workshop at IndianaUniversity for comments on an early draft. Wealso thank Bill Fletcher for help with data collec-tion and John McCammon for computer pro-gramming. This research was funded by the Na-tional Science Foundation (SBR-9631520), theUniversity Research Council at Vanderbilt Uni-versity, and the Carrie Chapman Catt Center forWomen and Politics at Iowa State University.

Holly J. McCammon

Vanderbilt UniversityKaren E. Campbell

Vanderbilt University

Ellen M. Granberg

Vanderbilt UniversityChristine Mowery

Vanderbilt University


ing attention to the role of social move-ments, noting simply the presence or ab-sence of movement activity and rarely not-ing variations in the organizational strength,strategies, and ideologies of movements. Inaddition, some movement researchers haveturned their attention to the systematic studyof movements and their outcomes, but thetheoretical focus of this work has also beensomewhat narrow (Amenta, Dunleavy, andBernstein 1994; Banaszak 1996; Giugni,McAdam, and Tilly 1998). Movement re-searchers who examine the success of move-ments either focus primarily on the strate-gies of the movements themselves (Morris1993) or, in examining the context of mobi-lization, limit their consideration to politicalinstitutions and actors, that is, to the politi-cal opportunity structure (Amenta, Car-ruthers, and Zylan 1992; Gelb 1987).

We argue that a model of movement suc-cess must consider not only the mobilizationof the movements but the broad context inwhich those movements operate, includingpolitical and other social dynamics that canaffect movement success. We elaborate onthis below, but first we point out that amodel of the political success of movementsmust theorize the impact of movements andtheir contexts on political decision-makers.Bringing about political or policy change—in the case of the suffragists, the expansionof voting rights—requires a willingness onthe part of political decision-makers to sup-port such change. A model of movement suc-cess, then, must specify the circumstancesfostering such willingness on the part of po-litical actors.

Sociological theories of the state and ofpolicymaking have long recognized the needto theorize the interests and actions of stateactors to understand the policymaking pro-cess (Alford and Friedland 1985). Yet, thefew recent studies examining movementsuccess fail to acknowledge this point ex-plicitly and thus fail to draw on the diversetheories of the state to understand the fullrange of factors that can influence politicaldecision-makers.

As movement researchers shift their focusaway from the determinants of movementmobilization to the determinants of move-ment outcomes, they bring with them theconcept of “political opportunity structure.”

Defining the political process model,McAdam (1982) states that “ any event orbroad social process that serves to under-mine the calculations and assumptions onwhich the political establishment is struc-tured occasions a shift in political opportu-nities” (p. 41, emphasis in original). But re-cently, political opportunity structure is typi-cally considered to encompass only dynam-ics involving electoral politics and the state.For instance, Jenkins and Klandermans(1995) write that “ social movements developin a context defined by the state and the rep-resentation system, which afford opportuni-ties for mobilization and set limits on the ef-fectiveness of movement strategies” (p. 7,emphasis added; also see Skocpol 1992:41).McAdam (1996:27), surveying the recent lit-erature, lists various “dimensions of politi-cal opportunity,” all of which are character-istics of the formal political structure, its ca-pacities, and the configuration of politicalelites. Others as well limit their view of theopportunity structure to the formal politicallandscape of party politics and the structureand action of the state (e.g., Amenta et al.1992; Amenta et al. 1994; Banaszak 1996;Brockett 1991; Costain 1992; Kriesi 1995;Tarrow 1998).

But this narrow definition of opportunitystructure represents only one view offeredby state theorists of the factors that can in-fluence policymaking. The focus on politi-cal structures and dynamics, typically ex-cluding other contextual influences, repre-sents a polity- or state-centered theory ofpolicymaking (Evans, Rueschemeyer, andSkocpol 1985; Skocpol 1980). This viewholds that the interests of political decision-makers, particularly those of state actors, arebased on preserving or expanding their in-stitutional authority, and their lawmakingdecisions reflect this orientation (Alford andFriedland 1985; Rueschemeyer and Evans1985). A political opportunity for movementsuccess arises, then, when political circum-stances shift such that political actors arewilling to support policy change becausethey perceive the change will strengthen orpreserve their own institutional positions.Empirical research in this area clearly dem-onstrates that formal political dynamics likethese influence political decision-makingand thus can influence the political success


of social movements (Amenta and Poulsen1996). But the underlying assumption thatpolitical decision-making and movementsuccess are unaffected by circumstances be-yond these formal political dynamics is toonarrow.

Society-centered approaches to politicaldecision-making, such as feminist (Connell1990; MacKinnon 1989), class (Block 1977;Domhoff 1998), and racial (Quadagno 1994)theories of the state, offer a starting point forbroader theories of the types of opportunitystructures that may influence movement out-comes. Society-centered theories assumethat gender, economic class, and race rela-tions can and do influence policymaking. Weuse polity- and society-centered theories toargue that not only did opportunities forgranting suffrage to women emerge fromchanging dynamics in formal politics that al-tered the political interests of political deci-sion-makers, but opportunities emerged ingendered ways as well.

Specifically, we posit that shifting genderrelations produced a gendered opportunityfor women’s suffrage by altering attitudesamong political decision-makers about theappropriate roles of women in society. Thatis, changing gender relations altered expec-tations about women’s participation in thepolity, and these changes in gendered expec-tations increased the willingness of politicaldecision-makers to support suffrage.

Thus, we argue that, on the one hand,shifts in political circumstances altered thepolitical calculus on which decision-makersbased their actions, providing a political op-portunity for suffrage. On the other hand,changing gender relations also caused politi-cal decision-makers to alter their viewsabout the proper roles for women in society,and these changing attitudes about gender—not changing attitudes about the political vi-ability of a particular stance on suffrage—provided a gendered opportunity for suffragesuccess. Thus, political dynamics andchanging gender relations both influencedwhether political actors voted for suffrage,but through different mechanisms: onethrough changing political interests; theother through changing attitudes aboutwomen’s roles in society.

Gender theorists point out that gender canpermeate all organizations and processes in

society (Acker 1992; Lorber 1994). Whilewe find in the history of suffrage that theopportunity structures helping the suffragistswin voting rights typically took one of twoforms—political or gendered—we also findthat in some cases political opportunities in-volved gendered considerations as well. Thatis, formal political interests were intertwinedwith gendered expectations, often expecta-tions about how women as voters would casttheir ballots. Thus, consideration by socialmovement researchers of only political op-portunity structures is simply too narrow anapproach to understanding the full range ofopportunity structures that foster movementsuccess. Others have pointed to a need to re-fine our understanding of movement oppor-tunity structures by recognizing that factorsbeyond formal political dynamics can influ-ence movements and their success (Gamsonand Meyer 1996; Goodwin and Jasper 1999:53; Koopmans 1999; Rucht 1996; Taylor1999). The opportunity structures confrontedby the suffragists offer a chance to view thisbroader range of opportunity structures.

Our work advances theorizing of socialmovement success, but we also anticipatethat it will add to the historical literature onsuffrage success. The work of historiansstudying women’s suffrage often parallelsthat of sociologists studying social move-ments. Historians explaining suffrage haveemphasized the political circumstances nec-essary for winning the vote (Buenker 1971;Edwards 1997) or the movement strategiesneeded for success (Degler 1980; Flexner1975). Some have pointed to changing gen-der relations as a contributing factor to suf-frage victories (Kraditor 1965; Morgan1972). Other historians, though, contend thatthe states granting suffrage had little in com-mon and that these states gave women vot-ing rights for idiosyncratic reasons (Beeton1986; Larson 1971a).1

We have collected data from primary andsecondary sources on all state suffragemovements and their contexts. These data

1 Given that western states (and territories)were the first to pass full suffrage, much of thehistorical literature on suffrage success concernssuffrage in the U.S. West. For a fuller treatmentof this literature see McCammon and Campbell(2001).


represent the only attempt thus far to sys-tematize and analyze the historical knowl-edge on the state suffrage movements, andour analysis allows us to compare states thatdid and did not grant suffrage. Contrary tothe argument that states granting suffragehad little in common, we identify a numberof characteristics common to the suffragestates. Our comparative analysis also revealsthat no single cause can explain women’ssuffrage success. Women did not win thevote simply because of particular politicalcircumstances, or solely because of the strat-egies used by the suffragists, or simply be-cause of changing gender relations. A com-bination of factors was required to broadendemocracy to women.

A DEFINITION OF

MOVEMENT SUCCESS

We define movement success in terms of“political or policy outcomes” (Staggenborg1995:341), a definition that falls withinGamson’s (1975) notion of “ new advan-tages” and Kitschelt’s (1986) “ structural im-pact.” Success for the suffragists was win-ning voting rights, a goal that, when ach-ieved, would fundamentally redefinewomen’s role in society by giving them “po-litical citizenship” (T. Marshall 1950). In 29states, women achieved full, primary, orpresidential suffrage prior to passage of theNineteenth Amendment.2 Table 1 lists thestates and the years in which each stategranted women these forms of suffrage. Wefocus on full, primary, and presidential suf-frage because they represent the state suf-frage movements’ most significant victories(Banaszak 1996:84). In a one-party South,winning the franchise in primary electionswas tantamount to gaining full suffrage(Harper [1922] 1985:637), and winningpresidential suffrage was considered a majorvictory by the suffragists (Buechler 1986).

Some states granted suffrage through leg-islative action alone. In other states, boththe legislature and the electorate voted onthe matter. Thus, two groups—state legisla-

2 We include only the 48 contiguous states inour analysis because full data for Alaska and Ha-waii are not available for the time period stud-ied.

Table 1. Years in Which States and TerritoriesPassed Full, Presidential, and PrimarySuffrage for Women

Year

Full Presidential PrimaryState Suffrage Suffrage Suffrage

Arizona 1912 — —

Arkansas — — 1917

California 1911 — —

Colorado 1893 — —

Illinois — 1913 —

Idaho 1896 — —

Indiana — 1919 —

Iowa — 1919 —

Kansas 1912 — —

Maine — 1919 —

Michigan 1918 1917 —

Minnesota — 1919 —

Missouri — 1919 —

Montana 1914 — —

Nebraska — 1919 —

Nevada 1914 — —

New York 1917 — —

North Dakota — 1917 —

Ohio — 1919 —

Oklahoma 1918 — —

Oregon 1912 — —

Rhode Island — 1917 —

South Dakota 1918 — —

Tennessee — 1919 —

Texas — — 1918

Utah 1870, 1895 a — —

Washington 1883, 1910 a — —

Wisconsin — 1919 —

Wyoming 1869 — —

Sources: Flexner (1975), NAWSA (1940), andnumerous state-specific sources (see discussion intext on page 59).

Note: Other forms of partial suffrage weregranted to women in some states prior to the pas-sage of the Nineteenth Amendment, includingschool, tax and bond, and municipal suffrage, butthey are not listed.

a Full suffrage was passed in the first year listed,was rescinded, and then was passed again in the sec-ond year listed.


tors and the electorate, both male—wereimplicated in the political decision-makingnecessary to grant women suffrage. Amodel of suffrage success must considerhow the suffrage movements, along withthe contexts in which they operated, influ-enced the willingness of these two male po-litical bodies to support broadening democ-racy to women.3

A CAUSAL MODEL OF

THE POLITICAL SUCCESS OF

A SOCIAL MOVEMENT

Both agency and structure are likely to in-fluence a movement’s political outcomes(Amenta et al. 1992; Gamson and Meyer1996). For the suffragists, we posit that bothgendered and political opportunity structuresinfluenced the political decision-makers thatgave women the vote. But the suffragemovements themselves also were activeagents in winning suffrage, and we considerthe influence on movement success of move-ment mobilization and ideological framing.


We argue that changing gender relations canprovide a gendered opportunity for move-ment success. Gender relations refer to “ thesocial organization of the relationship be-tween the sexes” (Scott 1986:1053; also seeLorber 1992). According to Acker (1992)gender relations are a “pervasive ordering ofhuman activities, practices, and social struc-tures in terms of differentiations betweenwomen and men” (p. 567). This social order-ing also represents a balance of power be-tween the sexes that is maintained by so-cially constructed and widely held under-standings of the appropriate (and different)roles for women and men in society (Connell1987:96–97; Scott 1986). During the nine-teenth century, a clear cultural demarcationbetween women’s and men’s appropriate so-cial spheres emerged, with women’s placedefined as the private sphere of child-rear-

ing and domestic work and men’s place de-fined as the public sphere of politics andbusiness (Welter 1966). Such beliefs ex-cluded women from having a formal voiceand thus formal power in politics. But thesuffragists found gendered opportunities inchanges or variations in gender relations thataltered existing views about the proper rolesof women. These variations occurred in avariety of contexts but, in each case, pro-duced gendered opportunities that ultimatelyfostered suffrage success.

In particular, the rise of the “new woman”offered a gendered opportunity to the suf-fragists. Around the turn of the century,women increasingly received extensive edu-cation, worked outside the home, enteredprofessional careers, had fewer children, di-vorced, and became involved in variouscharitable and political activities (Giele1995; Matthews 1992). Women were mov-ing into traditionally male domains, and thesocial order between the sexes began a trans-formation. The growing presence of this newwoman in the public sphere—in factories,offices, universities, and the professions—gradually weakened the widely held nine-teenth-century assumption that women’s ap-propriate place was in the home.

Not only did such changes create a grow-ing population of independent women whowere often predisposed to suffrage argu-ments (DuBois 1975), but the blurring ofthis public/private distinction helped per-suade the population more generally, includ-ing male political decision-makers in statelegislatures and the electorate, of women’sability to participate in the public sphere, in-cluding politics (S. Marshall 1998; Morgan1972). Caruso (1986), in her study of themovement in Michigan, tells us that therewas a “ realization that women were alreadyplaying a public role” and this “helped de-fuse the fear of women engaging in activi-ties that obviously were not connected to thehome” (p. 269). Ivie (1971) writes of thesuffrage movement in Oklahoma that “pub-lic opinion had changed sufficiently to makepossible effective suffrage work. Womenwere involved in civic affairs, more[women] were economically independent,and greater numbers of women were openlystating their opinions. Oklahoma was readysoil for suffrage work” (p. 68; also see

3 In some cases, governors needed to approvenew suffrage legislation. They usually agreedwith legislatures, but in a few cases (e.g., in Ari-zona in 1903), governors vetoed suffrage legis-lation.


Alexander 1970:22; Fus 1972:101–102, 133;Giele 1995:162; Lerner 1981:365–66; White1974:359). We argue that where womenmoved into previously male spheres of ac-tivity, gender attitudes became more egali-tarian and suffrage success was more likelyas lawmakers and the male electorate actedon these changing views.

Another gendered opportunity emergesfrom previous suffrage successes. In somestates, suffragists won partial suffrage in theform of municipal, tax, or school suffrage.Also, some states bordered states in whichfull, primary, or presidential suffrage hadbeen enacted. These legal changes repre-sented a fundamental shift in gender rela-tions in that they expanded women’s politi-cal rights; such changes previously in a stateor in a neighboring state also altered viewsabout women’s capacity to participate inpolitics. It might be argued that because suchchanges involved a legal change, it is a po-litical opportunity. But we maintain that atits core it is a gendered opportunity (albeitwith a political catalyst) because the changein law redefined gender relations by allow-ing women formal access to the polity and,as the historical record suggests, politicaldecision-makers began to view gender rela-tions differently. As the public witnessedwomen voting in minor elections locally orin major elections in neighboring states withcompetence and good results, views towardwomen’s political participation liberalizedand acceptance of suffrage rights grew(Fleming 1990:40; Jensen 1973:266; Larson1971b:17–18; Smith 1975:26). We hypoth-esize that where the suffragists had gainedsuch rights, they were more likely to succeedwhen pressing for full, primary, and presi-dential suffrage.

A third gendered opportunity exists inpopulations with a particular ethnic makeup.Some studies of women’s suffrage suggestthat certain ethnic groups strongly opposedsuffrage for women, and they did so, for themost part, on the basis of traditional beliefsabout the role of women in society(McDonagh and Price 1985; McDonald1987). Irish- and Italian-Americans, whosefamily practices at the time tended to bemore traditional and grounded in the beliefthat women’s appropriate place was in thedomestic sphere, were less receptive to ar-

guments that women should have the vote(McDonagh and Price 1985). We hypoth-esize that where there were high proportionsof Irish and Italian immigrants, suffragistsconfronted a significant hurdle to winningsuffrage because legislators and electorateswould be less willing to support suffragegiven the prevalence of conservative genderrelations and attitudes. But where thesegroups made up smaller percentages of thepopulation, more liberal gender relationsprevailed and an opening, or gendered op-portunity, for winning the vote existed.4

The fourth gendered opportunity concernsthe western frontier states. Many of thewestern states were the first to grant full suf-frage to women (see Table 1). Some histori-ans argue that the arduous life on the fron-tier produced unusual gender relations thatcompelled women and men to labor side byside to contend with the harsh and as yet un-tamed physical environment (Cole 1990;Turner 1972). As Cole (1990) states, “ thefrontier provided women with opportunity:[T]here they could be fiercely independent,capable, and durable” and they were “ freefrom the constraints which bound their east-ern sisters” (pp. 262–63). Thus, the sharedand more equal circumstance of women andmen in the West fostered a frontier egalitari-anism. Again, because gender practicesshaped gender ideologies, this increased thechances that decision-makers would supportsuffrage.

A fifth gendered opportunity is both gen-dered and political. World War I and its im-mediate aftermath generated an opportunityfor the suffragists—it provided a circum-stance that increased the number of suffragepolitical allies. Political opportunity theo-rists posit that when a movement can claimallies in the polity—political elites sympa-thetic to the movement’s demands and will-ing to act on the movement’s behalf—themovement is more likely to succeed(McAdam 1996; Tarrow 1998). During the

4 We also examined the percentage of RomanCatholics as a variable because Catholics alsotended to have more traditional family practices(Hackett 1995). This measure, however, overlapswith the Irish/Italian measure. The Catholic vari-able in a separate analysis did not significantlyinfluence suffrage success.


war, the suffragists often set aside their suf-frage work and contributed instead to thewar effort. They raised funds for overseashospitals and helped organize studentnurses. And more generally, women assumedduties in the economy—in factories and onfarms—previously done by men (Flexner1975). Toward the end of the war and imme-diately after, politicians and male voterswere more willing to support women’s suf-frage, acknowledging not only women’s ef-forts during the war but the contradiction offighting a war for democracy in Europewhen half of the U.S. population was dis-franchised (Sinclair 1965). To legitimatetheir incumbency and political office, manypoliticians felt pressure to make their posi-tion consistent with arguments in favor ofthe war.

World War I, then, was a gendered oppor-tunity in that the war brought suffragists andwomen generally, as Flexner (1975) states,“out of their homes into new spheres of ac-tion” (p. 298). These new roles altered gen-dered relations and changed attitudes aboutwomen’s fitness for the public sphere. Butthe opportunity was also political in thatpoliticians understood the harm that an in-consistent position on democracy could doto their political well-being. The war thusrepresents both a gendered and a politicalopportunity, and we hypothesize that duringand just after World War I, states shouldhave been more likely to enact suffrage.

Political Opportunity Structures

There are three political opportunities forsuffrage; they involve the formal politicalinterests of political decision-makers orvariations in state structures. However, thefirst two of these (party conflict and prohi-bition laws) also involve policymakers’ con-siderations of how women might cast theirballots if given the vote. And while theseopportunities do not involve changing atti-tudes about gender relations—and in thissense, are not gendered opportunities as wedefine them—they do include gendered ex-pectations about women’s behavior.

The first of these political opportunitiesconcerns instabilities or conflicts among po-litical elites, particularly among politicalparties. Political opportunity theorists point

out that such instabilities compel parties tosearch for additional sources of politicalsupport, given that periods of conflict canthreaten the tenure of political actors hold-ing public office (McAdam 1996; Piven andCloward 1977:28–29). Third-party challeng-ers, such as the Populists, Progressives, Pro-hibitionists, and Socialists, sought tobroaden their party’s base and, in some in-stances, supported women’s suffrage, believ-ing that women voters, in turn, would sup-port their parties (Berman 1987; Marilley1996). This political dynamic was evidentamong the Populists in Colorado in the1890s (Marilley 1996:124–25). We hypoth-esize that where third parties presented sig-nificant challenges to Democratic or Repub-lican legislators, a political opportunity forsuffrage, one rooted in political conflict, ex-isted. But this political opportunity also con-tained a gendered element: Not only didthird-party challengers conduct an analysisof the policy positions most likely to getthem elected, they incorporated into thisanalysis expectations about the “ women’svote,” thus bringing gender to bear on theirpolitical calculations.5

The second of these political opportunitiesstems from involvement of the liquor andbrewing industries in debates over suffrage.These industries opposed suffrage, fearingthat female voters would favor prohibition(Catt and Shuler [1926] 1969). Prohibition,of course, would deeply hurt profit-makingin these industries. These industries not onlyworked to persuade the public to oppose suf-frage, but heavily lobbied politicians to voteagainst it. With passage of state prohibitionlaws, however, the liquor and brewing indus-tries’ opposition subsided. We theorize thatthe waning of economic resistance to suf-frage provided “ space” for political deci-

5 A similar dynamic may be at work in com-petitive elections between the major parties.However, a measure of the competitiveness ofgubernatorial elections between Democrats andRepublicans (the only measure available) was nota significant predictor of suffrage. This measurealso captures a possible influence on governors’decisions to support or veto suffrage legislation:A competitive gubernatorial election might in-crease a governor’ s willingness to support suf-frage. The nonsignificant effect, however, sug-gests that this is not the case.


sion-makers to support suffrage. Their politi-cal calculations were no longer constrainedby a monied interest opposing suffrage. Thesuffragists, then, encountered a political op-portunity for success. But gendered expec-tations also factored into this political oppor-tunity: The absence of concern about thewomen’s vote (following prohibition) pro-duced the political “ space” in which pol-icymakers could now favor suffrage.6

These last two political opportunities—third-party challenges and prohibition—aredifferent from the gendered and political op-portunity that World War I offered the suf-fragists. Third-party challenges and the pas-sage of prohibition clearly presented politi-cal opportunities for suffrage success be-cause the historical record indicates that thepolitical calculus of decision-makerschanged under these circumstances. But be-cause there is no evidence of changing be-liefs about gender roles, third-party chal-lenges and prohibition did not offer gen-dered opportunities for suffrage, only politi-cal opportunities tempered by expectationsabout the women’s vote.

The final political opportunity concernsthe degree of access to the polity experi-enced by the suffragists; this opportunitydoes not have a gendered component. Al-though political opportunity theorists oftenuse the existence of voting rights as an indi-cator of the openness of the polity, we mustconsider another indicator of political accessbecause suffragists did not have votingrights (Amenta et al. 1994; Tarrow 1998).Even without the vote, suffragists could anddid lobby state legislators to introduce a billor resolution in the legislature that, if passedby the legislature (and in many cases theelectorate as well), would grant voting rightsto women. The procedural ease or difficulty

of winning the passage of such legislation,then, offers an indicator of political open-ness. The procedure required to change vot-ing policy was substantially less complex insome states in that fewer decision-makerswere involved in the decision-making pro-cess. For instance, in territorial Wyomingonly one positive legislative vote wasneeded to grant women suffrage. In Nevada,however, a favorable vote on a resolution fora referendum on a state constitutionalamendment was required in two consecutivelegislatures which met only every other year,and only then was the measure taken to theelectorate.

Thus, not only is this last political oppor-tunity not gendered, it offers a different kindof political opportunity for suffrage successin that, where it existed, fewer decision-makers were involved in granting the voteto women. The other two political opportu-nities offer an opportunity for politicalchange by altering the political interests ofdecision-makers in ways that make themmore inclined to enact policy change. Onepolitical opportunity is a quantitative shift inthe number of decision-makers; the othersare qualitative shifts in decision-makers’ in-terests. We hypothesize that states with sim-pler procedures for changing suffrage lawoffered a more accessible policymaking pro-cess that heightened the chance that suffragewould be passed.7

Resource Mobilization and

Cultural Framing Theory

We now turn our focus from the context inwhich movements operate to the movementsthemselves. Resource mobilization and fram-

6 Admittedly, this is a weaker instance of po-litical opportunity than the others. The absenceof economically motivated opposition to suffragemay reduce lawmakers’ resistance, but it does notexplain why lawmakers voted for suffrage. Also,we examined the impact of other measures of li-quor industry opposition to suffrage (e.g., per-centage employed in brewing and liquor indus-tries, the presence of a state brewers’ association,and percentage of German immigrants). Thesemeasures were not significant factors in suffragesuccess.

7 Only full suffrage required a constitutionalamendment (although not in all states). Thus, theprocedural ease of policy change may influencethe suffragists’ capacity to gain full suffrage butnot presidential and primary suffrage. Also, be-cause the territories required only one legislativevote on suffrage and no referendum, we includeda dummy variable indicating the territories in ourmodel, but it did not significantly predict suf-frage. We examined the impact of another mea-sure of the openness of the polity—the availabil-ity of the initiative and referendum—but foundno significant effect on suffrage (also see Ban-aszak 1998).


ing perspectives offer insights into the ac-tions of movement activists and the ways inwhich those actions influence political deci-sion-makers’ support for movement agendas.Contrasting these perspectives with thoseemphasizing opportunity structures high-lights the distinction between the influencesof agency (the movement itself) and struc-ture (the movement’s context). We considerboth because we believe that both influencemovement outcomes. We hypothesize thatthe ability of a movement to effect politicalchange hinges in part on the same factors thatresource mobilization theorists deem crucialto movement emergence and development—organization and key strategies (McCarthyand Zald 1977; Piven and Cloward 1977)—because successful movement mobilizationaids a movement’s ability to bring about po-litical change (Gamson 1975).

Every state except Wyoming had a statesuffrage association, and therefore the sim-ple presence of such an organization is notlikely to explain whether a state suffragemovement was successful.8 The extent of or-ganizing, however, may influence suffragepolitical success. The larger the movement,the more capable it should be of disseminat-ing its message and thus of convincing leg-islators and the electorate to vote for suf-frage. In some states, the suffrage move-ments organized men’s or college women’ssuffrage leagues in addition to state suffrageassociations, and some states had multiplestate suffrage leagues. We use a measure ofthe number of organizations in a state as oneindicator of the extent of organizing. An-other measure of the extent of organizing ismembership in state suffrage associations,which varied widely across states (NationalAmerican Woman Suffrage Association[NAWSA] 1893–1917, 1919). Utah and NewHampshire had some of the largest per capitamemberships, while South Carolina andIdaho had some of the smallest. We expectthat the larger the per capita membership,the more successful the movement.

Until recently, little has been written onthe relationship between movement strate-

gies and movement outcomes (Banaszak1996; Klandermans 1989; Morris 1993).The history of suffrage movements makesclear that suffragists relied almost exclu-sively on what R. Turner (1970) labels “per-suasion strategies,” or “ the use of strictlysymbolic manipulation without substantialrewards or punishments” (p. 149). In mostcases, the suffragists relied on argument topersuade others that their goals were wor-thy. Rarely did they rely on disruptive tac-tics.9

Movement researchers recognize the im-portance of struggles over meanings and be-liefs in movements. Snow et al. (1986) referto “ frame bridging” and Gamson (1988) re-fers to “ cultural resonance” to describe aprocess in which movement participants, inthe public rationales for their demands, usethemes and ideas that tap into widely heldbeliefs. This resonance with popular beliefshelps to build movement strength and effec-tiveness by expanding support for themovement’s cause (McCammon 1995;Quadagno 1992).

Suffrage movement histories suggest thata shift in the suffragists’ ideological ration-ales for the vote occurred as the suffragistslearned that they could link their argumentsto existing beliefs espousing women’s fit-ness for roles in the private or domesticsphere (Baker 1984; Buechler 1986; Mar-illey 1996). Kraditor (1965) writes that thesuffragists began to rely less on “ justice ar-guments” and instead began to use “expedi-ency arguments.” The earlier justice argu-ments used by the suffragists were based inliberal individualism—suffragists assertedthat women deserved political rights equalto those of men because, like men, theywere citizens. Such arguments, however, di-rectly challenged the widely acceptedboundary separating men’s and women’sspheres in that they attempted to redefinewomen’s roles, particularly by definingwomen’s participation in politics (or in thepublic sphere) as acceptable (Baker 1984).

8 Because the state associations organized atdifferent points in time, we examined the effectof the presence of a state organization on move-ment success but found no significant impact.

9 Early in the movement, suffragists engagedin civil disobedience by attempting to vote ille-gally. Because this strategy was used primarilyin the early 1870s, it is not a focus of our re-search. Analyses reveal, however, that illegalvoting did not affect the likelihood of suffrage.


But over time, suffragists began to use ex-pediency arguments. What we call a “ sepa-rate-spheres argument” emphasized thatstate policies increasingly regulated the do-mestic sphere and that women could bringknowledge of the domestic sphere to the po-litical arena in determining, for instance,how food, water, domestic violence, child-ren’s schooling, and even alcohol abuseshould be regulated. Marilley (1996) makesa similar point when she focuses on theemergence of “home protection” argumentsamong the suffragists. The strategy of es-pousing separate-spheres arguments, we ex-pect, should have produced success for thesuffragists because, unlike justice argu-ments, separate-spheres arguments did notovertly challenge the traditional roles ofwomen and men. Such arguments simplypointed to the advantage of allowing womento help regulate the private sphere. Thus, po-litical decision-makers should find these ar-guments more convincing than those basedon justice.

A second hypothesis about successfulstrategies used by the suffragists concernstwo types of tactics used by the suffragists—insider and outsider strategies (Ornstein andElder 1978). Both of these strategies, whereused, should help the suffragists win the vote.Insider strategies are the political activitiesthat the suffragists engaged in to influencepolitical “ insiders,” that is, state legislators.Suffragists lobbied lawmakers to introducesuffrage bills and resolutions into the legis-lature or to cast a favorable vote on suffrage,worked to elect sympathetic legislators, androutinely appealed to state political parties,particularly the Democrats and Republicans,to endorse suffrage. Banaszak (1996:139)argues that such political strategies were keyfor suffragist success. We also argue that theywere important because convincing the leg-islature to approve suffrage was necessaryfor winning the vote in every state.

Outsider strategies, on the other hand, aretactics aimed at persuading the general pub-lic of the movement’s goals, and they targetcitizens or “ outsiders” to the polity. Suchstrategies included formal and informal pub-lic speeches—the latter involving “ streetspeaking” (soapbox speeches on street cor-ners) and “ autotours” (statewide speakingtours in which speeches were given from au-

tomobiles)—parades, booths at fairs, store-front advertising, and pamphleting. Whileoutsider strategies could influence legisla-tors, their impact may be most discernible instates in which a public referendum on thesuffrage question was required and thus inwhich public opinion was crucial to the out-come. This was the case in most states vot-ing on full suffrage for women.

To close this discussion of suffragist strat-egies, we consider the impact of fundraisingon political success. Fundraising does not fitneatly into either the insider or outsider clas-sification. Successful fundraising could sup-port both insider and outsider work, for in-stance, by affording the suffragists trips tothe state capitol or party conventions tolobby politicians or allowing the suffragiststo purchase space to advertise in newspapersor literature to distribute. But also, becauseincreased funding can translate into manydifferent types of activities—whatever wasdeemed critical by suffrage leaders—fund-raising becomes an important strategy thatshould foster movement success. In effect, itgave the suffragists flexibility because theycould target insiders or outsiders for theirpersuasion tactics, whoever was deemedmore important. The histories of suffrage arereplete with instances of fundraising activi-ties, and the national suffrage organization(NAWSA) was adamant in its communica-tions to state suffrage associations that theyengage in a variety of fundraising activities.We expect that fundraising heightened suf-fragists’ chances of political success.

Finally, we briefly consider another“agent” that may have influenced the suc-cess of the suffragists. Anti-suffrage organi-zations mobilized in many states to opposevoting rights for women. Members of thiscountermovement argued that suffrage poseda threat to the “ sanctity of the home” andthat involvement in politics would overbur-den women, cause discord between husbandand wife, and violate the natural and biologi-cal differences between women and men(Camhi 1994). Where this resistance to suf-frage existed, political decision-makers mayhave been persuaded by its arguments andmay have voted against giving women thevote. Thus, where the anti-suffrage organi-zations were most active should be wherethe suffragists were least successful.


DATA AND METHODS

We use discrete-time methods in event his-tory analysis to examine the factors leadingto suffragists’ political success between1866 and 1919 (when Congress passed theNineteenth Amendment). All data are annual(i.e., calendar year), state-level data. Our de-pendent variable indicates years in whichstates granted full, primary, or presidentialsuffrage, the suffragists’ most significant po-litical successes. The variable is coded as 0for years prior to the enactment of suffrageand 1 for the year in which suffrage wasgranted. Sources used in the construction ofthe dependent variable include: Flexner(1975), NAWSA (1940), and numerousstate-specific sources.10

We collected a large portion of our data onthe state suffrage movements and their con-texts from an extensive search of secondarymaterials on each of the state movements(over 650 sources in all). We also relied onvolumes 3, 4, and 6 of The History of Wo-man Suffrage (Stanton, Anthony, and Gage ,vol. 3, [1886] 1985; Anthony and Harper,vol. 4, [1902] 1985; Harper, vol. 6, [1922]1985), which were written by movementparticipants and contain lengthy descriptionsof the state suffrage movements. In addition,we examined The Hand Book of the NationalAmerican Woman Suffrage Association andProceedings of the Annual Convention from1893 to 1919, the years of its publication forour period. For six states (Arizona, Dela-ware, Maine, New Hampshire, New Mexico,and North Dakota), few secondary materials

exist, and we conducted archival research inthe states to supplement secondary accounts.For these states, we also examined theWoman’s Journal from 1870 to 1919, a suf-frage newspaper published by NAWSA andits predecessors that routinely reported statesuffrage events. Unless other sources arenoted, the reader can assume that the sourceof data for a variable came from this exten-sive collection process.

We used content analysis to code the in-formation in each article, book, dissertation,or archival source that we read. Three cod-ers collected data, and Krippendorff ’s (1980)alpha ranged from .91 to .95 on selectedsources for which we assessed interrater re-liability.

The rise of the new woman is indicatedwith an index that combines three measures:(1) the proportion of college and universitystudents who are female (U.S. Departmentof Commerce various years; U.S. Office ofEducation various years), (2) the proportionof lawyers and doctors who are female (U.S.Bureau of the Census various years; U.S.Department of Commerce various years),11

and (3) the number of prominent women’sorganizations active in a state (viz., the Con-sumers’ League [Nathan 1926], the GeneralFederation of Women’s Clubs [Skocpol1992:330], the National Congress of Moth-ers [Mason 1928:295], the National Wo-men’s Trade Union League [Dye 1980; Na-tional Women’s Trade Union League(NWTUL) various years], and the Woman’sChristian Temperance Union [WCTU]). Wecombined these three measures by summingtheir standardized values (α = .64).12

10 Even though states passing presidential orprimary suffrage are still “at risk” of enacting theother forms of suffrage, to avoid bias in the esti-mated standard errors that can result if events arenot independent (as may be the case when suf-frage is enacted more than once in a state), weanalyze the possibility of only one event per state(Allison 1984:54). This affects the coding of thedependent variable for 3 of the 29 states thatpassed suffrage. Michigan passed presidentialsuffrage in 1917 and full suffrage in 1918. Wechose to analyze the passage of full suffrage inMichigan because of its wider impact. Utah andWashington both passed full suffrage and thenrescinded it. Both later passed full suffrage againand did not later rescind it. For these states, weanalyze the passage of permanent suffrage.

11 For two variables (proportion of lawyers anddoctors who are female and percentage of Irishor Italian immigrants), only decennial data areavailable. For these variables we linearly inter-polated data for the intervening years.

12 We acknowledge that, from a resource mo-bilization perspective, these measures of the newwoman may also be measures of potential lead-ers for the suffrage movements. However, giventhat we include more direct measures of move-ment organizational strength and strategies in theanalysis (i.e., the likely consequences of goodleadership), the contributions of measures of thenew woman to effective leadership (and thus tomovement success) have been controlled in theanalysis. Also, it could be argued that our mea-


Legal changes expanding women rightsare indicated by (1) the earlier passage in astate of municipal, tax, or school suffrage(coded as a simple count of the number offorms of partial suffrage in existence; samesources as full, presidential, and primary)and (2) the proportion of neighboring statesthat enacted full, primary, or presidentialsuffrage.13 Each of the legal variables islagged one year because we assume thatthere will be a brief delay in the effects of achange in law.

The presence of Irish and Italian immi-grants is measured by the number of Irish-and Italian-born immigrants living in a stateper 1,000 in the population (U.S. Bureau ofthe Census various years).

We use a dichotomous measure to indicatethe western frontier states (defined on thebasis of census categories). The measure iscoded 1 if the state is in the West and 0 oth-erwise. The western states are: Arizona,California, Colorado, Idaho, Kansas, Mon-tana, Nebraska, Nevada, New Mexico,North Dakota, Oklahoma, Oregon, SouthDakota, Utah, Washington, and Wyoming.

World War I is a dichotomous variablecoded 1 in 1917 and 1918, the years of U.S.involvement in the war, and 0 otherwise.The term is lagged one year because of ourexpectation that the effects of World War I

on suffrage success will be experienced laterin the war and in its immediate aftermath.

Third-party legislative challenges are in-dicated by the percentage of seats in the statelegislature held by third parties. This vari-able is lagged one year. Although the mea-sure is not a direct indicator of third-partychallenges during elections, it reflects theoutcome of elections. We lag the variable toapproximate the influence of a third-partychallenge during an election in the previousyear.14

The passage of a state prohibition law iscoded 1 for years following the passage of aprohibition law and 0 otherwise (Cashman1981).

The accessibility of the polity measuresthe procedural difficulty of winning suf-frage. It is indicated with a variable coded 1through 5, with 1 representing the simplestprocedures and 5 the most difficult (a fullerdescription of this variable is available onrequest from the authors).

We measure the extent of suffrage orga-nizing in a state by: (1) the number of suf-frage organizations in existence in a givenyear (including state associations, collegewomen’s suffrage leagues, and men’s suf-frage leagues), and (2) members (per100,000 in the population) in the NAWSA-affiliated state suffrage associations(NAWSA 1893–1917, 1919). Membership isestimated from dues paid by state associa-tions to NAWSA (Banazak 1996:231–33).

The suffragists’ use of separate-spheresarguments is coded 1 if a separate-spheresargument was made (in a public speech orpublic document) in a given year, and 0 oth-erwise.15

sure of women’ s organizations measures a non-suffrage organizational resource for the suffragemovement because some of the these organiza-tions, especially the WCTU, campaigned for suf-frage. We excluded these measures (first thewomen’ s organization measure and then just theWCTU measure) from our index, and each of therevised indices remained significant and positive.

13 Including a measure of the passage of suf-frage in contiguous states provides a crude mea-sure of a spatial diffusion process whereby eventsin one state spread to neighboring states. If sucha process exists in our data, however, without amore sophisticated control, our estimates couldbe biased (Doreian 1981). Thus, we included inanalyses not shown a spatial effects term con-structed by multiplying a normalized weight ma-trix for our states (coded 1 for contiguous statesand 0 for noncontiguous states) by the predictedvalue for suffrage success (Deane, Beck, andTolnay 1998). Inclusion of this term rather thanthe simple measure did not alter our findings inany meaningful way. Thus, in the results pre-sented here we use only the simpler measure.

14 Most legislatures during this period met ev-ery other year. In a few states, elections and ses-sions occurred in the same year. A nonlaggedmeasure of third-party electoral success, how-ever, did not significantly predict the passage ofsuffrage. Because legislatures cannot enact lawin nonsession years, we included an indicator foryears in which legislatures were in session. Themeasure was not significant, suggesting that ourother variables better explain when states werelikely to grant suffrage.

15 We examined the impact of justice argu-ments on suffrage success as well (not shown),but the variable did not significantly influencethe passage of suffrage.


Insider strategies used by the suffragistsinclude lobbying state legislators, writingthem letters, giving speeches in the legisla-ture, presenting petitions to the legislature,campaigning for pro-suffrage candidates,and seeking endorsements from politicalparties at state party conventions. Outsiderstrategies include holding formal publicmeetings (usually with an admission fee andlecturers), giving informal public speeches(such as street speaking and autotours), par-ticipating in parades, setting up booths atfairs, and distributing leaflets door to dooror on public streets. The insider and outsiderstrategy variables are counts of the differenttypes of strategies used in a given year.

Fundraising activity on the part of the suf-fragists is indicated by any activity that wasdesigned to raise money for the movement,such as selling tickets to suffrage lecturesand plays, soliciting donations, and holdingvarious types of sales to raise money. If thesuffragists engaged in any such activities ina given year the measure is coded 1 and 0otherwise.

The presence of anti-suffrage activity iscoded 0 when no activity occurred, 1 whenout-of-state anti-suffrage groups were ac-tive, 2 when a few state anti-suffragists wereactive, and 3 when a state anti-suffrage or-ganization existed.

We also include in our models dichoto-mous variables indicating the various de-cades in our period of analysis (the decadeof the 1910s is the reference category).Many states passed suffrage in the last de-cade of our analysis (see Table 1). Includingthe decade measures allows us to determinewhether, after controlling for other factors,any period effects remain.

RESULTS

Table 2 presents results from our event his-tory analyses of the circumstances leading tosuffrage success. Models 1 and 2 include allvariables discussed above. Model 1 containsa measure of the number of suffrage organi-zations in a state, while Model 2 contains ameasure of suffrage association member-ship.16 Model 3, our final model, includes

only the significant variables from Models 1and 2.17 Models 4 through 6 are variationson Model 3.

The impact of gendered opportunity struc-tures on movement success is shown in thefirst two panels (labeled “Gendered Oppor-tunity Structures” and “Gendered and Politi-cal Opportunity Structure” ) for Models 1through 3. Although not all gendered oppor-tunity variables are significant, there is sub-stantial evidence that gendered opportunitiesfor suffrage success existed. The presence ofthe new woman—educated, professional,and politically and civically active women—significantly enhanced the likelihood thatthe suffragists would win the vote. Further-more, Model 4 shows that all three of the in-dividual measures used to construct ournew-woman index are significant predictorsof suffrage success. These results suggestthat the presence of such women demon-strated to the larger population that womenwere fit for a significant political role in thepublic sphere and convinced legislators andthe electorate to support women’s suffrage.The presence of the new woman thus playedan important role in helping the suffragistswin the vote.

Moreover, the passage of full, presiden-tial, and primary suffrage in one or moreneighboring states significantly encouragedthe passage of these types of suffrage in aparticular state. This suggests anothergendered opportunity rooted in the legal cli-mate. Laws granting women broader citi-zenship rights helped to redefine appropri-ate gender roles. The redefinition of legalgender roles in one state appeared to alterthinking about gender roles by political ac-tors in neighboring states, resulting in thepassage of suffrage there as well. Interest-ingly, the passage of various forms of par-tial suffrage (school, tax, and municipalsuffrage) in a state did not increase the like-lihood of suffrage success in that state. Thissuggests that for a gendered opportunity of

16 We do not include these two measures in thesame model because they measure the same con-

cept and data on membership are only availablebeginning in 1892.

17 Comparing the log-likelihoods for Models 1and 3 shows that Model 3 is a significant im-provement over Model 1 (likelihood-ratio chi-square equals 26.3, which is significant at p ≤.001).


Table 2. Maximum-Likelihood Coefficients from an Event History Analysis of Factors Influencingthe Passage of Women’s Suffrage in U.S. States, 1866 to 1919

Independent Variable Model 1a Model 2b Model 3c Model 4c Model 5 c,d Model 6


New-woman index 1.234* 1.096* 1.002* .— .977* .—(.327) (.272) (.230) (.224)

Proportion female college .— .— .— .105* .— .—students (.030)

Proportion female lawyers .— .— .— .027* .— .—and doctors (.011)

Number of women’ s .— .— .— .596* .— .446*

organizations (.309) (.267)

Proportion of neighboring 5.100* 4.353* 4.736* 3.824* 4.175* 5.650*

states with full, presidential, (1.822) (1.971) (1.081) (1.180) (1.010) (1.086)or primary suffrage (lagged)

Number of school, municipal, .254 .291 .— .— .— .—and tax suffrage laws (lagged) (.348) (.357)

Irish and Italian immigrants .008 –.001 .— .— .— .—per 1,000 population (.015) (.015)

Western state –.043 .104 .— .— .— .—(.963) (.973)

Gendered and Political Opportunity Structure

World War I years (lagged) 1.723* 1.964* 2.037* 2.171* 2.312* 1.821*

(.899) (.874) (.654) (.698) (.693) (.638)

Political Opportunity Structures

Percent third-party seats .020 .013 .— .— .— .—in legislature (lagged) (.035) (.035)

State prohibition law 2.783* 2.439* 2.279* 1.869* 1.978* 1.361*

(.975) (.887) (.661) (.704) (.631) (.558)

Accessibility of the polity –.746* –.689* –.528* –.620* .323 –.433*

(.315) (.307) (.242) (.250) (4.861) (.225)

Resource Mobilization and Cultural Framing

Number of suffrage –.376 .— .— .— .— .—organizations (.430)

Suffrage association member- .— .011 .— .— .— .—ship per 100,000 population (.024)

Number of insider strategies .113 .150 .— .— .— .—(.369) (.366)

Number of outsider strategies .373 .262 .— .— .845 .—(.340) (.321) (2.349)

Outsider strategies used .— .— .— .— –.631 .—in a referendum state (2.351)

Fundraising activity 1.856* 1.740* 1.938* 1.910* 1.831* 1.987*

(.788) (.783) (.633) (.635) (.666) (.551)

Separate-spheres arguments 2.153* 1.955* 1.719* 1.688* 1.659* 1.424*

used by suffragists (.769) (.742) (.600) (.630) (.583) (.528)

Anti-suffrage activity –.244 –.242 .— .— .— .—(.249) (.248)

(Table 2 continued on next page)


this nature to occur, the enactment of suf-frage must have been of the same caliber,that is, full, presidential, or primary. Minorforms of suffrage apparently did not trans-form attitudes about women’s politicalrights and clearly did not lead to the pas-sage of major forms of suffrage.

The presence of Irish and Italian immi-grants had no significant negative effect onwinning the vote for women. Thus, the ab-sence of these groups did not provide agendered opportunity for suffrage success.Several other studies (e.g., Lerner 1981;Mahoney 1969) also provide evidence thatimmigrants were no less likely to supportsuffrage than were native-born residents.Although we do not have a direct measureof immigrant attitudes, our results generallyare in line with these findings. Also, the

western frontier did not provide a genderedopportunity for the passage of suffrage. Itmay be that the new-woman measures bet-ter explain suffrage success in the West:Educated and professional women were of-ten more prevalent in the West than in otherregions (U.S. Bureau of the Census variousyears; U.S. Department of Commerce vari-ous years; U.S. Office of Education variousyears).18

Decade1860s .— .— .— .— .— 1.559

(1.461)

1870s –3.901 .— –3.598 –4.471 –2.986 –6.750(57.714) (33.607) (54.522) (34.092) (32.598)

1880s –5.678 .— –4.986 –6.819 –4.742 –7.089(50.268) (30.165) (44.746) (29.837) (32.432)

1890s –1.381 –.747 .057 –.336 .408 –.076(1.727) (1.658) (.769) (.902) (.781) (.817)

1900s –10.177 –8.933 –8.893 –9.963 –8.549 –8.598(46.366) (27.837) (28.561) (47.142) (29.076) (31.552)

Constant –8.343* –8.256* –7.738* –11.682* –9.733* –6.513*

(1.608) (1.572) (1.275) (2.142) (4.994) (1.303)

Number of cases 1,921 1,161 2,078 2,078 2,078 2,358

Year beginning period of analysis 1872 1892 1872 1872 1872 1866

Likelihood-ratio chi square 168.1* 142.4* 178.3* 182.2* 174.1* 158.0*

Sources: See text on pages 59 to 61.

Note: Numbers in parentheses are standard errors.a Analysis begins in 1872 because female college student data begin in 1872. Also, professional women

data begin in 1870, and immigrant and third-party data are unavailable for some territories.b Analysis begins in 1892 because suffrage association membership data begin in 1892. Also, female

college student data begin in 1872, professional women data begin in 1870, and immigrant and third-partydata are unavailable for some territories.

c Analysis begins in 1872 because female college student data begin in 1872. Also, professional womendata begin in 1870.

d In Model 5 only, the accessibility of the polity variable is an interaction term indicating the use ofoutsider strategies in states in which the referendum was required to change a constitutional amendment.

*p ≤ .05 (one-tailed tests)

(Table 2 continued from previous page)

Independent Variable Model 1a Model 2b Model 3c Model 4c Model 5 c,d Model 6

18 Even though full suffrage was more likelyto be enacted by western states compared withstates in other regions, in analyses not shown, wefound that the “western” variable did not signifi-cantly influence the passage of full suffrage (ex-cluding presidential and primary suffrage fromthe dependent variable) once other factors hadbeen taken into account.


States were significantly more likely toenact suffrage toward the end of or just afterWorld War I (see panel labeled “Genderedand Political Opportunity Structure” ). Al-though this effect indicates a political oppor-tunity for suffrage, it also adds to the evi-dence that gendered opportunities were in-strumental in bringing about women’s suf-frage.

The fact that the new-woman variable andthe measures of suffrage in neighboringstates and World War I are the significantgendered opportunity indicators suggeststhat gendered opportunities for suffragestemmed particularly from women’s grow-ing presence in the public sphere. Thesemeasures, unlike most of the other genderedopportunity measures (e.g.,, the measures ofimmigrants and western states) are clear in-dicators of women’s movement into the pub-lic sphere.

Among the measures of political opportu-nity structures, the passage of state prohibi-tion laws and the accessibility of the policy-making process significantly influencedwhether the suffragists won citizenshiprights (third panel, Models 1 through 3).These results, combined with those for theWorld War I variable, indicate that politicalopportunities also helped suffragists gain thevote. Increased political support for the suf-fragists during the war, reduced liquor indus-try opposition after the passage of state pro-hibition laws, and simpler enfranchisementprocedures all worked favorably for the suf-fragists.

Third-party representation in the legisla-ture did not provide a significant politicalopportunity for the enactment of suffrage.Perhaps during periods of third-party con-flict, some political actors supported suf-frage, hoping to appeal to future women vot-ers, while others opposed suffrage, fearingharm to (re-)election chances.

The results for the resource mobilizationand ideological framing measures suggestthat larger suffrage movements were nomore successful in winning the vote thansmaller movements (fourth panel, Models 1through 3). Neither measure of the extent oforganization—the number of suffrage orga-nizations (Model 1) or the size of member-ship (Model 2)—is significant. This findingis not surprising for two reasons: Gamson

(1975:51) found a similar pattern in his earlywork, and some suffragists recognized theimportance of small, quiet movements. Forinstance, Abigail Scott Duniway, an activistin the northwest, saw the merit of what shecalled the “ still hunt”—quiet persuasion onmany fronts so as not to arouse the opposi-tion (Moynihan 1983; also see Trout 1920).These results, along with Gamson’s, suggestthat movement researchers need to rethinkthe role of large movements in winning po-litical gains.

Also, the use of insider or outsider strate-gies does not significantly predict suffragesuccess.19 The nonsignificant results for in-sider strategies run counter to Banaszak’s(1996) claim that the lobbying efforts of thesuffragists were key to their success. Model5 includes a term indicating the use of out-sider strategies in states where a public ref-erendum was required to amend the stateconstitution.20 Where the procedure forchanging voting rights entails a vote by theelectorate, outsider strategies—designed tosway public opinion—may be more effectivein bringing about suffrage success. But theresults show that the use of such strategiesin referendum states did not significantly in-crease the suffragists’ likelihood of winningvoting rights.

Two strategies, however, when employedby the suffragists, did help them achievetheir goals—fundraising and the use of sepa-rate-spheres arguments (fourth panel, Mod-els 1 through 3). The significant result forfundraising suggests that this activity waskey because it afforded the suffragists manyways of influencing the actions of politicaldecision-makers. The significant result forthe separate-spheres measure suggests thatwhen suffragists framed rationales for suf-frage along the lines that women would usethe vote to protect their children and their

19 We disaggregated the insider and outsiderstrategy measures into their constituent parts andincluded these more specific measures in analy-ses. None of the individual measures was signifi-cant.

20 The interaction term is constructed by mul-tiplying the outsider strategy variable by a di-chotomous variable coded 1 for states requiringa referendum and 0 otherwise. The interactionreplaces the accessibility of the polity variable inthe analysis.


homes, they were effective because such ar-guments were consistent with establishedgender beliefs.

Two of the significant measures in thesemodels have missing data: women’s repre-sentation among college students andamong professionals. Both measures havemissing data at the beginning of our timeperiod.21 Because missing data in event his-tory analysis can introduce censoring,which can then produce sample selectionbias and thus biased estimates (Yamaguchi1991:3–9), Model 6 excludes these twovariables from the analysis to ascertainwhether censoring affects our findings.However, the results are substantively un-changed once the earlier years are includedin the analysis.22

Finally, the results for the decade mea-sures (the fifth panel of Table 2) suggest thatour substantive variables explain period dif-ferences in the pace of suffrage reform overtime. None of the decade measures is signifi-cant, indicating that once other factors havebeen taken into account, the 1910s are nodifferent from the other decades. The flurryof suffrage successes in the 1910s probablyresults from a combination of many of thefactors shown here to be significant—theemergence of the new woman, the increas-ing number of states having already passedsuffrage, World War I, state prohibition laws,and greater fundraising and separate-spheresframing on the part of the suffragists. (Ourindicators show that both of these last mea-sures were increasing over time.)23

CONCLUSION

Susan B. Anthony declared, in her last pub-lic address in Washington, D.C. in 1906, that“ failure is impossible” (Harper 1908:1409).She was right: In 1920, with ratification ofthe Nineteenth Amendment, women won theright to vote in all elections across theUnited States. But for five decades, suffrag-ists in the state movements had battled forthe vote with varying degrees of success.Our goal has been to uncover the circum-stances that produced political success forsome of the state movements. The historiesof the state suffrage movements provide arare opportunity to compare, systematically,multiple movements working toward thesame general goal. Our findings reveal thatseveral circumstances were necessary forsuffrage success.

Not only did resource mobilization, cul-tural framing, and political opportunitystructures help to produce movement suc-cess, but gendered opportunity structureswere important as well. In the case of thesuffragists, changing gender relations fos-tered movement success by altering beliefsamong political decision-makers about theproper role of women in society. Thesechanged beliefs, we argue, made legislatorsand the electorate more willing to vote forsuffrage. Some of the political opportunitiesthat benefited the suffragists also had gen-dered elements. In fact, the political oppor-tunity provided by World War I was so inter-twined with a gendered opportunity that wecould not analyze them separately. Clearly,gender theorists are correct when they statethat gender “ is integral to many societal pro-cesses” (Acker 1992:565).

Our most important theoretical conclusionis that social movement scholars must rec-ognize that other types of opportunity struc-tures, beyond those stemming from formalpolitical dynamics and the formal politicalinterests that they generate, can also influ-ence movement success. Our findings thatgendered opportunities fostered suffragesuccess and that political opportunities weregendered in various ways demonstrate thatmore than simply formal politics matter formovement success.

Opportunity structures can take a varietyof forms. In the context of suffrage, we

21 Data for the proportion of professionalwomen begin in 1870 and data for the proportionof female college students begin in 1872.

22 One event is left-censored because of thesemissing data (passage of full suffrage in Wyo-ming in 1869). Although censoring occurs in theanalyses in other models as well, we found noevidence of bias.

23 Analyses including interaction terms be-tween movement mobilization and opportunitystructure variables revealed no significant inter-action effects. Recursive regressions showed nomeaningful instability in our significant coeffi-cients. Also, multicollinearity does not affect theresults: For any substantially correlated pairs ofpredictors, excluding one of the variables fromthe analysis did not affect the results of the re-maining variable.


found evidence of political and gendered op-portunities and even a combined genderedand political opportunity. Researchers study-ing other movements may find differenttypes of opportunity structures. Amongmovements working to redefine broad cul-tural definitions of women’s and men’s rolesin society, gendered opportunities—opportu-nities emerging from changing gender rela-tions and altered views about gender—arelikely to be crucial to movement success.However, a social movement struggling toredefine the position of a racial group in so-ciety, such as the civil rights movement’s at-tempt to achieve political inclusion for Afri-can Americans, is likely to find opportunitystructures in changing race relations that al-ter people’s thinking about the appropriaterelations between races. Similarly, changingeconomic relations between employers andworkers may provide an opportunity struc-ture for working-class political success.

Thus, rather than using the narrower “po-litical opportunity structure” to refer to thecontextual factors that influence movementsuccess, researchers would do well to use thebroader and simpler term, “ opportunitystructure.” Then, to determine the precisenature of the opportunities—be they gen-dered, racial, ethnic, class or combinationsof these—researchers must carefully exam-ine the evidence in the particular context fortheir movement(s) and scrutinize the dynam-ics shaping the willingness of key politicaldecision-makers to support policy change.We draw once again on McAdam (1982:41,italics in original), but with an important re-vision: “ [A]ny event or broad social processthat serves to undermine the calculations andassumptions” of key political decision-mak-ers presents the possibility of an opportunitystructure for movement success.

Our findings about the role of genderedopportunities in producing suffrage success,along with the way in which the suffragistsframed their rationales for expanded politi-cal rights, reveal a paradox in the circum-stances leading to suffrage success. Compet-ing logics appear to have been at work. Onthe one hand, the suffragists were more suc-cessful when they used an ideological ratio-nale that resonated with widely accepted buthighly traditional beliefs about women’s ap-propriate roles in society. The suffragists’

use of separate-spheres arguments, whichheld that women should be allowed to votebecause they could protect and help regulatethe domestic sphere, significantly aided thesuffragists in their bid for the vote.

On the other hand, the presence of the newwoman (an independent, civically active,and educated woman), along with expandedpolitical rights in neighboring states andwomen’s contributions during the war,helped the suffragists win voting rights.These circumstances successfully chippedaway at the traditional belief that women’sonly appropriate place in society was in thedomestic sphere. Here, more egalitarian gen-der relations created an opportunity for suf-frage success, unlike the separate-spheresarguments that tapped into support for tradi-tional gender relations. That both dynamics,one rooted in traditional beliefs about gen-der roles and the other rooted in the emer-gence of more egalitarian relations, could beoperative during these years is not surpris-ing when one considers the fundamentalchanges that women’s lives were undergoing(Flexner 1975). In such a period of transi-tion, competing logics were simultaneouslyat work.

Finally, our findings show that not onlydid opportunity structures help the suffrag-ists win the vote, but the actions of the suf-fragists themselves were important. The suf-fragists were clearly agents in the redefini-tion of democracy that occurred during theseyears. They were active in all states, and thedegree to which they mobilized financial re-sources and the manner in which theyframed rationales for the vote significantlyinfluenced whether they were successful inwinning the vote. But the outcome was notentirely within their control. Gendered andpolitical dynamics transpiring in the broadercontext also mattered. In the end, then, weargue, as have others, that both agency andstructure determine a movement’s politicalsuccess.

Holly J. McCammon is Associate Professor ofSociology at Vanderbilt University. Her researchinterests concern social movements, and politi-cal and legal change. In addition to studying thepolitical successes of the suffragists, she is inter-ested in the mobilization and strategy choices ofthese movements. Her previous work has focusedon the ability of various women's groups to bring


about changes in women's protective legislationand on the labor movement's use of legal mobili-zation as a movement strategy.

Karen E. Campbell is Associate Professor of So-ciology at Vanderbilt University. Her work withHolly McCammon on U.S. women's suffragemovements reflects her broad interest in genderinequality. She is also investigating popular ex-planations of the gender gaps in wages and childcare responsibilities, and studying how statesvary in their regulation of nurse practitioners. Inaddition, she has published articles on genderdifferences in workplace and neighborhood-based networks.

Ellen M. Granberg is a Ph.D. candidate in theDepartment of Sociology at Vanderbilt Univer-sity. Her interests include historical methods, so-cial psychology, deviance, and identity. She iscurrently completing a dissertation on the iden-tity implications of maintaining weight loss overtime.

Christine Mowery is a Ph.D. candidate in theDepartment of Sociology at Vanderbilt Univer-sity. Her dissertation examines the formation ofthe National Woman's Party's feminist collectiveidentity. More broadly, her interests include gen-der, religion, and social movements. She is cur-rently a visiting instructor of sociology at NorthCarolina State University.

REFERENCES

Acker, Joan. 1992. “Gendered Institutions: FromSex Roles to Gendered Institutions.” Contem-porary Sociology 21:565–69.

Alexander, Thomas G. 1970. “An Experiment inProgressive Legislation: The Granting of Wo-man Suffrage in Utah in 1870.” Utah Histori-cal Quarterly 38:20–30.

Alford, Robert R. and Roger Friedland. 1985.Powers of Theory: Capitalism, the State, andDemocracy. New York: Cambridge UniversityPress.

Allison, Paul D. 1984. Event History Analysis:Regression Longitudinal Event Data. NewburyPark, CA: Sage.

Amenta, Edwin, Bruce G. Carruthers, andYvonne Zylan. 1992. “A Hero for the Aged?The Townsend Movement, the Political Me-diation Model, and U.S. Old-Age Policy,1934–1950.” American Journal of Sociology98:308–39.

Amenta, Edwin, Kathleen Dunleavy, and MaryBernstein. 1994. “ Stolen Thunder? HueyLong’ s ‘Share Our Wealth,’ Political Media-tion, and the Second New Deal.” American So-ciological Review 59:678–702.

Amenta, Edwin and Jane D. Poulsen. 1996. “So-

cial Politics in Context: The Institutional Poli-tics Theory and Social Spending at the End ofthe New Deal.” Social Forces 75:33–61.

Anthony, Susan B. and Ida Husted Harper.[1902] 1985. The History of Woman Suffrage.Vol. 4. Reprint, Salem, NH: Ayer.

Baker, Paula. 1984. “The Domestication of Poli-tics: Women and American Political Society,1780–1920.” American Historical Review 89:620–47.

Banaszak, Lee Ann. 1996. Why Movements Suc-ceed or Fail: Opportunity, Culture, and theStruggle for Woman Suffrage. Princeton, NJ:Princeton University Press.

———. 1998. “Use of the Initiative Process byWoman Suffrage Movements.” Pp. 99–114 inSocial Movements and American Political In-stitutions, edited by A. N. Costain and A. S.McFarland. Lanham, MD: Rowman and Little-field.

Beeton, Beverly. 1986. Women Vote in the West:The Woman Suffrage Movement, 1869–1896.New York: Garland.

Berman, David R. 1987. “ Male Support forWoman Suffrage: An Analysis of Voting Pat-terns in the Mountain West.” Social ScienceHistory 11:281–94.

Block, Fred. 1977. “The Ruling Class Does NotRule: Notes on the Marxist Theory of theState.” Socialist Revolution 7:6–28.

Brockett, Charles D. 1991. “The Structure of Po-litical Opportunities and Peasant Mobilizationin Central America.” Comparative Politics23:253–74.

Buechler, Steven M. 1986. The Transformationof the Woman Suffrage Movement: The Caseof Illinois, 1850–1920. New Brunswick: Rut-gers University Press.

Buenker, John D. 1971. “ The Urban PoliticalMachine and Woman Suffrage: A Study in Po-litical Adaptability.” The Historian 33:264–79.

Camhi, Jane Jerome. 1994. Women against Wo-men: American Anti-Suffragism, 1880–1920.Brooklyn, NY: Carlson.

Caruso, Virginia Ann Paganelli. 1986. “A His-tory of Woman Suffrage in Michigan.” Ph.D.dissertation, Department of English: AmericanStudies, Michigan State University, Lansing,MI.

Cashman, Sean Dennis. 1981. Prohibition: TheLie of the Land. New York: Free Press.

Catt, Carrie Chapman and Nettie Rogers Shuler.[1926] 1969. Woman Suffrage and Politics:The Inner Story of the Suffrage Movement. Re-print, Seattle, WA: University of WashingtonPress.

Cole, Judith K. 1990. “A Wide Field for Useful-ness: Women’ s Civil Status and the Evolutionof Women’ s Suffrage on the Montana Frontier,1864–1914.” American Journal of Legal His-


tory 34:262–94.Connell, R. W. 1987. Gender and Power: Soci-

ety, the Person and Sexual Politics. Stanford,CA: Stanford University Press.

———. 1990. “ The State, Gender, and SexualPolitics: Theory and Appraisal.” Theory andSociety 19:507–44.

Costain, Anne N. 1992. Inviting Women’s Rebel-lion: A Political Process Interpretation of theWomen’s Movement. Baltimore, MD: JohnsHopkins University Press.

Deane, Glenn, E. M. Beck, and Stewart E. Tol-nay. 1998. “ Incorporating Space into SocialHistories: How Spatial Processes Operate andHow We Observe Them.” International Re-view of Social History 43:57–80.

Degler, Carl N. 1980. At Odds: Women and theFamily in America from the Revolution to thePresent. New York: Oxford University Press.

Domhoff, G. William. 1998. Who Rules Amer-ica? Power and Politics in the Year 2000.Mountain View, CA: Mayfield.

Doreian, Patrick. 1981. “Estimating Linear Mod-els with Spatially Distributed Data.” Sociologi-cal Methodology 11:359–88.

DuBois, Ellen. 1975. “ The Radicalism of theWoman Suffrage Movement: Notes Towardthe Reconstruction of Nineteenth-CenturyFeminism.” Feminist Studies 3:63–71.

Dye, Nancy Schrom. 1980. As Equals and As Sis-ters: Feminism, the Labor Movement, and theWomen’s Trade Union League of New York.Columbia, MO: University of Missouri Press.

Edwards, Rebecca. 1997. Angels in the Machin-ery: Gender in American Party Politics fromthe Civil War to the Progressive Era. NewYork: Oxford University Press.

Evans, Peter B., Dietrich Rueschemeyer, andTheda Skocpol. 1985. Bringing the State BackIn. New York: Cambridge University Press.

Fleming, Sidney Howell. 1990. “Solving the Jig-saw Puzzle: One Suffrage Story at a Time.”Annals of Wyoming 62:23–74.

Flexner, Eleanor. 1975. Century of Struggle: TheWoman’s Rights Movement in the UnitedStates. Cambridge, MA: Belknap.

Fus, Dennis A. 1972. “Persuasion on the Plains:The Woman Suffrage Movement in Nebraska.”Ph.D. dissertation, Department of Speech, In-diana University, Bloomington, IN.

Gamson, William A. 1975. The Strategy of So-cial Protest. Homewood, IL: Dorsey.

———. 1988. “ Political Discourse and Collec-tive Action.” International Social MovementResearch 1:219–44.

Gamson, William A. and David S. Meyer. 1996.“ Framing Political Opportunity.” Pp. 275–90in Comparative Perspectives on Social Move-ments: Political Opportunities, MobilizingStructures, and Cultural Framings, edited by

D. McAdam, J. D. McCarthy, and M. N. Zald.New York: Cambridge University Press.

Gelb, Joyce. 1987. “Social Movement ‘Success’ :A Comparative Analysis of Feminism in theUnited States and the United Kingdom.” Pp.267–89 in The Women’s Movements in theUnited States and Western Europe, edited byM. F. Katzenstein and C. M. Mueller. Phila-delphia, PA: Temple University Press.

Giele, Janet Zollinger. 1995. Two Paths to Wo-men’s Equality: Temperance, Suffrage, and theOrigins of Modern Feminism. New York:Twayne.

Giugni, Marco G. 1998. “Was It Worth the Ef-fort? The Outcomes and Consequences of So-cial Movements.” Annual Review of Sociology98:371–93.

Giugni, Marco G., Doug McAdam, and CharlesTilly, eds. 1998. From Contention to Democ-racy. Lanham, MD: Rowman and Littlefield.

Goodwin, Jeff and James M. Jasper. 1999.“ Caught in a Winding, Snarling Vine: TheStructural Bias of Political Process Theory.”Sociological Forum 14:27–54.

Hackett, David G. 1995. “Gender and Religionin American Culture, 1870–1930.” Religionand American Culture: A Journal of Interpre-tation 5:127–57.

Harper, Ida Husted. 1908. The Life and Work ofSusan B. Anthony. Vol. 3. Indianapolis, IN:Hollenbeck.

———. [1922] 1985. The History of Woman Suf-frage. Vol. 6. Reprint, Salem, NH: Ayer.

Ivie, Mattie L. 1971. “Woman Suffrage in Okla-homa, 1890–1918.” Master’ s thesis, OklahomaState University, Stillwater, OK.

Jenkins, J. Craig and Bert Klandermans, eds.1995. The Politics of Social Protest: Compara-tive Perspectives on States and Social Move-ments. Minneapolis, MN: University of Min-nesota Press.

Jensen, Billie Barnes. 1973. “Colorado WomanSuffrage Campaigns of the 1870s.” Journal ofthe West 12:254–71.

Kitschelt, Herbert P. 1986. “ Political Opportu-nity Structures and Political Protest: Anti-Nuclear Movements in Four Democracies.”British Journal of Political Science 16:57–85.

Klandermans, Bert. 1989. “ Introduction: Organi-zational Effectiveness.” International SocialMovement Research 2:383–94.

Koopmans, Ruud. 1999. “Political. Opportunity.Structure. Some Splitting to Balance theLumping.” Sociological Forum 14:93–105.

Kraditor, Aileen S. 1965. The Ideas of the Wo-man Suffrage Movement, 1890–1920. NewYork: Columbia University Press.

Kriesi, Hanspeter. 1995. “The Political Opportu-nity Structure of New Social Movements: ItsImpact on Their Mobilization.” Pp. 167–98 in


The Politics of Social Protest: ComparativePerspectives on States and Social Movements,edited by J. C. Jenkins and B. Klandermans.Minneapolis, MN: University of MinnesotaPress.

Krippendorff, Klaus. 1980. Content Analysis: AnIntroduction to Its Methodology. Beverly Hills,CA: Sage.

Larson, T. A. 1965. “Woman Suffrage in Wyo-ming.” Pacific Northwest Quarterly 56:57–66.

———. 1971a. “Emancipating the West’ s Dolls,Vassals, and Hopeless Drudges: The Origins ofWomen Suffrage in the West.” Pp. 1–16 in Es-says in Western History in Honor of ProfessorT. A. Larson, vol. 37, edited by R. Daniels.Laramie, WY: University of Wyoming Publi-cations

———. 1971b. “ Woman Suffrage in WesternAmerica.” Utah Historical Quarterly 38:7–19.

Lerner, Elinor. 1981. “ Immigrant and WorkingClass Involvement in the New York City Wo-man Suffrage Movement, 1905–1917: A Studyin Progressive Era Politics.” Ph.D. disserta-tion, Department of Sociology, University ofCalifornia, Berkeley, CA.

Lorber, Judith. 1992. “ Gender.” Pp. 748–54 inEncyclopedia of Sociology, vol. 2, edited by E.F. Borgatta and M. L. Borgatta. New York:Macmillan.

———. 1994. Paradoxes of Gender. New Haven,CT: Yale University Press.

MacKinnon, Catharine. 1989. Toward a FeministTheory of the State. Cambridge, MA: HarvardUniversity Press.

Mahoney, Joseph F. 1969. “Woman Suffrage andthe Urban Masses.” New Jersey History 87:151–72.

Marilley, Suzanne M. 1996. Woman Suffrage andthe Origins of Liberal Feminism in the UnitedStates, 1820–1920. Cambridge, MA: HarvardUniversity Press.

Marshall, Susan E. 1998. “ The Gender Gap inVoting Behavior: Evidence from a Referen-dum on Woman Suffrage.” Research in Politi-cal Sociology 8:189–207.

Marshall, T. H. 1950. Citizenship and SocialClass. Cambridge, England: Cambridge Uni-versity Press.

Mason, Martha Sprague. 1928. Parents andTeachers: A Survey of Organized Cooperationof Home, School, and Community. Boston,MA: Ginn and Company.

Matthews, Glenna. 1992. The Rise of PublicWoman: Woman’s Power and Woman’s Placein the United States, 1630–1970. New York:Oxford University Press.

McAdam, Doug. 1982. Political Process and theDevelopment of Black Insurgency, 1930–1970.Chicago, IL: University of Chicago Press.

———. 1996. “ Conceptual Origins, Current

Problems, Future Directions.” Pp. 23–40 inComparative Perspectives on Social Move-ments: Political Opportunities, MobilizingStructures, and Cultural Framings, edited byD. McAdam, J. D. McCarthy, and M. N. Zald.1977. New York: Cambridge University Press.

McAdam, Doug, John D. McCarthy, and MayerN. Zald. 1988. “Social Movements.” Pp. 695–737 in The Handbook of Sociology, edited byN. J. Smelser. Newbury Park, CA: Sage.

McCammon, Holly J. 1995. “The Politics of Pro-tection: State Minimum Wage and MaximumHours Laws for Women in the United States,1870–1930.” Sociological Quarterly 36:217–49.

McCammon, Holly J. and Karen E. Campbell.2001. “Winning the Vote in the West: The Po-litical Successes of the Women’ s SuffrageMovement, 1866–1919.” Gender and Society15:56–83.

McCarthy, John and Mayer Zald. 1977. “ Re-source Mobilization and Social Movements: APartial Theory.” American Journal of Sociol-ogy 48:1212–41.

McDonagh, Eileen L. and H. Douglas Price.1985. “ Woman Suffrage in the ProgressiveEra: Patterns of Opposition and Support inReferenda Voting, 1910–1918.” American Po-litical Science Review 79:415–35.

McDonald, David K. 1987. “Organizing Wom-anhood: Women’ s Culture and the Politics ofWoman Suffrage in New York State, 1865–1917.” Ph.D. dissertation, Department of His-tory, State University of New York, StonyBrook, NY.

Morgan, David. 1972. Suffragists and Demo-crats: The Politics of Woman Suffrage inAmerica. East Lansing, MI: Michigan StateUniversity Press.

Morris, Aldon D. 1993. “Birmingham Confron-tation Reconsidered: An Analysis of the Dy-namics and Tactics of Mobilization.” Ameri-can Sociological Review 58:621–36.

Moynihan, Ruth Barnes. 1983. Rebel for Rights:Abigail Scott Duniway. New Haven, CT: YaleUniversity Press.

Nathan, Maud. 1926. The Story of an Epoch-Making Movement. Garden City, NJ: Double-day, Page, and Co.

National American Woman Suffrage Association(NAWSA). 1893–1917, 1919. The Hand Bookof the National American Woman Suffrage As-sociation and Proceedings of the Annual Con-vention. New York: National American Wo-man Suffrage Association.

———. 1940. Victory: How Women Won It. NewYork: H. W. Wilson.

National Women’ s Trade Union League. 1911,1915, 1917, 1919, 1922. Proceedings of theBiennial Convention of the National Women’s


Trade Union League of America. Chicago, IL:William C. Faehse.

Ornstein, Norman J. and Shirley Elder. 1978. In-terest Groups, Lobbying and Policymaking.Washington, DC: Congressional QuarterlyPress.

Piven, Frances Fox and Richard A. Cloward.1977. Poor People’s Movements: Why TheySucceed, How They Fail. New York: Vintage.

Quadagno, Jill. 1992. “ Social Movements andState Transformation: Labor Unions and Ra-cial Conflict in the War on Poverty.” Ameri-can Sociological Review 57:616–34.

———. 1994. The Color of Welfare: How Rac-ism Undermined the War on Poverty. NewYork: Oxford University Press.

Rucht, Dieter. 1996. “ The Impact of NationalContexts on Social Movement Structures: ACross-Movement and Cross-National Com-parison.” Pp. 185–204 in Comparative Per-spectives on Social Movements: Political Op-portunities, Mobilizing Structures, and Cul-tural Framings, edited by D. McAdam, J. D.McCarthy, and M. N. Zald. New York: Cam-bridge University Press.

Rueschemeyer, Dietrich and Peter B. Evans.1985. Pp. 44–77 in Bringing the State Back In,edited by P. B. Evans, D. Rueschemeyer, andT. Skocpol. New York: Cambridge UniversityPress.

Scott, Joan W. 1986. “ Gender: A Useful Cat-egory of Historical Analysis.” American His-torical Review 91:1053–75.

Sinclair, Andrew. 1965. The Better Half: TheEmancipation of the American Woman. NewYork: Harper and Row.

Skocpol, Theda. 1980. “ Political Response toCapitalist Crisis: Neo-Marxist Theories of theState and the Case of the New Deal.” Politicsand Society 10:155–201.

———. 1992. Protecting Soldiers and Mothers:The Political Origins of Social Policy in theUnited States. Cambridge, MA: Belknap.

Smith, Ann Warren. 1975. “Anne Martin and aHistory of Woman Suffrage in Nevada, 1869–1914.” Ph.D. dissertation, Department of His-tory, University of Nevada, Reno, NV.

Snow, David A., E. Burke Rochford Jr., StevenK. Worden, and Robert D. Benford. 1986.“ Frame Alignment Processes, Micromobili-zation, and Movement Participation.” Ameri-can Sociological Review 51:464–81.

Staggenborg, Suzanne. 1995. “Can Feminist Or-ganizations Be Effective?” Pp. 339–55 inFeminist Organizations: Harvest of the NewWomen’s Movement, edited by M. M. Ferreeand P. Y. Martin. Philadelphia, PA: TempleUniversity Press.

Stanton, Elizabeth Cady, Susan B. Anthony, andMatilda Joslyn Gage, eds. [1886] 1985. His-tory of Woman Suffrage. Vol. 3. Reprint, Sa-lem, NH: Ayer.

Tarrow, Sidney. 1998. Power in Movement: So-cial Movements and Contenious Politics. 2ded. New York: Cambridge University Press.

Taylor, Verta. 1999. “Gender and Social Move-ments: Gender Processes in Women’ s Self-Help Movement.” Gender and Society 13:8–33.

Trout, Grace Wilbur. 1920. “Side Lights on Illi-nois Suffrage History.” Journal of the IllinoisHistorical Society 13:145–79.

Turner, Frederick Jackson. 1972. “ The Signifi-cance of the Frontier in American History.”Pp. 1–18 in The Turner Thesis: Concerningthe Role of the Frontier in American History,edited by G. R. Taylor. Lexington, MA:Heath.

Turner, Ralph H. 1970. “Determinants of SocialMovement Strategies.” Pp. 145–64 in HumanNature and Collective Behavior: Papers inHonor of Herbert Blumer, edited by T. Shib-utani. Englewood Cliffs, NJ: Prentice-Hall.

U. S. Bureau of the Census. 1872, 1883, 1897,1902, 1914, 1923. Census of the United States.Population. Washington, DC: GovernmentPrinting Office.

U. S. Department of Commerce. 1919, 1920,1922, 1923. Statistical Abstract of the UnitedStates. Washington, DC: Government PrintingOffice.

U. S. Office of Education. 1872–1900, 1902–1914, 1916, 1917. Annual Report of the Com-missioner of Education. Washington, DC:Government Printing Office.

Welter, Barbara. 1966. “The Cult of True Wom-anhood: 1820–1860.” American Quarterly18:151–74.

White, Jean Bickmore. 1974. “Woman’ s Place Isin the Constitution: The Struggle for EqualRights in Utah in 1895.” Utah HistoricalQuarterly 42:344–69.

Yamaguchi, Kazuo. 1991. Event History Analy-sis. Newbury Park, CA: Sage.

SOCIAL MOVEMENTS AND POLICY IMPLEMENTATIONSOCIAL MOVEMENTS AND POLICY IMPLEMENTATIONSOCIAL MOVEMENTS AND POLICY IMPLEMENTATIONSOCIAL MOVEMENTS AND POLICY IMPLEMENTATIONSOCIAL MOVEMENTS AND POLICY IMPLEMENTATION 7 17 17 17 17 1


Social Movements and Policy

Implementation: The Mississippi

Civil Rights Movement and the

War on Poverty, 1965 to 1971

Kenneth T. Andrews

Harvard University

This study of the Mississippi civil rights movement and the War on Poverty examinesthe relationship between social movements and policy implementation. A “movementinfrastructure” model is developed that focuses on organizational structure, re-sources, and leadership to account for the impact of social movements on policyimplementation. A two-tiered research design is employed that includes (1) a quanti-tative analysis of poverty programs in Mississippi counties from 1965 to 1971, and(2) case studies that show the complex interaction between the civil rights move-ment, resistance by whites, local powerholders, and federal agencies. The quantita-tive analysis shows that counties with strong movement infrastructures generatedgreater funding for Community Action Programs. The case studies show that move-ments were excluded from the initial formation of these programs as local whitesattempted to preempt civil rights activists. However, in counties with strong move-ment infrastructures, activists were able to gain access to decision-making bodiesand shape the content of poverty programs.

and 1980s, but the study of movement out-comes did not. . . . [The result is] that westill know very little about the impact of so-cial movements on social change” (p. 276).Furthermore, the question of movement im-pact addresses one of the most importantconcerns of movement participants—the ef-ficacy of social movements.

I have two major objectives in this paper.First, I provide a conceptual framework foranalyzing movement outcomes. Most dis-cussions focus on the analytic problems ofestablishing whether movements createchange, but how movements generate politi-cal change must also be examined. I identifyand compare the major theoretical modelsused to explain the relationship betweenmovements and political change. I argue thatour understanding of the influence of socialmovements will be greatly improved by de-lineating models that specify how move-ments generate institutional change(McAdam and Snow 1997). I propose a“ movement infrastructure” model that fo-

ocial movement scholars agree thatthe question of a social movement’s

impact on political change is important andunderstudied. Over the past four decades,leading scholars have reviewed the relevantliterature on social movements and havenoted the limited amount of systematic re-search on outcomes (Diani 1997; Eckstein1965; Giugni 1998; Marx and Woods 1975;McAdam, McCarthy, and Zald 1988; Tarrow1998). Burstein, Einwohner, and Hollander(1995) observe that, “ the field of socialmovements grew tremendously in the 1970s

S

Direct all correspondence to Kenneth T.Andrews, Department of Sociology, HarvardUniversity, Cambridge, MA 02138 ([email protected]). I have benefited from com-ments and criticisms by Michael Biggs, MarshallGanz, Peter Marsden, Ziad Munson, Shuva Paul,Gerald Platt, Michael Schwartz, Alex Trillo,Mary Vogel, Charlie Zicari, and Bob Zussman.Some of the data collection was supported by anNSF Doctoral Dissertation Improvement Grant(9625597).


cuses on the organizational structure, re-sources, and leadership of a movement toexplain its impact on the political process.Second, I present an extensive analysis ofthe effects of the civil rights movement inMississippi on the implementation of pov-erty programs at the local level. I investigatewhether local movements directly and indi-rectly shaped the implementation of federalpolicy in Mississippi. First, a quantitativeanalysis of poverty program funding exam-ines the impact of movement organization,white countermobilization, social, political,and economic factors on funding from 1965to 1971. Two case studies follow that assessthe impact of local civil rights movementson the form and content of poverty programsin their communities.

CONCEPTUALIZING

MOVEMENT OUTCOMES

Outcomes as Changes in

Political Institutions

I focus on outcomes rather than success. Re-cent research has identified methodologicaland theoretical problems with studying suc-cess (Amenta and Young 1999; Giugni1998). Success implies the attainment of spe-cific, widely shared goals, but the goals ofmost social movements are contested by par-ticipants and observers. Goals also changeover the course of a movement. Studyingoutcomes avoids these problems and allowsscholars to focus on unintended and negativeconsequences as well as successes.

The analysis here pertains to politicalmovements and institutional outcomes in thepolitical arena. Political movements involvea sustained challenge to existing power rela-tions, and they employ disruptive, nonrou-tine tactics that publicly challenge the distri-bution and uses of power in the broader so-ciety (Gamson 1990; McAdam 1982;Schwartz 1976; Tilly 1978). This focus ex-cludes movements focused on changes inter-nal to a group and its members. Because po-litical movements also directly or indirectlymake claims on the state, I focus on institu-tional outcomes. Typically, political move-ments attempt to build organizations andchange the culture and consciousness oftheir members or the broader public. In fact,

a movement’s impact on institutions oftendepends on its ability to build organizationsand shape collective identities (Mueller1987). These movements, however, seekchange in political institutions, and thosechanges may take a variety of forms such as:(1) gaining access to the decision-makingprocess, (2) altering an institution’s goalsand priorities, (3) securing favorable poli-cies, (4) insuring that those policies areimplemented, or (5) shifting the distributionof institutional resources to benefit themovement’s constituents (Burstein et al.1995; Gamson 1990; Kreisi et al. 1995;Schumaker 1975).1 Overall, a focus on insti-tutional outcomes makes sense because itencompasses the long-term goals of manysocial movements. In addition, a focus oninstitutional outcomes has a methodologicaladvantage because in many cases these out-comes are more easily measured than cul-tural, attitudinal, and psychological out-comes. Political outcomes provide an impor-tant indicator of “ the results of [the civilrights] movement in the lives of blacksoutherners” (Button 1989:4).

Opportunity Structures,

Institutional Arenas,

and Key Actors

Political process theories note that the emer-gence of social movements is patterned bybroad changes in the “political opportunitystructure” (McAdam 1982). This observationpoints to one of the methodological chal-lenges for research on movement outcomes:If changes in the opportunity structure facili-tate the emergence of a social movement,then those same changes may account for theapparent impact of a movement (Amenta,Dunleavy, and Bernstein 1994). The impor-tance of opportunity structures has been es-tablished, but few scholars would argue thatthey have a singular and deterministic effecton social movements (Goldstone 1980;Kitschelt 1986). Rather, the emergence andmaintenance of a social movement is in partattributable to the internal dynamics of the

1 Movements can also influence “ reactive” out-comes such as preventing a policy that woulddamage the movement or its constituents (Kriesiet al. 1995).


movement itself. In addition, some scholarsargue that the impacts of movements on op-portunity structures should be studied. Forexample, McAdam (1996) notes that “ ourcollective failure to undertake any seriousaccounting of the effect of past movementson . . . political opportunities is as puzzlingas it is lamentable” (p. 36).

Any analysis of movement outcomesmust examine the structure and strategies ofthe relevant exogenous political actors andinstitutions. Movements make claims thatdirectly or indirectly impinge on othergroups. Thus, movements have complex andsometimes unexpected relationships withother groups that become allies or oppo-nents. They also mobilize within institu-tional settings that structure conflict andpossible outcomes. For example, federalagencies are constrained by their relation-ship to Congress and public opinion(Burstein 1999). These rules and resourcesshape the possible responses of state actorsto social movements.

Measuring Outcomes Over Time

For methodological and conceptual reasons,I need to measure multiple outcomes and tomeasure outcomes over time (Andrews1997; Banaszak 1996; Button 1989; Snyderand Kelly 1979). Movement outcomes overtime must be measured because movementschange their tactics and goals. For example,Katzenstein (1990) finds that feminist activ-ists in 1973 organized around the issue ofordination, but by 1983 the movement hadbroadened its analysis and goals to include“ running shelters for homeless women; do-ing prison work; organizing in the sanctuarymovement; joining in protests against US in-tervention in Central America; running em-powerment workshops, lesbian retreats, andconferences to build bridges betweenwomen religious and laywomen” (p. 41; alsosee Katzenstein 1998). Another reason formeasuring outcomes over time is that theform and degree of influence may vary overtime (Andrews 1997). By focusing on amovement’s immediate impact the move-ment’s influence could be over- or underes-timated.

In sum, analyzing movement outcomes in-volves: (1) examining different forms of po-

litical change (e.g., access, policy enact-ment, implementation), (2) analyzing oppor-tunity structures, institutional arenas and keyactors that shape movement dynamics, (3)incorporating temporal processes by measur-ing outcomes over time.

FOUR DIVERGENT VIEWS ON

MOVEMENTS AND OUTCOMES

Studies of the impact of social movementshave typically focused on the question ofwhether movements exert influence. In thosecases for which one can identify the influ-ence of movements on institutional changeindependent of other nonmovement factors,a second set of questions must be answered.First, the causal argument must be specified.What characteristics of a movement ormovement activity account for the impact?Second, the mechanisms of influence mustbe revealed. What is the process or mecha-nism by which a movement influences a po-litical institution? There are several promi-nent answers to these questions.

Analyses of movement outcomes will beimproved by systematically comparing andelaborating these contending models. In myview, no single model can account for theways movements generate change. Thisview stems from the variety of cases and po-litical contexts that have been studied as “ so-cial movements.” Nevertheless, there is arelatively limited set of possibilities, and ourunderstanding of movement impacts will beimproved by specifying those models as“ ideal types.” Scholars often operate with animplicit model that remains undertheorized.Elaborating these models allows researchersto ask how particular cases diverge from thetheoretical models. Most important, compar-ing different models can direct scholarshiptoward broader questions about variationacross movements and political contexts.2

I delineate four major approaches to therelationship between movements and out-comes.3 Each model singles out key ele-

2 For example, Piven and Cloward’ s (1977)explicit focus on “ poor people’ s movements”suggests that class composition is a key variable(also see Ragin 1989).

3 I focus on theories that explicitly examine themovement/outcome relationship. Other than my


ments that account for a movement’s impact,and each implies different mechanismsthrough which movements can exert influ-ence. These distinct ways of thinking aboutmovement impact are rarely made explicit orcontrasted with one another in sociologicalresearch. By explicating each, I aim toclarify the lines of debate in the field andplace my research within that debate.

Action-Reaction Models:

Disruption or Persuasion

In the first two models, which I call “action-reaction” models, mobilization has the mo-mentary potential to leverage changethrough its impact on political elites, elec-toral coalitions, or public opinion. Withinthe action-reaction approach, theorists de-scribe two possible routes whereby move-ments are influential.

In one route, movements are dramatic, dis-ruptive and threatening to elites, whichprompts a rapid response—typically eitherconcessions and/or repression. Piven andCloward (1977) have been the primary pro-ponents of this view arguing that “ the mostuseful way to think about the effectivenessof protest is to examine the disruptive effectson institutions of different forms of mass de-fiance, and then to examine the political re-verberations of those disruptions” (p. 24).For Piven and Cloward (1977), it is not clearthat protest has an independent impact be-cause it “wells up in response to momentouschanges in the institutional order. It is notcreated by organizers and leaders” (p. 36).Protest is one link in a sequence, and oncethe sequence is initiated protesters have littlecontrol over the policy response. The au-thors conclude that “ whatever influencelower-class groups occasionally exert inAmerican politics does not result from orga-nization, but from mass protest and the dis-ruptive consequences of protest” (Piven andCloward 1977:36).

Organizations, particularly mass-basedmembership organizations, are doomed tofailure because powerless groups can nevermobilize as effectively as dominant groupsin a society. As a result, organization canonly lessen the disruptive capacity and effi-cacy of protest (Piven and Cloward 1984,1992; also see Gamson and Schmeidler1984; Morris 1984). Elite reaction is ulti-mately focused in a self-interested way onending protest. Analyzing urban policychanges in the 1960s, Katznelson (1981) ar-gues that “ the targets of these public poli-cies were not objects of compassion, but offear born of uncertainty” (p. 3). Policy-makers caught off guard by protest, attemptto quickly assemble a strategy of repression,concessions, or a combination of the twothat will end the protest wave (Tarrow1993). Disruption models focus on the limi-tations of protest on policymaking beyondthe agenda-setting stage.

In the second version of the action-reac-tion model, movements are dramatic andgenerate support from sympathetic third par-ties that take up the cause of the movement.The intervening role of “ third parties,” “by-stander publics,” or “ conscience constitu-ents” is critical. In a classic essay, Lipsky(1968) argues that “ the ‘problem of the pow-erless’ in protest activity is to activate ‘ thirdparties’ to enter the implicit or explicit bar-gaining arena in ways favorable to protest-ers” (p. 1145). Lipsky claims that “ if protesttactics are not considered significant by themedia . . . protest organizations will not suc-ceed. Like the tree falling unheard in the for-est, there is no protest unless protest is per-ceived and projected” (p. 1151; also seeBenford and Hunt 1992). 4

Garrow (1978) argues that civil rightscampaigns, especially in Selma, Alabama,generated momentum for the 1965 VotingRights Act. For some theorists, repression isan intervening link. For example, Garrow

brief discussion of political opportunity structure,I do not focus on theories of political-institu-tional change including (1) pluralist or interestgroup theories, (2) state-centric theories, or (3)elite theories. Some scholars have contrastedthese theories with “movement theories” of po-litical change (Amenta, Caruthers, and Zylan1992; Quadagno 1992).

4 These models of movement influence areconnected to methodological strategies. For ex-ample, Rucht and Neidhardt (1998) argue thatmedia reported protest is a meaningful barometerof all protest: “ Insofar as we are interested inthose protests which are an input for the politicalsystem, media reported protests have a highervalidity than the whole range of actual protests”(p. 76).


argues that attacks by southern officials oncivil rights activists further solidified thesupport of bystanders. Burstein (1985)shows that the movement did not reverse thedirection of public opinion arguing thatmovements are probably unable to have sucha substantial impact on opinion. Rather, pro-test increased the salience of the civil rightsissue, and political representatives were ableto act on those louder and clearer signals(Burstein 1999). In this view protest is aform of communication, and persuasion isthe major way that movements influencepolicy (Mansbridge 1994).

These two versions of the action-reactionmodel differ: The first emphasizes disrup-tive and often violent action forcing a re-sponse from political elites; the second pro-poses that protest can mobilize sympatheticthird parties that advance the movement’sagenda by exerting influence on politicalelites. But both versions of the action-reac-tion model share the assumption that (1)large-scale dramatic events shape the pro-cess of change by, (2) mobilizing morepowerful actors to advance the movement’scause, and (3) that (implicitly) movementshave little or no direct influence beyondthis initial point. In both versions, the pri-mary focus is on public protest eventsrather than on organizations.

Access-Influence Model:

Routinization of Protest

The third major approach argues that the de-terminant of movement efficacy is the ac-quisition of routine access to the politythrough institutionalized tactics. This ap-proach typically describes a drift toward lessdisruptive tactics such as electoral politics,coalitions, lobbying and litigation. Organi-zation and leadership figure prominently inthis model. Organizational changes parallelthe tactical shift including increasing cen-tralization and bureaucratization of move-ment organizations. In short, social move-ment organizations evolve into interestgroups. In the “access-influence” model, theorganizational and tactical shifts are accom-panied by an increase in influence over rel-evant policy arenas. In contrast, the action-reaction model would predict that move-ment influence declines as tactics become

routinized and organizations become incor-porated. Most important, the access-influ-ence model argues that disruptive tacticshave little independent impact on institu-tional change. In their study of the impactsof black and Hispanic political mobilizationon a variety of policy outcomes, Browning,Marshall, and Tabb (1984) argue that protestand electoral strategies were used togethereffectively, but “ demand-protest strategiesby themselves produced limited results inmost cities” (p. 246).

Access-influence models also assert thatsecuring insider status is more consequentialthan pursuing a single, specific policy objec-tive. Rochon and Mazmanian (1993) arguethat the antinuclear movement, by advocat-ing a single piece of legislation, was unsuc-cessful. In contrast, the environmentalmovement, especially antitoxic groups, at-tempted to become a legitimate participantin the regulatory process. By gaining access,the movement has been able to have a sub-stantial, long-term impact on policy (also seeCostain 1981; Sabatier 1975).

The access-influence model has fewerproponents within the movement literaturethan the action-reaction models. However,the notion that “ routine” tactics are most ef-ficacious is consistent with pluralist theoriesof democracy that view the political systemas relatively open to citizen influence. In thismodel, organization-building (especiallyprofessionalization, bureaucratization, andcentralization) provides movements with thenecessary tools to operate in the interestgroup system where bargaining is the keymechanism of influence.

The Movement

Infrastructure Model

Finally, I propose a “movement infrastruc-ture” model. Three components of a move-ment’s infrastructure must be examined toexplain its influence on the policy process:leadership, organizational structure, and re-sources. Infrastructures that allow the move-ment to employ multiple mechanisms of in-fluence (including disruption, persuasion,and bargaining) will have the greatest impacton policy implementation. At a general level,the autonomy and continuity of the infra-structure are key factors explaining the long-


term viability and impact of the movement,sustaining a movement through shifts in thebroader political environment (Andrews1997; Rupp and Taylor 1990). A strongmovement infrastructure can spur politicalelites to initiate policy concessions in re-sponse to the perceived threat of the move-ment. That threat rests on the belief that amovement has the capacity to institute moresubstantial change through parallel, autono-mous institutions.

Leaders and organizations must be embed-ded in indigenous, informal networks. Suchlinks make leaders more responsive to theirconstituency and less easily co-opted (Mor-ris 1984). Robnett (1996) distinguishes be-tween formal leaders (e.g., ministers) and anintermediate layer of “bridge leaders,” whostand at nodal points within the informal net-works of a community. This type of leader-ship structure can generate ongoing tensionwithin a movement. However, it also canprovide advantages, such as innovation(Stepan-Norris and Zeitlin 1995). A differ-entiated leadership structure allows for com-munication to various audiences includingparticipants, potential recruits, opponents,and state actors (Klandermans 1997). A lead-ership structure with a diversity of skills andexperiences will be better able to use mass-based tactics as well as routine negotiationwith outside groups (Ganz 2000; Gerlachand Hine 1970).

The critical role of preexisting organiza-tion and resources has been established in theemergence of social movements. To persistover time, movements must forge new orga-nizational forms and establish independentresource flows (McAdam 1982; Schwartz1976). In the mobilization process, the infor-mal structure of relationships among activ-ists and organizations must be expansiveacross communities and subgroups. In thepolicymaking process, formal organizationsbecome a necessary vehicle for advancing agroup’s claims. Organizational structures canalter the routine operation of the politicalprocess when they are perceived as legiti-mate and/or threatening by established po-litical actors (Clemens 1997; Gamson 1990).

Movements that rely primarily on the“mobilization of people” rather than on fi-nancial resources are more likely to continueusing protest tactics (Schwartz and Paul

1992). As a result, their strategic and tacti-cal options are broader (Ganz 2000). Ulti-mately, movements require substantial con-tributions of volunteer labor to maintain or-ganizations and launch protest campaigns.This is seen most clearly at the local levelwhere movement organizations are lesslikely to maintain a paid, professional staff.

In the movement infrastructure model,strategy and tactics depend on a movement’sleadership, organization, and resources. Thiscontrasts with the action-reaction model thateither views protest and organization in con-flict with one another or pays little attentionto organization. Strategy and tactics are con-ceptualized broadly in the infrastructuremodel and range from protest to the buildingof counter-institutions.

In sum, strong movement infrastructureshave diverse leaders and a complex leader-ship structure, multiple organizations, infor-mal ties that cross geographic and socialboundaries, and a resource base that drawssubstantially on contributions from theirmembers for both labor and money. Thesecharacteristics provide movements withgreater flexibility that allows them to influ-ence the policy process through multiplemechanisms.

COMPARING THE MODELS

The movement infrastructure model buildson the insights of the prior three models.First, it assumes, like the action-reactionmodels, that there are key moments whenmovements can be especially efficacious.Further, it assumes that disruptive tactics areimportant for movements to have an impact,especially when disruptive tactics are cre-atively injected into routine political pro-cesses. The movement infrastructure modeldiffers from the others because it emphasizesthe building and sustaining of movement in-frastructures as an important determinant ofthe long-term impact of these movements (incontrast to short-term impacts, like agenda-setting). Furthermore, unlike the access-in-fluence model, these organizations have thegreatest impact when they maintain theirability to use both “outsider” and “ insider”tactics. Litigation, lobbying, and electoralpolitics can be effectively employed by so-cial movements. However, movements lose


key opportunities for leverage in the politi-cal process when they quickly adopt the tac-tics of “ interest groups” and abandon “ insur-gent” tactics.

Movements must be able to create lever-age through multiple mechanisms. The priorthree models focus on a single mechanismas the primary means by which movementscreate change (e.g., disruption, persuasion,or negotiation). The movement infrastruc-ture model accounts for the ability of move-ments to impact political change throughmultiple mechanisms, and this change canoccur when a movement’s leadership and or-ganization allow for strategic flexibility.

The pattern of outcomes for a movementmay depend on processes described by eachof these models. For example, both action-reaction models focus on agenda-setting asthe primary outcome that movements can in-fluence. In contrast, access-influence andmovement infrastructure models examinelater stages in the policymaking process. Ul-timately, researchers should use these mod-els to compare across different types of so-cial movements and political contexts. Theanalysis I present here demonstrates the util-ity of the movement infrastructure model asapplied to the Mississippi case.

RESEARCH DESIGN

The War on Poverty as an Outcome

The War on Poverty created a new set of op-portunities and constraints for the civil rightsmovement. These programs brought substan-tial resources into impoverished communi-ties, providing opportunities for blacks to in-fluence the shape and direction of policy. Atfirst glance, it is surprising how thoroughlylocal movements became involved in the Waron Poverty. After all, the publicly stated goalof the movement in the early 1960s was gain-ing access to electoral politics. However, anunderlying objective of the movement inMississippi was building local movementsthat could define and pursue their own goals(Payne 1995). The early movement organi-zations were not directly involved in the Waron Poverty. Nevertheless, local movementscontinued to operate in the post-1965 periodand attempted to shape the local implemen-tation of poverty programs. Many local ac-

tivists defined economic empowerment as anatural outgrowth of the political empower-ment pursued through voter registration. Infact, many believed that political powerwould be meaningless unless black commu-nities could generate viable economic pro-grams (Dorsey 1977).

There were several obstacles to movementinfluence. First, the objectives of federalagencies constrained the ability of localmovements to direct the War on Poverty. The“professionalization of reform” could reducethe participation and influence of the poorto a primarily symbolic role (Helfgot 1974;also see Friedland 1976). In addition, the ad-ministration of poverty programs requirednegotiations with many community groups,some of which were potential allies or oppo-nents of civil rights activists. While move-ment mobilization shaped the distributionand development of antipoverty programs inMississippi, these programs also shaped thedirection of local movements.5 Once the Waron Poverty was initiated, local movements inMississippi and across the country attemptedto secure resources and shape programs(Patterson 1994:146). Quadagno (1994)notes that a “crucial linkage . . . unquestion-ably did develop between the civil rightsmovement and the War on Poverty” (p. 28).

The poverty programs in Mississippi canbe examined as an outcome of the civilrights movement for four main reasons: (1)the poverty programs and the civil rightsmovement both targeted an overlappingarena of activity, (2) there was substantialand ongoing interactions between civil rightsactivists and the Office of Economic Oppor-tunity, (3) the programs provided benefits tothe movement’s primary constituency(blacks in the South), and (4) there is sig-nificant variation across states and countiesin local actors’ influence on the programs.

Study Design: Quantitative and

Qualitative Analyses

Mississippi is an important case for examin-ing the long-term impacts of the civil rights

5 I do not analyze the impacts of poverty pro-grams on the civil rights movement, e.g., whetherthe programs co-opted the movement (seeEisinger 1979).


movement. The state is widely known for itsinstitutionalization of the “ tripartite systemof domination”—a term Morris (1984) hasused to describe the political, economic, andpersonal bases of racial inequality in theU.S. South. On one hand, Mississippi can beviewed as a test case where the movementmet its most intense resistance. At the sametime, there is substantial variation within thestate across key variables: movement mobi-lization, countermovement, structural char-acteristics, and the implementation of pov-erty programs.

Follow Amenta’s (1991) suggestion of ana-lyzing subunits, I use counties as the unit ofanalysis to strengthen the theoretical valueof the study. This focus has substantive meritbecause the Mississippi movement targetedcounties as areas within which to organize.In addition, counties are the most importantlocal political unit in the South (Krane andShaffer 1992). Finally, poverty programswere instituted in Mississippi across coun-ties rather than across municipalities.

The research here combines two comple-mentary strategies: (1) a quantitative analy-sis of Mississippi counties that allows forprecise estimates of the distribution of pro-grams and funding, and (2) qualitative evi-dence from case studies using interview andarchival data. Most previous research on theWar on Poverty has focused on urban areas,riots, and the distribution of poverty pro-grams (Button 1978; Fording 1997). Beyondsingle case studies, few scholars have exam-ined the impacts of social movement pro-cesses on poverty programs. In my quantita-tive analysis, I ask whether movements hadan impact on poverty programs independentof other relevant factors.6 After establishingthat movements did have an impact on pov-

erty program funding, I use case studies toexamine the processes and form of conflictat the local level (i.e., the mechanismsthrough which local movement organiza-tions shaped the development of povertyprograms).

The primary sources are the records ofmovement organizations and informationfrom the Office of Economic Opportunity.These sources provide data on the key ac-tors, their activities, and their analyses of thepolitical landscape. For the case studies,written records are supplemented with par-ticipant interviews from published and un-published collections.

THE CIVIL RIGHTS MOVEMENT

AND THE WAR ON POVERTY:

NATIONAL AND STATE CONTEXTS

The National Context of the

War on Poverty 7

On August 20, 1964, President LyndonJohnson signed the Economic OpportunityAct, a key component of his Great Societyagenda. The initiation of the War on Povertycoincided with a set of national policy ini-tiatives of the early 1960s, including the1964 Civil Rights Act and the 1965 VotingRights Act—legislation that altered the po-litical context of the civil rights movement.The War on Poverty included a cluster ofprograms administered primarily through thenewly formed Office of Economic Opportu-nity (OEO). (Table 1 provides a list of acro-nyms used throughout this paper.) The Waron Poverty lacked a unified approach con-ceptually and administratively. For example,the 1964 legislation included plans forNeighborhood Youth Corps, Community Ac-tion Programs, Head Start, Volunteers in Ser-vice to America (VISTA), and the collegework-study program (Patterson 1994).Through the 1960s, OEO administered themajority of these programs, allowing themto bypass old-line agencies like the Depart-ment of Labor and local or state agencies.Over time, however, the major poverty pro-grams were phased out or shifted over to the

6 The data set is drawn from a larger study thatincludes measures of the civil rights movement,local countermobilization, contextual variables,federal intervention, and other outcomes. Thisdata set includes all Mississippi counties exceptHinds County, which includes Jackson, the capi-tol of Mississippi. The large size of HindsCounty makes it an outlier in some analyses. Inaddition, Jackson served as the organizationalcenter for state-level activities. My interest is inthe local forms of mobilization, and in HindsCounty these cannot be distinguished from state-level mobilization.

7 For overviews see Friedman (1977),Patterson (1994), Piven and Cloward (1993), andQuadagno (1994).


more conservative agencies, and in 1973OEO was eliminated (Quadagno 1994).

Among the various poverty programs, theCommunity Action Program (CAP) receivedthe greatest attention and became almostsynonymous with the War on Poverty.Policymakers pushing “ community actionhoped to stimulate better coordinationamong the melange of public and privateagencies delivering social services”(Peterson and Greenstone 1977:241). Thisobjective, however, was abandoned in favorof “ citizen participation.” OEO and localCAPs had little impact on the establishedagencies providing services to poor commu-nities. As a result, CAPs administered manyof the new antipoverty programs. CAPs werecoordinated at the local level through a CAPBoard that served as the overarching admin-istrative body and provided a point of poten-tial access for local movements. This open-ing paved the way for intense conflicts be-tween local groups attempting to gain accessto CAP boards in order to influence the flowof OEO funds.

The Mississippi Civil Rights

Movement

In Mississippi, the Student Nonviolent Co-ordinating Committee (SNCC) began devel-oping community projects in the early 1960saround voter registration (Carson 1981;Dittmer 1994; Payne 1995). These earlyprojects linked the small network of indig-enous NAACP leaders and an emerginggroup of grassroots leaders exemplified byAnnie Devine, Fannie Lou Hamer, andVictoria Gray (Payne 1995). Civil rightsprojects met intense repression across thestate from local law enforcement and localwhites. SNCC’s early efforts were expandedduring the 1964 Freedom Summer projectthat brought college students from across thecountry into the local movements. Two fea-tures of this early period stand out: (1) theintensity of white resistance and (2) the fo-cus on building local community organiza-tions and leaders.

Following Freedom Summer, the newlyformed Mississippi Freedom DemocraticParty (MFDP) challenged the all-white Mis-sissippi delegation to the Democratic Na-tional Convention in Atlantic City. This is

often portrayed as the final chapter of theMississippi movement as national attentionshifted away from the southern movementfollowing passage of the 1965 Voting RightsAct. However, key struggles took place atthe state and local levels concerning theimplementation of voting rights and socialpolicies. Both the NAACP and MFDP con-tinued to pursue a civil rights agenda after1965 in Mississippi. The period followingthe Atlantic City convention was marked byincreasing conflict between the two domi-nant organizations. In the electoral arena,both organizations supported candidates inlocal and state elections. Local branches ofboth organizations pursued school desegre-gation, organized boycotts and demonstra-tions, and pushed for expanded poverty pro-grams in their communities (McLemore1971; Parker 1990).

The Child Development Group of

Mississippi: Early Involvement in

the War on Poverty

From their origins, poverty programs in Mis-sissippi were closely tied to the dynamics ofthe civil rights movement. One of the earli-est and most celebrated programs, the ChildDevelopment Group of Mississippi (CDGM),administered Head Start centers across the

Table 1. List of Acronyms and Organizations

Acronym Organization

ACBC Associated Communities of BolivarCounty

CAP Community Action Program

CDGM Child Development Group ofMississippi

CMI Central Mississippi, Inc.

COFO Council of Federated Organizations

CORE Congress of Racial Equality

MFDP Mississippi Freedom DemocraticParty

NAACP National Association for theAdvancement of Colored People

OEO Office of Economic Opportunity

SNCC Student Nonviolent CoordinatingCommittee


state building directly on the movement’sbase of Freedom Schools and communitycenters. CDGM provided an entry point foractivists into the War on Poverty (Greenberg1969).

CDGM was formed by a small group ofpolicymakers and psychologists with looseconnections to the Mississippi movement.For example, Tom Levin, the first directorof the program, had participated in FreedomSummer through the Medical Committeefor Human Rights, a group providing medi-cal assistance to local projects. Despitethese ties, when proposals for CDGM werecirculated in early 1965, the response fromSNCC and MFDP’s state-level leadershipwas one of skepticism and opposition(Payne 1995). Many movement leaderswere suspicious of the federal governmentand the initiatives of white liberals follow-ing the challenge at Atlantic City (Dittmer1994). Thus, the state-level civil rights or-ganizations made little effort to supportCDGM.

Nevertheless, CDGM quickly diffusedthrough the local movement infrastructure.In April 1965, CDGM held its first statewidemeeting to begin developing the organiza-tion for the upcoming summer. At the firstmeeting, representatives from 20 communi-ties attended. By the second meeting in themiddle of April, that number had increasedto 64 (Greenberg 1969:18, 22). For the firstsummer, Payne (1995) reports that “on open-ing day of the eight-week session, eighty-four centers opened across the state, servingfifty-six hundred children” (p. 329). Green-berg (1969), the OEO staff person respon-sible for CDGM, claims that “CDGM stoodon the shoulders of COFO and its compan-ion projects which were active the preced-ing summer” (p. 28).

Holmes County illustrates the relation-ship that developed between the civil rightsmovement and CDGM at the local level. Aninspection during the second year of theprogram found that 102 of the 108 staffmembers in Holmes County were activemembers of MFDP. Reflecting the strengthof the local movement, the investigationfound that “ many of the Negroes in thecommunities around the centers have do-nated money and time to build buildings forthe centers and work with the programs”

(NA, RG 381, Box 108, July 30, 1966).8

Bernice Johnson, who worked with CDGMin Holmes County, recalled that communitycenters were used in the daytime for HeadStart and at night for the MFDP (BerniceJohnson, interviewed by author, June 20,1996). The same core groups of activistsparticipated in both activities. Investiga-tions across the state showed that CDGMstaff were affiliated with COFO, SNCC,CORE, NAACP, the Urban League, theDelta Ministry, and MFDP (NA, RG 381,Box 108, July 5, 1966).

The strong relationship between localmovements and CDGM made the Head Startprogram a target of opposition, includingviolence. The primary resistance came frominfluential Mississippi politicians, includingSenator James Eastland, who chaired the Ju-diciary Committee, and Senator JohnStennis, who chaired the AppropriationsCommittee. The opposition to CDGM reso-nated with growing fear from around thecountry that the War on Poverty was fund-ing black insurgency (Quadagno 1994).

CDGM acquired its second grant for the1966 summer after a massive mobilizationincluding a demonstration in which “ forty-eight black children and their teachers turnedthe hearing room of the House Education andLabor Committee into a kindergarten”(Dittmer 1994:375). After this, CDGM wasfunded at 5.6 million dollars. In response,

Governor Johnson and his allies came to seethat by setting up CAP agencies in Missis-sippi communities, local whites could pre-vent the flow of federal dollars into pro-grams like CDGM. Under continuing attackfrom segregationists, OEO was eager to rec-ognize any CAP agency in Mississippi, re-gardless of its composition. (Dittmer 1994:375)

This tactical shift is remarkable—that Mis-sissippi politicians opposed to federal anti-poverty programs would come to embracethem must be attributed to the threat posedby the Mississippi civil rights movement.9

8 Complete citations for archival material arelisted in the bibliography under ArchivalSources. “ NA” indicates the National Archivesand Records Administration, and “RG indicatesthe Record Group.

9 This opposition to federal intervention wasspecific to programs that would benefit black


OEO undermined the viability of CDGMby stipulating that in counties with a CAP,Head Start must be administered through thelocal CAP agency rather than a specialized,statewide program like CDGM. TurningHead Start over to CAPs gave local agenciesa high profile in the community. The policyalso undermined the movement’s control ofHead Start in Mississippi. OEO realized thatthis would shift the attention of movementactivists toward local CAPs. In November1966, OEO’s southeast regional directorwrote to OEO director Sargent Shriver ex-plaining that “CDGM . . . had a large num-ber of local poor people involved or hired.These same people can be expected to be-come involved in local CA[P] activities astheir concern or experience warrants” (NA,RG 381, Box 2, November 8, 1966). Thisbecame the main battleground as activists at-tempted to shape Community Action Pro-grams in Mississippi.

MEASURES AND MODELS:

THE FORMATION AND

FUNDING OF CAPS

Community Action Programs became thecentral component of the War on Poverty.Did local movements in Mississippi shapethe formation and funding of CAPs? If so,in what ways did they influence CAPs? Ianalyze the funding of Community ActionPrograms during two phases, the initial de-velopment phase from 1965 to 1968 and thelater phase of declining resources from 1969to 1971.10 The dependent variable is the

amount of CAP funding for each period. Theindependent variables include measures ofthe civil rights movement (black mobiliza-tion), white resistance to the movement(countermobilization), and local characteris-tics of the county (political and socioeco-nomic variables). (See Appendix Table A fora list of variables, descriptions, means andstandard deviations.)

Black mobilization is measured by threevariables. MFDP staff in 1965 and NAACPmembership in 1963 distinguishes betweenthe effects of the militant (MFDP) and mod-erate (NAACP) wings of the Mississippicivil rights movement. I measure black elec-toral mobilization by the number of blackcandidates running for office in 1967. Fewblack candidates won in these initial elec-tions following the Voting Rights Act. How-ever, the variable indicates the early consoli-dation of organizations and networks fo-cused on electoral politics.

Countermobilization by whites is mea-sured by three variables: incidents of violentresistance during Freedom Summer, thepresence of a Citizens’ Council organization,and the presence of a Ku Klux Klan organi-zation in the county.11 The formation of aCommunity Action Program required somesupport and participation from local whites,typically from the County Board of Supervi-sors. Hence, the areas that had most stronglyresisted the civil rights movement should bethe least likely to seek out or support federalprograms. In some counties, for example, lo-cal whites became targets of repression ifthey met with civil rights groups (Dittmer1994; Harris 1982).

Political characteristics of the county areexamined in terms of the political orienta-tion of the electorate and the organizationalcapacity of the local government. The parti-san loyalty of a county’s electorate is mea-sured by the percentage of votes cast forLyndon Johnson in 1964. Higher levels ofDemocratic loyalty may have been rewarded

Mississippians. Cobb (1990) notes that “ Deltaplanters were skilled in the pursuit and manipu-lation of federal assistance long before the NewDeal,” including flood control programs andcrop-reduction subsidies (p. 914).

10 The two dependent variables are the totalCAP grants for 1965–1968 and 1969–1971 (NA,RG 381, Box 14, n.d.). There were 18 CAPs inMississippi from 1965 to 1971 of which 8 weremulticounty agencies. I used two different strate-gies for estimating county-level expenditures formulticounty agencies. First, I divided the budgetevenly among the counties covered by the CAP.For the second estimation, I divided the budgetamong the counties proportional to the numberof households in each county with an income be-low $3,000 per year. The two estimates produced

similar results; I report the analysis using the“ proportional” estimates. Because participationin programs was based on economic eligibility,this strategy is a better, if not perfect, approxi-mation of the distribution.

11 These three variables are not highly corre-lated, and thus I treat them as distinct modalitiesof resistance.


with higher levels of funding. In addition, Iexamine the possible influence of local po-litical institutions on program implementa-tion by including the proportion of the laborforce employed in local government in 1964.I expect that counties with large political in-stitutions will be more likely to seek outpoverty program funding because of theirgreater organizational capacity (Mazmanianand Sabatier 1983).

Socioeconomic characteristics that mightinfluence the formation and funding ofCAPs include the local class structure, thelevel of poverty, and the population size. Iexamine the local class structure using threedifferent indicators: (1) the proportion of thelabor force employed in manufacturing, (2)the proportion of the labor force employedas professionals, and (3) landowner concen-tration for commercial farms. Measures ofclass structure are often used in studies ofthe policy process. James (1988) finds thatmanufacturing is a key component of thesouthern class structure that influences thelevel of racial inequality in political partici-pation. Hence, I expect manufacturing tohave a negative impact on poverty programfunding. Professionals were potential sup-porters of poverty programs, so I expect theproportion employed as professionals tohave a positive impact on CAP funding. Themeasure of landholding concentration esti-mates the predominance of the traditionalplantation economy. Roscigno andTomaskovic-Devey (1994) find that a simi-lar indicator is an important determinant oflocal political outcomes in North Carolina.The expected direction of the relationshipwith this variable is unclear: While southernplanters historically had opposed extensionsof the welfare state system into the localeconomy, the mechanization of farming co-incided with the rise of the civil rights move-ment and the initiation of the War on Pov-erty. This left many farm laborers unem-ployed, and poverty programs could havebeen viewed as a viable strategy for address-ing the social and economic consequences oftechnological change (Cobb 1990).

To measure poverty, I use the proportionof households with incomes below $3,000per year. In these models, using householdsor individuals produces similar results be-cause they are highly correlated (r = .994). I

also include a variable measuring the totalnumber of households. To measure povertyand the number of households, I use datafrom the 1960 census (rather than 1970) be-cause OEO would have used these data atthe time. (Analyses using the 1970 data pro-vide similar results.)12

The initial models were estimated usingOLS regression. However, in the final mod-els I conduct an additional test using a “ spa-tial error” model, which tests for spatial de-pendence in the model that can result fromthe geographic proximity of the units ofanalysis (see Amenta et al. 1994; Gould1991). The presence of spatial dependencecan lead to inflated significance tests(Anselin 1992; Doreian 1980). Theautocorrelation term in both models is sta-tistically significant. However, the profile ofresults for the remaining independent vari-ables is similar to that in the OLS models.13

RESULTS

Black Mobilization and Community

Action Programs

Table 2 shows that the measures of blackmobilization play an important role in thefunding of Community Action Programsduring both periods: The MFDP has a statis-

12 Similarly, I tested several measures of pov-erty, such as the number of households with in-come below $2,000 and $1,000 in these models.Each indicator produced comparable results. Thechosen indicator, households earning less than$3,000, is the closest approximation of the fed-eral poverty line. CAP grant applications re-quired that applicants list the percentage ofhouseholds earning below $3,000. When OEOinvestigated the composition of CAP boards orprogram employees to determine whether therewas sufficient representation of “ the poor,” thiswas the indicator they used. The indicator be-came so widely used that the movement em-ployed it, and in 1966, MFDP sent out a call fora statewide meeting of all persons earning below$3,000. Studies by political scientists and soci-ologists measuring poverty with 1960 data havealso employed this same indicator (e.g., Colby1985; Cowart 1969; Friedland 1976).

13 The independent variables that were signifi-cant in the OLS models remain significant, withthe exception of the presence of Ku Klux Klanorganizations.


tically significant effect in both periods, theNAACP variable is significant for 1965–1968 but not for 1969–1971, and the mea-sure of black electoral organization is posi-tive and statistically significant for 1969–1971.14

The models underscore the influence atthe local level of the more militant organiza-tions. Quadagno (1994) argues that “ OEO

promoted black moderates at the expense ofmore militant civil rights activists” (p. 43).Certainly, OEO attempted to do this. Butthese analyses show that local movements,especially militant groups, promoted the ex-pansion of OEO programs.

Next I discuss the remaining variables inthe CAP models. Then I provide an extendeddiscussion of movement influence. I arguethat we must look further to determinewhether movement activists, moderate ormilitant, played a direct role in the adminis-tration of CAPs. I draw on two case studiesof Mississippi counties to examine howmovements shaped poverty program forma-tion and funding.

Table 2. Unstandaradized Coefficients from the Maximum-Likelihood Regression of CommunityAction Program Funding (in $100,000s) on Selected Independent Variables: MississippiCounties, 1965–1968 and 1969–1971

CAP Grants CAP GrantsIndependent Variables 1965–1968 (S.E.) 1969–1971 (S.E.)

Black MobilizationMFDP membership, 1965 1.942*** (.506) 1.109*** (.243)

NAACP membership, 1963 (logged) 1.646** (.715) .451 (.350)

Number of black candidates, 1967 .— .566** (.211)

CountermobilizationViolent resistance during Freedom Summer –1.96*** (.577) –1.240*** (.283)

Citizens’ Council organization in county, 1956 –1.657 (2.401) –.657 (1.105)

Ku Klux Klan organization in county, 1964 –2.622 (2.324) –1.707 (1.116)

Political CharacteristicsPercentage voting for Lyndon Johnson, 1964 –.011 (.199) .—

Proportion employed in local government, 1964 690.229* (327.970) 415.988** (156.610)

Socioeconomic CharacteristicsProportion employed in manufacturing –.025 (.205) –.090 (.093)

Proportion professionals .431 (.374) .131 (.176)

Landowner concentration 5.251 (7.432) .107 (3.357)

Poverty, 1959 (proportion of households 43.769* (25.408) 11.447 (12.129) earning less than $3,000)

Total number of households, 1960 (in 1,000s) .907* (.424) .760*** (.206)

Spatial autocorrelation (λ) .551*** (.114) .383** (.136)

Constant –53.083* (26.425) –18.618 (12.416)

Fit .422 .562

Maximized log-likelihood (LIK) –1,226.3 –1,165.3

Akaike Information Criterion (AIC) 2,478.6 2,356.6

Note: Numbers in parentheses are standard errors. “ Fit” measures the squared correlation between thepredicted and observed values (Anselin 1992). Number of counties = 81.

*p < .05 **p < .01 ***p < .001 (one-tailed tests [except landowner concentration; see text])

14 The black candidate variable does not ac-count for the declining importance of NAACPmembership. Models omitting the black candi-date variable show a similar profile of coeffi-cients, and the NAACP variable remains nonsig-nificant.


Countermobilization:

Repression and Poverty Programs

The negative coefficients for violent resis-tance during Freedom Summer are statisti-cally significant in both models. Less CAPfunding went to those locales that had beensites of the most militant resistance to thecivil rights movement. One white leader inCoahoma County articulated the commonview that “ if the white leaders did not be-come involved then the alternative was moreFederal intervention with the county’s anti-poverty program being turned over to theNegroes” (Mosley and Williams 1967:8).Most counties had some white leaders whoshared this view, but they did not prevail incounties that had high levels of violent re-pression. Local white moderates were thetargets of white violence in some counties,but in counties that were relatively less re-pressive, moderate white leaders steppedforward to form poverty programs.

This interpretation is supported by evi-dence from the case studies and broaderhistorical material on the civil rights move-ment (see Cunnigen 1987; Jacoway andColburn 1982). In those cases in which localwhites supported the civil rights movement,they were often singled out for repression.For example, during Freedom Summer, theHeffners, a white couple, met with civilrights activists in their home in McComb.After this meeting, the Heffners were intimi-dated until they left the state (Dittmer 1994;Harris 1982). This type of repression wasnot limited to Freedom Summer. In 1966, forexample, a white Head Start teacher inPanola County “ received threatening phonecalls. On July 16, a letter was distributedaround the city of Batesville. It was signedKKK, listed some of the white teachers andaides working in the program and said theywould be given just one more opportunity toget out on their own. . . . As a result of thethreat, four white aides left the Head Startprogram” (NA, RG 381, Box 110, July 17,1966). OEO field reports and CDGMrecords document similar efforts to limitwhite support for the movement and the pov-erty programs.

Movement scholars often argue that re-pression has a negative impact on amovement’s ability to achieve its objectives

(Gamson 1990).15 This can occur when re-pression undermines the organizational ca-pacity of the movement, but in this case, Iargue that a different process is operating—repression diminished CAP funding by sup-pressing the mobilization of other groups.16

The Local Context:

Political Variables and Poverty

The pattern reported in Table 2 indicates thatpoverty was significantly associated withhigh levels of CAP funding during the firstperiod only. County size (measured by thenumber of households) also has a positiveeffect in both models. The proportion em-ployed in local government has a positiveand statistically significant relationship topoverty program funding in both periods.Partisanship and social class measures donot show statistically significant effects onpoverty program funding.

Most CAPs were initiated in the earlyyears following the 1964 legislation. As bud-get cuts were made through the late 1960s,the funding of new grants was minimal.OEO’s broad guideline was to make reduc-tions of “ approximately equal percentage”while allowing room for administrative dis-cretion (NA, RG 381, Box 2, October 14,1966). However, Table 2 reveals some im-portant shifts, including the declining role ofpoverty and the increasing role of countysize (measured by number of households).Overall, the results reported in Table 2 indi-cate that there was some continuity in thefunding of CAPs.

Local Movements and

Community Action Programs

The key finding from the regression models,then, is the significant positive impact of

15 The relationship between repression andprotest has received considerable attention in re-cent years and has been strongly influenced byTilly’ s (1978) early analysis (Koopmans 1997;Lichbach 1987; Rasler 1996). In contrast, little isknown about the impact of repression on out-comes.

16 I conducted a third case study that shedslight on this process. In Madison County, violentrepression endured longer than it did in most ofMississippi. Only one effort was made to estab-


black mobilization on the funding of CAPs.However, this finding is consistent with dif-ferent interpretations. One possibility is thatlocal movements were directly involved inthe formation of CAPs. However, anotherpossibility is that movements posed a threatthat mobilized other groups in the county todevelop poverty programs. These scenarioscorrespond to Gamson’s (1990) concepts ofsuccess and preemption: Success occurswhen movements gain access to the poli-cymaking process and generate substantivegains; movements are preempted when sub-stantive gains are achieved without access tothe policymaking process.17

The regression equations do not indicatewhich of the two interpretations apply inMississippi. The case studies show that thepattern was more complex. Initially move-ments were preempted, and this was followedby long struggles with varying degrees ofsuccess to achieve access to the policymak-ing process. Movements gained influence byemploying multiple strategies such as disrup-tive protest, negotiation with OEO officials,and administering independent poverty pro-grams. In short, the movement infrastructurein the community shaped the extent and formof influence that was ultimately achieved.

In 1965 and 1966, Community Action Pro-grams were formed without substantial par-ticipation from movement activists. Blackparticipation often involved traditional lead-ers not affiliated with the civil rights move-ments (neither the moderate NAACP nor themore militant MFDP representatives), suchas ministers and teachers. OEO was, in fact,aware of what it called the “Tom” problem.In early 1966, the southeast regional man-ager of CAP reported that in Mississippi

. . . the most frequent problem and the onewhich requires the most time in its solutionis representation. Boards on original submis-sion are almost always hand picked and

packed in favor of the Governor. Negro rep-resentation is always ‘Tom.’ . . . Protests al-most always follow the selection of such ini-tial Boards and resolution generally takesfrom 3 to 4 months. (NA, RG 381, Box 2,February 24, 1966)18

Even though they were aware of the prob-lem, OEO’s grant administrators often didnot have detailed information about the lo-cal situation and lacked “ the technical com-petence necessary to help with Board prob-lems” (NA, RG 381, Box 2, February 24,1966). This problem was particularly acutein the early years. During this period, OEOdepended on local movements to act as“whistle-blowers.”

THE CASE STUDIES

The Community Action Programs in Holmesand Bolivar Counties were formed with littledirect involvement from activists. However,this changed as each movement attempted toinfluence local CAPs. The cases differ withregard to the specific strategies deployed bylocal movements and the way that localelites responded to those efforts. In HolmesCounty, activists were able to secure posi-tions and influence within the CAP adminis-tration and staff. In Bolivar County, activ-ists used a variety of tactics to establish anindependent poverty program that operatedalongside the local CAP.

Holmes County

In the early 1960s, Holmes County devel-oped one of the most successful local civilrights movements in Mississippi (MacLeod1991; Payne 1995). The movement devel-oped an infrastructure with broad leadership,multiple organizations, indigenous re-sources, and strategic flexibility. A coregroup of activists emerged in the small com-munity of Mileston. Bernice Johnson, one ofthe first activists from the eastern part of thelish a CAP, and it was unsuccessful (Madison

County Herald, “CAP Meeting Saturday,” April27, 1967, p. 1).

17 There were some rare cases in whichNAACP leaders and liberal whites formed coali-tions at the local level—this occurred, for ex-ample, in Coahoma County. However, these coa-litions were stronger in state-level organizationslike the Loyalist Democrats (Dittmer 1994;Simpson 1982).

18 Governors could veto poverty programs un-less they were administered through a college oruniversity. Some programs, such as CDGM, wereadministered through historically black collegesto avoid the veto. Other programs were spon-sored by universities outside the South, such asthe Tufts Delta Health Center in Bolivar County.


county, remembers the diffusion process asfollows:

Well, they were constantly trying to get newmembers. I remember when I first startedgoing to Mileston, I encouraged the peoplein the community where I lived (which wasSunny Mount) to start having a meeting. . . .We were constantly going from communityto community, from church to church, ask-ing people to allow us to come into yourchurch. . . . “Set up a community meeting.Elect you some officers—a president, a sec-retary, a treasurer or what have you—desig-nate a certain time for your communitymeeting.” (Rural Organizing and CulturalCenter 1991, p. 70)

By 1964, most of the small communities inthe county had held meetings sponsored bythe MFDP that culminated in a monthlycountywide meeting (Mississippi Depart-ment of Archives and History, MFDPRecords, Reel 3, n.d.). Sue Lorenzi, a com-munity organizer, reported weekly meetingsin 15 different communities in 1966 (StateHistorical Society of Wisconsin [SHSW],Alvin Oderman Papers, August 27, 1966).

The movement infrastructure includedmultiple venues for leadership development.Salamon (1971:440), who conducted fieldresearch in Holmes County in 1969, esti-mated that there were approximately 800 for-mal leadership positions in movement orga-nizations held by 600 different individuals.19

Financial resources were modest. How-ever, they were derived from local activitiesincluding collections at monthly meetings,plate dinners, and set donations fromchurches of, for example, $100 a year. TheFDP office was sustained by local collec-tions—in 1966, “over $500 was raised . . . forits phone, rent, lights, some supplies” (SHSW,Alvin Oderman Papers, August 27, 1966).While the vast majority of resources weregenerated internally in the form of labor, themovement periodically employed outsidehelp from sources like legal aid organizationsor national civil rights organizations.

Ed Brown, one of the early SNCC work-ers in Holmes County, described the local

movement

. . . as opposed to placing the emphasis onconfrontational politics we had placed theemphasis on organizing so that in the in-stances where there were confrontationsthere was sufficient organizational strengthbehind it to make the whites think, youknow, twice before doing anything.(Tougaloo College, Tom Dent Collection,July 2, 1979)

The Holmes County movement was aloosely coordinated confederation of move-ments across the county that expanded therepertoire of skills at the local level andbrought local activists into contact with stateand national politics.

Initial efforts to form a CAP in HolmesCounty bypassed the strong movement infra-structure. In the fall of 1965, a committeeappointed by the Board of Supervisors be-gan plans to join Central Mississippi, Inc.(CMI), a multicounty CAP. OEO’s SoutheastRegional Office was skeptical of CMI’s ini-tial proposal. Bob Westgate, an OEO staffmember, noted that

. . . although there are three Negroes on eachof the [five], seven member county boards,I have my doubts of their real value to theirpeople, whether they were really “elected”by their people, and suggest that they shouldbe checked by someone from this office. Atleast eight of the 15 Negro members are de-pendent upon the white power structure fortheir jobs or welfare pension payments (fiveprincipals or teachers, two on welfare andone maid). (NA, RG 381, Box 5, December11, 1965)

Westgate sought information through COREand NAACP contacts, but neither organiza-tion could provide contacts because they didnot have organizations in the counties.Originally, CMI had submitted a proposalreporting that 25 percent of the populationwas black, but OEO required an increase inthe number of “ minority representatives”when it discovered that the population in thesix counties was actually 58 percent black(NA, RG 381, Box 5, December 11, 1965).

OEO was also concerned that “eight of the20 white board members are ‘ Johnson colo-nels’—men who contributed funds and sup-port during Governor Johnson’s campaign”(NA, RG 381, Box 5, December 11, 1965).The governor exercised considerable power

19 The 1960 census reports 19,488 black per-sons (71.9 percent of the total population) livingin Holmes County (U.S. Bureau of the Census1963).


over CAPs because he had to sign off ongrants and the organization’s charter. WithCMI, Johnson “allegedly held up the sign-ing of the charter until these eight [support-ers] were appointed on the board.” Thepresident and vice president of the CMIboard were Johnson loyalists, and they hadstrong ties to the local political structure. Forexample, Ringold, the president, was the at-torney for the Board of Supervisors (themost powerful local political body in Mis-sissippi) in Montgomery County (NA, RG381, Box 5, December 11, 1965).

Because the formation of CMI occurredoutside the public arena, it could not be con-tested by local activists. Daisy Lewis, direc-tor of the Holmes County Community Cen-ter, observed that “CAP came into HolmesCounty unexpected before the poor Negroand poor white had the chance to take partin it or decide if it would help our county ornot. . .” (Tougaloo College, Ed King Papers,Box 11, 1966). A group of approximately 40white leaders held a planning meeting inFebruary 1966 to coordinate efforts. TheLexington Advertiser reported that “ leaderswere told that they have a choice of thecounty conducting it’s own anti-poverty pro-gram and ‘ taking the Negroes along with us’or not acting and have the ‘Negroes and civilrights workers’ take over” (“ Anti-PovertyProgram Discussed,” February 24, 1966, p.1). Despite being caught off guard, themovement quickly mobilized to participatein the program. On March 7, a public meet-ing was held with approximately 500 blacksand 30 whites in attendance (Lexington Ad-vertiser, “ Holmes CAP Advisory Group,”March 10, 1966, p. 1). Activists brought aseries of demands including the dissolutionof the existing board. A compromise wasreached in which six additional memberswere elected to a temporary advisory com-mittee. Other changes were made, includingthe election of a 31-member permanent ad-visory committee that would elect a six-member Board of Directors. In addition,each Head Start center would elect a sepa-rate advisory committee. Because the smallcommunities throughout the county were al-ready organized, the movement could electa majority to the advisory committee and in-fluence key policy decisions of the Commu-nity Action Program (Salamon 1971).

The Holmes County movement thus re-structured the organization of poverty pro-grams during the course of a single meeting.These policies ensured a high level of move-ment participation in future program imple-mentation. By securing access to the admin-istration of CAP, the civil rights movementwas able to maintain control of Head Startcenters through an independent, delegateagency. In addition, CAP initiated severalprojects that went beyond job training to ad-dress rural poverty in Holmes County. Whilethe poverty programs provided services,they also provided jobs—the programs con-stituted the single largest employer inHolmes County (Salamon 1971).

Bolivar County

In the mid-1960s, the Bolivar County move-ment was weaker than that in HolmesCounty. Community organizers had beguncampaigns in some towns (e.g., Shaw), butseveral communities had no movement ac-tivity. The movement was held together by aloose network of activists, but it did not havethe regular meetings, diverse organizations,or comprehensive presence that HolmesCounty did. Nevertheless, civil rights activ-ists mobilized a successful, widespread cam-paign to secure an independent, parallel pro-gram. This campaign became a major ve-hicle for building a movement infrastructurein Bolivar County.

As in Holmes County, the initial plans fora CAP occurred without movement partici-pation. The Bolivar County Community Ac-tion Committee was formed in 1965 withkey support from local elites including theBoard of Supervisors and the Chamber ofCommerce. As editor of the Bolivar-Com-mercial and President of the Chamber ofCommerce, Cliff Langford provided consid-erable support for the program. From its be-ginning, local activists criticized the pro-gram for excluding movement participationand appointing conservative blacks to theCAP board. As was the case in many CDGMcounties, mobilization crystallized in early1966 when local leaders in Bolivar Countylearned that Head Start could no longer beadministered through CDGM. Consistentwith its new policy, OEO recommended thatthe Head Start program be shifted to the lo-


cal CAP. The CDGM group formed a localorganization called the Associated Commu-nities of Bolivar County (ACBC). A cam-paign was launched that simultaneously at-tacked the local CAP for excluding move-ment activists and demanded the continua-tion of Head Start through the establishedCDGM program. A similar strategy was usedin Sunflower County (Mills 1993). Blackmembers of the CAP board were singled outas “Toms” appointed by the “power struc-ture.” The CDGM group was outraged thatone of the black ministers appointed to theCAP board had denied CDGM access to sev-eral churches in 1965 (SHSW, Amzie MoorePapers, Box 2, n.d.b and Box 3, n.d.).

The primary leader of the challenge wasAmzie Moore, one of the early NAACPleaders in Mississippi. However, the leader-ship included a large number of local minis-ters (from churches in which Head Startcenters operated) and the staff from HeadStart programs throughout the county.These efforts also received support from theDelta Ministry and MFDP. The challengecould quickly mobilize throughout the orga-nizational infrastructure that had been usedto operate Head Start. The local movement,calling itself the “Committee of the Poor inBolivar County,” held mass meetings, cir-culated a petition, and operated the CDGMcenters for approximately 1,200 children ona volunteer basis through the spring of 1966(NA, RG 381, Box 40, March 17, 1966;SHSW, Amzie Moore Papers, Box 2, Janu-ary 19, 1966). The volunteer programsdemonstrated the commitment of the localmovement and posed an ongoing challengeto the legitimacy of the funded project inthe county. One OEO investigator notedthat the petition “ is a forceful and dramaticexpression of the feelings of these people ofBolivar County. It does show that there is agood deal of organization at the grass rootslevel” (NA, RG 381, Box 40, March 3,1966). In addition to the local activities,leaders went to Washington, D.C. to lobbyOEO to maintain the CDGM-based programin Bolivar County.20

The Bolivar County CAP tried to respondto charges that its board was unrepresenta-tive by holding open meetings at the locallevel to discuss program objectives and con-solidate support. These meetings providedan opportunity for representatives of theCDGM-based group to publicly criticize theCAP board and build support for their chal-lenge (SHSW, Amzie Moore Papers, Box 2,n.d.a; SHSW, Amzie Moore Papers, Box 2,March 13, 1966). These events culminatedin a meeting between CAP and the CDGMgroup in March at which the Bolivar CAPvoted down a proposal to transfer funds tothe CDGM group and allow it to administeran independent program. This forced OEOto make a decision regarding the two groups(SHSW, Amzie Moore Papers, Box 2, 1966;SHSW, Amzie Moore Papers, Box 2, March22, 1966).

OEO was initially opposed to having par-allel organizations and favored a reorgani-zation of the existing CAP board. Summa-rizing an extensive investigation, an OEOreport emphasized that “ it is crucially im-portant that the Bolivar County CommunityAction Committee be given every consider-ation for funding” (NA, RG 381, Box 40,1965). Despite initial support of the localCAP, Bill Seward concluded his investiga-tion for OEO that

. . . although representing less than a thirdof the Negro population, [the CDGM group]is a potent and vocal force that must be rec-ognized and included in any further OEOprograms. . . . [F]urther postponement [offunding] will raise the level of emotionaldiscontent of the Negro/poor from one offrustration, channeled into constructive ef-fort, to one of frustration resulting in overtdemonstration. In other words, there hadbetter be a Head Start and quick before thelid blows. (NA, RG 381, Box 40, March 17,1966)

This analysis led Seward to recommend di-viding the funds evenly between CDGM andthe local CAP (NA, RG 381, Box 40, March31, 1966; U.S. Senate 1967). In April, thiswas the compromise that OEO reached inBolivar County—two separate Head Start

20 Although there is no precise estimate of themovement’ s size, the “ Outline of ImportantEvents” cited above reports “ approximately7,000 signatures” on the petition (SHSW, Amzie

Moore Papers, Box 2, n.d.b; for a copy of the pe-tition see SHSW, Amzie Moore Papers, Box 2,n.d.c).


programs with separate staffs and adminis-trations were funded (SHSW, Amzie MoorePapers, Box 3, April 14, 1966).

Despite initial opposition, support withinOEO increased for the movement-basedprogram which had applied for funding asthe Associated Communities of BolivarCounty (ACBC). A 1967 report noted that“ preliminary evaluations indicate that theACBC programs are probably better thanthe CA[P]’s.” Even though the BolivarCounty CAP was making efforts to sub-sume ACBC within its program, OEO rep-resentatives in Mississippi stated that “ourposition will be to support and maintainACBC as a separate entity” (NA, RG 381,Box 5, January 13, 1967). The BolivarCounty movement leveraged a responsefrom OEO because of its sustained mobili-zation using conventional and disruptivetactics. In the 1966 year-end report, thesoutheast regional director singled outBolivar County because of the “ lessening ofover-all community tensions” (NA, RG381, Box 2, December 30, 1966).

The Bolivar County movement was able touse sustained protest to secure an autono-mous poverty program. Despite initial oppo-sition, OEO officials came to see the dupli-cation of administrative staff and costs aspreferable to an ongoing challenge to theirlegitimacy in Bolivar County. The move-ment’s challenge depended on an expansivenetwork of activists that could run Head Startcenters, coordinate mass meetings, and ne-gotiate the grant-writing process with OEO.

Movements in Holmes County andBolivar County were successful at main-taining movement-controlled Head Startcenters. In addition, both movements poseda credible threat that compelled local politi-cal elites to establish well-funded Commu-nity Action Programs. However, the coun-ties differed in important respects. InHolmes County, activists achieved greaterimpact on the structure of CAP by capitaliz-ing on a strong movement infrastructure. InBolivar County, activists protected move-ment-affiliated Head Start programs butceded control to the broader CAP program.This outcome resulted from the relativelygreater opposition and the less developedinfrastructure in Bolivar County comparedwith Holmes County.

CONCLUSION

A striking finding of this study is the extentto which movements shaped the implemen-tation of local poverty programs. While thisinfluence was certainly less than local activ-ists would have desired, it was nonethelessconsiderable. The quantitative analysisshows that local movements had a positiveimpact on the amount of CAP funding inMississippi counties. The case studies sup-port my interpretation of the quantitativeevidence and show how movements influ-enced the formation of Community ActionPrograms by carving out areas of adminis-trative control.

I propose that researchers specify moreprecisely how movements shape socialpolicy. Even in this small case study, I dem-onstrate several ways that local movementsinfluenced policy implementation in Mis-sissippi, including the disruption of pro-gram operations, negotiation with agencyofficials, and symbolic and persuasive pro-test activities. The greatest influence mayhave occurred indirectly when movementsprompted local white politicians to activelypursue grants for poverty programs.

In terms of the movement-outcome mod-els, the evidence indicates that the impactsof the movement were cumulative, ratherthan momentary as suggested by the action-reaction models. The movement posed athreat, but the threat was based on the abilityof the movement to distribute federal pro-grams independent of local agencies. Localmovements used a variety of conventionaltactics, but they did not abandon the politicsof protest—marches and boycotts were orga-nized in local communities throughout thelate 1960s. Rather, movements were most in-fluential when they built local organizationsthat allowed for an oscillation between mass-based tactics and routine negotiation withagency officials.

The action-reaction models cannot accountfor the sustained interactions between localmovements and OEO officials during theimplementation of poverty programs. Duringmuch of this period, local activists and pro-gram officials collaborated to establish pov-erty programs. The access-influence modelwould suggest a drift toward greater profes-sionalization of the movement and the aban-


donment of protest tactics, but this did notoccur as the Bolivar County case illustrates.

The movement infrastructure model showshow movements exert influence throughmultiple causal mechanisms. The three mostcrucial mechanisms observed in this studyare (1) direct implementation of poverty pro-grams, (2) indirect influence by challengingthe political authority of local elites, and (3)disruptive and persuasive protest that com-pelled OEO to act on behalf of the move-ment. These forms of influence all derivefrom the organizational capacity of localmovements. Direct program implementationrequired an extensive leadership cadre thatcould maintain ongoing ties to OEO offi-cials, other programs throughout the state,and community members. In BolivarCounty, activists secured independent pro-grams over the initial opposition of OEO ad-ministrators and the local CAP. The move-ment-affiliated centers (formerly CDGM)continued to operate Head Start programs in1966 without funding, illustrating the under-lying strength of the local organization. Thesecond form of influence flowed from thefirst. Because local movements were capableof operating poverty programs indepen-dently, they undermined the authority of lo-cal officials who had historically adminis-tered social programs. Finally, local move-ments used protest, including disruptive pro-test in Bolivar County, to bring additionalpressure to bear on OEO and to mobilize na-tional support. OEO officials came to seethis as an inevitable part of the implementa-tion process in Mississippi with civil rightsgroups acting as whistle-blowers.

As an analytic strategy, I have addressedthe long-standing problems of studyingmovement outcomes, opening questionsabout the variation in movement infrastruc-tures and the ability of movements to influ-ence policy. The strategy employed is lessgeneralizable than studies based on a repre-sentative sample of social movement orga-nizations or campaigns (e.g., Gamson 1990),yet it avoids some of the problems inherentin studies analyzing multiple movements,such as limited measures of movement im-pact. While the Mississippi movement is ex-ceptional in some respects, the movementemployed organizational forms and strate-gies that are comparable to those of many

other social movements, including labormovements (Fantasia 1989; Ganz 2000), theenvironmental justice movement (Bullard1990), and many women’s movements(Ferree and Martin 1995; Whittier 1995).

One might reasonably ask whether therewere distinctive aspects of the War on Pov-erty in Mississippi that make this instance ofpolicy implementation and the findings pre-sented here unique? First, the high degree ofpublic participation required by the povertyprograms facilitated the movement’s accessto the programs while bringing the move-ment’s opponents into the implementationprocess. Second, the high level of local au-tonomy permitted in the formation and man-agement of projects allowed the movementto pursue local efforts to influence povertyprograms rather than pursue a nationalstruggle in which the movement would havehad to target federal actors, especially Con-gress and the Presidency. Local variation inpolicy implementation is common for socialpolicies in the United States (Amenta et al.1994; Banaszak 1996; Clemens 1997), butthe model might require modification to ac-commodate variation in political context(Amenta, Halfmann, and Young 1999).Third, the central role of racial politics in thedevelopment of the War on Poverty is seen,for example, in the ongoing efforts of OEOto showcase racial integration in its pro-grams (Quadagno 1994). However, this con-flict reveals dynamics of a more general na-ture in that the long-term goals of programadministrators and movement activists oftenconflict. To address these concerns in a moremeaningful way requires similar analyses ofother social movements across a variety ofpolicy arenas.

The growing body of research on move-ment outcomes calls for more explicit devel-opment of the causal arguments concerningmovement impact. I provide a preliminarymap of these arguments that allows for moresystematic, comparative research: Futurestudies of outcomes can address these issuesby (1) using quantitative analysis of out-comes across time and policy arenas, (2)giving greater attention to the process andmechanisms of impact through case studies,and (3) synthesizing across specific findingsto explain variation across movements andpolitical contexts.


Appendix Table A. Descriptions and Means and Standard Deviations of Variables Used in Analysis

Variable Name Variable Description Source Mean S.D.

Black Mobilization

MFDP membership, Number of MFDP staff/ Tougaloo College, Ed King 1.81 2.55 1965 contact persons in county. Papers, Box 11, August 23, 1965.

NAACP membership, Number of members in Library of Congress, NAACP 1.06 1.89 1963 (logged) NAACP 1963, (logged). Papers, Box 75, 1963.

Number of black Black candidates running Mississippi Department of candidates, 1967 for office in 1967 county Archives and History, MFDP 1.38 2.73

and state elections. Records, Reel 2, n.d.; TougalooCollege, Rims Barber Papers,Box 1, August 4, 1967.

Countermobilization

Violent resistance Number of incidents of McAdam (1988:257–82); Holt 1.00 2.31 during Freedom physical attack on civil (1965:207–52); summary of Summer rights workers, June– incidents also included in SNCC

August 1964. and CORE papers.

Citizens’ Council organ- Presence of organization Citizens’ Council of America .69 .46 ization in county in county, January, 1956. (1956).

Ku Klux Klan organ- Presence of organization U.S. House of Representatives .48 .50 ization in county in county, c. 1964. (1965).

Political Variables

Percentage voting for Percentage of votes cast U.S. Bureau of the Census (1967). 11.61 8.61 Lyndon Johnson, 1964 for Johnson in 1964.

Proportion employed Proportion of labor force U.S. Bureau of the Census (1967). .021 .004 in local government, employed by local govern- 1964 ment in 1964.

County Socioeconomic Variables

Proportion employed Proportion of labor force U.S. Bureau of the Census (1963). 20.42 8.98 in manuafacturing employed in manufacturing.

Proportion professionals Proportion of labor force U.S. Bureau of the Census (1963). 24.27 6.3employed as professionals.

Landowner concen- Proportion of all commer- U.S. Bureau of the Census (1969). .42 .22 tration cial farm land owned by

owners of 500 acres ormore, 1964.

Poverty, 1959 Proportion of households U.S. Bureau of the Census (1963). .59 .11earning less than $3,000in 1959.

Total number of house- Number of households in U.S. Bureau of the Census (1963). 5.64 4.17 holds (in 1,000s) county in 1960.

Dependent Variables

CAP grants 1965–1968 Total CAP grants for NA, RG 381, Box 14, n.d. multiple 5.28 13.05 (in $100,000s) 1965–1968. files by organization name.

CAP grants 1969–1971 Total CAP Grants for NA, RG 381, Box 14, n.d. multiple 3.30 6.74 (in $100,000s) 1969–1971. files by organization name.

Kenneth T. Andrews is Assistant Professor ofSociology at Harvard University. His current re-search examines questions about the impacts ofsocial movements. He is completing a book,“Freedom is a Constant Struggle” : The Dynam-ics and Consequences of the Mississippi Civil

Rights Movement (Forthcoming, University ofChicago Press), that traces the development ofthe movement and its impact on electoral poli-tics, educational institutions, and social policies.He is also beginning a project on contemporaryenvironmental politics in the United States.


REFERENCES

Amenta, Edwin. 1991. “ Making the Most of aCase Study.” International Journal of Com-parative Sociology 32:172–94.

Amenta, Edwin, Kathleen Dunleavy, and MaryBernstein. 1994. “ Stolen Thunder? HueyLong’ s ‘Share Our Wealth,’ Political Media-tion, and the Second New Deal.” American So-ciological Review 59:678–702.

Amenta, Edwin, Bruce Caruthers, and YvonneZylan. 1992. “ A Hero for the Aged? TheTownsend Movement, the Politcical MediationModel, and U.S. Old Age Policy, 1934–1950.”American Jouranl of Sociology 98:308–39.

Amenta, Edwin, Drew Halfmann, and Michael P.Young. 1999. “The Strategies and Context ofSocial Protest: Political Mediation and the Im-pact of the Townsend Movement in Califor-nia.” Mobilization 4:1–23.

Amenta, Edwin and Michael P. Young. 1999.“Making an Impact.” Pp. 22–41 in How SocialMovements Matter, edited by M. Giugni, D.McAdam, and C. Tilly. Minneapolis, MN:University of Minnesota Press.

Andrews, Kenneth T. 1997. “The Impacts of So-cial Movements on the Political Process: AStudy of the Civil Rights Movement and BlackElectoral Politics in Mississippi.” AmericanSociological Review 62:800–19.

Anselin, Luc. 1992. SpaceStat Tutorial. Urbana,IL: University of Illinois.

Banaszak, Lee Ann. 1996. Why Movements Suc-ceed or Fail. Princeton, NJ: Princeton Univer-sity Press.

Benford, Robert and Scott Hunt. 1992. “Drama-turgy and Social Movements.” Sociological In-quiry 62:36–55.

Browning, Rufus, Dale Marshall, and DavidTabb. 1984. Protest Is Not Enough: TheStruggle of Blacks and Hispanics for Equalityin Urban Politics. Berkeley, CA: University ofCalifornia Press.

Bullard, Robert. 1990. Dumping in Dixie. Boul-der, CO: Westview.

Burstein, Paul. 1985. Discrimination, Jobs, andPolitics. Chicago, IL: University of ChicagoPress.

———. 1999. “ Social Movements and PublicPolicy.” Pp. 3–21 in How Social MovementsMatter, edited by M. Giugni, D. McAdam, andC. Tilly. Minneapolis, MN: University of Min-nesota Press.

Burstein, Paul, Rachel Einwohner, and JocellynHollander. 1995. “ The Success of SocialMovements: A Bargaining Perspective.” Pp.275–95 in The Politics of Social Protest, ed-ited by J. C. Jenkins and B. Klandermans. Min-neapolis, MN: University of Minnesota Press.

Button, James. 1978. Black Violence. Princeton,

NJ: Princeton University Press.———. 1989. Blacks and Social Change.

Princeton, NJ: Princeton University Press.Carson, Clayborne. 1981. In Struggle: SNCC and

the Black Awakening of the 1960’s. Cam-bridge, MA: Harvard University Press.

Citizens’ Councils of America. 1956. “ Missis-sippi Citizens’ Council Map.” The Citizens’Council, June, p. 6.

Clemens, Elisabeth. 1997. The People’s Lobby.Chicago, IL: University of Chicago Press.

Cobb, James. 1990. “‘ Somebody Done Nailed Uson the Cross’ : Federal Farm and WelfarePolicy and the Civil Rights Movement in theMississippi Delta.” Journal of American His-tory 77:912–36.

Colby, David. 1985. “Black Power, White Resis-tance, and Public Policy: Political Power andPoverty Program Grants in Mississippi.” Jour-nal of Politics 47:579–95.

Costain, Anne. 1981. “ Representing Women.”Western Political Quarterly 34:100–113.

Cowart, Andrew. 1969. “Anti-Poverty Expendi-tures in the American States.” Midwest Jour-nal of Political Science 13:219–36.

Cunnigen, Donald. 1987. “Men and Women ofGoodwill: Mississippi’ s White Liberals.”Ph.D. dissertation, Department of Sociology,Harvard University, Cambridge, MA.

Diani, Mario. 1997. “Social Movements and So-cial Capital.” Mobilization 2:129–47.

Dittmer, John. 1994. Local People: The Strugglefor Civil Rights in Mississippi. Urbana, IL:University of Illinois Press.

Doreian, Patrick. 1980. “ Linear Models withSpatially Distributed Data.” SociologicalMethods and Research 9:29–60.

Dorsey, L. C. 1977. Freedom Came to Missis-sippi. New York: Field Foundation.

Eckstein, Harry. 1965. “On the Etiology of Inter-nal Wars.” History and Theory 4:133–63.

Eisinger, Peter. 1979. “The Community ActionProgram and the Development of Black Lead-ership.” Pp. 127–44 in Urban Policy Making,edited by D. R. Marshall. Beverly Hills, CA: Sage.

Fantasia, Rick. 1989. Cultures of Solidarity. Ber-keley, CA: University of California Press.

Ferree, Myra Marx and Patricia Martin. 1995.“Doing the Work of the Movement: FeministOrganizations.” Pp. 3–23 in Feminist Organi-zations, edited by M. M. Ferree and P. Y. Mar-tin. Philadelphia, PA: Temple University Press.

Fording, Richard. 1997. “The Conditional Effectof Violence as a Political Tactic.” AmericanJournal of Political Science 41:1–29.

Friedland, Roger. 1976. “Class Power and SocialControl: The War on Poverty.” Politics andSociety 6:459–89.

Friedman, Lawrence. 1977. “The Social and Po-litical Context of the War on Poverty.” Pp. 21–


47 in A Decade of Federal Antipoverty Pro-grams, edited by R. H. Haveman. New York:Academic.

Gamson, William. 1990. The Strategy of SocialProtest. 2d ed. Belmont, CA: Wadsworth.

Gamson, William and Emilie Schmeidler. 1984.“Organizing the Poor.” Theory and Society 13:587–99.

Ganz, Marshall. 2000. “The Paradox of Power-lessness: Leadership, Organization, and Strat-egy in the Unionization of California Agricul-ture, 1959–1977.” American Journal of Soci-ology 105:1003–1062.

Garrow, David. 1978. Protest at Selma. New Ha-ven, CT: Yale University Press.

Gerlach, Luther and Virginia Hine. 1970. People,Power, Change. Indianapolis, IN: Bobbs-Merrill.

Giugni, Marco. 1998. “Was It Worth the Effort?The Outcomes and Consequences of SocialMovements.” Annual Review of Sociology24:371–93.

Goldstone, Jack. 1980. “The Weakness of Orga-nization: A New Look at Gamson’ s The Strat-egy of Social Protest.” American Journal ofSociology 85:1017–42.

Gould, Roger. 1991. “ Multiple Networks andMobilization in the Paris Commune, 1871.”American Sociological Review. 56:716–29.

Greenberg, Polly. 1969. The Devil Has SlipperyShoes: A Biased Biography of the Child De-velopment Group in Mississippi. London, En-gland: Macmillan.

Harris, David. 1982. Dreams Die Hard. NewYork: St. Martin’ s.

Helfgot, Joseph. 1974. “Professional Reform Or-ganizations and the Symbolic Representationof the Poor.” American Sociological Review39:475–91.

Holt, Len. 1965. The Summer That Didn’t End.New York: William Morrow.

Jacoway, Elizabeth and David Colburn, eds.1982. Southern Businessmen and Desegrega-tion. Baton Rouge, LA: Louisiana State Uni-versity Press.

James, David. 1988. “The Transformation of theSouthern Racial State.” American SociologicalReview 53:191–208.

Katzenstein, Mary. 1990. “ Feminism withinAmerican Institutions.” Signs 16:27–54.

———. 1998. Faithful and Fearless: MovingFeminist Protest Inside the Church and Mili-tary. Princeton, NJ: Princeton UniversityPress.

Katznelson, Ira. 1981. City Trenches. Chicago,IL: University of Chicago Press.

Kitschelt, Herbert. 1986. “Political OpportunityStructures and Political Protest.” British Jour-nal of Political Science 16:57–85.

Klandermans, Bert. 1997. The Social Psychologyof Protest. Cambridge, England: Blackwell.

Koopmans, Ruud. 1997. “Dynamics of Repres-sion and Mobilization: The German ExtremeRight in the 1990s.” Mobilization 2:149–65.

Krane, Dale and Stephen Shaffer. 1992. Missis-sippi Government and Politics. Lincoln, NE:University of Nebraska Press.

Kriesi, Hanspeter, Ruud Koopmans, JanDuyvendak, and Marco Giugni. 1995. New So-cial Movements in Western Europe. Minneapo-lis, MN: University of Minnesota Press.

Lichbach, Mark. 1987. “ Deterrence or Escala-tion? The Puzzle of Aggregate Studies of Re-pression and Dissent.” Journal of ConflictResolution 31:266–97.

Lipsky, Michael. 1968. “ Protest as a PoliticalResource.” American Political Science Review62:1144–58.

Mansbridge, Jane. 1994. “Politics of Persuasion.”Pp. 298–310 in The Dynamics of AmericanPolitics, edited by L. C. Dodd and C. Jillison.Boulder, CO: Westview.

MacLeod, Jay. 1991. “ Introduction: Racism, Re-sistance, and the Origins of the Holmes CountyMovement.” Pp. 1–20 in Minds Stayed onFreedom. Boulder, CO: Westview.

Marx, Gary and James Woods. 1975. “Strands ofResearch and Theory in Collective Behavior.”Annual Review of Sociology 1:363–428.

Mazmanian, Daniel and Paul Sabatier. 1983.Implementation and Public Policy. Glenview,IL: Scott, Foresman and Company.

McAdam, Doug. 1982. Political Process and theDevelopment of Black Insurgency. Chicago,IL: University of Chicago Press.

———. 1988. Freedom Summer. New York: Ox-ford University Press.

McAdam, Doug. 1996. “ Conceptual Origins,Current Problems, Future Directions.” Pp. 23–40 in Comparative Perspectives on SocialMovements, edited by D. McAdam, J. D.McCarthy, and M. N. Zald. New York: Cam-bridge University Press.

McAdam, Doug, John McCarthy, and MayerZald. 1988. “Social Movements.” Pp. 695–737in Handbook of Sociology, edited by N.Smelser. Newbury Park, CA: Sage.

McAdam, Doug and David Snow. 1997. SocialMovements. Los Angeles, CA: Roxbury.

McLemore, Leslie. 1971. “The Mississippi Free-dom Democratic Party: A Case History ofGrass-Roots Politics.” Ph.D. dissertation, De-partment of Political Science, University ofMassachusetts, Amherst, MA.

Mills, Kay. 1993. This Little Light of Mine: TheLife of Fannie Lou Hamer. New York: Dutton.

Morris, Aldon. 1984. The Origins of the CivilRights Movement. New York: Free Press.

Mosley, Donald C. and D. C. Williams. 1967. AnAnalysis and Evaluation of a Community Ac-tion Anti-Poverty Program in the Mississippi


Delta. State College, MS: College of Businessand Industry, Mississippi State University.

Mueller, Carol. 1987. “Collective Consciousness,Identity Transformation, and the Rise of theWomen in Public Office in the United States.”Pp. 89–108 in The Women’s Movements of theUnited States and Western Europe, edited byM. Katzenstein and C. Mueller. Philadelphia,PA: Temple University Press.

Parker, Frank. 1990. Black Votes Count. ChapelHill, NC: University of North Carolina Press.

Patterson, James. 1994. America’s Struggleagainst Poverty. 2d ed. Cambridge, MA:Harvard University Press.

Payne, Charles. 1995. I’ve Got the Light of Free-dom: The Organizing Tradition and the Mis-sissippi Freedom Struggle. Berkeley, CA: Uni-versity of California Press.

Peterson, Paul and David Greenstone. 1977. “Ra-cial Change and Citizen Participation.” Pp.241–78 in A Decade of Federal AntipovertyPrograms, edited by R. H. Haveman. NewYork: Academic.

Piven, Frances Fox and Richard Cloward. 1993.Regulating the Poor. Updated ed. New York:Vintage.

———. 1977. Poor People’s Movements. NewYork: Pantheon.

———. 1984. “Disruption and Organization: AReply to Gamson and Schmeidler.” Theoryand Society 13:587–99.

———. 1992. “Normalizing Collective Protest.”Pp. 301–25 in Frontiers in Social MovementTheory, edited by A. Morris and C. Mueller.New Haven, CT: Yale University Press.

Quadagno, Jill. 1992. “ Social Movements andState Transformation: Labor Unions and Ra-cial Conflict in the War on Poverty.” Ameri-can Sociological Review 57:616–34.

———. 1994. The Color of Welfare. New York:Oxford University Press.

Ragin, Charles. 1989. “The Logic of Compara-tive Method and the Algebra of Logic.” Jour-nal of Quantitative Anthropology 1:373–98.

Rasler, Karen. 1996. “Concessions, Repression,and Political Protest in the Iranian Revolu-tion.” American Sociological Review 61:132–52.

Robnett, Belinda. 1996. “ African-AmericanWomen in the Civil Rights Movement, 1954–1965: Gender, Leadership, and Micromobiliza-tion.” American Journal of Sociology 101:1661–93.

Rochon, Thomas and Daniel Mazmanian. 1993.“ Social Movements and the Policy Process.”Annals of the American Academy of Politicaland Social Sciences 528:75–156.

Roscigno, Vincent and Donald Tomaskovic-Devey. 1994. “Racial Politics in the Contem-porary South.” Social Problems 41:585–607.

Rucht, Dieter and Friedhelm Neidhardt. 1998.

“ Methodological Issues in Collecting ProtestEvent Data.” Pp. 65–89 in Acts of Dissent, ed-ited by D. Rucht, R. Koopmans, and F.Neidhardt. Berlin, Germany: Edition Sigma.

Rupp, Leila and Verta Taylor. 1990. Survival inthe Doldrums. Columbus, OH: Ohio State Uni-versity Press.

Rural Organizing Cultural Center. 1991. MindsStayed on Freedom. Boulder, CO: Westview.

Sabatier, Paul. 1975. “ Social Movements andRegulatory Agencies.” Policy Sciences 6:301–42.

Salamon, Lester. 1971. “ Protest, Politics, andModernization in the American South: Missis-sippi as a ‘Developing Society.’” Ph.D. disser-tation, Department of Government, HarvardUniversity, Cambridge, MA.

Schumaker, Paul. 1975. “Policy Responsivenessto Protest Group Demands.” Journal of Poli-tics 37:488–521.

Schwartz, Michael. 1976. Radical Protest andSocial Structure. New York: Academic.

Schwartz, Michael and Shuva Paul. 1992. “Re-source Mobilization versus the Mobilization ofPeople.” Pp. 205–23 in Frontiers in SocialMovement Theory, edited by A. Morris and C.Mueller. New Haven, CT: Yale UniversityPress.

Simpson, William. 1982. “The Birth of the Mis-sissippi ‘ Loyalist Democrats’ (1965–1968).”Journal of Mississippi History 44:27–45.

Snyder, David and William Kelly. 1979. “Strate-gies for Investigating Violence and SocialChange.” Pp. 212–37 in The Dynamics of So-cial Movements, edited by M. Zald and J.McCarthy. Cambridge, MA: Winthrop.

Stepan-Norris, Judith and Maurice Zeitlin. 1995.“ Union Democracy, Radical Leadership, andthe Hegemony of Capital.” American Socio-logical Review 60:829–50.

Tarrow, Sidney. 1993. “Social Protest and PolicyReform: May 1968 and the Loi d’Orientationin France.” Comparative Political Studies25:579–607.

———. 1998. Power in Movement. 2d ed. Cam-bridge, England: Cambridge University Press.

Tilly, Charles. 1978. From Mobilization to Revo-lution. Englewood, CA: Prentice-Hall.

U.S. Bureau of the Census. 1963. Census of thePopulation: 1960. Vol. 1, Characteristics ofthe Population. Washington, DC: GovernmentPrinting Office.

———. 1967. County and City Data Book, 1967.Washington, DC: Government Printing Office.

———. 1969. Census of Agriculture, Mississippi.Part 33, sect. 1, vol. 1. Washington, DC: Gov-ernment Printing Office.

U.S. House of Representatives, Committee onUn-American Activities. 1965. Activities of theKu Klux Klan Organizations in the U.S. Wash-


ington, DC: Government Printing Office.U.S. Senate, Committee on Labor and Public

Welfare, Subcommittee on Employment, Man-power, and Poverty. 1967. “Examination ofthe War on Poverty.” 90th Cong., 1st sess.

Whittier, Nancy. 1995. Feminist Generations.Philadelphia, PA: Temple University Press.

ARCHIVAL SOURCES

Library of Congress. NAACP Papers, Box 75.1963. File: Mississippi State Conference,1956–1972. Report on Memberships and Free-dom Fund Contributions from MississippiBranches.

Mississippi Department of Archives and History.Mississippi Freedom Democratic PartyRecords, Reel 2. n.d. List of Candidates.

———. Mississippi Freedom Democratic PartyRecords, Reel 3. n.d. Item 6, “Black Politicsin the South,” David Emmons.

National Archives (NA). Record Group 381, Box 2.February 24, 1966. File: Southeast Region,1966; Regional Organizational Subject Files,1966–1969; CAP Office, Records of the Direc-tor. Memo from Dean to Shriver.

———. Record Group 381, Box 2. October 14,1966. File: Southeast Region, 1966, RegionalOrganizational Subject Files, 1966–1969; CAPOffice, Records of the Director. Letter fromSloan to Bozman.

———. Record Group 381, Box 2. November 8,1966. File: Regional Organizational SubjectFiles, 1966–1969, CAP Office, Records of theDirector. Letter from Sloan to Shriver.

———. Record Group 381, Box 2. December 30,1966. File: Southeast Region, 1966; RegionalOrganizational Subject Files, 1966–1969; CAPOffice, Records of the Director. Memo fromSloan to Shriver and Harding.

———. Record Group 381, Box 5. December 11,1965. File: Administrative, Mississippi; StateFiles, 1965–1968, CAP Office, Records of theDirector. Memo from Westgate to Gonzales.

———. Record Group 381, Box 5. January 13,1967. File: Administrative, Mississippi; StateFile, 1965–1968; CAP Office, Records of theDirector. Memo from Dean to Berry.

———. Record Group 381, Box 14. n.d. File:Multiple files by organization name, RegionIV, Mississippi, Grant Profiles 1965–1972,Policy Research Division, Office of Opera-tions, OEO.

———. Record Group 381, Box 40. 1965. File:Mississippi OEO Program (Compilation),July–September 1965. Memo from Redwine toDirector, CAP.

———. Record Group 381, Box 40. March 3,1966. File: Mississippi—OEO Program (Com-pilation), January–March 1966. PreliminaryReport—Bolivar County, Mississippi.

———. Record Group 381, Box 40. March 17,1966. File: Mississippi—OEO Program (Com-pilation), January–March 1966. Memo fromSeward to May.

———. Record Group 381, Box 40. March 31,1966. File: Mississippi—OEO Program (Com-pilation), January–March 1966. Memo fromBerry to May.

———. Record Group 381, Box 108. July 5,1965. File: CDGM 1965, Head Start—Missis-sippi. Letter from Scott to Haddad.

———. Record Group 381, Box 108. July 30,1966. File: Mississippi. Inspection Report,CDGM Area B.

———. Record Group 381, Box 110. July 17,1966. File: Mid-State Opportunity, Inc.Inspector’ s Field Report, Jenneman.

State Historical Society of Wisconsin (SHSW).Alvin Oderman Papers. August 27, 1966.News Sheet on the Lorenzi’ s—Holmes,County, MS, Sue Lorenzi.

———. Amzie Moore Papers, Box 2. N.d.a. File:6. Resident Participation, Bolivar CountyCAP, Inc.

———. Amzie Moore Papers, Box 2. N.d.b. File:8. An Outline of Important Events.

———. Amzie Moore Papers, Box 2. N.d.c. File:10. Petition from ACBC to Sargent Shriver.

———. Amzie Moore Papers, Box 2. 1966. File:8. Telegram from Associated Communities ofBolivar County to Shriver and Sloan.

———. Amzie Moore Papers, Box 2. Jaunary 19,1966. Minutes of the Proceedings, St. Peter’ sRock M. B. Church.

———. Amzie Moore Papers, Box 2. March 13,1966. File: 8. Letter from Velma Bartley.

———. Amzie Moore Papers, Box 2. March 22,1966. File: 8. A Report of the Meeting Heldwith CAP.

———. Amzie Moore Papers, Box 3. N.d. File:1. Chronological History.

———. Amzie Moore Papers, Box 3. April 14 ,1966. File: 1. Letter from Berry to Long,Smith, and Moore.

Tougaloo College. Ed King Papers, Box 11. Au-gust 23, 1965. File: 579. FDP Projects byCounty and Contacts.

———. Ed King Papers, Box 11. 1966. File: 607.Holmes County Community Newsletter.

———. Rims Barber Papers, Box 1. August 4,1967. File: Black Candidates, F.I.S. Newsletter.

———. Tom Dent Collection, July 2, 1979. In-terview of Ed Brown by Tom Dent.


96 American Sociological Review, 2001, Vol. 66 (February:96–124)

Studying Status:

An Integrated Framework

This paper reports development of an integrated framework for studying status. Theframework provides models and methods for addressing long-standing, unresolvedissues, such as (1) the emergence of status, (2) distinguishing between the status ofindividuals and the status of characteristics, and (3) measuring and understandingthe status gap between subgroups (between men and women, or between races). Theframework, which covers both small groups and large societies, and both task andnontask groups, utilizes ideas and insights from several literatures to identify threetypes of status, linked in precise ways to two kinds of personal characteristics(quantitative and qualitative). The three types of status are mathematically specified,and initial theoretical development is presented for all three, including, for each,formulation of measures, derivation of testable implications, and analysis of how tochange status and the status structure. Testable implications cover such phenomenaas status differences between group members, status gaps between subgroups, over-all status inequality, and status gains and losses from discrimination – all undervarying conditions, including the number and intercorrelation of status-conferringpersonal characteristics and the proportions in the subgroups. The new status theoryalso identifies two mechanisms involved in the phenomenon of “internalized oppres-sion.” The framework thus opens many avenues for future work, both theoreticalwork, deriving more and sharper implications, and empirical work, testing the im-plications and using the new measures for the status of persons and the status ofcharacteristics to assess key status phenomena in surveys and experiments.

1996) provide new urgency for obtainingsharper, more precise, and more reliableknowledge about the operation of status.

Although much has been learned, manybasic questions about status remain unan-swered; and the insights that could be mar-shaled for sustained inquiry reside in sepa-rate literatures. In this paper I develop an in-tegrated framework for studying status.1 Theframework provides models and methods foraddressing long-standing, unresolved issues.These issues include: (1) the emergence ofstatus; (2) how to distinguish between, andmeasure, the status of persons and the status

tatus processes are central to the so-cial life, and understanding status is a

central task for sociology. Status processesplay a part in the development of powerfulinequalities, which shape the structure ofgroups and societies as well as, directly andindirectly, the opportunities of individuals(Berger, Rosenholtz, and Zelditch 1980).Moreover, recent conjectures and prelimi-nary evidence suggesting that status may di-rectly affect physical health (Marmot 2000;Sapolsky 1993; Smith 1999; Wilkenson

S

Direct correspondence to Guillermina Jasso,Department of Sociology, New York University,269 Mercer Street, 4th Floor, New York, NewYork 10003-0831 ([email protected]). I am gratefulto the ASR Editors and anonymous reviewers formany helpful comments and suggestions.

Guillermina Jasso

New York University

1 Status, as used in this paper, refers to evalua-tions of the worth of individuals and characteris-tics; synonyms include “honor, esteem, respect,and prestige” (Zelditch 1968:250, 253).

STATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSIS 9 79 79 79 79 7

of characteristics; (3) whether quantitativeand qualitative characteristics operate differ-ently; (4) how to measure status gaps be-tween subgroups of a group or society; (5)how to assess the effects of the proportionsin different subgroups; (6) how to incorpo-rate multiple bases for status; (7) how statusprocesses differ in small groups and largesocieties; (8) how status processes differ intask groups and other kinds of groups; and(9) how status is shaped by the degree ofcorrelation among valued personal charac-teristics.

The proposed framework has three keyfeatures. First, it distinguishes between twokinds of characteristics (quantitative andqualitative) and between three types of sta-tus which are linked in precise ways to thetwo kinds of characteristics. Second, itbuilds carefully on previous work, incorpo-rating seminal ideas and insights into a con-sistent, coherent whole; chief among theseideas and insights are (1) the mathematicalspecification for the production of statusfrom quantitative characteristics (owed toGoode [1978] and Sørensen [1979]), (2) thelink between the status conferred on indi-viduals by quantitative characteristics andthe status acquired by qualitative character-istics (owed to Ridgeway [1991, 1997b],Ridgeway and Balkwell [1997], and Websterand Hysom [1998]), and (3) the multidimen-sional specification of the status of individu-als (analyzed by Barber [1968], Goode[1978], Rossi [1979], and Turner [1984,1995]). Third, the framework is fully mathe-matized, leading to precise predictions aboutthe magnitude of status conferred on indi-viduals and the magnitude of status obtainedby qualitative characteristics, and about theconditions conducive to greater or lesser sta-tus and larger or smaller status gaps, andproviding as well a set of measures ready foruse in empirical work.

For example, it is universally believed thatin almost all societies, men have higher sta-tus than women. But the status gap betweenthe sexes is also known to vary greatly, andthere has not been a theory-based way tomeasure it. The status-gap continuum ex-tends from societies in which women are notpermitted to vote or to own property to soci-eties in which women and men work inteams (on earth and in space, in war and in

peace), vote their conscience, and face nolegal bars to income and wealth. The newframework provides a coherent set of mod-els and tools, based on the status, prestige,and stratification literatures, which enableboth measurement of the status gap andanalysis of why and how the status gap var-ies in magnitude.

The new framework not only enables ex-amination of all the unresolved issues butalso yields unexpected results. It shows thatkey ideas and insights from the several sta-tus literatures, when combined together, pro-duce unexpected synergies, providing newtestable implications and opening new av-enues for status research. The initial set oftestable implications includes implicationsfor the effects on status and status structureof (1) the number and intercorrelation of per-sonal characteristics, (2) the availability ofinformation about personal characteristics,and (3) the proportions of a group in eachcategory of a qualitative characteristic. Sta-tus processes have a long reach, and the newstatus theory identifies two mechanisms thatmay be involved in the phenomenon of “ in-ternalized oppression” (Bourdieu 1997;Ridgeway 1997a:222; Stanton-Salazar1997), provides a new way to understandgains and losses from discrimination, andyields ceteris paribus implications for a widerange of behavioral and social phenomenaincluding coalition formation, defection,identity and reference-group processes, re-sponse rates and missing data in surveys,veiling customs, relative size and skill ofpolitical parties, and the tension between in-dividualism and collectivism.

Of course, the new framework for statusanalysis is preliminary, and further theoreti-cal and empirical inquiry will no doubt leadto modifications. To make the framework asuseful as possible, I take a threefold ap-proach: First, I lay the foundation carefully,formalizing the three basic status functions(corresponding to the three kinds of statusidentified by the framework), which becomethe three basic assumptions of the new sta-tus theory. Second, I provide a sampling ofthe kinds of theoretical development enabledby the framework, deriving implications forseveral special cases, including both smallgroups and large societies under a variety ofconditions. Third, I take a brief look at how


to change status and the status structure, in-cluding speculation linking measures andmechanisms identified by the framework topotential empirical applications across abroad spectrum of social contexts. Thisthreefold approach invites further work onthe content of the functions and assump-tions, on additional special cases for deriv-ing implications, and on empirical applica-tions, including surveys and experiments.

The next section provides an overview ofthe framework. It is followed by three sec-tions, focusing, respectively, on the threekinds of status.

OVERVIEW OF THE FRAMEWORK

FOR STATUS ANALYSIS

The objective is to develop a general frame-work for the study of status—a frameworkthat will cover status phenomena and statusprocesses in all settings (e.g., task groups aswell as nontask groups) and at both microand macro levels (e.g., in both small groupsand large societies), that distinguishesclearly between the status of individuals andthe status of characteristics and between sta-tus and its determinants and its conse-quences, and that enables both measurementand analysis. The framework should containthe basic building blocks which provide thestarting assumptions from which a variety oftestable implications can be derived andwhich provide measures ready for use em-pirically. Future work can then proceed onthree fronts—expanding and refining theframework, building theories and derivingmore and sharper implications, and testingthe implications and accumulating informa-tion on magnitudes and correlates of key sta-tus phenomena.

The framework draws insights and reason-ings from the many pertinent literatures, in-cluding literatures on social organization,social stratification, status organizing pro-cesses, and inequality.2 The emphasis

throughout is on developing a foundation forcoherent and fruitful synthesis. Of course,not all status topics are covered, but futurework can draw them in, expanding and re-fining the framework as needed, as well asestablishing links to other frameworks.

This section discusses the two basic ingre-dients in the framework: (1) a distinction be-tween two kinds of personal characteristics;and (2) a distinction between three types ofstatus.3

Two Types of Characteristics:

Quantitative and Qualitative

The new framework distinguishes betweenquantitative and qualitative characteristics.Quantitative characteristics are characteris-tics of which individuals can have “more” or“ less” ; qualitative characteristics, in con-trast, describe features of individuals thathave no inherent ordering but which can beused to classify them into groups or catego-ries. Quantitative characteristics may be car-dinal (like wealth) or ordinal (like beauty).Qualitative characteristics may be binary(like gender) or polytomous (like race andethnicity).

Until a decade ago, status research did notdistinguish between quantitative and quali-tative characteristics; any possible distinc-tiveness in status processes had not been no-ticed. For example, Berger et al. (1980) ob-served:

The key concept in the study of status orga-nizing processes is the status characteristic,any characteristic of actors around whichevaluations of and beliefs about them cometo be organized. Examples include age, sex,

2 Valuable contributions to these literatures in-clude Barber (1968); Berger et al. (1977); Bergeret al. (1980); Fararo (1989); Goode (1978);Homans (1967); Lipset (1968); Merton ([1949,1957] 1968); Parsons [1949] 1964; Ridgeway(1991, 1997b); Ridgeway and Balkwell (1997);Rossi (1979); Sørensen (1979); Skvoretz and

Fararo (1996); Stinchcombe (1968); Turner(1984, 1995); Veblen [1899] 1953; Wagner andBerger (1993); Weber [1922] 1978; Webster andHysom (1998); and Zelditch (1968).

3 For simplicity, the framework is presented interms of persons, personal characteristics, and thestatus of persons and of personal characteristics.Status processes, however, also operate at otherlevels of analysis, such as that of social entities.Orchestras differ in status, as do countries. Theframework is, with minor modifications, appli-cable to status processes at all levels of analysis.For example, replacing the word “person” withthe word “actor” extends the framework to cor-porate actors.

STATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSIS 9 99 99 99 99 9

race, ethnicity, education, occupation,physical attractiveness, intelligence quo-tients, reading ability. . . . (P. 479)

Ridgeway (1991) was the first to pose thequestion of how qualitative characteristicsacquire status value, and pioneered develop-ment of a theory of status construction inwhich cardinal characteristics produce statusfor qualitative characteristics (Ridgeway1991, 1997b; Ridgeway and Balkwell 1997).Webster and Hysom (1998) extended thetheory so that ordinal as well as cardinal—that is, all quantitative characteristics—canbe used to produce status for qualitativecharacteristics.

Quantitative characteristics:

Goods and bads. Within the set of quan-titative personal characteristics, most statusresearch has focused on characteristicswhich have the property that “more” is pre-ferred to “ less” ; for convenience, these willbe called “goods.” For simplicity, this analy-sis is couched in terms of goods, but, ofcourse, “ bads” (less is preferred to more)may also operate in status processes andtheir handling is straightforward.

Quantitative characteristics:

Cardinal versus ordinal goods, and

rank versus amount of cardinal

goods. Two related questions arise: whetherto distinguish between cardinal and ordinalgoods; and whether, among cardinal goods,amounts play a part in status processes oronly ranks. Note that the operation of cardi-nal and ordinal goods can only be distin-guished by incorporating amounts of cardi-nal goods; and, conversely, if only rank mat-ters in cardinal goods, then there is no dis-tinction between cardinal and ordinal goods.

As discussed below, the process of choos-ing a functional form for first-order statusled to the rank-based function that Sørensen(1979) proposed in his work on the status ofoccupations; and thus the status of individu-als is modeled as a function of rank only. Ashortcoming is that the distribution of aquantitative characteristic such as wealth istreated as flat, rather than, say, as a peakeddistribution with most individuals located insome region of the distribution. This short-coming is mitigated somewhat, however, bytwo things: First, the status function, as willbe seen, is not flat, being a nonlinear func-

tion of rank; and, second, ranks are flexiblein that situations with most people in someregion can be modeled by the use of tiedranks.

Nonetheless, this element of the frame-work merits further research. Such researchshould proceed along two lines. One ismathematical—searching for functionalforms capable of incorporating cardinal andordinal goods. The other is substantive—as-sessing the extent to which status processesare responsive to rank rather than to amountsof cardinal goods.

Three Types of Status:

S1, S2, and S3 Status

The framework distinguishes between threetypes of status. First-order status, denotedS1, is a property of individuals and is basedon quantitative personal characteristics. Sec-ond-order status, denoted S2, is a propertyof qualitative characteristics. Third-orderstatus, denoted S3, is a property of individu-als; it arises as a way to impute status to in-dividuals when S1 cannot be generated be-cause there is no information on quantitativecharacteristics, but may linger after informa-tion is obtained, combining both S1 and S2types of status. Each of the three types ofstatus operates in distinctive ways—eacharises and is maintained or altered via dis-tinctive processes, as elucidated below.

The work on S1— the first kind of status,which is a property of individuals and whichis based on quantitative characteristics—be-sides building in a general way on the sta-tus-relevant literatures, builds in a specificway on two seminal contributions to themathematical specification of status, Goode(1978) and Sørensen (1979). Similarly, thework on S2—the second kind of status,which is a property of qualitative character-istics—builds directly on Ridgeway (1991,1997b), Ridgeway and Balkwell (1997), andWebster and Hysom (1998). The work onS3—the third kind of status which is a prop-erty of persons and which arises when S1cannot be generated—builds on the multidi-mensionality aspect of status and prestige(Barber 1968; Goode 1978; Rossi 1979;Turner 1984, 1995).

Modeling S1 status requires a functionwhich yields a clear status metric and em-


bodies the properties discussed in the statusliteratures. Chief among these properties isthe property analyzed by Goode (1978) that,as rank increases, status increases at an in-creasing rate, rising steeply at higher ranks.The search for a function satisfying proper-ties desirable in an S1 function led to thefunction proposed by Sørensen (1979) tomeasure the status of occupations. The sim-plicity, elegance, and tractability of thisfunction made it an appealing choice.

Personal characteristics and

types of status. Quantitative characteris-tics play a special role in the framework, asthey form the basis for first-order status(S1). In this framework, building onRidgeway’s fundamental insight, qualitativecharacteristics cannot confer first-order sta-tus, but rather they must first acquire sec-ond-order status. Accordingly, qualitativecharacteristics are used to form subgroups;the subgroups are characterized by a sum-mary measure of the members’ first-orderstatus, and this summary measure in turn at-taches to each category of a qualitative char-acteristic, becoming its measure of second-order status (S2).

Summary measure of S1 within the

categories of a qualitative charac-

teristic. The summary measure of S1,which becomes a category’s measure of S2,

can be any measure of location, such as thearithmetic mean, the geometric mean, or themedian. For simplicity and concreteness, inthe theoretical development below, we usethe arithmetic mean. As will be seen, choiceof summary measure affects some results butnot others. Future research might investigateboth preference for, availability of, and ef-fects of alternative measures.

Information and types of status.

Information is of two kinds: (1) informationabout the quantitative characteristics of par-ticular individuals, and (2) informationabout the summary measure of first-orderstatus in the subgroups formed by qualita-tive characteristics. Information about indi-viduals’ ranks on quantitative characteristicsis used to produce first-order status (S1). In-formation about the average first-order sta-tus of the subgroups formed by qualitativecharacteristics is used to generate second-or-der status (S2) in the qualitative characteris-tics. In the absence of information about thequantitative characteristics of particular in-dividuals (i.e., when first-order status can-not be generated), second-order status isused to produce an imputed individual sta-tus, the third-order status (S3). But S3 maynot disappear when information about thequantitative characteristics of particular in-dividuals is obtained; the process by which

Table 1. Summary of the Three Types of Status in the Framework for Status Analysis

Three Types of Status

Characterization S1 S2 S3

Property of Person Qualitative Personcharacteristic

Produced by Quantitative S1 S2, and possibly S1characteristic

Information Rank on quantitative Correlation between S2required characteristic quantitative

characteristic andqualitative characteristic

Formula wrggg

G

ln1

11 −

=∑ S S2 1c cM= [ ]( ) w wS1 S2+

Note: Quantitative characteristics are orderable characteristics; they may be cardinal (like wealth) or ordi-nal (like attractiveness). Quantitative characteristics of which more is preferred to less are called “ goods” ;quantitative characteristics of which less is preferred to more are called “ bads.” Qualitative characteristicsare unorderable characteristics (e.g., race and sex).

In the formulas above, g denotes a good (indexed from 1 to G), w denotes the weight, r denotes rank on aquantitative characteristic, c denotes a category of a qualitative characteristic, M(.) denotes the average(mean, median, etc.), and bold characters denote vectors.

STATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSISSTATUS ANALYSIS 101101101101101

S3 combines both S1 and S2, with S2 lin-gering even in the face of S1—a form of dis-crimination—is analyzed below.

Summary table of three types of

status. Table 1 provides a summary of thethree types of status, including the formulasto be presented below. The table may be help-ful not only as an outline of the expositionbut also as a guide to further theoretical work(e.g., assessing alternative functional formsfor S1) and empirical work (e.g., formulat-ing information conditions in experiments).

Status Matrices

Corresponding to each type of status is amatrix containing each actor’s evaluations ofthe status of persons (S1 and S3) and the sta-tus of characteristics (S2). To illustrate, con-sider S1 status in a collectivity of N persons.Each individual (called “Self” ) accords S1status (or makes prestige payments, inGoode’s [1978] evocative phrase) to everyindividual (to “Others” and also to “Self” ).Group members may differ in the first-orderstatus they accord to any given Other. Dif-ferent individuals may value different quan-titative personal characteristics; for example,one individual may value wealth, while an-other may value beauty, and a third mayvalue both. Moreover, different goods maybe weighted differently; for example, one in-dividual may weight wealth two-thirds andbeauty one-third, while a second may do theopposite. Thus, the S1 status order is repre-sented by a matrix (Table 2).

The S1 and S3 matrices are square, eachmember of the collectivity represented byboth a row and a column. In the S2 matrix,each person occupies a row and each char-acteristic is represented by a set of columns,one column for each category of the charac-teristic.

If all the rows of a matrix are identical,the matrix collapses to a vector. An impor-tant area of research focuses on the pro-cesses by which individuals agree, or not, onstatus matters, for example, how individualschoose the quantitative characteristics theyuse in the S1 function, how individualsshape the societal S1 function and societiesin turn shape individuals’ S1 functions.4

The initial theoretical development belowfollows a twofold approach. First, as in moststatus research, we characterize groups by asingle status function, asking, for example,what status processes look like if all groupmembers share the same S1 function (i.e.,use the same quantitative characteristics asbases of evaluation and weight them thesame way). Second, we consider strategiesfor changing status structures, some ofwhich involve parallel status structures. Forexample, individuals make, and expect,prestige payments (in Goode’s words) basedon the S1 structure in their heads; con-versely, they receive prestige paymentsbased on the S1 structures in others’ heads.Important and interesting phenomena ac-company such interindividual differences inS1 structure.

FIRST-ORDER STATUS (S1)

S1 Assumption and Function

We begin with the general S1 function, writ-ten to accommodate multiple goods, denotedg, and differential weights, denoted w (withbold characters denoting vectors). Formally:

Assumption 1a (General First-Order StatusFunction): First-order status (S1) is aweighted function of goods,

S1 = S1(wg). (1)

Table 2. Self-Other S1 Status Matrix

S111 S112 S113 LLLLL S11J



MMMMM MMMMM MMMMM OOOOO MMMMM

S1N1 S1N2 S1N3 LLLLL S1NJ

Note: Each individual (i = 1 to N) accords S1 sta-tus to each individual (j = 1 to J). Each row repre-sents the S1 status accorded by one individual (toSelf and to Others), and each column represents theS1 status received by one individual. Thus, eachrow represents the S1 status structure in the mind ofone person. In the special case of consensus, thematrix collapses to a vector.

4 For insightful analysis of diffusion and con-sensus processes, see Berger et al. (1998), Blau(1977), Friedkin (1998), Ridgeway and Balkwell(1997), and Skvoretz and Fararo (1996).


The weights, which may take zero or posi-tive values, must sum to one; if zero, the as-sociated good does not produce status, andif one, the associated good is the only goodthat confers first-order status.

Following Goode (1978), we assume thatfirst-order status is a special kind of func-tion: it is not only an increasing function ofthe quantitative personal characteristics butalso it increases at an increasing rate. AsGoode (1978:142) observes, “prestige pay-ments rise steeply.” As discussed above, weadopt the function proposed by Sørensen(1979), which has the upwardly-concaveproperty as well as other appealing proper-ties. Accordingly, we specify S1 status as afunction of rank on a valued characteristic:

S1 =−

ln ,1

1 r(2)

where r denotes the relative rank (betweenzero and one) on the valued quantitative char-acteristic.5 For convenience, we will refer tothis function as the “ log-rank function.” Thelog-rank function ranges from zero to infin-ity, approaching but never reaching the valuezero. When working with small groups, therank r is approximated by i/(N+1), where idenotes the raw rank (the sequence of inte-gers from 1 to the group size N, with 1 as-signed to the lowest-ranking person).6 Theformula for the small-group case is:

S1 =+

+ −

ln .N

N i

1

1(3)

Extending Sørensen’s (1979) function tothe multiple-good case and making explicitthe assumption about the specific form of theS1 function:

Assumption 1b (Specific First-Order StatusFunction): First-order status is theweighted sum of good-specific S1 com-ponents, where each component is thelog-rank function of a good:

S1 =−

=∑w

rggg

G

ln .1

11

(4)

S1 Initial Theoretical Development

In this initial theoretical development, weanalyze the operation of first-order status inseveral settings, varying the number ofgoods and their association and modeling S1in both small groups and large societies. Thetheoretical results are empirically testableimplications. Of course, many more resultscan be obtained, and the framework can beapplied to many new arenas.

S1 in the special case of one-good

small groups. To examine how first-orderstatus operates, we begin with the simplestone-good case in small groups of size rang-ing from 2 to 12, with no tied ranks. Table 3presents the magnitudes of S1 for each mem-ber of such groups; the valued characteristiccan be any quantitative personal characteris-tic (e.g., wealth or beauty or athletic skill).7

The figures immediately reveal three impor-tant implications of the S1 function in theone-good case. First, the status of the low-est-ranking person (Member 1) declines asgroup size increases. Second, the status ofthe highest-ranking person (the last memberin each column) increases as group size in-creases. Third, average status increases asgroup size increases. Thus, the status differ-ence between the lowest- and highest-rank-ing members of the group increases steeplywith group size, from .7 S1 units in the dyadand 1.1 in the triad to almost 2.5 in the 12-member case.8

Many of the status phenomena in smallgroups involve interactions within subsetsof the members. The status difference ineach of the possible pairs of members of agroup is thus an important feature ofgroups. The number of pairs rises steeplyfrom one in the two-member group to 66 inthe 12-member group. To investigate dyadicstatus difference in small groups, Table 4presents the complete S1-difference struc-ture for the possible pairs in three groups,those of sizes 4, 5, and 6. Although the full

5 Formally, the first-order status function haspositive first and second derivatives.

6 The formula can incorporate tied ranks. Boththe sequence of integers and the set of ranks in-corporating tied ranks sum to the same quantity,N (N + 1) /2.

7 The figures in Table 3 are obtained by apply-ing the S1 formula to the one-good case; seeequations 2, 3, and 4 and the notes to Table 3.

8 Formally, the range of S1 is equal to ln(N).The average of S1 can also be expressed as afunction of group size, as shown in the note toTable 3.


structure could be reported in the triangleabove the diagonal, Table 4 presents com-pletely filled-out matrices; these have theadvantage that they directly show not onlythe set of S1 differences for the group butalso the set of S1 differences for each indi-vidual.

As expected from Table 3, the magnitudeof the smallest status distance decreases withgroup size, and the magnitude of the largeststatus distance increases with group size.Additionally, there are three main results inTable 4. First, the magnitude of the S1 dif-ference increases for successive ranks;Member 1 in the 4-member group is not onlyfarther away from Member 4 than fromMember 3 and farther away from Member 3than from Member 2 but also the incrementin the status distance increases for each suc-cessive rank (from .288 to .405 to .683).Second, a group member is always closer tothe individual below than to the individualabove. For example, in the 6-member group,Member 3 is closer to Member 2 than toMember 4, but Member 2 is closer to Mem-ber 1 than to Member 3. Thus, dyadic rela-tions are not symmetric. Third, a few (very

few) group members are equally distantfrom two group members, one below and theother above, with the one below being notimmediately below and the one above beingimmediately above. In Table 4 there arethree such persons, one in each group: Mem-ber 3 in the 4-member group, Member 4 inthe 5-member group, and Member 5 in the6-member group. These individuals mayplay pivotal parts in group dynamics, beinguniquely situated to bridge status distance intwo directions.

There is evidence from previous empiricalresearch that observable behaviors producedby status follow the predicted S1 patterns.For example, Bales (1999) documented theacts initiated by each member of a group, ingroups of size 3 to 8, coding both acts di-rected at the group and acts directed at an-other member. The number of acts initiatedincreases at an increasing rate with rank, andwho-to-whom matrices of acts initiated in-dicate that each member is closer in statusto the lower neighbor than to the higherneighbor, as predicted here.

S1 in the special case of two-good

small groups. Consider now the case in

Table 3. S1 Status in One-Good Small Groups: By Group Member and Group Size

Group Size

Member 2 3 4 5 6 7 8 9 10 11 12

1 .405 .288 .223 .182 .154 .134 .118 .105 .095 .087 .080

2 1.099 .693 .511 .405 .336 .288 .251 .223 .201 .182 .167

3 .— 1.386 .916 .693 .560 .470 .405 .357 .318 .288 .262

4 .— .— 1.609 1.099 .847 .693 .588 .511 .452 .405 .368

5 .— .— .— 1.792 1.253 .981 .811 .693 .606 .539 .486

6 .— .— .— .— 1.946 1.386 1.099 .916 .788 .693 .619

7 .— .— .— .— .— 2.079 1.504 1.204 1.012 .875 .773

8 .— .— .— .— .— .— 2.197 1.609 1.299 1.099 .956

9 .— .— .— .— .— .— .— 2.303 1.705 1.386 1.179

10 .— .— .— .— .— .— .— .— 2.398 1.792 1.466

11 .— .— .— .— .— .— .— .— .— 2.485 1.872

12 .— .— .— .— .— .— .— .— .— .— 2.565

Mean .752 .789 .815 .834 .849 .862 .872 .880 .887 .894 .899

Note: S1 status is a function of one valued quantitative characteristic. Group members are ordered fromlowest ranking to highest ranking on the valued characteristic. S1 status is given by the formula, ln[1/(1 – r)],where r denotes the relative rank and is approximated by [i/(N + 1)], where i, in turn, denotes the raw rankand N denotes the group size. The equivalent formula expressed directly in terms of the raw rank i and group

size N is: ln[(N + 1)/(N +1 – i)]. The formula for the arithmetic mean of S1 is: EN

NN( ln

!S1) =

+1.


which the members of a small group use twoquantitative characteristics as bases ofevaluation; as in the previous special case,there are no tied ranks. If the two goods areperfectly positively associated—that is, eachmember’s rank is the same on both charac-teristics—then the S1 structure remains thesame as in the one-good case. However, ifthe two goods are independent or negativelyassociated or imperfectly positively associ-ated, then S1 structure changes. Here we in-vestigate the case in which two equallyweighted goods are perfectly negatively as-sociated. Other special cases can be simi-larly studied.

Table 5 reports the status structure in thecase of two goods negatively associated andequally weighted, for groups of size 4, 5, and6. For each group member, the table reportsS1 derived from each of the two goods, de-noted s11 and s12. Member number denotesthe rank on the first good. Thus, for example,Member 1 in the 4-member group ranks low-est on the first good and highest on the sec-

ond good; Member 1 has a magnitude of .223on s11 (the lowest status) and a magnitude of1.609 on s12 (the highest status). The two-good S1 status is the unweighted average ofs11 and s12, namely, .916.

Table 5 shows that the status structure isdramatically compressed, although the aver-age status in each group remains the same(compare with Table 3). First, S1 no longerapproaches zero, but now has a floor ofln(2), or approximately .693.9 Second, thestatus of the top person—who is always theperson who ranks highest and lowest on thetwo characteristics—is reduced by theamount ln N ; for example, S1 for Member1 in the 5-member group is reduced from1.792 (as in Table 3, as well as for the sec-ond good’s status s12 in Table 5) to .987.Third, the range of S1 is substantially re-duced, for example, from 1.792 (1.946 –

Table 4. S1 Status Differences among Members of One-Good Groups: By Member Pairs in Groupsof Three Different Sizes

Group Sizeand Member Member 1 Member 2 Member 3 Member 4 Member 5 Member 6

N = 4: Six Pairs

Member 1 .— .288 .693 1.386 .NA .NAMember 2 .288 .— .405 1.099 .NA .NAMember 3 .693 .405 .— .693 .NA .NAMember 4 1.386 1.099 .693 .— .NA .NA

N = 5: Ten Pairs

Member 1 .— .223 .511 .916 1.609 .NAMember 2 .223 .— .288 .693 1.386 .NAMember 3 .511 .288 .— .405 1.099 .NAMember 4 .916 .693 .405 .— .693 .NAMember 5 1.609 1.386 1.099 .693 .— .NA

N = 6: Fifteen Pairs

Member 1 .— .182 .405 .693 1.099 1.792Member 2 .182 .— .223 .511 .916 1.609Member 3 .405 .223 .— .288 .693 1.386Member 4 .693 .511 .288 .— .405 1.099Member 5 1.099 .916 .693 .405 .— .693Member 6 1.792 1.609 1.386 1.099 .693 .—

Note: Group members are ordered from lowest ranking to highest ranking on the valued quantitative char-acteristic; each person’ s S1 status appears in Table 3 (see the previous page). Each row and/or column pro-vides the set of pairwise S1 status differences involving each member of the group. The full set of pairwiseS1 status differences for the entire group occupies the triangle above the diagonal (and is duplicated in thetriangle below the diagonal).

9 The lowest value of S1 is always ln(2) inodd-sized groups and approaches it in even-sizedgroups.


.154) to .347 (1.050 – .703) for the 6-mem-ber group. Fourth, all the groups are sym-metric, such that the S1 scores are symmet-ric about the midrange. Thus, except forodd-sized groups, each member has a fellowgroup member of identical status; for ex-ample, in the 4-member group, Members 1and 4 have identical status, and Members 2and 3 have identical status.

To the extent that a group’s status struc-ture defines its character, the groups in Table5 are dramatically different from their one-good counterparts in Table 3—they have lessinequality, less status distance, and more dy-adic symmetry.

To more carefully assess these two-goods/equally-weighted/negatively-associatedsmall groups, Table 6 presents the status-dis-tance matrices for the groups portrayed inTable 5. The results are striking. First,whereas the one-good status-distance matri-ces of Table 5 had large status distances—for example, ranging from .288 to 1.386 inthe 4-member group—the corresponding sta-tus distances in these two-good matrices aresmall—for example, ranging from 0 to .203in the 4-member group. Second, the status-distance matrices retain the property that thelargest status distance increases with groupsize, but the smallest status distance is now

independent of group size, being zero forevery group size. Third, all group membershave several identical status distances fromother members, in contradistinction to theone-good groups in which very few groupmembers were in this situation; for example,in the 4-member group, every member hasidentical status distances from two othermembers.

Figure 1 illustrates the contrast betweenthe one-good and the two-goods/negatively-associated small groups, displaying the fullset of S1 differences for the 4-member and5-member groups (i.e., with 6 and 10 statusdifferences, respectively). It is clear fromFigure 1 that the one-good group has con-siderably more status inequality.

S1 in the special case of large so-

cieties. Status phenomena and processesoccur in groups and societies of all sizes, andthus it is important to investigate the opera-tion of first-order status (S1) in large societ-ies as well as in small groups. To do so, weuse techniques from the study of probabilitydistributions (Stuart and Ord 1987).10 Sub-

Table 5. S1 Status in Small Groups with Two Negatively Associated, Equally Weighted Goods:By Group Member and Group Size

Group Size

4 5 6

Member s11 s12 S1 s11 s12 S1 s11 s12 S1

1 .223 1.609 .916 .182 1.792 .987 .154 1.946 1.050

2 .511 .916 .714 .405 1.099 .752 .336 1.253 .795

3 .916 .511 .714 .693 .693 .693 .560 .847 .703

4 1.609 .223 .916 1.099 .405 .752 .847 .560 .703

5 .— .— .— 1.792 .182 .987 1.253 .336 .795

6 .— .— .— .— .— .— 1.946 .154 1.050

Mean .815 .815 .815 .834 .834 .834 .849 .849 .849

Note: The status components s11 and s12 are each functions of a single valued quantitative characteristic(as in Table 3). Member number refers to rank on the first valued characteristic, corresponding to the statuscomponent s11 . The two characteristics are perfectly negatively associated. In this example, the two charac-teristics are weighted equally, and thus S1 is the unweighted average of s11 and s12. A direct formula for S1

in this case is given by: − −ln r r2 , where r denotes the relative rank on the first characteristic and is ap-

proximated by [i /(N + 1)], where i, in turn, denotes the raw rank, and N denotes the group size. The equiva-

lent formula expressed directly in terms of the raw rank i and the group size N is: ln ( ) ( )N i N i+ + −[ ]1 1 .

10 Formally, many of the formulas and resultsfor large societies may be thought of as the limit,as N increases to infinity, of the formulas and re-sults for small groups.


stantively, the model is set up as before. S1is a function of valued quantitative personalcharacteristics; the function is the log-rankfunction given in equations 2 and 4. Be-cause, as seen, the case in which S1 arisesfrom one good is identical to the case inwhich S1 arises from two perfectly posi-tively associated goods, we present an inte-grated theoretical development in which twogoods are used as bases of evaluation. Asbefore, let the two goods be weightedequally in the S1 function. Also as before,we distinguish between perfectly positivelyassociated goods and perfectly negativelyassociated goods; we also introduce a newcase, which arises naturally in the study ofprobability distributions—namely, the casein which the two goods are independent.

At the outset it is known that the distribu-tion associated with S1 in the positively as-sociated case is the exponential and that itsmean equals 1. Because the mean of the av-erage of two identical variates is equal to theoriginal mean, it is also known that the mean

of S1 in the negatively associated and inde-pendent cases equals 1.

The formula for S1 in the two-goods/nega-tively-associated case is straightforward toobtain. The two formulas for both the posi-tively and negatively associated cases, withequally weighted goods, may be expressed:

S1

,positively associated goods

,negatively associated goods.

=−

−

ln

ln

1

1

2

1 2

r

r(5)

As before, r denotes the relative rank; in thenegatively associated goods case, r is therelative rank on one of the two goods.11

When the two goods are independent, it isnot possible to obtain an algebraic formulafor S1. However, S1 can be numerically ap-proximated by interpolating from the cumu-

Table 6. S1 Status Differences among Members of Small Groups with Two Negatively Associated,Equally Weighted Goods: By Member Pairs in Groups of Three Different Sizes

Group Sizeand Member Member 1 Member 2 Member 3 Member 4 Member 5 Member 6

N = 4: Six PairsMember 1 . — .203 .203 0 .NA .NAMember 2 .203 . — 0 .203 .NA .NAMember 3 .203 0 . — .203 .NA .NAMember 4 0 .203 .203 . — .NA .NA

N = 5: Ten PairsMember 1 . — .235 .294 .235 0 .NAMember 2 .235 . — .059 0 .235 .NAMember 3 .294 .059 . — .059 .294 .NAMember 4 .235 0 .059 . — .235 .NAMember 5 0 .235 .294 .235 . — .NA

N = 6: Fifteen Pairs

Member 1 . — .255 .347 .347 .255 0Member 2 .255 . — .091 .091 0 .255Member 3 .347 .091 . — 0 .091 .347Member 4 .347 .091 0 . — .091 .347Member 5 .255 0 .091 .091 . — .255Member 6 0 .255 .347 .347 .255 . —

Note: Group members are ordered from lowest ranking to highest ranking on the first valued characteris-tic, as in Table 5; each person’ s S1 status appears in Table 5. Each row and/or column provides the set ofpairwise S1 status differences involving each member of the group. The full set of pairwise S1 status differ-ences for the entire group occupies the triangle above the diagonal (and is duplicated in the triangle belowthe diagonal).

11 In probability-distribution terms, the formu-las for S1 in equation 5 are quantile functions.


lative distribution function of the Erlangvariate.12

Table 7 reports the first-order status (S1)for individuals at selected relative ranks(from 0 to 1, in increments of .05), for thethree cases—positively associated goods,negatively associated goods, and indepen-dent goods. As expected from the work withsmall groups, the positively and negativelyassociated cases differ both with respect tothe lower extreme value of S1 and to thecompression of S1. In the independent-goods case, S1 retains the lower extremevalue of zero but, with respect to compres-sion, occupies an intermediate place be-tween the positively and negatively associ-ated cases. In all three cases, the majority ofthe population have S1 magnitudes belowthe mean of 1—59.4 percent in the indepen-dent case, 63.2 percent in the positively as-sociated case, and 67.7 percent in the nega-tively associated case.

A different way to gauge the three typesof S1 structures is to examine their probabil-ity density functions (pdf). Figure 2 presentsgraphs of the three pdf ’s. The graphs indi-cate several important features: First, thereare individuals of very high status in allthree types of S1 structure. Second, there areindividuals of very low status (approachingzero) only in the positively associated andindependent cases; as already known, in thenegatively associated case, S1 has a floor ofapproximately .7. Third, in the positivelyand negatively associated cases, the modeoccurs at the lower extreme value and in theindependent case it occurs at .5; that is, thereis a concentration of low-status persons.These graphs provide vivid depiction of thesociological insight that high status is rela-tively scarce and the philosophical insightthat societies differ dramatically accordingto whether valued characteristics are posi-tively or negatively associated.

How to Change S1 and the S1

Structure

Individuals are accorded status because theypossess quantitative characteristics which

Figure 1. S1 Status Differences in Small Groups

Note: In the two-good case, the goods are equally weighted and negatively associated. In the four-membergroup there are 6 pairs, and in the five-member group there are 10 pairs. In each of the two-good groups,there are two pairs with S1 status differences equal to zero.

One Good Two Goods One Good Two Goods

2.0

1.5

1.0

.5

0

S1

Sta

tus

Dif

fere

nce

Four-Member Group Five-Member Group

12 The sum of two independently and identi-cally distributed exponential variates is distrib-uted as an Erlang variate; the unweighted aver-age of two iid exponentials is also an Erlang. TheErlang is a member of the larger gamma family.


individuals and groups reward and use as astatus-conferring device. If individuals losethe valued characteristics, they lose their sta-tus, holding constant the goods in the S1function. Conversely, if individuals changetheir minds about what they find admirableand desirable, then receipt of S1 is altered.For example, suppose that Smith is the bestswimmer in the world. As long as some in-dividuals (swimming enthusiasts, say) valueswimming skill and Smith retains the skill,Smith will be accorded the highest status onthe S1 swimming component (by swimmingenthusiasts). However, if Smith loses theskill or if swimming enthusiasts disappear,Smith will lose status. Similarly, the strategyfor the film character who says, “ I don’ t getno respect,” is to promote admiration for, ordesirability of, some characteristic on which

he or she ranks high.Note that the strategy of changing the

quantitative characteristics used to conferstatus appears in two versions. In the firstversion, an individual attempts to changeothers’ valued goods, for example, by per-suading them that wealth is meaningless andshould not be used as a status-conferring de-vice or that writing poetry is the most admi-rable and desirable skill. In the second ver-sion, an individual changes his or her ownvalued goods, effectively ceasing to makeprestige payments, in Goode’s (1978) evoca-tive phrase, and thus putting an end to thismicro form of “ internalized oppression.”

To illustrate, consider two college room-mates in a fraternity house: One is wealthyand an athlete of no particular distinction;the other comes from a poor family and is a

Table 7. S1 Status in Three Kinds of Two-Good Large Societies, by Member’s Relative Rank

Member’ s Two Characteristics, Two Characteristics, Two Characteristics,Relative Rank Positively Associated Negatively Associated Independent

. 0 . 0 .693 . 0

.10 .105 .698 .266

.15 .162 .704 .342

.20 .223 .714 .413

.25 .288 .725 .481

.30 .357 .740 .549

.35 .431 .758 .618

.40 .511 .780 .689

.45 .598 .806 .762

.50 .693 .837 .840

.55 .798 .873 .922

.60 .916 .916 1.012

.65 1.050 .968 1.110

.70 1.204 1.030 1.220

.75 1.386 1.106 1.347

.80 1.609 1.204 1.498

.85 1.897 1.334 1.687

.90 2.303 1.524 1.945

.95 2.996 1.857 2.372

. 1 .∞ .∞ .∞Note: Member’ s relative rank is the relative rank corresponding to (a) both characteristics in the positively

associated case, (b) one characteristic in the negatively associated case, and (c) the composite rank in the

independent case. The formulas for S1 status are: (a) in the positively associated case, S1 1 1= −[ ]ln ( )r ; and

(b) in the negatively associated case, S1 2 1 2= −

ln r . In the independent case, S1 is approximated nu-merically from the cumulative distribution function of the S1 distribution (see Appendix Table A).


gifted athlete.13 If both roommates valuewealth only and use it to confer status, thenthe rich roommate will be accorded higherstatus than the other roommate—by both ofthem. The athlete roommate can change thisS1 structure by two methods: (1) S/he canpersuade him/herself and the rich roommatethat athletic skill is highly desirable andshould also be used to confer status; or(2) s/he can relinquish his/her love of richesand value only athletic skill. Under the firstmethod, if successful (and athletic skill andwealth are weighted equally), the S1 struc-ture would be transformed into one of per-fect equality—the two goods are perfectlynegatively associated, hence both room-mates would receive equal magnitudes ofS1. Under the second method, if successful,the athlete roommate would no longer ac-cord higher status to the rich roommate. Ofcourse, if the rich roommate still uses wealthas a basis of evaluation, there would be anew tension arising from their having dis-crepant S1 functions.

As shown, the kind of statistical associa-tion between two or more valued goods—whether they are positively associated, nega-

tively associated, or independent—power-fully affects first-order status. The associa-tion between any pair of goods is less opento alteration than the choice of valued goods,although it can be changed by changing thepopulation, through recruitment or expulsionof members.

SECOND-ORDER STATUS (S2)


First-order status is a property of persons—individuals obtain first-order status fromtheir quantitative characteristics. Individualsalso have qualitative characteristics and, byprocesses of status generalization (Berger etal. 1977; Ridgeway 1991, 1997b; Ridgewayand Balkwell 1997; Webster and Hysom1998), if in a group or society there is an as-sociation between the valued quantitativecharacteristics (the goods) and a qualitativecharacteristic, the qualitative characteristicacquires status by that association. This sta-tus acquired by qualitative characteristics wecall second-order status (S2).14

Goods negatively associated

Goods positivelyassociated

Goods independent

0 .5 1.0 1.5 2.0 2.5 3.0

S1 Status

Pro

bab

ilit

y D

en

sit

y F

un

cti

on

2.0

1.5

1.0

.5

0

Figure 2. S1 Status in Two-Good Large Societies

Note: Formulas for the probability density function are reported in Appendix Table A.

13 This example was suggested by an anony-mous reviewer.

14 In the analysis of second-order status, a newkind of association figures prominently. Whereasin the analysis of first-order status the focal as-


Thus, as discussed earlier, we assume thatqualitative characteristics cannot producestatus; they can only derive it from some al-ready produced status. Formally, letting cdenote a category of the qualitative charac-teristic and M(.) denote the average (mean,median, etc.):

Assumption 2a (Second-Order Status): Sec-ond-order status is obtained by a quali-tative characteristic if and only if theaverage first-order status differs acrossits categories. In such case, the second-order status attached to each category isequal to the average first-order status inthe category:

S2 (S1)c cM= [ ] . (6)

The second-order status gap follows im-mediately. Letting cA and cB denote two cat-egories of the qualitative characteristic andgap denote the gap in second-order statusbetween them:

Assumption 2b (Second-Order Status Gap):The S2 status gap is the difference in S2between the categories of a qualitativecharacteristic:

S2 (S1) (S1)A B

gap c cM M= [ ] − [ ] . (7)

The second-order status gap ranges fromzero to high positive values. For a givenqualitative characteristic, there is a set ofsecond-order status gaps defined as the setof gaps for all pairs of categories. For ex-ample, while gender has only one pair of cat-egories and hence only one possible second-order status gap, ethnicity could have 3, 6,10, 15 pairs of categories, and so on, andhence could have that many second-orderstatus gaps. If all gaps are equal to zero, then

the qualitative characteristic has not ac-quired second-order status.

Note that the process of generating S2 sta-tus may operate simultaneously for manyqualitative characteristics, each acquiring S2status (or not) depending on the average S1status in its categories. For example, gender,race, ethnicity, and religion all potentiallycan acquire S2 status simultaneously.


We focus here on two outcomes—second-or-der status (S2) and the gap in second-orderstatus between the categories of a qualitativecharacteristic (S2gap)—and on determinantsof these outcomes. We examine the effectsof (1) the number of valued goods in the S1function and the goods’ association, (2) theassociation between the valued good(s) andthe qualitative characteristic, and (3) therelative sizes of the categories of the quali-tative characteristic. The summary measureof S1 is specified as the arithmetic mean.

Sufficient conditions for second-

order status. To begin, look back at thelayout of first-order status in one-good smallgroups presented in Table 3. If there is a per-fect association between the good (i.e., thequantitative characteristic used to conferfirst-order status) and a binary qualitativecharacteristic, then the qualitative character-istic will acquire second-order status. Forexample, in the group of size 4, if Member 1is female and Members 2 through 4 aremale, or if Members 1 and 2 are female andMembers 3 and 4 male, or if Members 1through 3 are female and Member 4 male,then gender will acquire second-order status,such that the category “ male” will have ahigher magnitude of S2 than the category“ female.” And conversely, if the associationgoes in the opposite direction. Thus, thecombination of (1) a one-good S1 functionwith (2) a perfect association between thegood and a qualitative characteristic is suffi-cient to induce second-order status for thequalitative characteristic. Moreover, becausethe S1 structure of a one-good society isidentical to that in a society with several per-fectly positively associated goods, the com-bination of (1) perfectly positively associ-ated goods in the S1 function with (2) a per-fect association between the goods and a

sociation is between quantitative characteristics(as between wealth and beauty), in the analysisof second-order status the focal association is be-tween one quantitative characteristic and onequalitative characteristic. Put differently, giventhat the term “ goods” is used for quantitativecharacteristics of which more is preferred to lessand given that here we are not working with“bads,” the focal association in the analysis of S1is between two goods, whereas the focal associa-tion in the analysis of S2 is between a good and aqualitative characteristic.


qualitative characteristic is also a sufficientcondition for second-order status to arise.

Consider now a somewhat more compli-cated case, that of societies with negativelyassociated goods. Look at Table 5. If there isa perfect association between one of the twogoods and the qualitative characteristic andif group members are assigned in equal num-bers to categories of a binary qualitativecharacteristic (e.g., let Members 1 and 2 ofthe four-member group be men and Members3 and 4 be women), then the average S1among men equals the average S1 amongwomen. In this situation, gender does notacquire status. On the other hand, supposethat Members 1 and 4 are women and Mem-bers 2 and 3 are men—that is, eliminate theperfect association between the valued goodsand the qualitative characteristic—then av-erage S1 status differs across the two sexesand gender acquires second-order status.15

Combining the results obtained from in-spection of Tables 3 and 5 leads to statementof sufficient conditions for the emergence ofS2 status: Given a perfect association be-tween one quantitative good and the qualita-tive characteristic, singly sufficient condi-tions for the qualitative characteristic to ac-quire second-order status are: (1) S1 is basedon one good; (2) S1 is based on several per-fectly positively associated goods; (3) S1 isbased on two perfectly negatively associatedgoods, and the group size is odd; (4) S1 isbased on two perfectly negatively associatedgoods, and the subgroup split is not fifty-fifty.16

Note again that even in this case of per-fect association between one quantitativegood and the qualitative characteristic, there

are situations in which S2 will not arise, in-cluding the situation, obvious from Table 5,in which S1 is based on two perfectly nega-tively associated goods and the subgroupsplit is fifty-fifty.17


small groups. Consider a group of size 12(as in Table 3). First-order status (S1) isbased on either one good or several posi-tively associated goods, there are no tiedranks, and the qualitative characteristic isbinary. Next, suppose that there is a perfectassociation between the good(s) and thequalitative characteristic. The subgroup withthe lowest-ranking individuals on the quan-titative characteristic is called the bottomsubgroup, and the subgroup with the high-est-ranking individuals is called the top sub-group. Now imagine all possible subgroupsplits, ranging from the case in which thebottom subgroup has 1 member and the topsubgroup has 11 members to the oppositecase in which the bottom subgroup has 11individuals and the top subgroup has only 1.

Table 8 reports the average first-order sta-tus in each subgroup formed by the 11 pos-sible subgroup splits (with each split repre-sented by a row). As shown, average S1 dif-fers for the two subgroups in each row; forexample, in the case of a fifty-fifty split, av-erage S1 is .33 in the bottom subgroup and1.47 in the top subgroup. Thus, this situationunambiguously produces second-order sta-tus. If, in the example just given, the quali-tative characteristic is gender, then genderacquires second-order status, and if womenare the top subgroup and men are the bottomsubgroup, then the magnitudes of S2 are .33for the category “men” and 1.47 for the cat-egory “women.”

Table 8 also reports the S2 gap betweenthe two categories of the qualitative charac-teristic. As shown, the second-order status

15 The case of imperfect correlation betweenthe quantitative and qualitative characteristics isan important one to analyze. Although space con-straints prevent such an analysis here, two thingsare worth noting: First, a priori it is obvious fromequation 7 that the status gap will be smaller ifthe correlation is weaker; the bottom subgroupwill include higher scorers, so to speak, and thetop subgroup will include lower scorers, thus at-tenuating the status gap. Second, attenuation ofthe status gap, especially in situations of multiplequalitative characteristics acquiring S2, may leadto competition among alternative views for struc-turing status relations.

16 Work is underway to analyze the case of twoindependent goods.

17 This is a highly suggestive case for genderphenomena. For example, if heroism (or, say,wealth or hunting skill) is perfectly negativelyassociated with beauty and concomitantly per-fectly associated with sex and if the sex split isfifty-fifty, then the average S1 status will be thesame among both sexes and gender will not ac-quire S2 status. The new framework enablesfresh interpretation of many situations in historyand literature, including the rise of veiling cus-toms and their link to gender inequality.


gap increases as the relative size of the bot-tom subgroup increases. To illustrate, ifwomen are wealthier than men, such that thepoorest woman is richer than the richest man,then the S2 gap between men and women islowest when there is only one man and thereare 11 women (a gap of .89 S1 units) and it ishighest when there are 11 men and onewoman (a gap of 1.82 S1 units).

S2 acquisition is a general process, andthis analysis applies to any qualitative char-acteristic. For example, consider the statusof academic disciplines. Suppose that agiven interdisciplinary course enrolls first-year graduate students from discipline A andthird-year graduate students from disciplineB. If competence is valued and if it increaseswith years of study, then S2 status arises, fa-voring discipline B. Moreover, if the disci-plines differ in sex ratio and racial or ethniccomposition, then a train of further status in-equalities is set in motion.


large societies. Consider now S2 inlarge societies in which a binary qualitativecharacteristic is perfectly associated with thegood(s). As in the analysis of S1 in large so-cieties, techniques from the study of prob-

ability distributions make it possible to ob-tain many a priori results. Table 9 presents,as in Table 8, the S2 acquired by the two cat-egories of the qualitative characteristic andthe S2 gap. As shown, and consistent withthe previous results, in both categories, sec-ond-order status increases with the relativesize of the bottom subgroup. More impor-tant, the second-order status gap also in-creases with the relative size of the bottomsubgroup. This means, for example, thatgiven a racially divided society in which tworaces are, respectively, advantaged and dis-advantaged on the good, the status gap willbe substantially larger if the disadvantagedrace constitutes 90 percent of the populationand the advantaged race constitutes 10 per-cent of the population than if the disadvan-taged race constitutes 10 percent of thepopulation and the advantaged race consti-tutes 90 percent of the population.

To more fully appreciate the results, it isuseful to graph S1 and S2 (average S1) foreach subgroup split of interest. Figure 3 re-ports the graphs of S1 and S2 for the case inwhich the population is evenly split betweenthe two categories of the binary qualitativecharacteristic. As expected, S1 increases at

Table 8. S2 Status (Average S1 Status) in Two Subgroups of Group of Size 12

Bottom Subgroup Top Subgroup S2 StatusSize S2 = E(S1) Size S2 = E(S1) Gap

1 .080 11 .974 .894

2 .124 10 1.054 .931

3 .170 9 1.142 .973

4 .219 8 1.239 1.020

5 .272 7 1.347 1.074

6 .330 6 1.468 1.138

7 .394 5 1.607 1.214

8 .464 4 1.770 1.307

10 .636 2 2.218 1.583

11 .748 1 2.565 1.817

Note: Each group member’ s S1 status is reported in Table 3. Direct formulas for calculating S2 status(average S1 status) in the two subgroups and the S2 gap (letting N denote the group size and n denote thesize of the bottom subgroup) are:

S2 in bottom subgroup: S2bot nNN

N n= + −

−ln( ) ln

!

( )!1 ;

S2 in upper subgroup: S2topN nN N n= + − −−ln( ) ln ( )!1 ;

S2 gap: S2gapN

n N nN n

nN=

−−

−{ } +

( )ln[( )!] ln( !)

1.


an increasing rate as relative rank increases.A vertical dashed line divides the populationinto the two equal-sized subgroups. Averagefirst-order status is represented by horizon-tal lines; the two short lines are for S2 ineach of the two subgroups, and the long lineis for average S1 over the entire population.

In this equal-split case, the two categories’magnitudes of second-order status are equi-distant from the overall average S1 of 1; andthe S2 gap between the two categories is2ln(2). (There are precise relations betweeneach subgroup split and the S2 gap.)

This kind of analysis, carried out moreelaborately and examining the effects of all

the elements in play (e.g., the number ofgoods and their intercorrelations andweights, the correlation between good(s) andqualitative characteristics, and populationsubgroup splits) may prove useful in under-standing gender relations as well as inter-group relations across a wide variety of set-tings, including, for example, empires, colo-nial societies, and multiracial societies.

How to Change S2

We have described the process by whichqualitative characteristics acquire status.The question arises, how second-order sta-

Table 9. S2 Status (Average S1 Status) in Two Subgroups of Large Society

Bottom Subgroup Top Subgroup S2 StatusRelative Size S2 = E(S1) Relative Size S2 = E(S1) Gap

.05 .025 .95 1.051 1.026

.10 .052 .90 1.105 1.054

.15 .079 .85 1.162 1.083

.20 .107 .80 1.223 1.116

.25 .137 .75 1.288 1.151

.30 .168 .70 1.357 1.189

.35 .200 .65 1.431 1.231

.40 .234 .60 1.511 1.277

.45 .269 .55 1.598 1.328

.50 .307 .50 1.693 1.386

.55 .347 .45 1.798 1.452

.60 .389 .40 1.916 1.527

.65 .435 .35 2.050 1.615

.70 .484 .30 2.204 1.720

.75 .538 .25 2.386 1.848

.80 .598 .20 2.609 2.012

.85 .665 .15 2.897 2.232

.90 .744 .10 3.303 2.558

.95 .842 .05 3.996 3.153

Note: Subgroup S2 status (average S1 status) is based on S1 arising from a single valued characteristic orseveral perfectly positively associated characteristics. S1 status for representative persons in this populationis reported in the second column of Table 7. Formulas for calculating the S2 status (average S1 status) in thetwo subgroups and the S2 status gap in this large-population case (letting p denote the relative size of thebottom subgroup) are:

S2 in bottom subgroup: S2botp

p p= −

−

−

1

1 1

1ln ;

S2 in upper subgroup: S2top p= +

−

1

1

1ln ;

S2 gap: S2gap p p=

−

1 1

1ln .


tus changes, is maintained, or is discarded.According to the framework, S2 is totallydependent on S1 and on the configuration ofgoods, their weights, their association witheach other and with the qualitative charac-teristic. Thus, S2 cannot be directly af-fected; rather it is altered by altering the de-terminants of S1 and the association be-tween the goods and the qualitative charac-teristic.

To illustrate with an example based on thework of Ridgeway (1991, 1997b), Ridgewayand Balkwell (1997), and Webster andHysom (1998): Suppose that wealth is thevalued good and that wealth is perfectly as-sociated with gender, such that men arericher than women. Wealth produces first-order status, and first-order status generatessecond-order status—the category “ male”has greater S2 than the category “ female.”Strategies to make gender shed its second-order status include: (1) change the S1 func-tion (e.g., by eliminating wealth or by intro-ducing a second good negatively associatedwith wealth); and (2) change the associationbetween wealth and gender (e.g., by recruit-ing wealthy women and/or destitute meninto the society).

These strategies illuminate the underlyingdynamics, pointing to a mechanism thatmay play a part in the phenomenon of “ in-ternalized oppression” (Bourdieu 1997;Stanton-Salazar 1997). If women did notvalue wealth, gender would not have ac-quired S2. The members of the bottom sub-group are always in collusion, so to speak,with the members of the top subgroup. Byagreeing that wealth is desirable, membersof the bottom subgroup let wealth conferfirst-order status, and once wealth confersfirst-order status, given the association be-tween wealth and gender, gender acquiressecond-order status. Hence, a radical way toremove S2 from gender is to renouncewealth. The literature on utopian communi-ties provides trenchant insights into thisprocess for removing S2 from qualitativecharacteristics. Note, however, that ifwomen renounce wealth but men still valueit, then there will be two parallel statusstructures, with men making prestige pay-ments on the basis of wealth (to both menand women) and women making prestigepayments on the basis of something else (toboth men and women). The ensuing wind-falls and shortfalls in status constitute a

3

2

1

0

0 .25 .50 .75 1Relative Rank

S1

Sta

tus

Bottom Subgroup Top Subgroup

Average S1 status in population

S2 status inbottom subgroup

S2 status in top subgroup

S1 Status

Figure 3. S1 Status and S2 Status in a One-Good Society with Two Equal-Sized Subgroups

Note: Subgroup S2 status equals average S1 status in each subgroup. The vertical dashed line denotes thedivision of the society into two equal-sized subgroups. Horizontal lines denote average S1 status in eachsubgroup and in the population as a whole. S1 values are reported in Table 7 (see page 108) and S2 values inTable 9 (see page 113).


new avenue for research, with many newspecial cases to be analyzed, together withthe mechanisms for achieving a new con-sensus.

The second strategy is less drastic. Onecan continue to value wealth and enjoy itsuse, but one has to search for anomalous in-dividuals and import them into the society,thus reducing or eliminating the associationbetween wealth and gender.

THIRD-ORDER STATUS (S3)


We have described the processes for produc-ing first-order status—which is a property ofindividuals and arises from individuals’quantitative characteristics, together with in-formation about their ranks—and for pro-ducing second-order status—which is aproperty of qualitative characteristics andarises from first-order status. Now considerthe case in which first-order status cannot beproduced because there is no informationabout an individual’s ranks on quantitativecharacteristics. Suppose that wealth is thebasis of evaluation, but there is no informa-tion about an individual’s wealth rank. Inthis case, if the individual can be catego-rized with respect to a qualitative character-istic (e.g., can be classified as male or fe-male) and if there is information about theaverage S1 status in that category of thequalitative characteristic, then third-orderstatus is produced and imputed to the indi-vidual. Formally:

Assumption 3a (Third-Order Status): If in-formation about an individual’s ranks onquantitative characteristics (goods in theS1 function) is not available so thatfirst-order status cannot be produced,and if there is information about the av-erage S1 in the category of a qualitativecharacteristic corresponding to the indi-vidual, then third-order status (S3) isimputed to the individual. S3 for the jthperson in the cth category of a qualita-tive characteristic is equal to the second-order status of the category:

S3jc = S2c . (8)

The S3 formula may be generalized in anumber of ways. First, there may be infor-mation about average S1 status for subsetsformed by more than one qualitative charac-teristic (e.g., “black women,” “white men,”etc.). Second, if S1 is produced by more thanone quantitative characteristic, there may beseparate information about average S1 sta-tus derived from one good in one qualitativecharacteristic and average S1 status derivedfrom another good in another qualitativecharacteristic (e.g., average S1 status due towealth in each category of gender and aver-age S1 status due to beauty in each categoryof race).

There is another important process. S2,once produced, acquires a life of its own, soto speak. Even if information about the per-tinent quantitative characteristic(s) becomesavailable, there may be a reluctance to relin-quish the S2 component of S3, leading to ageneralized version of S3, denoted S3*. For-mally:

Assumption 3b (Generalized Third-OrderStatus): Given that third-order status hasbeen produced, a generalized version ofS3, S3*, is generated; this is a weightedsum of S1 and S2:

S3* = wS1 + wS2. (9)

(As before, bold characters indicate vectorsand the weights must sum to one.)

This process of combining S1 and S2 isrelated, in part, to the empirical traditionpioneered by Rossi (1979) in which the pres-tige of an individual is linked to a large setof the individual’s quantitative and qualita-tive characteristics. Thus, consistent with thetheoretical framework developed here, if S1is generated by schooling and earnings andS2 is acquired by race and gender, then S3may respond to all four characteristics. Aprimary research objective would be to as-certain whether, given information aboutschooling and earnings, race and genderhave no effect or whether, alternatively, S2,once generated, is long-lived.


Why would use of S2 survive introductionof the information required to produce S1?Look again at Figure 3. Notice that the sub-


group S2 lines intersect the S1 curve. To theleft of the intersection in each subgroup, S2status (average S1 status) is greater than theindividual’s first-order status (S1), and to theright, S2 status (average S1 status) is smallerthan the individual’s S1. This indicates thatthere are status gains and losses from the useof S2: Individuals whose S1 is lower thantheir subgroup’s S2 gain status from the useof S2, while individuals whose S1 is higherthan their subgroup’s S2 lose status from theuse of S2.18

Use of S2 is a form of discrimination—individuals are not assessed on their indi-vidual S1-pertinent quantitative characteris-tics but rather are treated as part of a sub-group and their subgroup’s characteristics areimputed to them. Thus, there are status gainsand losses from discrimination, and they maynot be what one expects. That is, it would bereassuring to find that all members of thebottom subgroup suffer from discriminationand that all members of the top subgroup gainfrom discrimination. But the reality is morecomplex, with some individuals in the bot-tom subgroup gaining status from discrimi-nation and some individuals in the top sub-group losing status from discrimination.19


small groups. To analyze S3 in smallgroups, we return to the case of a 12-membergroup which values one good and in whichthat good is perfectly associated with a bi-nary qualitative characteristic. The members’S1 was presented in Table 3, and the sub-groups’ S2 was presented in Table 8. Table10 reports the status gains and losses fromdiscrimination in each subgroup and the totalgain and loss, for all possible subgroup splitsfrom 1–11 to 11–1. For each subgroup, Table10 reports the Member ID numbers of thosewho gain and those who lose, and the per-centage of the subgroup who gain and lose.

For example, in the 3-9 subgroup split, thetwo lowest-ranking members (Member IDs 1and 2) gain from discrimination and the thirdmember of the bottom subgroup (Member ID3) loses from discrimination—within thesubgroup, then, 67 percent gain from dis-crimination and 33 percent lose from dis-crimination. Meanwhile, in the top subgroup,the bottom five members gain from discrimi-nation and the top four lose from discrimina-tion; thus, within the top subgroup, 56 per-cent gain from discrimination and 44 percentlose. Looking at the group as a whole, sevenmembers gain and five members lose, for to-tal percentages of 58 percent who gain fromdiscrimination and 42 percent who lose. Ofcourse, when the subgroup has only one per-son, as in the bottom subgroup of the 1–11split and the top subgroup of the 11–1 split,the person neither gains nor loses from dis-crimination; in those cases, the total gain andloss do not sum to 100.

Table 10 shows that the majority gainfrom discrimination in all subgroup splitsexcept the 8–4 (a tie) and the 11–1 (whichhas a plurality gaining from discrimination).If individuals care about their status, then inmost situations the majority will want to re-tain S3 even if information becomes avail-able to generate S1.

If affinities arise from similarity in inter-ests and if individuals care about their sta-tus, then there is a natural affinity betweenthose members of the bottom subgroup whogain from discrimination and those membersof the top subgroup who also gain from dis-crimination. Similarly, there is a natural af-finity between those members of the bottomsubgroup who lose from discrimination andthose members of the top subgroup who alsolose. Accordingly, the stage is set for coali-tions to form. These coalitions have the in-teresting property that proponents of dis-crimination are drawn from among the leastadvantaged of each subgroup, while oppo-nents of discrimination are drawn fromamong the most advantaged of each sub-group (regardless of the summary measureof S1 underlying S2). Thus, if the quantita-tive characteristic which confers S1 status iscorrelated with political skill, then oppo-nents of discrimination—although almost al-ways outnumbered—may by cunning winthe day.

18 Note that this result continues to hold evenif S2 is measured by the median of S1, or by anyother location measure, rather than by the meanof S1. In contrast, the proportions who gain andlose from discrimination depend on the summarymeasure of S1.

19 Note the importance of distinguishing be-tween use of S2 when S1 cannot be generatedand use of S2 when S1 is available. Note also thatsimilar reasonings may be used to analyze iden-tity and reference-group processes.


S3 in the special case of large so-

cieties. Next we analyze the status gainsand losses from discrimination in large soci-eties which value one good or several posi-tively associated goods. S1 in such societieswas reported in Table 7, and S2 in Table 9.Table 11 indicates that, for every subgroupsplit, majorities of both the bottom and topsubgroups gain from discrimination. Thesizes of these majorities differ between bot-tom and top subgroups and across the sub-group split. While in the top subgroup, thepercentages who gain and lose are constantacross all subgroup splits—63 percent al-ways gain and 37 percent always lose—inthe bottom subgroup, the percentage whogain increases with the relative size of thebottom subgroup, and the percentage wholose decreases.20

To flesh out Table 11, suppose that thebottom subgroup (or its leadership) decidesto fight against discrimination and the topsubgroup (or its leadership) decides to fightfor discrimination. Both the bottom and topsubgroups are vulnerable to defections; thatis, members may disagree with the leader-ship and refuse to participate in the fight oreven sabotage it. The figures in Table 11quantify the danger of defection. In the bot-tom subgroup, those who gain from dis-crimination are at risk of defecting; and inthe top subgroup, those who lose from dis-crimination are at risk of defecting. Accord-ingly, while the top subgroup has a constantrisk of defecting of approximately 37 per-cent of the membership, the bottom sub-group has a larger subset at risk of defect-ing—the lowest subset at risk of defecting isover half (50.2 percent in the .05 subgroupsplit)—and the subset at risk increases withthe subgroup’s relative size.

Thus, the bottom subgroup is more diffi-cult to discipline (to prevent defections) thanis the top subgroup. And the difficulty in-creases as the bottom subgroup increases inrelative size.

Table 10. Status Gains and Losses from Discrimination in a Group of Size 12


Gain Loss Gain LossS1i < E(S1) S1i > E(S1) S1i < E(S1) S1i > E(S1)

Total

Member Member Member Member Gain Loss

Size ID Percent ID Percent Size ID Percent ID Percent Percent Percent

1 .— .— .— .— 11 2–8 63.6 9–12 36.4 58.3 33.3

2 1 50.0 2 50.0 10 3–8 60.0 9–12 40.0 58.3 41.7

3 1–2 66.7 3 33.3 9 4–8 55.6 9–12 44.4 58.3 41.7

4 1–2 50.0 3–4 50.0 8 5–9 62.5 10–12 37.5 58.3 41.7

5 1–3 60.0 4–5 40.0 7 6–9 57.1 10–12 42.9 58.3 41.7

6 1–3 50.0 4–6 50.0 6 7–10 66.7 11–12 33.3 58.3 41.7

7 1–4 57.1 5–7 42.9 5 8–10 60.0 11–12 40.0 58.3 41.7

8 1–4 50.0 5–8 50.0 4 9–10 50.0 11–12 50.0 50.0 50.0

9 1–5 55.6 6–9 44.4 3 10–11 66.7 12 33.3 58.3 41.7

10 1–6 60.0 7–10 40.0 2 11 50.0 12 50.0 58.3 41.7

11 1–6 54.5 7–11 45.5 1 .— .— .— .— 50.0 41.7

Note: The entries in this table are obtained by comparing each person’ s S1 status from the rightmostcolumn of Table 3 (see page 103) with the subgroups’ S2 status (average S1 status) from Table 8 (see page112). When a subgroup has only one member, that member’ s S1 equals the average, and hence he or sheneither gains nor loses status from discrimination. In those cases (the top and bottom rows of the table), thepercentages of the entire group with status gains and losses from discrimination do not sum to 100; theresidual may be thought of as unaffected by discrimination.

20 Analysis of the formulas underlying thequantities in Table 11 (Appendix Table B) indi-cates that in the bottom subgroup, as the sub-group split approaches zero, the proportion whogain from discrimination approaches its lowerlimit of .5 and the proportion who lose from dis-crimination approaches its upper limit of .5.


Although the subgroup-specific patternsare monotonic (or constant, as in the top sub-group), they combine to form totals whichare nonmonotonic. As shown in Table 11, theoverall percentage who gain first decreasesas the relative size of the bottom subgroupincreases, then when the bottom subgroupcontains approximately 70 percent of thepopulation, the overall percentage who gainbegins to increase. Meanwhile, the overallpercentage who lose first increases, then,when the bottom subgroup contains approxi-mately 70 percent of the population—mir-roring the pattern among the percentage whogain—the percentage who lose begins to de-crease. However, although the percentageswho gain and lose vary, a majority alwaysgains. Thus, in a contest adjudicated by ma-jority rule, and in which everyone voted andeveryone voted for the platform under which

their own status would be highest, discrimi-nation would win.

However, as noted above, the proponentsof discrimination are drawn from the bottomranks of each subgroup, while the opponentsof discrimination are drawn from the topranks of each subgroup. Thus, if the status-conferring quantitative characteristic is cor-related with political skill, then opponents ofdiscrimination—though always outnum-bered—may be able to snatch victory.

A different way to approach these resultsis to assess the sources of support for andopposition to discrimination. Table 12 pre-sents, separately for the subset who gain sta-tus from discrimination and presumably sup-port it and the subset who lose status fromdiscrimination and presumably oppose it, thepercentage drawn from the bottom subgroupand the percentage drawn from the top sub-

Table 11. Status Gains and Losses from Discrimination in Large Societies


Gain Loss Gain Loss Total

RelativeS1i < E(S1) S1i > E(S1)

RelativeS1i < E(S1) S1i > E(S1) Gain Loss

Size Rank Percent Rank Percent Size Rank Percent Rank Percent Percent Percent

.05 0–2.5 50.2 2.5–5 49.8 .95 5–65 63.2 65–100 36.8 62.6 37.4

.10 0–5 50.4 5–10 49.6 .90 10–67 63.2 67–100 36.8 61.9 38.1

.15 0–8 50.7 8–15 49.3 .85 15–69 63.2 69–100 36.8 61.3 38.7

.20 0–10 50.9 10–20 49.1 .80 20–71 63.2 71–100 36.8 60.8 39.2

.25 0–13 51.2 13–25 48.8 .75 25–72 63.2 72–100 36.8 60.2 39.8

.30 0–15 51.5 15–30 48.5 .70 30–74 63.2 74–100 36.8 59.7 40.3

.35 0–18 51.8 18–35 48.2 .65 35–76 63.2 76–100 36.8 59.2 40.8

.40 0–21 52.1 21–40 47.9 .60 40–78 63.2 78–100 36.8 58.8 41.2

.45 0–24 52.5 24–45 47.5 .55 45–80 63.2 80–100 36.8 58.4 41.6

.50 0–26 52.8 26–50 47.2 .50 50–82 63.2 82–100 36.8 58.0 42.0

.55 0–29 53.3 29–55 46.7 .45 55–83 63.2 83–100 36.8 57.7 42.3

.60 0–32 53.7 32–60 46.3 .40 60–85 63.2 85–100 36.8 57.5 42.5

.65 0–35 54.2 35–65 45.8 .35 65–87 63.2 87–100 36.8 57.4 42.6

.70 0–38 54.8 38–70 45.2 .30 70–89 63.2 89–100 36.8 57.3 42.7

.75 0–42 55.5 42–75 44.5 .25 75–91 63.2 91–100 36.8 57.4 42.6

.80 0–45 56.2 45–80 43.8 .20 80–93 63.2 93–100 36.8 57.6 42.4

.85 0–49 57.2 49–85 42.8 .15 85–94 63.2 94–100 36.8 58.1 41.9

.90 0–52 58.3 52–90 41.7 .10 90–96 63.2 96–100 36.8 58.8 41.2

.95 0–57 60.0 57–95 40.0 .05 95–98 63.2 98–100 36.8 60.1 39.9

Note: The entries in this table are obtained by applying the mathematical formulas in Appendix Table B,which compare each person’ s S1 status, as in the second column of Table 7 (see page 108), with the sub-groups’ S2 status (average S1 status), as in Table 9 (see page 113).


group. The percentages from the bottom andtop subgroups sum to one. For example, inthe case where 25 percent of the populationis in the bottom subgroup (the fifth row),support for discrimination relies heavily onthe top subgroup, which constitutes 78.7 ofits base; concomitantly, opposition to dis-crimination also relies heavily on the topsubgroup, which provides 69.3 percent of itsconstituency.

The sources of support and opposition todiscrimination operate monotonically. As thepercentage in the bottom subgroup in-creases, the percentages of both the supportand opposition constituencies drawn fromthe bottom subgroup increase, and the per-centages drawn from the top subgroup de-crease.

A final way to examine the status gainsand losses from discrimination is to graphthe proportions of the entire populationwho are in each of the four subsets (thosewho gain and are in the bottom subgroup,those who lose and are in the bottom sub-group, those who gain and are in the topsubgroup, and those who lose and are inthe top subgroup) as a function of the rela-tive size of the bottom subgroup. Figure 4presents these plots. As shown, the twosubsets from the top subgroup decrease lin-early, the gain subset from the bottom sub-group increases nonlinearly, and the losssubset from the bottom subgroup increasesthroughout most of the range but thenshifts direction (at a subgroup split of ap-proximately .955).

Table 12. Sources of Support for and Opposition to Discrimination, by Population Subgroup Split

Support for Discrimination Opposition to Discrimination

Percentage Percentage Percentage Percentage Percentagein Bottom from Bottom from Top from Bottom from TopSubgroup Subgroup Subgroup Subgroup Subgroup

5 4.0 96.0 6.7 93.4

10 8.1 91.9 13.0 87.0

15 12.4 87.6 19.1 80.9

20 16.8 83.2 25.0 75.0

25 21.3 78.7 30.7 69.3

30 25.9 74.1 36.1 63.9

35 30.6 69.4 41.4 58.6

40 35.5 64.5 46.5 53.5

45 40.4 59.6 51.4 48.6

50 45.5 54.5 56.2 43.8

55 50.7 49.3 60.8 39.2

60 56.0 44.0 65.4 34.6

65 61.4 38.6 69.8 30.2

70 66.9 33.1 74.1 25.9

75 72.5 27.5 78.4 21.6

80 78.1 21.9 82.6 17.4

85 83.7 16.3 86.8 13.2

90 89.2 10.7 91.1 8.9

95 94.7 5.3 95.4 4.6

Note: The population subgroup split is represented by the percentage in the bottom subgroup in the leftmostcolumn. The sources of support for discrimination sum to 100 percent, as do the sources of opposition todiscrimination. Thus, for example, in a society split into two equal-sized subgroups, the majority of support-ers for discrimination are drawn from the top subgroup (54.5 percent), and the majority of opposers to dis-crimination are drawn from the bottom subgroup (56.2 percent). The entries in this table are obtained byapplying the formulas in Appendix Table B.


Because the proportions in the four sub-sets sum to one, Figure 4 enables assessmentof which sets dominate, by population split.For example, in societies in which the pro-portion in the bottom subgroup is less than alittle over half, the largest subset in the soci-ety is the one composed of top-subgroupmembers who gain from discrimination.Similarly, at the point where the top-gainand bottom-gain subsets intersect (i.e., whenthe proportion in the bottom subgroup is alittle over half), the two subsets (top-gainand bottom-gain) are of equal size.

It is especially interesting to examine Fig-ure 4 at the points corresponding to an equalpopulation split. As shown, the smallest sub-set consists of individuals from the top sub-group who lose from discrimination (18 per-cent), and the largest subset consists of indi-viduals from the top subgroup who gain fromdiscrimination (32 percent); the two subsetsfrom the bottom subgroup occupy intermedi-ate places (26 percent in the bottom-gain sub-set and 24 percent in the bottom-lose sub-set). Given that the gender split is approxi-mately fifty-fifty, it would be useful to reex-amine data generated by experiments on gen-der and status in light of these results.

How to Change (or Keep) S3

Third-order status arises because of the lackof information on quantitative personal char-acteristics; but provision of such informationdoes not guarantee its elimination. Nonethe-less, provision of the information may beuseful in eliminating S3—for those whowish to eliminate it. Perhaps the most in-triguing result is that some members of thedisadvantaged subgroup gain from discrimi-nation, suggesting another mechanism pro-ducing and maintaining “ internalized op-pression” (Bourdieu 1997; Ridgeway 1997a:222; Stanton-Salazar 1997). Thus, devisingstrategies to change S3 entails devising strat-egies for all actors.

Those who gain status from discriminationhave as their objective to maintain S3 as afunction exclusively of S2; introducing S1,even mildly weighted, diminishes their sta-tus. Accordingly, a useful strategy for themis to, first, prevent the free flow of informa-tion, second, discount the information or dis-pute its accuracy, and, third, make argumentsfor ignoring individual characteristics andfocusing instead on subgroup membership orother communal considerations. Because

Loss in bottomsubgroup

.75

.50

.25

0

0 .25 .50 .75 1

Relative Size of the Bottom Subgroup

Pro

po

rtio

n in

th

e P

op

ula

tio

n

Gain in topsubgroup

Loss in topsubgroup

Gain in bottomsubgroup

Figure 4. Status Gains and Losses from Discrimination

Note: For a given population subgroup split (represented by the relative size of the bottom subgroup), theproportions in the four subsets sum to one. Formulas are reported in Appendix Table B; related values arereported in Table 11 (see page 118).


there are people in both subgroups who gainfrom discrimination, they will form partner-ships; these partnerships can then be used tounderscore the amicable relations that obtainbetween the two subgroups.

The strategy for those who lose statusfrom discrimination is exactly the oppo-site—except that they will still form partner-ships with their natural allies in the othersubgroup. For these individuals, the goal isto increase information, make it widelyavailable, express confidence in it, extol in-dividualism.

Finally, a little insight for political activ-ists and organizers, who often must go intounfamiliar situations: In both subgroups, thelowest-ranking individuals (on the quantita-tive characteristic that confers S1 status) arelikely to support discrimination, and thehighest-ranking individuals are likely to op-pose discrimination. Thus, lobbyists for dis-crimination should seek out low-rankingpersons, and lobbyists against discriminationshould seek out high-ranking persons.

Further Implications

S3 status provides fertile ground for sub-stantial further theorizing. One avenue in-volves the effects of status processes onidentity phenomena. Suppose that activationof a subgroup identity produces status gainsand losses, exactly as in the analysis of dis-crimination. Then, among other things, twoceteris paribus implications follow: First, inself-report surveys, the higher-ranking ineach subgroup formed by a qualitative char-acteristic (such as race or ethnicity) will beless likely to answer subgroup-identificationquestions; thus, average schooling, skills,and income will be underestimated in allsubgroups. Second, in face-to-face inter-views, if interviewer characteristics or be-havior activate subgroup identity, there willbe a tendency for higher-ranking prospec-tive respondents to decline to participate inthe survey; thus, nonresponse will be dis-proportionately greater among persons withhigher schooling, skill, and income.

Finally, suppose that the cross-subgroupcoalitions discussed above lead to formationof political parties—an “ individualistic”party comprising those who lose from dis-crimination (or from subgroup identifica-

tion) and a “collectivistic” party comprisingthose who gain from discrimination (or fromsubgroup identification). Then the individu-alistic party is predicted to be smaller buthigher-skilled.21

CONCLUDING NOTE

This paper presented an integrated frame-work for studying status. The frameworkcombines ideas and insights from several lit-eratures in order to address long-standing,unresolved issues in status research, such as:(1) the emergence of status; (2) how to dis-tinguish between, and measure, the status ofindividuals and the status of characteristics;(3) how to measure and understand the sta-tus gap between subgroups (e.g., betweenmen and women, or between races); and (4)how to distinguish the operation of quantita-tive and qualitative characteristics in theproduction of status.

By identifying three distinct types of sta-tus and linking them in distinctive ways toquantitative and qualitative characteristics,the framework makes it possible to analyzea wide variety of status phenomena in abroad range of groups and societies.

The new status theory yields many test-able implications and, based on the work todate, appears capable of yielding manymore implications beyond the ones pre-sented here. The initial set includes impli-cations for the effects of (1) the number andintercorrelation of personal characteristics,(2) the availability of information aboutpersonal characteristics, and (3) the propor-tions of a group in each category of a quali-tative characteristic. The derived implica-tions cover such phenomena as status dif-ferences between group members, statusgaps between subgroups, and overall in-equality in the status structure, all undervarying conditions. For example, the analy-ses suggest that under certain given condi-tions: (1) status inequality is lower if thevalued goods are negatively correlated; (2)in a two-subgroup society, the least advan-taged from both subgroups gain status fromdiscrimination, and the most advantaged

21 The “Florida phase” of the 2000 U.S. presi-dential election may provide a case in point (seeNagourney and Barstow 2000).


from both subgroups lose status from dis-crimination, leading to cross-subgroupcoalitions; (3) the status gap between twosubgroups increases with the relative size ofthe disadvantaged subgroup; (4) when twosubgroups are fighting for and against dis-crimination, it is more difficult to preventdefections in the bottom subgroup than inthe top subgroup; and (5) opponents of dis-crimination are outnumbered.

The new framework for status analysisopens many avenues for future work—refin-ing the basic status functions, building theo-ries and deriving more and sharper implica-tions, and testing the implications and usingthe new measures for the status of personsand the status of characteristics to assess keystatus phenomena and their correlates in sur-veys and experiments.

Ahead, a further integration looms on thehorizon. The status analyzed here refers toevaluations of the worth of individuals andcharacteristics (footnote 1). This kind ofstatus is related to the other major kind ofstatus of interest in social science—the“ status” in status attainment (Sewell andHauser 1972, 1992)—in that the status instatus attainment consists of the character-istics which confer the evaluation kind ofstatus. Ultimately, this evaluation status

cannot be fully understood without under-standing how individuals come to have thecharacteristics they bring to the socialarena—and, importantly, the two processesmay be dynamically linked. Thus, a morecomplete framework than the one devel-oped in this paper would integrate the twokinds of status, providing fertile groundfor new substantive and methodologicalsynergies.

Guillermina Jasso is Professor of Sociology atNew York University. Recent work includes:“Trends in the Experience of Injustice” (SocialJustice Research, 2000), “The Changing Skill ofNew Immigrants to the United States” (with M.R. Rosenzweig and J. P. Smith, in Issues in theEconomics of Immigration, edited by G. J.Borjas, University of Chicago Press, 2000),“Some of Robert K. Merton’s Contributions toJustice Theory” (Sociological Theory, 2000),“The New Immigrant Survey Pilot (NIS-P)” (withD. S. Massey, M. R. Rosenzweig, and J. P. Smith,Demography, 2000), “How I Became a Theo-rist,” (Sociological Theory, 2000), “AssortativeMating among Married New Legal Immigrants tothe United States” (with D. S. Massey, M. R.Rosenzweig, and J. P. Smith, International Mi-gration Review, 2000), and “Rule Finding aboutRule Making: Comparison Processes and theMaking of Norms” (in Social Norms, edited byM. Hechter and K.-D. Opp, Russell Sage, 2001).

Appendix Table A. Principal Functions and Parameters of the S1 Status Distribution,in Four Special Cases

VariateCase Family f(x) F(x) Q(α) Q(0) E(X) Q(1)

One Characteristic Exponential e–x 1 – e–x ln1

1−

α

0 1 ∞

Two Characteristics:

Positively associated Exponential e –x 1 – e–x ln1

1−

α

0 1 ∞

Negatively associated Unnamed4

42

e

e

x

x

−

−1

42−

e xln

2

1 2−

αln2 1 ∞

Independent Erlang 4xe–2x 1 2 12− +[ ]−e xx ( ) — 0 1 ∞

Note: This appendix table presents principal functions (probability density function f(x), cumulative distributionfunction F(x), and quantile function Q(α)) and parameters of the S1 status distribution arising in four special casesdefined by the configuration of valued personal characteristics. In the three two-good cases, the two goods areequally weighted.


REFERENCES

Bales, Robert Freed. 1999. Social InteractionSystems: Theory and Measurement. NewBrunswick, NJ, and London, England: Trans-action.

Barber, Bernard. 1968. “ Stratification, Social:Introduction.” Pp. 288–96 in International En-cyclopedia of the Social Sciences, vol. 15, ed-ited by D. L. Sills. New York: Macmillan.

Berger, Joseph, Hamit Fisek, Robert Norman,and Morris Zelditch. 1977. Status Character-istics and Social Interaction: An ExpectationStates Approach. New York: Elsevier.

Berger, Joseph, Cecilia Ridgeway, M. HamitFisek, and Robert Z. Norman. 1998. “The Le-gitimation and Delegitimation of Power andPrestige Orders.” American Sociological Re-view 63:379–405.

Berger, Joseph, Susan J. Rosenholtz, and MorrisZelditch. 1980. “Status Organizing Processes.”Annual Review of Sociology 6:479–508.

Blau, Peter M. 1977. Inequality and Heterogene-ity. New York: Free Press.

Bourdieu, Pierre. 1997. “Masculine DominationRevisited.” The Goffman Prize Lecture. Trans-lated and edited by Loic Wacquant. BerkeleyJournal of Sociology 41:189–203.

Fararo, Thomas J. 1989. The Meaning of GeneralTheoretical Sociology: Tradition and Formal-

ization. Cambridge, England: Cambridge Uni-versity Press.

Friedkin, Noah E. 1998. A Structural Theory ofSocial Influence. Cambridge, England: Cam-bridge University Press.

Goode, William J. 1978. The Celebration of He-roes: Prestige as a Control System. Berkeley,CA: University of California Press.

Homans, George C. 1967. “Fundamental SocialProcesses.” Pp. 27–78 in Sociology: An Intro-duction, edited by N. J. Smelser. New York:Wiley.

Lipset, Seymour Martin. 1968. “ Stratification,Social: Social Class.” Pp. 296–316 in Interna-tional Encyclopedia of the Social Sciences,vol. 15, edited by D. L. Sills. New York:Macmillan.

Marmot, Michael. 2000. “Multilevel Approachesto Understanding Social Determinants.” Pp.349–67 in Social Epidemiology, edited by L.F. Berkman and I. Kawachi. Oxford, England:Oxford University Press.

Merton, Robert K. [1949, 1957] 1968. SocialTheory and Social Structure. 1968. Enlargeded. New York: Free Press.

Nagourney, Adam and David Barstow. 2000.“ G.O.P.’ s Depth Outdid Gore’ s Team inFlorida.” The New York Times, December 22,

Appendix Table B. Formulas for Calculating Gains and Losses from Discrimination

Quantity Bottom Subgroup Top Subgroup

(1) Relative rank at which S1 = E(S1) 11 1

1

1

− −

−

e p

p

p1

1−

− p

e

(2) Those who gain status from discrimination,1

1 1

1

1

− −

−

e p

p

p1 1

1−( ) −

peas a proportion of the population

(3) Those who lose status from discrimination,p

e p

p

p− +

−

−

11 1

1

1

1− p

eas a proportion of the population

(4) Those who gain status from discrimination,1

1 11

1

− −

−

e p

p

p

p

11

−eas a proportion of the subgroup

(5) Those who lose status from discrimination,p

e p

p

p

p− +

−

−

11 1

1

1

1

eas a proportion of the subgroup

Note: In the formulas, p denotes the relative size of the bottom subgroup. The four proportions given inrows 2 and 3—representing those who gain and lose status from discrimination, in both subgroups. as pro-portions of the population—sum to one. In each subgroup, the two proportions in rows 4 and 5—represent-ing those who gain and lose status from discrimination, as a proportion of the subgroup—sum to one.


2000, pp. A1, A29. (www.nytimes.com)Parsons, Talcott. [1949] 1964. Essays in Socio-

logical Theory. Rev. ed. New York: FreePress.

Ridgeway, Cecilia L. 1991. “ The Social Con-struction of Status Value: Gender and OtherNominal Characteristics.” Social Forces 70:367–86.

———. 1997a. “The Conservation of Gender In-equality.” American Sociological Review 62:218– 35.

———. 1997b. “Where Do Status-Value BeliefsCome From? New Developments.” Pp. 137–58in Status, Networks, and Structure, edited byJ. Szmatka, J. Skvoretz, and J. Berger.Stanford, CA: Stanford University Press.

Ridgeway, Cecilia L. and James Balkwell. 1997.“Group Processes and the Diffusion of Status-Beliefs.” Social Psychology Quarterly 60:14–31.

Rossi, Peter H. 1979. “Vignette Analysis: Uncov-ering the Normative Structure of ComplexJudgments.” Pp. 176–86 in Qualitative andQuantitative Social Research: Papers inHonor of Paul F. Lazarsfeld, edited by R. K.Merton, J. S. Coleman, and P. H. Rossi. NewYork: Free Press.

Sapolsky, Robert. 1993. “ Endocrinology Al-fresco: Psychoendocrine Studies of Wild Ba-boons.” Recent Progress in Hormone Research48:437–68.

Sewell, William H. and Robert M. Hauser. 1972.Education, Occupation, and Earnings:Achievement in the Early Career. New York:Academic.

———. 1992. “A Review of the Wisconsin Lon-gitudinal Study of Social and PsychologicalFactors in Aspirations and Achievements,1963–1990.” Working Paper 92-01, Center forDemography and Ecology, University of Wis-consin, Madison.

Sørensen, Aage B. 1979. “A Model and a Metricfor the Analysis of the Intragenerational Sta-tus Attainment Process.” American Journal ofSociology 85:361–84.

Skvoretz, John and Thomas J. Fararo. 1996. “Sta-tus and Participation in Task Groups: A Dy-namic Network Model.” American Journal ofSociology 101:1366–1414.

Smith, James P. 1999. “ Healthy Bodies and

Thick Wallets: The Dual Relation betweenHealth and Economic Status.” Journal of Eco-nomic Perspectives 13:145–66.

Stanton-Salazar, Ricardo D. 1997. “ A SocialCapital Framework for Understanding the So-cialization of Racial Minority Children andYouths.” Harvard Educational Review 67:1–40.

Stinchcombe, Arthur L. 1968. “Stratification, So-cial: The Structure of Stratification Systems.”Pp. 325–32 in International Encyclopedia ofthe Social Sciences, vol. 15, edited by D. L.Sills. New York: Macmillan.

Stuart, Alan and J. Keith Ord. 1987. Kendall’sAdvanced Theory of Statistics. Vol. 1, Distri-bution Theory. 5th ed. Originally authored bySir Maurice Kendall. New York: Oxford Uni-versity Press.

Turner, Jonathan H. 1984. Societal Stratification:A Theoretical Analysis. New York: ColumbiaUniversity Press.

———. 1995. Macrodynamics: Toward a Theoryon the Organization of Human Populations.ASA Rose Monograph Series. New Bruns-wick, NJ: Rutgers University Press.

Veblen, Thorstein. [1899] 1953. The Theory ofthe Leisure Class: An Economic Study of Insti-tution. Rev. ed. Reprint, New York: NewAmerican Library.

Wagner, David G. and Joseph Berger. 1993.“Status Characteristics Theory: The Growth ofa Program.” Pp. 23–63 in Theoretical Re-search Programs: Studies in the Growth ofTheory, edited by J. Berger and M. Zelditch Jr.Stanford, CA: Stanford University Press.

Weber, Max. [1922] 1978. Economy and Soci-ety: An Outline of Interpretive Sociology.Translated and edited by G. Roth and C.Wittich. Berkeley, CA: University of Califor-nia Press.

Webster, Murray, Jr. and Stuart J. Hysom. 1998.“ Creating Status Characteristics.” AmericanSociological Review 63:351–78.

Wilkenson, Richard G. 1996. Unhealthy Societ-ies: The Afflictions of Inequality. London, En-gland: Routledge.

Zelditch, Morris, Jr. 1968. “Status, Social.” Pp.250–57 in International Encyclopedia of theSocial Sciences, vol. 15, edited by D. L. Sills.New York: Macmillan.

DISTRIBUTIVE JUSTICE PERCEPTIONSDISTRIBUTIVE JUSTICE PERCEPTIONSDISTRIBUTIVE JUSTICE PERCEPTIONSDISTRIBUTIVE JUSTICE PERCEPTIONSDISTRIBUTIVE JUSTICE PERCEPTIONS 125125125125125


Justice Processes:

Specifying the Mediating Role of

Perceptions of Distributive Justice

Extant theories suggest that individuals’ perceptions of the fairness of their paycausally intervene between the salary/wages they receive and their emotional re-sponses (e.g., satisfaction) to that level of pay. In addition, it has been argued thatthe impact of an event evaluated by an individual as unfair depends on the impor-tance of fairness to that individual—an unfair event has a greater effect for thosewho place greater importance on distributive justice. Despite the centrality of thesearguments in the justice literature, current research has not adequately tested them.In this article, the authors propose a general theoretical model based on these argu-ments. A structural equation model is then estimated using data from a nationalsample of Protestant clergy. The findings support both the mediating role of minis-ters’ perceptions of distributive justice and the moderating role of the importance ofjustice in explaining their level of pay satisfaction.

the extent to which an individual’s percep-tion that a reward is unfair causally mediatesthe effect that reward has on cognitive, be-havioral, and emotional responses.

We model the causal relationships be-tween the reward conditions an individualfaces (especially relative to what he or shebelieves is fair), the importance the indi-vidual places on justice, the individual’s per-ception of the degree of injustice present inthe situation, and the individual’s emotionalresponse to the situation.1 Although our

ustice is a rather murky concept in so-cial psychology, perhaps because of its

various uses in common discourse. In gen-eral, the term distributive justice refers to thefairness of the outcomes or rewards that anindividual or group receives. This can becontrasted to procedural justice (i.e., the fair-ness of the process by which outcomes aredetermined) and interactional justice (i.e., thefairness of interpersonal treatment). Al-though the focus of research has recentlyshifted somewhat toward the latter two formsof justice, our understanding of distributivejustice processes is by no means complete. Acritical question that remains unanswered is

J

C. Wesley Younts

University of IowaCharles W. Mueller

University of Iowa

Direct all correspondence to C. WesleyYounts, Department of Sociology, University ofIowa, W140 Seashore Hall, Iowa City, IA 52242([email protected]). The LouisvilleInstitute provided grant support for collection ofthe data used in this study. We thank BarryMarkovsky, Lisa Troyer, Stacey DeCoster, sev-eral anonymous ASR reviewers, and the ASR Edi-tors for their helpful comments on early drafts.An earlier version of this paper was presented atthe annual meeting of the American SociologicalAssociation, Chicago, Illinois, August 1999.

1 Although most theories claim that justice per-ceptions can lead to cognitive, behavioral, andemotional responses (Jasso 1986, 1993; Markand Folger 1984), we focus here on emotionalresponses, and in particular on individuals’ ex-pressions of (dis)satisfaction regarding their re-wards. An individual’ s emotional response (e.g.,satisfaction, anger) is distinct from the affectivecomponent of justice evaluations, often referredto as the “ sense of injustice” or “ injustice experi-ence” (Jasso 1993). It is also important to distin-guish the perceptual components of justice evalu-ations (e.g., justice perceptions) from the emo-tional consequences of such evaluations (e.g., re-ward satisfaction)—an individual’ s assessment


theoretical model is based on the argumentscontained in extant theoretical perspectives,the pivotal concept in our model—“ justiceperceptions”— is seldom explicitly defined,and its causal position within the justiceevaluation process is usually only loosely, ifat all, specified. Our work adds to the theo-retical literature by more explicitly definingand specifying the causal position of thiscritical construct.

The looseness in how the concept of jus-tice perceptions is often used is rarely chal-lenged by justice theorists because therehave been few attempts to empirically assesswhether justice perceptions actually servethe mediating role suggested in the theoreti-cal literature. In response to this oversight,we estimate a structural equation model thatrelates workplace rewards (including theircomparison with the rewards believed to befair) to pay satisfaction through individuals’perceptions of distributive injustice. We usesurvey data collected from a national sampleof Protestant ministers.

JUSTICE PERCEPTIONS

For social psychologists, the notion that per-ceptions mediate the effects of concrete so-cial conditions on individuals’ responses isby no means novel—it is one of the corner-stones of the field. The symbolic inter-actionist tradition has long held such a con-ception of social reality, represented by suchconstructs as the “Thomas theorem” (Tho-mas and Thomas 1928) and the “ looking-glass self” (Cooley 1902). Each of these ab-stract concepts shares the general notion thatsocial behavior is based on the actor’s inter-pretation, or “ definition,” of the situation,and thus behavior cannot be directly pre-dicted from the characteristics of the objec-tive stimuli an actor faces (also see Goffman1959).

Although the mediating role of percep-tions has long been accepted as an orientingprinciple in sociological social psychology,its relevance for understanding distributivejustice processes has not been fully ex-ploited theoretically, nor has it been empiri-

cally demonstrated. The majority of socialpsychological theories employing the dis-tributive justice concept recognize (at leastimplicitly) that individuals’ perceptions ofinjustice in a particular situation intervenebetween the reward conditions they face andthe cognitive, behavioral, and emotional out-comes of interest. However, as Hegtvedt andMarkovsky (1995:272) argue, the perceptualcomponent of justice evaluations has re-ceived inadequate theoretical attention.Since the classical equity (Adams 1965;Homans 1974; Walster, Walster, and Bers-chied 1978) and relative deprivation per-spectives (Crosby 1976; Folger 1986) begandominating the justice literature more than30 years ago, primary attention has beenpaid to specifying (1) the comparison pro-cesses that are assumed to lead to percep-tions of injustice (i.e., justice perceptions asa dependent variable, or “ends” ), and (2) thecognitive, behavioral, and emotional conse-quences of injustice (i.e., justice perceptionsas an independent variable, or “ means” ).Both “ ends” and “ means” conceptions ofjustice contain, at their heart, the assumptionthat it is an individual’s perception of thedegree of injustice existing in a situation thatdrives his or her emotional response to re-ward conditions—if an individual does notperceive a situation as unfair, he or she willnot express negative emotions regarding thatsituation (Hegtvedt and Markovsky 1995;Törnblom 1992).

Extant research supports the importance ofjustice perceptions as both a dependent andan independent variable.2 Unfortunately,

of the fairness of rewards is conceptually and em-pirically distinct from his or her expressed satis-faction with them.

2 Most empirical investigations of distributivejustice processes also tend to follow either the“ ends” or the “ means” conceptions of justice.Research using justice perceptions as ends ofteninvolves experiments in which researchers ma-nipulate the rewards individuals receive andthose received by a social referent or implied bynormative standards; subjects are then asked toreport their perceptions of the degree of injusticein the situation (Anderson 1976; Jasso 1978;Mellers 1982). Studies treating justice percep-tions as means often use surveys in which mea-sures of justice perceptions predict individuals’reward satisfaction and their organizational com-mitment (McFarlin and Sweeney 1992; Muellerand Wallace 1996; Sweeney and McFarlin1993).


however, there is a lack of empirical evi-dence directly demonstrating that justiceperceptions act as a mediator in justiceevaluation processes. Given the significanceof this issue and the absence of demonstra-tive data, we believe it is important to em-pirically test the assumed mediating role ofjustice perceptions that has been at the rootof distributive justice theories since their in-ception.

Specifying a Causal Model of

Justice Evaluation Processes

We have developed a general theoreticalmodel (see Figure 1) in which justice per-ceptions explicitly mediate the effects of re-ward conditions on individuals’ emotionalresponses to these conditions. At the centerof our model are four distinct constructs thatmake up the justice evaluation process—thejustice evaluation, justice perceptions, jus-tice importance, and the emotional responseto perceived injustice. Below, we explicitlydefine these concepts and specify their cau-sal position within the model. As shown inFigure 1, we assume that the reward condi-tions individuals face relative to the rewardsthat are believed to be fair (i.e., the justiceevaluation) directly affect individuals’ per-

ceptions of distributive injustice in the situ-ation. However, we also assume that the im-portance of justice to the individual moder-ates the effect of unfair rewards on justiceperceptions, such that the more importantjustice is, the greater the effect of injusticeon the individual. Finally, we assume thatjustice perceptions directly affect the indi-vidual’s emotional response to the situationand mediate the effects of justice evalua-tions.

Reward Conditions and Justice

Evaluations

In our model, reward conditions refer to theabsolute magnitude of the rewards (e.g., pay,benefits) existing in a given situation. Ac-cording to a basic “ self-interest model,” thereward conditions faced by an actor have animportant direct impact on his or her ex-pressed emotions—the more rewards an in-dividual receives, the more “ positive” theemotional reaction. For instance, Randalland Mueller (1995) found partial empiricalsupport for a parsimonious self-interestmodel relating the absolute magnitude ofworkplace rewards and privileges to em-ployees’ job satisfaction and organizationalcommitment.

JUS

TIC

E E

VA

LU

AT

ION

PR

OC

ES

SReward conditions(pay)

Pay justice evaluation(actual pay vs. just pay)

Distributive justice perceptions:

• Pay distributive justice

• Global distributive justice

Emotional response(pay satisfaction)Justice importance

Pay justice evaluation × Justice importance

Control variables

εε

Figure 1. Causal Model of Justice Evaluations

Note: All exogenous variables are allowed to correlate. Disturbances for pay and global distributive jus-tice perceptions are allowed to correlate.


Contrary to the self-interest model, our“ justice model” emphasizes a comparisonprocess, suggesting that it is the magnitudeof rewards relative to some standard of fair-ness that generates subjective states of injus-tice and determines the resulting reactions.3

This view of the “ relativity” of rewards isexemplified by classical equity theories(Adams 1965; Homans 1974; Walster et al.1978) and relative deprivation formulations(Crosby 1976). More recent sociologicalperspectives have expanded the scope ofpossible referents, allowing comparisons toother individuals, groups, prior experiences,imagined alternatives, and normative stan-dards (Folger 1986; Jasso 1986; Markovsky1991). We follow Markovsky (1985) in as-suming, as a scope condition, that individu-als evaluating the fairness of rewards al-ready possess a principle of fairness withwhich to assess outcomes. Therefore, the rel-evant questions are: Given an activated stan-dard of distributive justice, how do particu-lar levels of over- or underreward (relativeto that standard) translate into specific de-grees of perceived distributive injustice?And, how are specific degrees of perceivedinjustice translated into measurable emo-tional reactions in specific situations (Jassoand Wegener 1997:394)?

Jasso (1980) has labeled the comparisonof actual rewards to just rewards the justiceevaluation; we adopt this term to representthe rewards individuals’ receive relative totheir activated justice standard. Specifi-cally, Jasso (1980) proposes that an indivi-dual’s sense of injustice can be expressed

by the following justice evaluation (JE)equation:4

JE =

ln ,

actual share

just share (1)

where ln is the natural logarithm, “ actualshare” is the amount of reward received inthe situation, and “ just share” is the amountof reward believed to be fair by the indi-vidual (for a more detailed discussion of thedetermination of the just share, see Jasso andWegener 1997:399).

Distributive injustice is assumed to be dis-tressing (Adams 1965; Homans 1974; Mar-kovsky 1988; Törnblom 1992) and is ex-pected to lead to cognitive, behavioral, andemotional responses (Jasso 1986; Mark andFolger 1984). Focusing on emotional re-sponses, we can specify an alternative to theself-interest model—the greater the discrep-ancy between the actual and just shares, themore “negative” the individual’s emotionalresponse to those rewards (Jasso 1993).Consistent with this argument, justice re-searchers have empirically demonstratedthat the greater the discrepancy between thepay/benefits received and those felt to be de-served, the more likely an individual is tomake a formal complaint (Markovsky 1985),the lower his or her satisfaction with thoserewards (Randall and Mueller 1995), and thegreater the resentment he or she will directat the perceived perpetrator (Folger et al.1983). With reference to the theoreticalmodel presented in Figure 1, we assume thatan individual’s justice evaluation has a totalcausal effect on his or her emotional re-sponse to the situation, net of the absolutelevel of rewards received.

The Mediating Role of

Justice Perceptions

We use the term justice perceptions to referto an individual’s assessment of the magni-

3 We recognize that an individual may base theoutcome he or she feels is fair (i.e., the just share)on a principle of “ self-interest,” such that the in-dividual feels he or she deserves the majority ofrewards (Leventhal, Karuza, and Fry 1980; Mes-sick and Sentis 1979). Although this allows a re-lationship between self-interest and notions offairness, the distinction between the justice andself-interest models remains. The justice modeltreats self-interest as one of the possible prin-ciples for determining the just share againstwhich actual outcomes are compared, and thiscomparison is assumed to lead to perceptions offairness and subsequently to observable reac-tions. On the other hand, the self-interest modeltreats the absolute magnitude of rewards as a di-rect determinant of individuals’ responses.

4 Several specific functional forms have beensuggested to represent the properties of justiceevaluations (Anderson 1976; Jasso 1980; Mar-kovsky 1985, 1991; Mellers 1982). We adoptJasso’ s (1980) specification because of its gen-eral acceptance in the sociological literature andits mathematical properties (for refinements ofthis mathematical model, also see Jasso andWegener 1997).


tude of distributive injustice in a particularsituation.5 Our primary contribution here isour assessment of the argument that justiceperceptions mediate the impact of rewardconditions and distributive justice evalua-tions on individuals’ reactions to injustice.Three basic assumptions are embodied in theargument that a variable (Y) mediates the ef-fects of another variable (X) on a given re-sponse variable (Z): (1) X is assumed to havea direct causal effect on Y, (2) Y is assumedto have a direct causal effect on Z, and (3)the effect of X on Z is assumed to be indi-rect, working through X’s effect on Y.

The first assumption suggests that justiceevaluations directly impact an individual’sperceptions of injustice in the situation. Al-though equity researchers have focused pri-marily on the link between “objective” ineq-uities and behavioral reactions, Adams(1965) stressed that perceptions play an im-portant role in linking the two. As Törnblom(1992) has noted, “perceived discrepanciesbetween the perceived actual and perceivedideal (expected) match between inputs andoutputs [are] assumed to result in subjectiveexperiences of inequity” (p. 181, italicsadded).

Contemporary theories of distributive jus-tice make even finer distinctions betweenthe concrete, evaluative, subjectively per-ceived, and responsive components of thejustice evaluation process. Jasso (1980) ex-plicitly states, with respect to her justiceevaluation function, that the ratio of actualshare to just share (in parentheses in equa-tion 1) is “an exclusively cognitive magni-tude, completely devoid of any emotionalcontent” (p. 6), thus categorically distin-guishing between distributive justice evalu-ations and their subjective impact on the in-dividual. Similarly, Markovsky (1985,

1991) treats the following as distinct ele-ments of the justice evaluation process: (1)the actual reward and the reward standardactivated in a situation, (2) the cognitive as-sessment of their fit, (3) the subjective feel-ing of justice or injustice resulting from thiscomparison, and (4) the overt behavioral re-sponses of the actor.

Although these perspectives differ some-what in how justice evaluations, justice per-ceptions, and reactions to injustice are con-ceptualized, each perspective acknowledges(at least implicitly) that perceptions of injus-tice are a crucial part of the process thatleads individuals to react to injustice. In themajority of perspectives, however, the defi-nition and causal position of the justice per-ception construct is somewhat ambiguous.Our conceptualization remedies this, as il-lustrated in Figure 1: Discrepancies betweenthe rewards an individual receives and thosehe or she believes are fair (i.e., the value ofthe justice evaluation) directly affect theindividual’s perception of the degree of in-justice in the situation.

The direct causal link between justiceevaluations and justice perceptions is sup-ported by an impressive body of experimen-tal research that has shown that manipulateddiscrepancies in rewards have a substantialimpact on individuals’ perceptions of thefairness of those rewards (Anderson 1976;Greenberg 1993; Hegtvedt, Thompson, andCook 1993; Mellers 1982). Likewise, ex-periments in which measures of justice per-ceptions are treated as a “ manipulationcheck” (i.e., a means of validating the gen-eral efficacy of the manipulated experimen-tal factors) are based on the logic that indi-viduals must perceive reward conditions asunfair in order to react as theoretically ex-pected. In such cases, a “ successful” ma-nipulation supports the justice evaluation →justice perception relationship (Hegtvedt1990; Markovsky 1985, 1988). Finally, em-pirical support for the specific justice evalu-ation function in equation 1 is provided bydata reported by Jasso (1980), Markovsky(1985), and Randall and Mueller (1995).

The foregoing theoretical arguments andresearch findings lead to our first hypothesis,represented in Figure 1 by the causal pathleading directly from pay justice evaluationsto justice perceptions.

5 We recognize that perceptual processes mayalso come into play at two other points in the jus-tice evaluation process: (1) an individual’ s per-ceptions of the amount of rewards received maydiffer from those actually received, and (2) anindividual’ s determination of the rewards that aredeserved in the situation (i.e., the “ just reward” )may differ from “objectively” determined stan-dards of fairness, such as the rewards receivedby comparison others, received in the past, orimplied by existing normative standards (Jassoand Wegener 1997:407–408).


Hypothesis 1: The greater the degree of in-justice in a distributive justice evalua-tion, the lower the perceived distributivejustice.

Theoretical interest in perceptual pro-cesses in the distributive justice literature isalso reflected in the emergence of several“ individual-difference” constructs. For in-stance, theorists have suggested that indi-viduals differ in terms of “ justice impor-tance” (Markovsky 1985), “equity sensitiv-ity” (Huseman, Hatfield, and Miles 1987),and “ expressiveness” (Jasso and Wegener1997). Although these concepts differ ac-cording to the particular individual and situ-ational characteristics they consider impor-tant and in the manner in which they are ex-pected to impact the justice evaluation pro-cess, each provides further support for thebasic contention that “ justice is in the eye ofthe beholder” (Markovsky 1985:822)—whatthe individual perceives in terms of fairnessgoverns his or her response to the situation.Determining how such individual-level fac-tors affect the justice evaluation process(e.g., through additive or interaction effects)is critical for correctly specifying and esti-mating empirical models of distributive jus-tice processes.

We begin to address this issue by incor-porating into our model Markovsky’s(1985) concept of “ justice importance,”which he defines as the degree to which anindividual values fairness in the given situa-tion and for the particular actor on whosebehalf the justice evaluation is being made.Markovsky suggests that justice importancemoderates the effect of reward discrepan-cies (captured by the justice evaluation) onthe amount of injustice perceived—themore the evaluator cares about justice, thegreater the impact an injustice will have.6

Markovsky (1985) has provided experimen-tal evidence that increasing the degree towhich an individual identifies with his orher workgroup (and thus presumably, themore important it is that the group be fairlyrewarded) increases the impact of a givenpay discrepancy on the number of com-plaints the individual makes on behalf ofthe workgroup. Similarly, Gartrell andPaille (1997) found that employees who feltresponsible for wage-cut decisions (andthus who were presumably more indifferentto the fairness of wage cuts) perceived ac-tual wage cuts as more fair than employeeswho were not responsible.

These are the only two studies we havefound, however, that empirically investigatethe potential moderating effects of justiceimportance for justice evaluations, and nei-ther study employed direct measures of jus-tice importance. This means not only that asignificant theoretical insight remains sup-ported by only a few empirical studies, but,just as important, that statistical models thatdo not control for the interactive effects ofjustice importance are potentially misspec-ified. Thus, we include the moderating im-pact of justice importance in our model ofjustice processes, represented in Figure 1 bythe direct causal path leading from the inter-action term (pay justice evaluation × justiceimportance) to justice perceptions. Thisleads to our second hypothesis.

Hypothesis 2: The greater the importance ofjustice, the greater the effect of a givendegree of injustice on perceptions ofdistributive justice.

The second assumption regarding the me-diating role of justice perceptions suggeststhat an individual’s perception of injustice ina situation directly affects his or her emo-tional reaction to that situation. We representthis assumption in Figure 1 by including adirect causal path leading from justice per-ceptions to the emotional response. Thiscausal link is supported by a vast body ofsurvey research in which measures of justiceperceptions have been used to predict a vari-ety of emotional responses, including job

justice importance interacts with the value of thejustice evaluation to predict perceptions of injus-tice (Markovsky 1985:838).

6 Formally, Markovsky (1985) defines the baseof the logarithmic function (equation 1) as theevaluator’ s indifference to injustice (i.e., the in-verse of justice importance) for the specific com-parison being made. As such, the greater the im-portance of justice to the evaluator, the smallerthe base of the logarithmic function and thesteeper the slope of the relationship between re-ward discrepancies and the amount of injusticefelt by the evaluator (Markovsky 1985:828). Dis-cursively, this translates into the assumption that


and reward satisfaction, organizational com-mitment, anger, and resentment (Folger andKonovsky 1989; McFarlin and Sweeney1992; Mueller and Wallace 1996; Sweeneyand McFarlin 1993). This causal path leadsto our third hypothesis.

Hypothesis 3: The lower the perceived dis-tributive justice, the more negative anindividual’s emotional response.

The final assumption regarding the medi-ating role of justice perceptions suggests thatthe effects of justice evaluations are indirect,working through perceptions of injustice.We represent this assumption in Figure 1 byexcluding direct paths from pay justiceevaluations to emotional reactions, leadingto our fourth hypothesis.

Hypothesis 4: Perceptions of distributivejustice mediate the effect of the justiceevaluation on an individual’s emotionalresponse.

Because the proposed interaction betweenjustice importance and justice evaluations isassumed to directly affect justice percep-tions, our causal model also implies that theinteraction effect should be mediated byjustice perceptions. We represent this as-sumption in Figure 1 by excluding directpaths from the interaction term (pay justiceevaluations × justice importance) to theemotional response, leading to our final hy-pothesis.

Hypothesis 5: Perceptions of distributivejustice mediate the effect of the pay jus-tice evaluation × justice importance in-teraction on an individual’s emotionalresponse.

Despite the seemingly “obvious” nature ofthese mediating propositions and the sub-stantial empirical evidence supporting eachof the direct causal paths in our model (rep-resented in Hypotheses 1 through 3), thereare few, if any, examples of experimental orsurvey research that directly test the mediat-ing role of justice perceptions we propose inHypotheses 4 and 5. To test their role re-quires that researchers simultaneously con-sider the reward conditions faced by indi-viduals, their justice evaluations, how im-portant justice is to them, and their percep-tions of the fairness of those conditions in

predicting individuals’ emotional reactions;this has yet to be done.

It is important to note, however, that wedo not mean to criticize or refute the myriadempirical investigations of justice processes.To the contrary, these studies provide impor-tant empirical justification for the individualassumptions of our model. Rather, we sug-gest simply that previous research has fo-cused on aspects of the justice evaluationprocess other than the mediating role of jus-tice perceptions, such that there is a generallack of evidence directly assessing the me-diating role of justice perceptions in thestudy of distributive justice. Below, we re-port results from a national survey of Prot-estant clergy that bear on the mediating roleof justice perceptions and the moderatingrole of justice importance in explaining min-isters’ reported pay satisfaction.

DATA AND METHODS

The data analyzed here come from a 1996national survey of parish ministers in theUnited Church of Christ (UCC) and Chris-tian Church (Disciples of Christ) (DOC).Both are Protestant denominations in themoderate to liberal range with respect to re-ligious conservatism. Details of the surveyinstrument and sample are provided by Mc-Duff and Mueller (1999, 2000). Of the 4,500ministers who were mailed the question-naire, a total of 2,780 (62 percent) returnedthem, a reasonable response rate for a na-tional-level mailed survey (Dillman 1991).To ensure sufficient cases for analyzing sub-groups (not at issue in the current paper), allfemale ministers were sampled, while a 50-percent random sample of male ministerswas taken. Because females were “ doubleover-sampled,” weights have been intro-duced so that coefficients estimated from theentire sample are not biased by female over-representation. Also, we restricted our analy-ses to the subset of ministers who reportedthat they are not in interim positions. Afterapplying the sampling weights and eliminat-ing cases listwise for missing data, the ef-fective sample size is 2,350 (1,953 malesand 397 females). In general, a comparisonof characteristics of the sample with thoseof the population of parish ministers in UCCand DOC churches indicates that there are


few differences, suggesting that this sampleis representative of the population (seeMcDuff and Mueller 1999, 2000).

Logic of Analysis

The hypotheses presented above are as-sessed by estimating a structural equationmodel (Jöreskog and Sörbom 1996) basedon the theoretical model presented in Figure1, as specified with the measures we de-scribe below. LISREL with maximum-like-lihood estimation produces reliability-cor-rected estimates of direct and indirect effectsand the relevant asymptotic standard errors.As described below, some of the constructsare captured with multiple-item measures,although most are based on single indicators.Finally, several of the indicators are mea-sured as ordinal scales and, thus, violate theassumption of interval scales when Pearsoncorrelations and MLE are used. We con-ducted several sensitivity checks of the re-sults using a smaller model, a polychoriccorrelation matrix, and weighted-least-squares estimation. These checks suggestthat our results are robust to the violation ofthis assumption.7

Variables and Measures

Full details of each of the measures used inthe analyses are provided in Appendix A,and descriptive statistics for all of the vari-

ables included in the analysis are presentedin Table 1. The variables are introduced be-low by tracing the causal model in Figure 1from left to right, that is, from the exogenousto the endogenous variables.

Controls. The estimated model in-cludes three sets of exogenous control vari-ables. The first set includes characteristics ofthe work setting—church budget, churchsize, and denomination. The first two vari-ables are common in models of work satis-faction and are potentially important sourcesof symbolic rewards, as large and well-funded churches tend to represent high-sta-tus positions in the ministerial occupationallabor market (McDuff and Mueller 2000).Also, there are qualitative differences be-tween denominations that might influencejustice processes, such as the status of min-isters in the congregation.

The second set of controls includes vari-ous characteristics of the respondents as em-ployees and as individuals—gender, maritalstatus, full-time employment status, beingthe primary wage-earner for their household,occupational tenure, and organizational ten-ure. Again, these are common controls formodels of work satisfaction that partial outpotentially confounding effects. In general,ministers in this sample are more likely thannot to be married, to be employed full-time,and to be the primary wage-earner; they tendto have been in the profession for around 18years and in their present position for around7 years.

The third set of controls focuses on thedegree of autonomy allowed ministers on thejob, an aspect of workplace conditions (otherthan the rewards considered below) that mayaffect respondents’ overall feelings of fair-ness and satisfaction. As suggested by recentresearch and theory, interactional and proce-dural aspects of the workplace, althoughconceptually distinct from distributive jus-tice, may exert an independent effect on per-

7 When variables in such models are ordinallyscaled, Bollen (1989) and Jöreskog and Sörbom(1996) recommend using polychoric correlationsto estimate an asymptotic covariance matrix andweighted-least-squares to estimate the structuralequation model. However, without very largesample sizes the asymptotic covariance matrix isoften poorly estimated and/or the solution willnot converge, even after a large number of itera-tions. Jöreskog and Sörbom (1996) argue thatmaximum-likelihood estimation (MLE) is thebetter alternative in such cases, and Bollen(1989) concludes (based on Monte Carlo simula-tions) that assuming interval-level data and us-ing Pearson correlations (with MLE) does notsubstantially alter the results when there are morethan four categories for the variables. Facing theproblem we have just described, Matsueda(1992) and Heimer and Matsueda (1994) estimatetheir full model with MLE, but then assess therobustness of their results by estimating smaller

parts of that model using polychoric correlationsand weighted-least-squares. We followed thisstrategy because our attempt to estimate the fullmodel with weighted-least-squares failed to con-verge after over 2,000 iterations. The results ob-tained by this strategy (available on request fromthe authors) were consistent with those reportedin Table 3 based on the MLE strategy.


Table 1. Descriptive Statistics for Variables inthe Analysis: Protestant Ministers,1996

Variable Mean S.D.

Emotional ResponsePay satisfaction 3.21 1.13

Work SettingChurch budget 2.28 .76

Church size 2.15 .73

Denomination .69 .46

Worker CharacteristicsFemale .17 .38

Married .87 .34

Employed full time .86 .35

Primary wage-earner .69 .46

Years in ministry 18.65 9.91

Tenure 7.07 6.00

Working ConditionsAutonomy a 4.29 .52

Workplace RewardsPay (in 1996 dollars) $33,713.08 $14,242.03

Formal benefits 7.84 2.76

Informal benefits 3.50 1.84

Distributive JusticePay justice evaluation 1.31 .30 (× 10–03) b

Justice importance 2.51 .73 (× 10–03) b

Pay justice evaluation × 1.32 .26 Justice importance (× 10–03) b

Distributive justice perceptions: Pay 2.38 .72

Global a 3.25 1.02

Note: Sample size is for the weighted sample af-ter listwise deletion of cases with missing data; N =2,350.

a Statistics are based on the simple sum of theitems divided by the total number of items.

b Statistics are calculated on the centered vari-able.

ceptions of distributive fairness (Brocknerand Wiesenfeld 1996; Randall and Mueller1995; van den Bos, Vermunt, and Wilke1997). By including autonomy as an exog-enous variable in the model, we control forthe potentially confounding effects of intrin-sic characteristics of the work, as well asprocedural and interactional factors of theworkplace, on distributive justice percep-tions and pay satisfaction. Ministers in thesample report, on average, that they have arelatively high degree of autonomy, with amean score of 4.29 (S.D. = .52) on our five-point autonomy scale (which ranges fromlow to high autonomy). Assuming that min-isters prefer more autonomy in their work,we expect autonomy to positively affect jus-tice perceptions and pay satisfaction in ourmodel.

Reward conditions. Because we focuson respondents’ evaluations and perceptionsof distributive (as opposed to procedural orinteractional) fairness, and the effects ofthose evaluations and perceptions on respon-dents’ pay satisfaction, we include severalvariables as indicators of the concrete re-wards received by respondents in their jobs.Perhaps the most obvious workplace rewardis pay—ministers in this sample earned anaverage of about $33,713 per year in 1996(S.D. = $14,242). To tap into workplace re-wards other than pay that may impact re-spondents’ perceptions of distributive injus-tice and thus indirectly affect their pay satis-faction, we also include the number of bothformal and informal benefits. Ministers inthis sample reported receiving an average of7.84 out of 13 possible formal benefits (S.D.= 2.76) and an average of 3.5 out of 9 pos-sible informal benefits (S.D. = 1.84),weighted by their frequency.

Justice evaluations. The first variableof interest in the distributive justice processis the distributive justice evaluation. Thisconstruct is measured as a pay justice evalu-ation, by taking the natural log of the ratioof the income reported by the respondent(i.e., the actual share) to the income the re-spondent says he or she deserves (i.e., thejust share).8 This operationalization of the

8 Although we do not have data pertaining tothe specific referent used by respondents in oursample, this measure of the just reward elimi-

nates the need for such information by asking re-spondents to report the level of pay they deservegiven their own experience, effort, education,and training. This allows the specific referent tovary across respondents without biasing the mea-sure of just rewards. Again, this strategy for mea-suring the just reward is justified by adopting


justice evaluation function proposed byJasso (1980) is based on the statistical mod-els employed by Markovsky (1985) andRandall and Mueller (1995). As coded here,the pay justice evaluation variable becomesmore negative as an individual’s salarymoves farther below the deserved salary(i.e., the greater the degree of underreward),and it becomes more positive as an indi-vidual’s salary moves farther above the de-served salary (i.e., the greater the degree ofoverreward). As suggested by Jasso (1980),the justice evaluation will equal 0 for an in-dividual who receives the salary he or shesays is deserved. For our sample, the uncen-tered mean pay justice evaluation is –.28(S.D. = .30), indicating that, on average,ministers in our sample are slightly under-rewarded.

As Randall and Mueller (1995) have dem-onstrated, the characteristics of the rewardsbeing evaluated must be carefully consid-ered within the context in which they areevaluated. Specifically, several justice re-searchers have suggested that, under particu-lar conditions, slight degrees of overrewardmay not lead to negative emotional reactions(Hegtvedt 1990; Hegtvedt et al. 1993; Jasso1986; Markovsky 1988). There are severalreasons that we do not expect overrewardedministers to perceive injustice in their pay orto express dissatisfaction with it. First, be-cause most ministers receive low pay rela-tive to other professionals, it is unlikely thatmany respondents in our sample will beoverrewarded. In fact, 85 percent of minis-ters in the sample were underrewarded,slightly more than 12 percent were justly re-warded, and only about 2 percent were over-rewarded, indicating a general state ofunderreward.9 Second, because most minis-

ters are not in direct competition with othersfor the salary and other benefits they receive,and because they are not directly responsiblefor determining their own pay, it is likelythat those who are overrewarded would beable to justify this fact through external at-tributions (Hegtvedt et al. 1993).

Based on this reasoning, the results re-ported by Randall and Mueller (1995), andthe coding of the variable, we expect payjustice evaluations to have a positive, mono-tonic effect on justice perceptions, such thatincreasingly negative justice evaluationsshould be associated with lower perceivedjustice (Hypothesis 1).

Justice importance. As conceptual-ized, Markovsky’s (1985) concept of justiceimportance can be directly operationalizedvia attitudinal survey items. We measure thisconstruct on a five-point scale indicating theimportance to the respondent of being fairlyrewarded given the work done (ranging from“ not at all important” to “ of great impor-tance” ). The uncentered mean justice impor-tance for respondents in this sample is 3.89(S.D. = .73).

Pay justice evaluation x Justice im-

portance interaction. We operational-ize the pay justice evaluation × justice im-portance interaction effect proposed in Hy-pothesis 2 by creating a variable computedas the product of the justice evaluation andjustice importance variables. This is consis-tent with Bollen’s (1989:128) recommenda-tion for including single-indicator interac-tion terms in LISREL models. Also, becauseof high collinearity between the interactionterm and the two variables composing it, wehave centered the pay justice evaluation andjustice importance variables at their meansprior to computing values for the interactionterm, as recommended by Neter et al.(1996).

Distributive justice perceptions.

We include two indicators of distributivejustice perceptions, one specific to pay and

overrewarded for each type of reward rangesfrom 0 percent to 11.1 percent). This adds to ourconfidence that the results reported here haveimplications for justice evaluations in othersamples and that there will be a monotonic rela-tionship between pay justice evaluations andeach of the endogenous variables we consider.

Markovsky’ s (1985) scope condition that indi-viduals already possess a standard of fairnesswith which to evaluate the rewards they receive,and because we are interested in the effects ofthe justice evaluation given such a standard,without concern for the source of that standard.

9 This frequency distribution is strikingly simi-lar to that reported by Randall and Mueller(1995:186) for a sample of registered nurses.Across a variety of qualitative, nontransferablerewards, an average of 2.9 percent of the respon-dents in their sample reported being over-rewarded (the percentage who reported being


one a more global distributive justice mea-sure (which includes pay plus other benefitsof being a minister). The pay-specific mea-sure taps the degree to which respondentsperceive that they have been under- oroverrewarded in terms of pay and benefits;the midpoint (2.5) represents perfect justice.The mean of 2.38 (S.D. = .72) indicates thatrespondents generally perceive themselvesto be slightly underrewarded. Global dis-tributive justice perceptions are measured bythree items from the Price-Mueller indexthat tap the degree to which respondents per-ceive themselves to be fairly rewarded giventheir effort, experience, and responsibilities(Price and Mueller 1986). Table 1 indicatesthat ministers generally agree that they arefairly rewarded, with a mean of 3.25 on afive-point Likert scale. However, there issubstantial variation in these perceptions(S.D. = 1.02), and this variation is more im-portant to our analysis than the overallmeans. Because we do not make any as-sumptions about the causal effect of one per-ception on the other, but we do expect themto be empirically related, we allow the dis-turbances for these two justice perceptionsto be correlated in the estimated structuralequation model.

Emotional response. We investigatethe effect of justice evaluations and percep-tions on respondents’ pay satisfaction, ourindicator of an emotional response to injus-tice. Pay satisfaction exemplifies what Jasso(1986) has called a “ responsive” conse-quence of experienced injustice (as opposedto purposive behaviors designed to reduceinjustice), and it includes both cognitive andaffective elements (Mark and Folger 1984;Randall and Mueller 1995). We chose thisspecific facet of job satisfaction (rather thanoverall job satisfaction) because (1) it pro-vides a cleaner test of the theoretical model(focusing attention on distributive ratherthan other forms of justice) and (2) follow-ing neoclassical economic theory, pay is per-haps the most universally (positively) valuedworkplace reward. This is important, be-cause, as Jasso and Wegener (1997) convinc-ingly argue, whether an individual frames aparticular outcome as a positively or nega-tively valued object has direct implicationsfor the form and magnitude of the sense ofinjustice that may result. In the analyses to

follow, we assume that respondents framepay as a positively valued goal-object.10

Our measure of pay satisfaction rangesfrom 1 to 5, with lower values representinglower pay satisfaction (i.e., a more “nega-tive” emotional response) and higher valuesrepresenting higher pay satisfaction (i.e., amore “positive” emotional response). As in-dicated in Table 1, on average, ministers inthis sample were relatively satisfied withtheir pay (mean = 3.21), although there issubstantial variation (S.D. = 1.13).

RESULTS

Table 2 gives the LISREL-estimated corre-lation coefficients (corrected for unreli-ability) and Table 3 presents the LISREL-es-timated path coefficients for the model pro-posed in Figure 1. Our main concerns are to(1) demonstrate the general efficacy of thecausal model proposed, (2) determinewhether justice importance interacts withjustice evaluations as proposed by Mar-kovsky (1985), and (3) determine whether(and to what extent) justice perceptions me-diate the effects of the exogenous variables.

In assessing the general efficacy of thecausal model, it is important to first considerwhether the expected relationships betweenthe exogenous and intervening variables arefound (see Table 3). Focusing on workplacerewards, both pay and informal benefitshave the expected positive effects on payand global justice perceptions. Unexpect-edly, however, formal benefits have a nega-tive effect on both pay and global justiceperceptions, a finding we discuss below. Au-tonomy, our indicator of working conditionsother than pay, has the expected positive ef-fect on both pay and global justice percep-tions. Moreover, attesting to the general ef-ficacy of the model, about 71 percent of thevariance in pay satisfaction is accounted forby all causally antecedent variables, and

10 The survey instrument we used contained aquestion asking respondents to indicate the im-portance they place on receiving “good pay.” Ourassumption that pay is positively valued by re-spondents is supported by the fact that the meanvalue for the pay importance variable is 3.80(S.D. = .75), nearly one and two-thirds standarddeviations above the scale midpoint.


Tab

le 2

.C

orre

lati

on C

oeff

icie

nts

bet

wee

n V

aria

ble

s in

th

e A

nal

yses

: P

rote

stan

t M

inis

ters

, 199

6

Var

iabl

es(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)(1

2)(1

3)(1

4)(1

5)(1

6)(1

7)(1

8)(1

9)

(1)

Pay

sat

isfa

ctio

n1.

00.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—

(2)

Chu

rch

budg

et.1

81.

00.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

(3)

Chu

rch

size

.15

.73

1.00

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

(4)

Den

omin

atio

n.0

1–.

02.0

61.

00.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

(5)

Fem

ale

–.08

–.21

–.19

.02

1.00

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

(6)

Mar

ried

.09

.20

.19

–.03

–.31

1.00

.—.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—

(7)

Em

ploy

ed f

ull

tim

e.0

1.4

9.4

5.0

1–.

15.0

71.

00.—

.—.—

.—.—

.—.—

.—.—

.—.—

.—

(8)

Pri

mar

y w

age

earn

er.0

5.2

9.2

9.0

2–.

14–.

09.4

01.

00.—

.—.—

.—.—

.—.—

.—.—

.—.—

(9)

Yea

rs i

n m

inis

try

.09

.23

.22

.00

–.39

.18

.11

.11

1.00

.—.—

.—.—

.—.—

.—.—

.—.—

(10)

Ten

ure

.07

.16

.15

.05

–.17

.09

.08

.02

.45

1.00

.—.—

.—.—

.—.—

.—.—

.—

(11)

Aut

onom

y.2

7.0

8.0

5.0

1.0

1.0

4–.

03.0

1.0

5.1

41.

00.—

.—.—

.—.—

.—.—

.—

(12)

Pa y

.25

.71

.63

.06

–.22

.18

.51

.34

.28

.18

.11

1.00

.—.—

.—.—

.—.—

.—

(13)

For

ma l

be n

e fit

s.0

9.5

0.4

6.2

7–.

04.0

9.4

8.3

2.1

2.0

6.0

6.5

01.

00.—

.—.—

.—.—

.—

(14)

Info

rma l

be n

e fit

s.2

0.0

5.0

9.0

1.0

4–.

07.0

8.0

9–.

05–.

03.1

9.0

9.1

61.

00.—

.—.—

.—.—

(15)

Pay

jus

tic e

eva

lua t

ion

.39

.37

.34

.05

–.07

.07

.34

.28

.04

.03

.06

.49

.35

.10

1.00

.—.—

.—.—

(16)

Just

ice

impo

rta n

c e–.

13.1

2.1

3.0

1–.

01–.

01.0

9.0

7.0

8.0

0.0

6.1

4.1

3.0

0.0

11.

00.—

.—.—

(17)

Pay

jus

tic e

eva

lua t

ion

×.0

9 –

.02

–.04

.01

.01

–.02

–.06

–.03

.01

.01

.06

–.02

–.06

.00

–.13

–.15

1.00

.—.—

Just

ice

impo

rta n

c e

(18)

Glo

bal

dist

ribu

tive

.83

.18

.16

–.02

–.11

.10

.05

.06

.10

.07

.30

.28

.11

.26

.41

–.06

.08

1.00

.—ju

stic

e

(19)

Pay

dis

trib

utiv

e ju

stic

e.6

5.1

4.1

2–.

03–.

01.0

8.0

2.0

6.0

6.0

3.1

5.2

2.0

5.1

6.4

4–.

10.0

5.6

71.

00

Not

e: C

orre

lati

ons

a re

c om

pute

d on

the

we i

ghte

d sa

mpl

e , f

or th

e c e

nte r

e d v

a ria

ble s

, and

cor

rec t

e d f

or u

nre l

iabi

lity

(a s

ge n

e ra t

e d b

y th

e L

ISR

EL

est

ima t

ion

soft

wa r

e ); N

=2,

350.


Table 3. Unstandardized LISREL Coefficients for the Justice Model Illustrated in Figure 1:Protestant Ministers, 1996

Mediating Variables

Pay Distributive Global DistributiveVariables Justice Perception Justice Perception Pay Satisfaction

Work Setting

Church budget .011 –.010 .074*

(.029) (.039) (.030)

Church size –.014 –.009 .006(.027) (.036) (.028)

Denomination –.046 –.069 .081**

(.029) (.040) (.030)

Worker CharacteristicsFemale –.109** –.153** .081*

(.039) (.053) (.041)

Married .064 .116* .008(.041) (.056) (.043)

Employed full time –.287*** –.323*** –.165***

(.047) (.064) (.049)

Primary wage-earner –.038 –.102* .011(.031) (.043) (.033)

Years in ministry .003 .005* .001(.002) (.002) (.002)

Tenure –.002 –.002 .001(.002) (.003) (.003)

Working ConditionsAutonomy .127*** .404*** .062*

(.028) (.038) (.030)

Workplace RewardsPay (in $1,000) .004** .009*** .001

(.001) (.002) (.002)

Formal benefits –.022** –.017* –.004(.006) (.008) (.006)

Informal benefits .047*** .099*** –.004(.007) (.010) (.008)

Distributive JusticePay justice evaluation 1.124*** 1.271*** .127*

(.051) (.070) (.059)

Justice importance –.080*** –.068** –.124***

(.018) (.024) (.019)

Pay justice evaluation × .226*** .395*** .026 Justice importance (.051) (.069) (.053)

Distributive Justice Perceptions:

Pay .— .— .216***

(.027)

Global .— .— .827***

(.023)

R2 .326 .271 .714

Note: Numbers in parentheses are standard errors. Model fit information: L2 = 482.412; d.f. = 98; stan-dardized Root Mean Square Residual = .0195; Goodness-of-Fit Index = .983; Adjusted Goodness-of-Fit In-dex = .949; Normed Fit Index = .980.

*p < .05 **p < .01 ***p < .001 (two-tailed tests)


each of the goodness-of-fit indices (see noteto Table 3) indicates a respectable corre-spondence between the model and the data.

Hypotheses 1 and 2 refer to the first set ofcausal paths directly linking justice evalua-tions and the justice evaluation × justice im-portance interaction to justice perceptions.Supporting Hypothesis 1, our measure ofpay justice evaluations has a significantpositive effect on both types of justice per-ceptions and, in support of Hypothesis 2, ourindicator for the pay justice evaluation × jus-tice importance interaction significantly in-fluences both pay and global justice percep-tions. As suggested by Markovsky (1985),the positive sign for this interaction termmeans that justice evaluations impact justiceperceptions more strongly when justice ismore important to the person (a graph of thisinteraction is available on request from theauthors). Unexpectedly, however, justice im-portance has a significant negative main ef-fect on both types of justice perceptions (wediscuss possible reasons for this below).Most important, these findings support thefirst assumption required for justice percep-tions to serve a mediating role as proposed—respondents’ justice evaluations directly af-fect their perceptions of distributive justice.

Recall that the second assumption regard-ing the mediating role of justice perceptionsrequires a direct effect of justice percep-tions on pay satisfaction (Hypothesis 3).

Turning to the estimates for the pay satis-faction equation in Table 3, both pay andglobal justice perceptions significantly anddramatically increase satisfaction withone’s pay, supporting Hypothesis 3. Withrespect to the other exogenous variables inour model, we find that the pay justiceevaluation has a significant positive directeffect, justice importance again has an un-expected significant negative main effect,and the direct effect of the interaction termis not significant. The causal importance ofjustice perceptions is highlighted in thattheir standardized coefficients (β = .137 andβ = .705 for pay and global justice percep-tions, respectively) are substantially largerthan those for any of the other variablesused to predict pay satisfaction.

The third assumption regarding the medi-ating role of justice perceptions requires thatthe effects of the pay justice evaluation andthe pay justice evaluation × justice impor-tance interaction be indirect, affecting paysatisfaction primarily through their impacton justice perceptions. In Table 4, we de-compose the total causal effects of the keyexogenous and intervening variables in ourmodel into direct and indirect effects. As re-ported above, autonomy, pay justice evalua-tions, and justice importance have signifi-cant direct effects on pay satisfaction. Theindirect effects of these variables (throughjustice perceptions), however, are 85 per-

Table 4. Direct, Indirect, and Total Causal Effects of Selected Variables on Pay Satisfaction:Protestant Ministers, 1996

Standardized LISREL Coefficients

Variables Direct Effect Indirect Effect Total Causal Effect a

Pay .007 .107*** .114**

Formal benefits –.010 –.046* –.056**

Informal benefits –.006 .150*** .144***

Autonomy .029* .169*** .198***

Pay justice evaluation .033* .342*** .375***

Justice importance –.080*** –.047*** –.128***

Pay justice evaluation × .006 .086*** .091***

Justice importance

a Coefficients in column 3 represent the total causal effect rather than the total effect of the variable. Thetotal causal effect does not include the covariation between the variable and pay satisfaction that is due tothe variable’ s correlation with other exogenous variables that also influence pay satisfaction. The total effectof the variable is the zero-order correlation with pay satisfaction and can be found in column 1 of Table 2.

*p < .05 **p < .01 ***p < .001 (two-tailed tests)


cent, 91 percent, and 38 percent of the totalcausal effects, respectively, supporting Hy-pothesis 4. We discuss possible reasons forthe substantial unmediated (direct) effect ofjustice importance below. Also supportive ofthe proposition that justice perceptions is themediator, the effects of pay, formal benefits,informal benefits, and the pay justice evalu-ation × justice importance interaction are en-tirely indirect, working through the two jus-tice perceptions. The fact that the effect ofthe interaction term is entirely mediated byjustice perceptions offers strong support forHypothesis 5.

DISCUSSION

These results provide substantial support forthe theoretical model presented in Figure 1and the five hypotheses we have drawn fromit. While the absolute level of pay and infor-mal benefits have significant and positivetotal effects on pay satisfaction, these effectsare entirely indirect through the individual’sperceptions of distributive justice. Thesefindings are especially relevant for pay (94percent of its total causal effect is indirect),given the inability of previous researchers todistinguish between the relative efficacy ofself-interest and justice models of work sat-isfaction (Randall and Mueller 1995). Con-sistent with the justice model, we find thatthe absolute level of rewards received by in-dividuals does not exhibit a direct effect onpay satisfaction net of justice evaluationsand perceptions. Randall and Mueller ’s(1995) failure to include justice perceptionsin their statistical models may account fortheir inability to adjudicate between the self-interest and distributive justice explanations.At a minimum, our findings suggest that fu-ture investigations of justice evaluation pro-cesses should include measures of the cru-cial mediating variable—justice percep-tions—to avoid reaching faulty substantiveconclusions.

We find an unexpected significant nega-tive direct effect of formal job benefits onboth pay and global justice perception mea-sures, and a similar negative total causal ef-fect (but no direct effect) on pay satisfac-tion. The fact that respondents who receivegreater formal benefits perceive beingunderrewarded, and express lower pay sat-

isfaction as a result, suggests an interestingparadox. It may be that receiving substan-tial formal benefits (e.g., insurance, vaca-tion) leads ministers in this sample to be-lieve that they could, and, in fact, should,receive greater pay from their parish. Obvi-ously, this interpretation is speculative, butit is supported by the positive correlation (r= .385, p < .001) between formal benefitsand the amount of pay respondents believeis just (the denominator of the pay justiceevaluation). Future research investigatingdistributive justice in work settings shouldinclude a wide variety of workplace re-wards, and should attempt to disentanglethe interrelationships between the rewardsrespondents receive and those they believethey deserve (also see Jasso and Wegener1997).

Also supporting the model presented inFigure 1, we find that pay justice evaluationshave a positive total causal effect on pay sat-isfaction, primarily through their effects onjustice perceptions and, consistent withMarkovsky’s (1985) argument, this effect ismoderated by the importance placed on jus-tice by the respondent. That is, for ministersin this sample, the more importance theyplace on being fairly rewarded, the greaterthe effect of discrepancies between their ac-tual and deserved pay on their perceptionsof distributive justice and, indirectly, ontheir pay satisfaction.

The main anomaly found in this study isthe direct negative main effect of justiceimportance on pay satisfaction—net of allother variables, ministers who place a highvalue on being fairly rewarded tend to per-ceive themselves as being underrewardedand to express lower pay satisfaction. Al-though justice importance has significantnegative effects on both pay and global jus-tice perception measures, its effect on paysatisfaction is predominantly direct (62.5percent of the total effect is unmediated).Our measure of justice importance asks re-spondents about the importance of beingfairly rewarded in general (not restricted topay or job benefits). Conceptually, it is pos-sible that the importance one places on dis-tributive justice affects not only the transla-tion of reward discrepancies into perceivedinjustice in concrete situations (i.e., the in-teraction effect proposed here), but also


may serve as a more general schema thatsensitizes individuals to issues of distribu-tive fairness beyond their pay and job ben-efits.11 Thus, individuals who place a highvalue on distributive justice may also bemore likely to attend to a wider range ofjustice evaluations (e.g., intangible rewards,workplace procedures, interactions with su-pervisors) and justice referents. In our sam-ple, ministers who place a high value onfair pay may be more sensitive to the factthat their pay is quite low when comparedwith pay in other professions (the ministers’average salary for 1996 was less than$34,000 per year). Future research shouldinvestigate both the moderating effects ofjustice importance for justice evaluations inspecific situations, and the broader implica-tions of this overarching value for justiceperceptions and for individuals’ emotionalresponses. Because this is one of only ahandful of studies that empirically consid-ers this intriguing concept, much remains tobe learned about the role of justice impor-tance in the justice evaluation process.

CONCLUSION

We have argued that justice perceptions—the degree of injustice that an individual per-ceives to exist in a situation—mediate theimpact of rewards (both absolute and rela-tive to those the individual believes are fair)on individuals’ emotional responses to injus-tice. Although this argument is basic to al-most all theories of distributive justice, itappears to be a “ taken-for-granted” assump-tion, as most theories leave the definitionand causal role of justice perceptions im-plicit. To counter this, we have presented atheoretical model of the justice evaluationprocess that (1) more explicitly defines andspecifies the causal position of justice per-ceptions, (2) explicitly incorporates themoderating effects of justice importancesuggested by Markovsky (1985), and (3)demonstrates the relevance of perceptions asintervening mechanisms in the justice evalu-ation process.

Most important for our purposes, past re-search has not directly demonstrated that

justice perceptions do, in fact, act as a me-diator in the justice evaluation process. Ex-tant empirical studies focus on the effects ofabsolute rewards, justice evaluations, or jus-tice perceptions on individuals’ reactions,but rarely on all three simultaneously. Wehave filled this empirical void by analyzingdata from a national sample of Protestantministers that contains measures of all ofthese theoretical constructs. Based on esti-mates from a structural equation model, wehave found considerable support for the pro-posed justice evaluation model. The effectsof (1) the absolute level of workplace re-wards (e.g., pay, informal benefits), (2) jus-tice evaluations (i.e., discrepancies betweenactual and deserved pay), and (3) the payjustice evaluation × justice importance inter-action on pay satisfaction are all primarilyindirect, mediated by perceptions of pay-specific and global distributive justice. Sup-port for our distributive justice model alsoprovides support for the more general notionwithin social psychology that individuals’perceptions (e.g., definitions of the situa-tion) play an important mediating role in de-termining their resultant behaviors and emo-tions.

Our findings also have practical implica-tions for understanding justice processes.Many professionals rely on their ability tosatisfy others’ sense of fairness to maintainrelatively harmonious interpersonal interac-tions. Understanding the crucial mediatingrole of justice perceptions means that thistask becomes more complex—not onlymust one consider the level of rewards of-fered relative to available normative stan-dards and social referents (e.g., coworkers),but also individuals’ perceptions of the fair-ness of those rewards and the individual-level factors (e.g., justice importance) thatimpact reactions to specific reward discrep-ancies. For instance, Mueller and Wallace(1996) argue that the “paradox of the con-tented female worker” (Phelan 1994), inwhich women whose work conditions andrewards are objectively below those of malecounterparts often feel just as satisfied as(or even more satisfied than) men, is partlydue to the fact that women do not perceivethe differences as unfair. Also, the ability tomanipulate factors such as responsibility(Gartrell and Paille 1997) and social identi-

11 We thank an anonymous ASR reviewer forsuggesting this interpretation.


fication (Markovsky 1985), which havebeen associated with the importance indi-viduals place on distributive fairness, mayprove particularly effective for managersand other professionals facing potentially“explosive” situations (e.g., pay raises/cuts,layoffs).

While our empirical model was estimatedbased on data from a relatively restrictivepopulation of professionals (i.e., ministersfrom two Protestant denominations), thereis reason to be guardedly confident that thegeneral model would also apply to othergroups of professionals and nonprofession-als. Justice researchers have generated animpressive body of research illustrating theimportance of distributive justice percep-tions for individuals’ attitudes, behaviors,and emotions in a vast array of work orga-nizations. This includes evidence that therewards employees receive affect percep-tions of fairness in a wide variety of set-tings, including workers at a plywood coop-erative facing pay-cuts (Gartrell and Paille1997), public university employees facingpay-raise decisions (Dulebohn andMartocchio 1998), and teachers in theUnited States and South Korea (Mueller,Iverson, and Jo 1999). There is also evi-dence that justice perceptions have substan-tial effects on job and/or pay satisfactionfor a broad range of occupations, includingfirst-line manufacturing employees (Folgerand Konovsky 1989), engineers in a publicutility company (Sweeney and McFarlin1993), bank employees (McFarlin andSweeney 1992), supervisors in Chinesejoint-venture hotels (Leung et al. 1996), andCanadian lawyers (Mueller and Wallace1996). Finally, Randall and Mueller (1995)provide evidence of the total effect of jus-tice evaluations on job satisfaction, organi-zational commitment, intent to stay with anemployer, and turnover for a sample of reg-istered nurses. The sampling universe cov-ered by these studies is impressive, rangingfrom salaried professionals to hourly manu-facturing workers. The fact that the evi-dence in each case supports a portion of ourmodel gives us reason to be confident thatour entire model should apply to othersamples. Because ours is the first study toconsider each of these relationships simul-taneously, however, such generalizability is

speculative and awaits further corroborationand theoretical refinement.

A final word of caution regarding our re-sults is warranted. The data we use to test thetheoretical model are cross-sectional, andtherefore we cannot make unambiguouscausal inferences despite our estimation of astructural equation model. In fact, some theo-rists have questioned the presumed causalrole of justice perceptions, suggesting in-stead that individuals first experience a nega-tive affective response, then “ search” for thecause of this negative affect in the situation,leading them to perceive that their rewardsare “unfair” (Scher and Heisse 1993). In ourcase, this argument implies that workplacerewards lead to pay satisfaction, and onlythen to evaluations of pay justice and to per-ceptions of distributive injustice. Future re-search, grounded in carefully controlled ex-periments, may assist in disentangling thecausal order of the cognitive and affectiveelements of justice evaluations. However,such fine-grained analysis is far beyond theaims of the current paper and the data used inthis study. Instead, we provide an explicit andgeneral theoretical model, based on the pre-dominant theoretical position and prior em-pirical evidence, that stresses the mediatingrole of justice perceptions in translating re-ward conditions into emotional responses,and find substantial support for the assump-tions represented by this model.

C. Wesley Younts is a Ph.D. candidate in theDepartment of Sociology and Assistant Directorof the Center for the Study of Group Processes atthe University of Iowa. His research focuses onbuilding and testing theories of basic social psy-chological processes. His recent work includesarticles on status characteristics and expectationstates theory (American Journal of Sociology1997, pp. 692–732; Social Psychology Quarterly,forthcoming), theories of distributive and proce-dural justice (Social Justice Research, forthcom-ing), and power in exchange networks. He is alsointerested in the etiology of crime and delin-quency, specifically, in social influence and so-cial learning as distinct explanations of peer in-fluences and in the structural characteristics ofcommunities that motivate and shape criminalbehavior.

Charles W. Mueller is Professor of Sociology atthe University of Iowa. His recent research fo-cuses on justice perceptions in the workplace


Unless otherwise noted, all variables were measuredon five-point Likert scales (1= strongly disagree to5 = strongly agree). Reverse coded items are denot-ed [R]. Coefficient alpha and factor loadings [ ] arereported for multi-item scales.

EXOGENOUS CONTROL VARIABLES

Work Setting

Church budget. “ What is the budget of your con-gregation?” (1) Under $50,000, (2) $50,000 to$100,000, (3) Over $100,000.

Church size. “ What is the size of your presentcongregation?” (1) 0 to 100, (2) 101 to 300, (3)more than 300.

Denomination. A dummy variable, with UCCcoded 1 and DOC coded 0.

Worker Characteristics

Female. A dummy variable, coded 1 if respon-dent is a female, 0 if respondent is a male.

Married. A dummy variable, coded 1 if respon-dent indicated that he or she was currently mar-ried, 0 if single, divorced, separated, or widowed.

Full-time. A dummy variable, coded 1 if currentposition is 40 or more hours per week, 0 if less.

Primary wage-earner. A dummy variable, coded1 if the respondent indicated he or she earned 51percent or more of the household income, 0 oth-erwise.

Years in the ministry. Total years as a minister.

Tenure. Years serving in present congregation.

Workplace Rewards

Pay. A variable constructed from a single item,which asked each respondent to indicate theircurrent yearly income from parish ministry be-fore taxes: (1) less than $10,000, (2) $10,000 to$15,999, (3) $16,000 to $20,000, (4) $21,000 to$25,999, (5) $26,000 to $30,999, (6) $31,000 to$35,999, (7) $36,000 to $40,999, (8) $41,000 to$50,999, (9) more than $60,000. We then com-puted each respondent’ s pay as the midpoint ofthe salary range they indicated, except in the caseof category (9), in which a Pareto estimate wasassigned (see Parker and Fenwick 1983 for de-tails on this procedure).

Formal benefits. The total number (ranging from0 to 13) of the following benefits the respondentreceives based on written contract with the con-gregation: (a) health insurance, (b) dental insur-ance, (c) medical co-payment plan, (d) vacation,(e) maternity/paternity leave, (f) life insurance,(g) disability insurance, (h) sabbaticals, (i) pen-sion/annuity, (j) travel expenses, (k) continuingeducation, (l) allowable reimbursement policy,(m) malpractice insurance.

Informal benefits. The total number of the fol-lowing benefits the respondent receives, weight-ed by their frequency (0 = never, 1 = sometimes,2 = often), that are dependent on the good will ofthe congregation and not in a written contract: (a)parsonage furnishings, (b) gifts of food, (c) din-ner invitations by parishioners, (d) special occa-sion gifts, (e) use of a parishioner’ s vacationhome, (f) use of a car, (g) reduced insurancerates, (h) reduced mortgage rates, (i) country cluband/or other club memberships.

Working Conditions

Autonomy. The mean of four items: (a) “ I have aconsiderable amount of freedom to do my job”[.82], (b) “ I influence the things that affect me onthe job” [.78], (c) “ I have input deciding whattasks or what parts of tasks I will do” [.71], and(d) “ I control the scheduling of my own work”[.67]. The composite scale has a reliability α =.83, and ranges from 1 to 5 with higher valuesrepresenting greater autonomy.

DISTRIBUTIVE JUSTICE VARIABLES

Pay justice evaluation. A variable constructed bytaking the natural log of the ratio of the respon-dent’ s value for pay (see above) and for just pay.The latter was constructed from the respondent’ sresponse to the question, “ Thinking in terms ofyour current position, what annual salary fromyour job (base salary plus housing or fair rentalvalue of parsonage) do you feel you deserve, onethat would be fair and just given your education,training, experience, position and input into yourwork?” (1) $10,000 to $15,999, (2) $16,000 to$20,999, (3) $21,000 to $25,999, (4) $26,000 to$30,999, (5) $31,000 to $35,999, (6) $36,000 to$40,999, (7) $41,000 to $50,999, (8) $51,000 to$60,999, (9) $61,000 to $70,999, (10) $71,000 ormore. We then assigned each respondent the mid-

(Social Psychology Quarterly, 1996, pp. 338–49;Social Justice Research, 2000, pp. 1–24), factorsdifferentially affecting organizational and pro-fessional commitment (Work and Occupations,

2000, pp. 89–116; Social Psychology Quarterly,1999, pp. 325–46), and temporary labor marketsfor clergy (Research in the Sociology of Work,2000, pp. 211–30).

Appendix A. Variable Names and Definitions

(Appendix A continued on next page)


REFERENCES

Adams, James S. 1965. “ Inequity in Social Ex-change.” Pp. 267–99 in Advances in Experi-mental Social Psychology, vol. 2, edited by L.Berkowitz. New York: Academic.

Anderson, Norman H. 1976. “Equity Judgmentsas Information Integration.” Journal of Per-sonality and Social Psychology 33:291–99.

Bollen, Kenneth. 1989. Structural Equations withLatent Variables. New York: Wiley.

Brockner, Joel and Batia M. Wiesenfeld. 1996.“An Integrative Framework for Explaining Re-actions to Decisions: Interactive Effects ofOutcomes and Procedures.” PsychologicalBulletin 120:189–208.

Cooley, Charles Horton. 1902. Human Natureand the Social Order. New York: Scribner.

Crosby, Faye. 1976. “A Model of Egoistic Rela-tive Deprivation.” Psychological Review83:85–113.

Dillman, Donald. 1991. “The Design and Admin-istration of Mail Surveys.” Annual Review ofSociology 17:225–49.

Dulebohn, James and Joseph J. Martocchio.1998. “ Employees’ Perceptions of the Dis-tributive Justice of Pay Raise Decisions: APolicy Capturing Approach.” Journal of Busi-ness and Psychology 13:41–64.

Folger, Robert. 1986 “ A Referent CognitionsTheory of Relative Deprivation.” Pp. 33–55 inRelative Deprivation and Social Comparison:The Ontario Symposium, vol. 4, edited by J. M.Olson, C. P. Herman, and M. P. Zanna.Hillsdale, NJ: Lawrence Erlbaum.

Folger, Robert and Mary A. Konovsky. 1989.“Effects of Procedural and Distributive Justiceon Reactions to Pay Raise Decisions.” Acad-

emy of Management Journal 32:115–30.Folger, Robert, David Rosenfield, Karen

Rheaume, and Chris Martin. 1983. “ RelativeDeprivation and Referent Cognitions.” Journalof Experimental Social Psychology 19:172–84.

Gartrell, C. David and Bernard E. Paille. 1997.“ Wage Cuts and the Fairness of Pay in aWorker-Owned Plywood Cooperative.” SocialPsychology Quarterly 60:103–17.

Goffman, Erving. 1959. Presentation of Self inEveryday Life. Garden City, NY: Anchor.

Greenberg, Jerald. 1993. “Stealing in the Nameof Justice: Informational and InterpersonalModerators of Theft Reactions to Underpay-ment Inequity.” Organizational Behavior andHuman Decision Processes 54:81–103.

Hegtvedt, Karen A. 1990. “The Effects of Rela-tionship Structure on Emotional Responses toInequity.” Social Psychology Quarterly53:214–28.

Hegtvedt, Karen A. and Barry Markovsky. 1995.“ Justice and Injustice.” Pp. 257–80 in Socio-logical Perspectives on Social Psychology, ed-ited by K. S. Cook, G. A. Fine, and J. S.House. Boston, MA: Allyn and Bacon.

Hegtvedt, Karen A., Elaine A. Thompson, andKaren S. Cook. 1993. “ Power and Equity:What Counts in Attributions for ExchangeOutcomes?” Social Psychology Quarterly56:100–119.

Heimer, Karen and Ross L. Matsueda. 1994.“Role-Taking, Role-Commitment, and Delin-quency: A Theory of Differential Social Con-trol.” American Sociological Review 59:365–90.

Homans, George C. 1974. Social Behavior: Its

point of the category they selected, except forcategory (10), in which a Pareto estimate of themedian value ($104,802 for females and $88,792for males) was assigned (see Parker and Fenwick1983 for details on this procedure).

Justice importance. Measured by a single itemthat asks respondents to rate the importance ofbeing fairly rewarded for the job they do: (1)“ Not important at all,” (2) “ Of little importance,”(3) “Of some importance,” (4) “Quite important,”(5) “ Of great importance.”

Pay justice evaluation ××××× justice importance. Aninteraction term created by multiplying the val-ues of the pay justice evaluation and justice im-portance variables.

Pay distributive justice perception. Measured bya single item that asks respondents to choose the

statement that best describes their beliefs abouttheir current pay and benefits? (1) “ I receivemuch less than I deserve,” (2) “ I receive less thanI deserve,” (3) “ I receive exactly what I deserve,”(4) “ I receive more than I deserve,” and (5) “ Ireceive much more than I deserve.”

Global distributive justice perception. The meanof three items: (a) “ I am fairly rewarded for theamount of work I put in” [.90], (b) “ I am fairlyrewarded considering the responsibilities I have”[.94], and (c) “ I am fairly rewarded in view ofmy experience” [.86]. The composite scale rang-es from 1 to 5, with higher values representinggreater perceived fairness (α = .93).

EMOTIONAL RESPONSE VARIABLE

Pay satisfaction. Measured by a single item: “ Iam satisfied with my pay.”

(Appendix A continued from previous page)


Elementary Forms. New York: Harcourt,Brace, and World.

Huseman, Richard C., John D. Hatfield, and Ed-ward W. Miles. 1987. “A New Perspective onEquity Theory: The Equity Sensitivity Con-struct.” Academy of Management Review12:222–34.

Jasso, Guillermina. 1978. “ On the Justice ofEarnings: A New Specification of the JusticeEvaluation Function.” American Journal ofSociology 83:1398–1419.

______. 1980. “ A New Theory of DistributiveJustice.” American Sociological Review 45:3–32.

______. 1986. “A New Representation of the JustTerm in Distributive Justice Theory: Its Prop-erties and Operation in Theoretical Derivationand Empirical Estimation.” Journal of Math-ematical Sociology 12:251–74.

______. 1993. “Choice and Emotion in Compari-son Theory.” Rationality and Society 5:231–74.

Jasso, Guillermina and Bernd Wegener. 1997.“Methods for Empirical Justice Analysis: PartI. Framework, Models and Quantities.” SocialJustice Research 10:393–430.

Jöreskog, Karl G. and Dag Sörbom. 1996.LISREL 8: User’s Reference Guide. Chicago,IL: Scientific Software International.

Leung, Kwok, Peter B. Smith, Zhongming Wang,and Haifa Sun. 1996. “ Job Satisfaction in JointVenture Hotels in China: An OrganizationalJustice Analysis.” Journal of InternationalBusiness Studies 27:947–62.

Leventhal, Gerald S., Jurgis Karuza Jr., and Wil-liam R. Fry. 1980. “ Beyond Fairness: ATheory of Allocation Preferences.” Pp. 167–218 in Justice in Social Interaction, edited byG. Mikula. New York: Springer.

Mark, Melvin M. and Robert Folger. 1984. “Re-sponses to Relative Deprivation: A ConceptualFramework.” Pp. 192–218 in Review of Per-sonality and Social Psychology, vol. 5, editedby P. Shaver. Beverly Hills, CA: Sage.

Markovsky, Barry. 1985. “Toward a MultilevelDistributive Justice Theory.” American Socio-logical Review 50:822–39.

______. 1988. “ Injustice and Arousal.” SocialJustice Research 2:223–34.

______. 1991. “Prospects for a Cognitive-Struc-tural Justice Theory.” Pp. 33–58 in Social Jus-tice in Human Relations, vol. 1, edited by R.Vermunt and H. Steensma. New York: Ple-num.

Matsueda, Ross L. 1992. “Reflected Appraisals,Parental Labeling, and Delinquency: Specify-ing a Symbolic Interactionist Theory.” Ameri-can Journal of Sociology 97:1577–1611.

McDuff, Elaine and Charles W. Mueller. 1999.“Social Support and Compensating Differen-

tials in the Ministry: Gender Differences inTwo Protestant Denominations.” Review ofReligious Research 40:307–30.

———. 2000. “The Ministry as an OccupationalLabor Market: Intentions to Leave an Em-ployer (Church) versus Intentions to Leave aProfession (Ministry).” Work and Occupations27:89–116.

McFarlin, Dean B. and Paul D. Sweeney. 1992.“Distributive and Procedural Justice as Predic-tors of Satisfaction with Personal and Organi-zational Outcomes.” Academy of ManagementJournal 35:626–37.

Mellers, Barbara A. 1982. “Equity Judgment: ARevision of Aristotelian Views.” Journal ofExperimental Psychology: General 111:242–70.

Messick, David M. and Keith P. Sentis. 1979.“Fairness and Preference.” Journal of Experi-mental Social Psychology 15:418–34.

Mueller, Charles W., Roderick Iverson, andDong-Gi Jo. 1999. “Distributive Justice Evalu-ations in Two Cultural Contexts: A Compari-son of U.S. and South Korean Teachers.” Hu-man Relations 52:869–93.

Mueller, Charles W. and Jean E. Wallace. 1996.“ Justice and the Paradox of the Contented Fe-male Worker.” Social Psychology Quarterly59:338–49.

Neter, John, Michael H. Kutner, ChristopherNechtsheim, and William Wasserman. 1996.Applied Linear Regression Models. Chicago,IL: Irwin.

Parker, Robert Nash and Rudy Fenwick. 1983.“ The Pareto Curve and Its Utility for Open-Ended Income Distributions in Survey Re-search.” Social Forces 61:872–85.

Phelan, Jo. 1994. “The Paradox of the ContentedFemale Worker: An Assessment of AlternativeExplanations.” Social Psychology Quarterly57:95–107.

Price, James and Charles W. Mueller. 1986.Handbook of Organizational Measurement.Marshfield, MA: Pitman.

Randall, Christine S. and Charles W. Mueller.1995. “ Extensions of Justice Theory: JusticeEvaluations and Employee’ s Reactions in aNatural Setting.” Social Psychology Quarterly58:178–94.

Scher, Steven J. and David R. Heisse. 1993. “Af-fect and the Perception of Injustice.” Pp. 223–52 in Advances in Group Processes, vol. 10,edited by E. J. Lawler, B. Markovsky, K.Heimer, and J. O’Brien. Greenwich, CT: JAIPress.

Sweeney, Paul D. and Dean B. McFarlin. 1993.“Workers’ Evaluations of the ‘Ends’ and the‘Means’ : An Examination of Four Models ofDistributive and Procedural Justice.” Organi-zational Behavior and Human Decision Pro-


cesses 55:23–40.Thomas, W. I. and Dorothy S. Thomas. 1928.

The Child in America: Behavior Problems andPrograms. New York: Knopf.

Törnblom, Kjell. 1992. “The Social Psychologyof Distributive Justice.” Pp. 177–236 in Jus-tice: Interdisciplinary Perspectives, edited byK. R. Scherer. Cambridge, England: Cam-bridge University Press.

van den Bos, Kees, Riel Vermunt, and Henk A.M. Wilke. 1997. “Procedural and DistributiveJustice: What Is Fair Depends More on WhatComes First Than on What Comes Next.”Journal of Personality and Social Psychology72:95–104.

Walster, Elaine, George Walster, and EllenBerschied. 1978. Equity Theory and Research.Boston, MA: Allyn and Bacon.


ferences) on beliefs. They use statisticalmodels that are mathematically incapable ofproducing such estimates. (5) G&H ignoregender differences in trends in belief in lifeafter death. In fact, results for men contra-dict G&H’s conclusions.

CRITICISMS

Differences “Not Significant”

with 12 Cases

G&H are correctly concerned about differ-ences in belief among Jewish denomina-tions. Before 1990, the GSS did not ask Jewstheir denomination. G&H assert, “ Differ-ences among Orthodox, Conservative, andReform Jews are not significant” (G&H, p.831). However, their reported “not signifi-cant” differences for the Orthodox are mis-leading, because it is based on only 12 Or-thodox Jewish respondents in the cumulativeGSS. For example, in a sample of only 12respondents, sample estimates of the popu-lation proportion who believe in the afterlifewould not be significantly different from 43percent (p ≤ .05, two-tailed test) if 2, 3, 4, 5,6, 7, 8, or 9 respondents were to report thatthey believe in the afterlife (based on bino-mial exact confidence intervals estimatedwith Stata 6.0). G&H do not report the num-ber of respondents who are Orthodox Jews.

Erroneous Statements about

Jewish Religious Doctrine

G&H write, “Belief in life after death is nota question of orthodoxy (sic.) for Jews as itis for Christians. As belief in the afterlife in-creases for Jews, they are innovating at leastas much as they are conforming” (p. 828).G&H offer no support for this assertion. Incontrast, consider the following statementsfrom the Oxford Dictionary of the JewishReligion (also see Asheri 1978:195–97;Gillman 1997):

By the time of the Talmud [approximatelythe year 200 of the common era], the con-cept of an afterlife had become highly de-veloped and had a number of components.

Comment on Greeley and Hout, ASR,December 1999

True Facts, True Stories,

and True Differences

Ross M. Stolzenberg

University of Chicago

G

Direct Correspondence to Ross M.Stolzenberg, Department of Sociology, Univer-sity of Chicago, 1126 E. 59th Street, Chicago, IL60637 ([email protected]).

1 I discuss only those problems that undermineG&H’ s core conclusions. One might object alsoto the accuracy, documentation, and appropriate-ness of the “ true story” with which G&H begintheir article, which concerns a conversation be-tween a rabbi and a priest at a Chicago funeral(p. 813).

Comment and Reply

reeley and Hout (1999, hencefor-ward G&H) make extended arguments

about the religions and religious beliefs ofJews and Catholics. They present hypothesesabout trends in those beliefs, which theyclaim to support with analyses of the 1973–1998 General Social Survey (GSS) and withexamination of the traditional teachings ofJudaism and Catholicism. Their argumentsare crippled by critical errors in their statis-tical analyses and in their exposition of Jew-ish religious doctrine.1 In particular: (1) Intheir comparisons of Orthodox, Conserva-tive, and Reform Jews, G&H claim to find“no significant differences” in beliefs in lifeafter death. They do not mention that theGSS includes data on only 12 Orthodox Jew-ish respondents. (2) G&H draw key substan-tive conclusions from statistically nonsig-nificant parameter estimates. They reportneither statistical significance tests nor thesmall numbers of cases for these estimates.(3) G&H misrepresent traditional Jewishdoctrine about life after death. Their asser-tions are inconsistent with well-known pri-mary and secondary sources, including thetraditional Jewish prayer book. (4) G&Hclaim to estimate cohort and age effects (dif-

COMMENT AND REPLYCOMMENT AND REPLYCOMMENT AND REPLYCOMMENT AND REPLYCOMMENT AND REPLY 147147147147147

Souls continued after death and received ei-ther reward or punishment based on theperson’ s conduct during life (divine retribu-tion). The righteous were assigned to thegarden of Eden to receive their reward, andthe wicked were assigned to Geihinnom toreceive their punishment, which was gener-ally not supposed to last more than twelvemonths. The Pharisees developed the doc-trine of bodily resurrection, which was oneof the major doctrinal disputes between thePharisees and Sadducees. . . . In the MiddleAges, Maimonides included resurrection anddivine reward and punishment in his Thir-teen Principles of Faith. (Werblowsky andWigoder 1997:193)

Maimonides’ [1135–1204] Thirteen Prin-ciples of Faith found their way into theprayer book in two versions: a prose versionin which each principle is introduced with“ I believe with perfect faith” (ani ma’amin),and a rhymed version known as Yigdal.(Werblowsky and Wigoder 1997:692)

The traditional prayer book offers a directview of Jewish orthodoxy. The afterlife ismentioned prominently and explicitly inOrthodox and Conservative prayer ser-vices.2 In addition to Yigdal (Harlow1985:326; Scherman 1984:12–13) and otherprayers, the Eighteen Benedictions(Amidah) asserts belief in the afterlife. Con-sider a popular translation for ConservativeJews: “Thou . . . callest the departed to ev-erlasting life. . . . Faithful art Thou, O Lord,who callest the dead to life everlasting”(Silverman [1946, 1973] 1984:96). A trans-lation meant for Orthodox Jews reads, “He. . . resuscitates the dead with abundantmercy . . . causes death and restores life”(Scherman 1984:269). For centuries, theAmidah has been recited once or more ineach Sabbath, holiday, and daily prayer ser-vice for the morning, afternoon, andevening (Elbogen [1913,1972] 1993;Harlow 1985; Scherman 1984). Contrary toG&H’s assertion (p. 814), the afterlife ofthe individual is explicit in Jewish ortho-doxy. For example, in memory of the dead(El Maleh Rachamim), Orthodox Jews ask,

“May his/her resting place be in the Gardenof Eden” (Scherman 1984:455). And, priorto Christianity, “The Book of Daniel (12.2)mentions the resurrection of the righteous”(Werblowsky and Wigoder 1997:193).

Trends Not Significant

G&H’s Table 1 (p. 816) gives the percent-age of persons believing in life after death,by survey year and religion from 1973 to1998. G&H write that Table 1 shows, “Be-lief increased monotonically among Jewssince the early 1970s, reinforcing the im-pression of significant change garnered fromcomparing the end points” (p. 816). G&Hreport neither number of cases nor signifi-cance tests here. My Table 1 (above) pre-sents the number of cases for Jews from theGSS data. To test for association betweentime period and afterlife belief in the 1970sand 1980s, I compute chi-square tests on thecounts in the first four rows (i.e., the 4 × 2tabulation of beliefs in 1973–1975, 1976–1980, 1982–1985, and 1986–1990). ThePearson χ2 is 5.28, the likelihood χ2 is 5.49;neither is significant at p ≤ .05 or even p ≤.10. Thus, we cannot reject the null hypoth-esis that the proportion of Jews who believein the afterlife remained stable throughoutthe 1970s and the 1980s.

For the 1990s, the story is the same: Theproportion who believe in 1991–1994 is notsignificantly different from the proportionwho believe in 1996–1998 (p ≤ .10). Nor canwe reject the hypothesis that the proportion

2 My statements about Jewish orthodoxy do notapply to theological positions taken by ReformJudaism, which rejects key tenets of traditionalJudaism and, by its very name, is neither “ tradi-tional and established,” nor “Orthodox.”

Table 1. Numbers and Percentages of JewishRespondents Believing in Life afterDeath: GSS, 1973 to 1998

Number of Respondents

GSS Don’ t PercentageSurvey Year Believe Believe Believing

1973–1975 46 11 19

1976–1980 52 19 27

1982–1985 39 20 34

1986–1990 68 37 35

1991–1994 32 34 52

1996–1998 34 44 56

All years 271 165 38


believing never declined from one intervalto the next. So G&H’s Table 1 does not sup-port their claim of monotonic increases inafterlife beliefs among Jews throughout the1970s and 1980s, or after 1990. Indeed, thedata are more consistent with the hypothesisthat these percentages remained approxi-mately stable from the early 1970s throughthe 1980s, increased substantially about1990, when the wording of GSS questionsabout Jewish identity changed, and then re-mained steady.

Ignoring Gender Differences

in Effects

Previous research shows strong gender dif-ferentiation in religious beliefs and practices(de Vaus and McAllister 1987; Mueller andJohnson 1975). G&H neither consider norcalculate gender differences in independentvariable effects. To assess the consequencesof ignoring these gender differences, myTable 2 recomputes G&H’s analyses (fromtheir Table 2, p. 818) separately for men and

Table 2. Recalculation, by Gender, of G&H’s Table 2: Logistic Regression Coefficients with Z-Statistics and Wald Test Results

a

Males Females

Tests/Cohorts Protestant Catholic Jewish Protestant Catholic Jewish

Test of the Null Hypothesis b

P-value .093 .232 .348 .244 .001 .001

Number of cases 5,831 2,445 196 8,411 3,285 219

χ2 12.25 9.29 7.83 9.13 24.76 24.24

d.f. 7 7 7 7 7 6

Birth Cohorts

1900–1909 –.603 –.451 –.419 –3.200* –.602 .229(–.493) (–1.245) (–1.913) (–2.274) (–1.877) (1.064)

1900–1909 –.419 –.451 –.603 .229 –.602 –3.200*

(–1.913) (–1.245) (–.493) (1.064) (–1.877) (–2.274)

1910–1919 –.370* –.323 –.370 .104 –.531* –1.101(–2.136) (–1.254) (–.432) (.619) (–2.184) (–1.258)

1920–1929 –.208 –.093 –.347 –.048 –.519* –1.651(–1.473) (–.449) (–.493) (–.357) (–2.761) (–1.859)

1940–1949 .214 .242 .329 –.118 .419* 2.427*

(1.622) (1.234) (.570) (–.951) (2.430) (2.940)

1950–1959 .166 .297 1.266* .091 .606* 3.263*

(1.120) (1.361) (1.967) (.632) (3.186) (3.601)

1960–1969 .325 .575* 1.681 .100 .554* 3.868*

(1.802) (2.289) (1.902) (.605) (2.558) (3.790)

1970–1978 .427 .643 2.381* .372 .666* .NA(1.548) (1.900) (2.274) (1.592) (2.202)

Constant 1.545* 1.221* –1.149* 1.921* 1.331* –2.104*

(13.688) (7.204) (–1.981) (17.747) (8.915) (–2.753)

Log-likelihood –2,757.000 –1,304.083 –12.878 –3,498.209 –1,666.300 –119.540

Note: Standard errors can be obtained by dividing the coefficients by the z-scores. Calculations areunweighted; weighted calculations lead to identical significance test results and nearly identical coefficients.G&H’ s Table 2 appears on page 818 of their December 1999 article. The number of cases in this table differslightly from those in G&H’ s table; see text. Two small coding errors are corrected (see footnote 3).

a For the Wald tests, coefficients for age dummies are not shown.b Tests for the null hypothesis that the cohort coefficients equal 0.*p ≤ .05 (two-tailed tests)


women.3 For example, to reassess G&H’sassertion that their Table 2 logit analyses“ show substantial increases [in belief in theafterlife] from earlier to more recent cohortsfor Catholics and Jews” (p. 819), I use thesegender-specific analyses to test the null hy-potheses that all year-of-birth (cohort) vari-ables have coefficients equal to 0. Cohortvariables are not significant for Jewish andCatholic males, but are significant for Jew-ish and Catholic females. These results arenot caused by small sample sizes: ForCatholic males, the sample size is large(2,445), and results are not significant, evenat p ≤ .20. For Jewish females, the numberof cases is small (219), and the significancelevel of the test is very stringent (p ≤ .0005).Tests for specific cohort variables show thesame thing. For Catholic males, only thedummy variable for the 1960–1969 birth co-hort has a statistically significant coefficient(p ≤ .05). For Jewish males, only coeffi-cients for only the 1950–1959 and 1970–1978 cohorts are significant. This absence ofyear-of-birth effects for Catholic and Jewishmen is entirely inconsistent with G&H’s ar-guments.

Confounding Age, Period,

and Cohort

In their Tables 2, 5, and 7, G&H claim toestimate cohort differences and to distin-guish them from age differences in afterlifebeliefs (henceforward “cohort effects” and“age effects” ). For example, in their Table 2analyses, G&H seek to “ completely purgethe effects of age from the data” (p. 817) andthereby “dismiss the conjecture that the in-creases across cohorts in belief in life afterdeath among Catholics and Jews are an arti-fact of the increased religiousness associatedwith age” (p. 819). Unfortunately, G&H con-found cohort, age, and period (survey year)effects.

The root cause of G&H’s confounded age,period, and cohort effects is the well-knownmulticollinearity of age, period, and cohort.If any two of these quantities are known,then the third is known also, as follows:

YOB = Period – Age;

Period = YOB + Age;

Age = Period – YOB;

where YOB represents year-of-birth (or co-hort), Period is the survey year of the GSS,and Age is the respondent’s years of age atthe time of the survey.

G&H solve their collinearity problems bysimply ignoring period effects; ignoring pe-riod effects does not make them go away(Blalock 1967; Mason et al. 1973:243). Thatis, absent some defensible assumptions toidentify age, period, and cohort effects, it ismathematically impossible to know if G&Hhave found age and cohort effects, age andperiod effects, period and cohort effects, ora combination of all three. G&H couldpresent ancillary evidence to support an as-sumption that period or cohort effects areabsent. Or they could include age, period,and cohort variables but constrain sets of co-horts, periods, or ages to have equal effects.For example, in response to a critic’s com-plaint of a previous underidentified age–pe-riod–cohort analysis, Knoke and Hout(1976) use the second strategy to support aclaim of “constant cohort effects” (p. 905).But G&H simply ignore period effects.(G&H’s two-knot linear spline transforma-

3 I also correct some coding errors. G&H indi-cate that their Table 2 describes analyses basedon respondents who report their religion as Prot-estant, Catholic, or Jewish (p. 817); who wereborn in or after 1900 (per their Table 2); whowere at least 18 years old when surveyed (pertheir Table 2); and who have no missing data onage, birth year, and response to the GSS questionabout belief in the afterlife. G&H report on20,715 respondents—14,489 Protestants, 5,794Catholics, and 432 Jews. Applying those samecriteria to the GSS data, I obtain 322 fewer re-spondents (14,242 Protestants, 5,730 Catholics,and 421 Jews). I am able to replicate G&H’ snumbers exactly (i.e., 432 Jews) if I include re-spondents who were born before 1900, and if Iexclude respondents to the 1998 GSS who re-ported their age as 18 years. Thus, I speculatethat G&H did not actually exclude those born be-fore 1900 from these analyses, and that in ma-nipulating the 1998 data, they included onlythose whose age was greater than (not greaterthan or equal to) 18 years. Respondents born be-fore 1900 would be grouped with those who wereborn in 1930–1939, which is the omitted (or ref-erence) category in G&H’ s dummy variable rep-resentation of the distribution of birth years.


tion of COHORT is no solution: In each ofthe three YOB intervals defined by the knotsthe model is linear and additive.)

To see the consequences of these interde-pendencies for G&H’s analyses, considerlogit models 1 and 2 below, which are esti-mated here for Jewish females in the GSS.Equation 1 is a continuous-variable, linearspecification of G&H’s analyses from theirTable 2 (p. 818). If additional independentvariables are added, equation 1 is also thesame model that G&H estimate in theirTables 5 and 7 (except for the continuousand linear parameterizations of YOB andAge). Y = 1 if the respondent believes in theafterlife; Y = 0 indicates otherwise. Z-statis-tics are presented in parentheses below thecoefficients. The asterisks indicate statisticalsignificance at p ≤ .05.

Logit (Y = 1) = –107.043 + .054 YOB(–2.556*) (2.569*)

+ .028 Age;(1.303)

Log-likelihood = –1224.174;N = 195. (1)

Equation 1 appears to do what G&H claimto do with a dummy variable specificationof the same model: It produces a significantpositive net association between cohort andprobability of afterlife belief, and it finds annonsignificant net association between ageand probability of afterlife belief (z = 1.303;p ≤ .05).

But the absence of age effects in equation1 is illusory. YOB incorporates age (becauseYOB = Period – Age), so equation 1 merelydivides the effect (or association) of age be-tween Age and YOB. The part of that effectthat is not absorbed into the effect of YOB ispositive, not much bigger than its standarderror (z = 1.303), and not statistically signifi-cant. The remaining effects of Age are hid-den, but not absent, from equation 1.

To see the full, negative, and significantpresence of the Age variable in equation 1,consider equation 2.

Logit(Y = 1) = –107.043 + .054 Period(–2.556*) (2.569*)

– .026 Age;(–2.799*)

Log-likelihood = –1224.174;N = 195 (2)

In equation 2, independent variables are Pe-riod and Age only; the coefficient of Age isnegative and significant (z = 2.799). If Ishow that equation 1 can be derived com-pletely and exactly from equation 2, then Ialso show that equation 1 incorporates thestatistically significant negative effects ofage, in contradiction of G&H. The algebraicmanipulations below accomplish the neces-sary derivation:

Logit(Y = 1) = –107.043 + .054 Period– .026 Age,

= –107.043 + .054 Period– .026 Age+ (.054 Age – .054 Age),

= –107.043+ .054 (Period – Age)– .026 Age + .054 Age,

= –107.043 + .054 YOB+ .028 Age;

which is the right hand side of equation 1.So equation 1 equals equation 2. The twoequations are simply different ways to writeexactly the same model.

Thus, it is no surprise that equations 1and 2 have identical log-likelihood statisticsand identical constant terms with identicalstandard errors and identical z-statistics.The coefficient, z-statistic, and standard er-ror of YOB in equation 1 are identical to thecoefficient, z-statistic, and standard errorfor Period in equation 2. If these two equa-tions are simply different ways to write thesame model, then it cannot be true thatequation 2 does show an effect of age,while equation 1 does not show an effect ofage. The same effect or association is re-vealed in equation 2 and is disguised inequation 1.

The chain rule for partial derivatives pro-vides an elegant and general method for re-vealing disguised effects of this type in anydifferentiable equation. Applying the chainrule for partial derivatives to equation 1 pro-duces the following result: Where βAge is thecoefficient for Age in equation 1 and βYOB isthe coefficient for YOB, then


∂[Logit(Y = 1)]/∂Age =

{∂[Logit(Y = 1)]/∂YOB}(∂YOB/∂Age)

+ βAge ,

= –βYOB + βAge ,

= (.054)(–1) + .028 ,

= –.026.

This partial derivative shows that equation 1can be used to calculate the effect of Age,and that effect is identical to the effect ob-tained from inspection of equation 2.

Thus, G&H have confounded age, period,and cohort. Their measures of cohort differ-ences are therefore confounded with the veryeffects they claim to have purged from theiranalyses. Ironically, the confounding of age,period, and cohort in their analyses meansthat their statistical results provide as muchsupport for a theory of age and period effectsas for the theory of age and cohort effectsthat they espouse.

CONCLUSIONS

The shortcomings of G&H’s paper rangefrom the specific to the general, and fromfactual to theoretical to technical. Some ofthese problems invalidate important parts ofG&H’s argument, while others cast doubt onvirtually all of their conclusions. Startingwith the specific, the shortcomings ofG&H’s paper are most severe in its treatmentof Jews: G&H tell us that there are no sig-nificant differences among Orthodox, Con-servative, and Reform Jews in afterlife be-liefs, but their significance test lacks powerand is misleading because it is made on adata set with only 12 Orthodox Jewish re-spondents. If the Orthodox sample size prob-lem is ignored and possible differencesamong Jewish denominations are over-looked, then we must face the failure of theGSS to support G&H’s claim, that “Belief[in the afterlife] increased monotonicallyamong Jews since the early 1970s” (p. 816).If we accept as true G&H’s incorrect claimof monotonic trends for Jews, and if weoverlook the virtually powerless test of dif-ferences among Jewish denominations, thenwe face the inaccurate statement of tradi-tional Jewish religious doctrine—at odds

even with the traditional Jewish prayerbook—that misinforms G&H’s thinkingabout changes in Jewish beliefs in the after-life. If we ignore all the problems specific totheir analyses of Jews, G&H’s analyses ofboth Catholics and Jews still suffer fromconfounding of age, period, and cohort ef-fects. And, finally, even if age–period–co-hort confounding and all other problems areignored, there remains the problem of gen-der effects. Although G&H’s models do notpermit analysis of gender differences intrends, the GSS data do not support theirclaims for either Catholic men, for whomsamples are large, or for Jewish men. Theseproblems—and perhaps others as well—would seem to overwhelm G&H’s claims tohave described trends in afterlife beliefs ofJews, or to have explained trends in the af-terlife beliefs of Catholics and Jews.

Ross M. Stolzenberg is Professor of Sociologyat the University of Chicago, and is the currenteditor of the journal, Sociological Methodology.His forthcoming article in the American Journalof Sociology explores the effects of husbands’and wives’ hours of employment on their ownand each others’ health. His chapters in theforthcoming Handbook of Data Analysis (Sage,forthcoming) consider regression analysis andrelated statistical procedures. He is co-author,with Mary Blair-Loy and Linda J. Waite, of astudy on religious participation over the lifecycle (e.g., “Religious Participation in EarlyAdulthood,” ASR, 1995, vol. 60, pp. 84–103).

REFERENCES

Asheri, Michael. 1978. Living Jewish: The Loreand Law of the Practicing Jew. New York:Everest.

Blalock, Hubert. 1967. “ Status Inconsistency,Social Mobility, Status Integration and Struc-tural Effects.” American Sociological Review35:790–801.

De Vaus, David and Ian McAllister. 1987. “Gen-der Differences in Religion: A Test of theStructural Location theory.” American Socio-logical Review 52:472–81.

Elbogen, Ismar. [1913, 1972] 1993. Jewish Lit-urgy: A Comprehensive History. Translated byR. P. Scheindlin. Reprint, Philadelphia, PAand New York: Jewish Publication Society

Gillman, Neil. 1997. The Death of Death: Resur-rection and Immortality in Jewish Thought.Woodstock, VT: Jewish Lights Publishing.


Greeley, Andrew and Michael Hout. 1999.“ Americans’ Increasing Belief in Life afterDeath: Religious Competition and Accultura-tion.” American Sociological Review 64:813–35.

Harlow, Jules, ed. 1985. Siddur Sim Shalom: APrayerbook for Shabbat, Festivals and Week-days. New York: Rabbinical Assembly, UnitedSynagogue of America.

Knoke, David and Michael Hout. 1976. “Replyto Glenn.” American Sociological Review41:905–908.

Mason, Karen, William Mason, H. H. Wins-borough, and W. Kenneth Poole. 1973. “SomeMethodological Issues in Cohort Analysis ofArchival Data.” American Sociological Review38:242–58.

Reply to Stolzenberg

Getting to the Truths

That Matter

Andrew Greeley

University of Chicago andUniversity of Arizona

Michael Hout

University of California, Berkeley

Our article (Greeley and Hout 1999,henceforward G&H) had two main

points: (1) Between 1973 and 1998, beliefin life after death increased for Catholic andJewish adults, and (2) these trends reflectdynamics internal to the Catholic and Jew-ish religions, not diffusion of beliefs fromProtestants to them. Nothing in Stolzen-berg’s comment warrants revising these con-clusions.

Stolzenberg (2001, henceforward Stolzen-berg) tenders five objections: (1) He notesthat the GSS provides only sparse data onJewish denominations (only 12 respondentsare Orthodox Jews). A larger sample and asignificant result would only have furthersupported our argument. (2) He says we mis-represented traditional Jewish doctrine, but

Mueller, Charles and Weldon Johnson. 1975.“Socioeconomic Status and Religious Partici-pation.” American Sociological Review 40:785–800.

Scherman, Nosson, ed. 1984. Siddur Ahavas Sha-lom / The Complete Artscroll Siddur. PocketSize—Nusach Ashkenaz. Translated by N.Scherman. New York: Mesorah Publications.

Silverman, Morris, ed. [1946, 1973] 1984. SiddurTephilat Yisrael L’Shabbat V’ShaloshRegalim: Sabbath and Festival Prayer Book.Translated by M. Silverman. Reprint, NewYork: The Rabbinical Assembly of Americaand The United Synagogue of America.

Werblowsky, R. J. Zwi and Geoffrey Wigoder.1997. The Oxford Dictionary of the Jewish Re-ligion. New York: Oxford University Press.

our passage that he quotes refers to Jewishrespondents in the GSS, not to doctrine. Thiscriticism is moot. (3) He asserts that the in-crease in belief in life after death is “not sig-nificant” for Jews, but the data as he presentsthem show a statistically significant in-crease. New calculations show even moreevidence of change. (4) He says we ignoregender differences. That is simply wrong;our multivariate models included all the sta-tistically significant effects of gender. Hecalculates separate tests for men’s andwomen’s cohort differences and finds thatthe cohort differences are not significantamong men. The separate tests unnecessar-ily reduce the sample size for each test. Co-hort patterns are nearly identical for men andwomen so we were correct in consideringthem together. (5) Stolzenberg manipulatesthe well-known equation:

Cohort = Year – Age,

to show that the data could be mathemati-cally explained as a function of year and age(we used cohort and age). But his alterna-tive formulation implies that people losefaith as they age. This makes little substan-tive sense in the sociology of religion; ourformulation is more sensible. None of hisobjections stand up to scrutiny, and our con-clusions need no revision.

RESPONSE

Now we respond to each of Stolzenberg’scriticisms:

Direct correspondence to Andrew Greeley,NORC, 1155 East 60th Street, Chicago, IL 60637([email protected]). We thank Claude S. Fischer,Leo A. Goodman, and Martín Sánchez Jankowskifor helpful comments. The views in this reply areour own.


stated in a transitional paragraph—was tomove the reader along, with a modicum ofcontext, from the data on Catholics’ beliefsto the data on Jews’ beliefs. A full review ofthe complexities of these debates would havebeen beyond the scope of our article (and ar-guably beyond the scope of the American So-ciological Review). Stolzenberg avers thatwe “offer no support for this assertion” (p.146). The GSS data are our support. The ob-servation that a majority of Jewish GSS re-spondents interviewed prior to 1990 did notbelieve in life after death establishes that thisbelief is not “customary or conventional”—the meaning of orthodoxy we intended.2

Similarly, the increase from 19 percent be-lievers to 38 percent over time and from 15to 68 percent across cohorts justifies our ref-erence to innovation among American Jewsin the twentieth century.

Findings Not Significant

Stolzenberg breaks the 1973–1998 time se-ries into two periods: 1973–1990 and 1991–1998. He presents an incomplete partition ofthe overall association between time and be-lief among Jews; his tests account for 4 ofthe 5 degrees of freedom in the original table(Stolzenberg, pp. 147–48). To complete thepartition, he should test the hypothesis of nochange between the 1973–1990 period andthe 1991–1998 period (Goodman 1981).Chi-square statistics reject this null hypoth-esis (L2 = 24.30; χ2 = 24.56; d.f. = 1; p <.001). This test is strong evidence that Jews’beliefs increased significantly from the firstto the second period (as Stolzenberg haschosen to divide the data). Had he finishedthe partition and presented this test, he couldnot have claimed that increasing belief is“not significant.”

Table 1 presents a full accounting of theassociation between time and belief amongJews in the GSS. The full GSS time series

Differences “Not Significant”

with 12 Cases

Evidence of differences among Orthodox,Conservative, and Reform Jews would sup-port our conclusion that increasing belief inlife after death among Jews reflects an inter-nal dynamic, not diffusion from Protestants.The data are sparse (we agree), and we can-not reject the null hypothesis. But our argu-ment does not depend on that test. We pre-sented two other pieces of evidence reject-ing diffusion and supporting our conclusionthat change reflects dynamics internal to theJewish community: (1) Jews married to Prot-estants are no more likely to believe in lifeafter death than are Jews with Jewishspouses, and (2) belief increases signifi-cantly as synagogue attendance increases.These two tests are sufficient to make ourcase, so a question of the statistical powerof the third test is a trivial issue.

False Statements about

Jewish Religious Doctrine

The quoted passage is not a false statementabout Jewish religious doctrine. It does notrefer to religious doctrine; it refers to theGSS respondents. Elsewhere in our article (p.814), we draw attention to the antiquity ofJewish teaching on life everlasting. In goingover the data we did have in mind the waythat many teachers and writers in the Reformand Reconstructionist traditions share theGSS respondents’ skepticism (Ellenson1998:72, 76–78),1 but we had no intention ofraising a theological question. Our aim—

1 Ellenson (1998:77) notes, for example, thatGates of Power, a Reform prayer book, replaces“ m’chayei hametim” (“giving life to the dead” )in the second blessing in the Amidah with“ m’chayeh hakol” (“giving life to everything” ).The Reconstructionist prayer book KolHaneshamah inserts “m’chayeh kol chai” (“nur-turing the life of every thing” ). Stolzenberg ad-dresses the beliefs of the Orthodox and Conser-vative Jewish traditions, but except for a mentionin his footnote 2, he does not address the beliefsof Reform Jews. Yet Reform beliefs are relevantto our analysis: Of the 152 Jews for whom weknow both denomination and belief in life afterdeath, 51 percent are Reform, 41 percent areConservative, and 8 percent are Orthodox.

2 The Random House Dictionary of the En-glish Language (1987) offers seven definitionsof “orthodox.” Definitions 1 and 2 refer to de-nominations. Definition 3 is “customary or con-ventional.” As sociologists we feel we are onsolid footing when judging whether opinions are“customary or conventional,” but we would notattempt to discern what is “ sound or correct” (thefourth definition of “orthodox” ) for respondents.


contains 17 data points.3 Change over timein this “ total” table is statistically significant(L2 = 33.62; d.f. = 16; p = .006), as shown inthe top row.4 Our original analysis was basedon six periods (1973–1975, 1976–1980,1982–1985, 1986–1990, 1991–1994, 1996–1998). The six-period cross-classificationalso shows evidence of statistically signifi-cant change (p < .001). The ratio of the six-period L2 to the total L2 shows that our six-period aggregation preserves 89 percent ofthe total association.

Partition 2 shows the result regarding thechange between 1973–1990 and 1991–1998;belief is significantly higher in 1991–1998

than in 1973–1990 (p < .001) as previouslynoted. The within-period L2 (5.69) is the sumof the 3 degree-of-freedom and 1 degree-of-freedom tests that Stolzenberg (pp. 147–48)reports. Because the within-period compo-nent is larger than 3.64 (the critical value ofchi-square with one degree of freedom),there might be additional evidence of sig-nificant change if the right 1 degree-of-free-dom contrast is examined. Indeed, our firstperiod—1973–1975—differs significantlyfrom the rest of Stolzenberg’s first period—1976–1990 (L2 =3.92; d.f. = 1; p = .048).This rather small refinement is important be-cause Stolzenberg suggests that the signifi-cant increases in positive responses to theGSS question on life after death were in-duced by a change in the questionnaire (p.148).5 Evidence that beliefs increased sig-

3 There was no GSS in 1979, 1981, 1992, 1995,or 1997, and the life after death question was notasked in 1974, 1977, 1982, and 1985.

4 The Pearson chi-square test also rejects thenull hypothesis. We use the likelihood ratio test(L2) because it is additive across most of the par-titions in Table 1.

Table 1. Partitioning the Likelihood Ratio Statistic for the Association between Survey Year andBelief in Life after Death: GSS, Jewish Adults in the United States, 1973 to 1998

PercentageComparisons L2 d.f. p-Value of Baseline

Total: All 17 years 33.62 16 .006 —

Partition 1: Six Periods Between 29.99 5 < .001 89

Within 3.63 11 .979 11

Partition 2: Two Periods Between 24.30 1 < .001 81

Within 5.69 4 .224 19

Partition 3: Three Periods Between 28.22 2 < .001 94

Within 1.77 3 .622 6

Partition 4: Change between 1987 and 1988 Split 1973–1990 into 1973–1987 and 1988–1990 .92 1 .338 —

Split 1986–1990 into 1986–1987 and 1988–1990 .0049 1 .944 —

Partition 5: Uniform Association Linear increase in log-odds 28.27 1 < .001 94

Residual 1.72 4 .788 6

Note: N = 440; cases of “ no opinion” are excluded.a The 17 survey years include 1973, 1975, 1976, 1978, 1980, 1983, 1984, 1986, 1987, 1988, 1989, 1990,

1991, 1993, 1994, 1996, and 1998.b The six periods are 1973–1975, 1976–1980, 1982–1985, 1986–1990, 1991–1994, and 1996–1998.c The two periods are 1973–1990 and 1991–1998.d The three periods are 1973–1975, 1976–1990, and 1991–1998.

5 Questions about current Jewish denominationand denomination of origin were added to theGSS in 1988.


nificantly 12 years before the questionnairechanged weakens his claim.

The definitive test of whether the observedchange in belief is an artifact of the 1988questionnaire change, though, is to break thetime series right at 1988. This test can bedone by splitting the full 1973–1990 seg-ment into two parts or by splitting 1986–1990 into two parts. The result is the sameboth ways: No significant difference be-tween 1973–1987 and 1988–1990 (p = .338)nor between 1986–1987 and 1988–1990 (p= .944). Thus, the observed changes in Jews’statements about life after death cannot beattributed to a change in the questionnaire.

The first four tests in Table 1 ignore thetime sequence of the surveys. Our originalconclusions, though, refer to a monotonic in-crease in belief. A general test of monotonicchange can be tedious because monotonicitycan take many functional forms. A verysimple model—uniform association—isavailable to test for linear change in the log-odds of belief. A good fit for uniform asso-ciation is not necessary, but it is sufficient toestablish a case for a monotonic trend.6 Par-tition 5 in Table 1 shows the result. The uni-form association model fits extraordinarilywell as the small residual L2 (1.72) and aplot of observed and expected percentagesshow (see Figure 1).7 The trend parameter ispositive and significant (p < .001). The good

fit of uniform association and the 37-pointincrease between the first period and the lastare strong evidence in favor of our conclu-sion that Jews’ belief in life after death in-creased monotonically over the period cov-ered by the GSS.

Ignoring Gender

We did not ignore gender; Stolzenberg ismistaken on this point. Models 2 to 6 of ourmultivariate analysis of Catholics’ beliefs in-clude it (see G&H, Table 5, pp. 824–25) asdoes our first multivariate model for Jews(Model 2 in G&H, Table 7, p. 829). Wedropped gender from Models 3 to 7 for Jewsbecause the data provided no evidence thatgender differences among Jews were signifi-cant.8

Stolzenberg splits the GSS into separatemale and female samples before testing forcohort differences. The only justification forthis practice would be a test that shows sig-nificant differences between men’s andwomen’s cohort effects. Stolzenberg presentsno such test. There are two ways to do thetest, and both indicate that the differences arenot significant. Adding the gender × cohortinteraction terms that define the differencebetween his model and ours results in likeli-hood ratios of 10.26 for Catholics (d.f. = 7, p= .17) and 10.48 for Jews (d.f. = 6, p = .11).9

6 A linear trend in log-odds is one form ofmonotonic trend but there are many others.

7 The line in Figure 1 is curved because theuniform association model is linear in log-oddsbut not in probablities.

0

15

30

45

60

1970 1980 1990 2000Year

Per

cen

tag

e B

elie

vin

g in

Lif

e af

ter

Dea

th

Figure 1. Observed Percentage Believing in Life after Death, by GSS Survey Year, 1973 to 1998

8 The gender coefficient in Model 2 is onlyone-fourth as large as its standard error.

9 All Jewish women in the most recent cohortsay they believe in life after death—forcing per-fect prediction and reducing the d.f. by 1.


Wald tests yield nearly identical results forCatholics (W = 10.27, d.f. = 7, p = .17) and amore definitive conclusion for Jews (W =6.24, d.f. = 6, p = .40).10 Lacking evidence ofsignificant gender differences in cohort ef-fects, Stolzenberg reduced the power of histests of cohort differences against the nullhypothesis and got specious null results.

Figure 2 helps interpret how Stolzenberg’sfindings and ours could come from the samedata. He splits men and women while wecombine them; Figure 2 plots all of the co-efficients by cohort and religion. All of thecoefficients (men, women, and total) followthe pattern of increasing belief across co-horts that we noted in discussing our origi-nal Table 2. As Stolzenberg notes, the 95-percent confidence intervals for several ofthe men’s coefficients overlap zero leadinghim to conclude that they are not significant.However, those same confidence intervalsalso overlap the corresponding coefficients

for the total sample and the 95-percent con-fidence intervals for the correspondingwomen’s coefficients; that means that thegender differences are not significant. Mostimportant, all but three of the gender-spe-cific coefficients are closer to the corre-sponding total coefficient than to zero (fur-ther indicating that men and women shouldbe combined). Cohort differences in thecombined model are significant at the p <.001 level (the Wald tests are 26.45 forCatholics and 35.75 for Jews with d.f. = 7for each).

Confounding Age, Period,

and Cohort

Stolzenberg presents several ways of look-ing at the well-known equation

Cohort = Year – Age,

but none of them constitutes evidence thatour age–cohort model is wrong in any way.Our model is both internally consistent andconsistent with the sociology of religion lit-erature; the alternative models—though for-mally equivalent to ours—entail implausible

10 In producing Table 2 in G&H, we errone-ously included persons born prior to 1900 andcoded them as if they were born 1930–1939; thecalculations in this paragraph correct this error.

1970–1980

1960–1969

1950–1959

1940–1949

1930–1939

1920–1929

1910–1919

1900–1909

–6 –3 0 3 6Cohort Coefficient

–2 –1 0 1 2

Cohort Coefficient

Yea

r o

f B

irth

Jewish Catholic

Men Women Total

Figure 2. Cohort Coefficients for Men, Women, and All Adults, by Religion

Note: The Horizontal bars around the men’ s and women’ s coeficients indicate the 95-percent confidenceintervals for these coefficient.


counteracting effects and have no basis inthe literature.

The age–period–cohort problem is not ascomplicated as the proofs and derivatives inStolzenberg make it seem. The “ identifica-tion problem” is simply the choice of how toconnect the symbols in Figure 3a. A re-searcher can connect the symbols that repre-sent people born in the same year (the age–cohort model) or connect the symbols thatrepresent people interviewed in the sameyear (the age–period model).11 Figure 3bshows the age–cohort perspective; Figure 3cshows the age–period perspective. From ei-ther perspective, the younger people areclearly more likely to believe in life afterdeath than the older people are. At issue ishow that pattern developed. Perhaps peopletend to be more believing as they grow olderand more recent cohorts started out with ahigher propensity to believe; that is the age–cohort perspective. The alternative is that theyounger people are temporary believers whotend to lose faith as they age, but the timesare so religious that they override this ten-dency; that is the age–period perspective.

The two perspectives account for the GSSdata equally well. Choosing between them isa matter of citing other evidence and decid-ing which of the arguments is more inter-nally consistent. The sociology of religionliterature supports the age–cohort perspec-tive; cross-sectional and panel studies (e.g.,Stolzenberg, Blair-Loy, and Waite 1995) in-dicate that other facets of religiosity increasewith age. The age–cohort perspective also issimpler; it can be summarized in twocomplementary statements. The age–periodalternative relies on an implausible combi-nation of negative age effects andcountervailing period effects strong enoughto override them. In this it gets no supportfrom the literature; no religious scholar weknow of has argued that people lose faith asthey age, and many have argued in favor ofnegative period effects. Our age–cohort per-spective plainly works better as a sociologi-cal model, even though the age–period per-

5 15 25 35 45 55 65 75 85

5 15 25 35 45 55 65 75 85

Per

cen

tag

e B

elie

vin

g in

Aft

erlif

e

Age

75

50

25

0

1933

1943

1953

1963

Year of Birth:1973

1903

1913

b. Age–Cohort Perspective

Per

cen

tag

e B

elie

vin

g in

Aft

erlif

e

Age

75

50

25

0

a. Blank Perspective

1988

1978

5 15 25 35 45 55 65 75 85

Age

c. Age–Period Perspective

1973

1983

1993

GSS Survey Year:1998

Per

cen

tag

e B

elie

vin

g in

Aft

erlif

e 75

50

25

0

Figure 3. Percentage Believing in Life afterDeath Expected under Model 7 ofGreeley and Hout: Three Views

Source: Greeley and Hout (1999:829–31).

Note: Symbols indicate data points that are in thesame cohort.

11 The symbols in Figure 3a display the per-centages believing in life after death expectedunder our Model 1 in Table 7 of G&H (p. 829),but the points we raise here are general ones thatapply to any age–period–cohort analysis.


spective results in the same expected prob-abilities.

SUMMARY AND CONCLUSION

We have countered all five criticisms raisedby Stolzenberg. His criticism that we use alow-power test of denominational differ-ences is moot because we have other, stron-ger evidence for our argument. Our use ofthe lowercase “ orthodox” in a transitionalparagraph was not an erroneous statementabout Orthodox Judaism, just a reference tothe observed variance in the beliefs of allJews in the GSS. The change over time inbeliefs among Jews is statistically signifi-cant over the whole time series and betweenthe periods that Stolzenberg contrasts; fur-ther analysis supports our conclusion thatJews’ beliefs in a life after death increasedmonotonically between 1973 and 1998. Wedid not ignore gender; we included its sta-tistically significant effects in our multivari-ate analyses. We did not test for cohort ef-fects for men and women separately becausetests indicate that the cohort effect is thesame for men and women (the main cohorteffect is statistically significant). Stolzen-berg’s manipulation of the age–period–co-hort identities leaves him with an implau-sible model that implies that unspecified re-cent events have counteracted peoples’ ten-dency to lose faith as they age. Our age–co-hort model provides a substantive explana-tion that is free of overriding effects and ac-cords well with the sociological literature onaging and religion.

We made no “critical errors.” There is noreason to change the findings or conclusionsof our original paper in light of this comment.

Andrew Greeley is Visiting Professor of Sociol-ogy at the University of Chicago and at the Uni-versity of Arizona. His most recent sociologybook is The Catholic Imagination (University ofCalifornia Press, 1999). He is currently workingon a book about religion in Europe at the begin-ning of the third Millennium.

Michael Hout is Professor of Sociology at theUniversity of California, Berkeley, where heteaches courses on research methods and socialinequality. In addition to his work with AndrewGreeley, he and Claude Fischer are directing aproject that will address American social andcultural trends in the twentieth century.

REFERENCES

Ellenson, David. 1998. “Modern Liturgies.” Pp.72–78 passim in My People’s Prayer Book:Traditional Prayers, Modern Commentaries,edited by L. A. Hoffman. Woodstock, VT:Jewish Lights Publishing.

Goodman, Leo A. 1981. “Criteria for Determin-ing Whether Certain Categories of a Cross-Classification Table Should Be Combined.”American Journal of Sociology 87:612–50.

Stolzenberg, Ross M. 2001. “ True Facts, TrueStories, and True Differences” (Comment onGreeley and Hout, ASR December 1999).American Sociological Review 66:146–52.

Stolzenberg, Ross M., Mary Blair-Loy, andLinda J. Waite. 1995. “Religious Participationover the Early Life Course: Age and FamilyLife Cycle Effects on Church Membership.”American Sociological Review 60:84–103.

Documents

SOCIAL JUSTICE AND SOCIOLOGY 1 Social Justice and ... · roots in a sociology committed to social justice, to cultivate and extend the long-standing “countersystem” approach to