Outgroup Favoritism and the Theory of System Justification: A Paradigm forInvestigating the Effects of Socioeconomic Success on Stereotype Content

John T. Jost

Stanford University

And in the momentyou arebornsinceyou doWtknow anybetter,everystickandstoneandeveryfaceis white,and sinceyou havenotyet seenamirror, you supposethatyou are,too. It comesasa greatshockaroundtheageof S or 6 or 7 todiscoverthat theflag to which you havepledgedallegiance,alongwith everybodyelse,has notpledgedallegianceto you.It comesasa greatsurprisethat GaryCooperkilling off the Indians,whenyou wererootingfor GaryCooper.that theIndianswere you. (Interviewwith JamesBaldwin, The Priceof theTicket)

It is not only the discipline of psychology tbat hasundergonea cognitiverevolutionin the latterhalfof the20thcentury.Socialscienceingeneralhasshifted towardan increasedrelianceon mentalstatesandprocessesinexplaining the behaviorsof individuals and groups.Sociologistsandpolitical scientists,for example,havefocused more and more on expectations,attitudes,beliefs,perceptions,decisions,andjudgmentsinexplain-ing suchdiversephenomenaas political socialization,group dynamics,votingbehavior,andotherresponsestostructuresof status,power,and prestige(e.g.,Berger&Zelditch, 1998; Howard, 1994; Iyengar & McGuire,1993; Ridgeway, 2001). Anthropologists, too, havemovedincreasinglytowardbeliefs,construals,interpreta-tions,and otherintentionalstatesin their descriptionsofcultural systemsand practices (e.g., Geertz, 1983;Shweder& LeVine, 1984;Sperber,1990).Evenphiloso-phershaveadoptedtbelanguageof cognitivesciencetothepoint wheretraditional metaphysicalandepistemo-logical approacheshavealmostdisappeared(e.g.,Gold-man, 1988;Kornblith, 1994;Solomon, 1992).

In the field of organizational behavior, social-cognitive constructs such as attributions, accounts,scripts,andjustificationshavebeenusedto shedlight onsuchappliedtopicsasjob satisfaction,division of labor,employeerelations,taskdesign,andcorporatestrategy(e.g.,Baron& Pfeffer, 1994;Martin, 1982;Salancik&Pfeffer,1978;Weick, 1993).Neo-institutionalisttheoriesof organizationshavefurtheranalyzedthe ways in whichideasandsymbolsare usedto structureand legitimatebusinessculturesandspreadinfluence(e.g., Powell &DiMaggio, 1990). Somewhat improbably given thesubjectmatter,cognitive theoriesof social movements

and revolutionshavebecomeparadigmatic(e.g.,Eyer-man & Jamison,1991;Moore, 1978; Snow& Oliver,1995),evenamongMarxist scholars,who aretradition-ally among the least individualistic of social theorists(e.g., Blster, 1985). Theseintellectual developments,spreadas they areacrossa variety of disciplines,meanthat constructssuch as attitudes,thoughts,and beliefshaveproveduseful indeedfor explainingbehaviorthatclearly falls outsideof the original domainof cognitivepsychology— in this case,behaviorthat is collective,coordinated,and downrightpolitical.

More precisely, one might say that in the post-cognitiverevolution world, sociologists,psychologists,anthropologists,political scientists,andorganizationaltheoristsall accept the fundamentalassumptionthatsocial systemsare maintainedat leastin part throughattitudesandbeliefsthat supportthem.In the languageofsocial cognition, researcherswould say that consciousandunconsciousthoughtprocessesplay apivotal role inthe acceptanceor rejection of particular social andpolitical forms (Jost, 1995). Onevariablein particular,theappraisalof legitimacy,hasemergedas animportantsocial-psychologicalpredictorof responsesto inequality(e.g.,Major, 1994;J. Martin, 1993; Tajfel & Turner,1986;Tyler, 1997).Historical and ethnographicstudiesmakeabundantlyclearthat, in theabsenceof militant ortotalitarian rule, authorities,procedures,and socialarrangementsare stableandenduringto the extentthattheyareperceivedashavinglegitimacy(e.g.,Gun,1970;Moore, 1978). Inequalityamonggroupsand individualsis acceptedand perpetuated,evenby thosewho standtolosethemostfrom it, solongas it is perceivedasfair andlegitimate.This is one of the startingpointsof the theory

89

90 JOST

of systemjustification (e.g., Jost& Banaji, 1994),whichis intendedas a social-cognitivetheory of intergrouprelationsandpolitical behavior,and it is also oneof thecentralcontentionsof this chapter.

THE JUSTIFICATION PRINCIPLE INSOCIAL PSYCHOLOGY

In their ground-breakingwork on The Social Construc-

tion of Reality,BergerandLuckmann(1967) observedthat “the institutional world requireslegitimation, that is,ways by which it can be ‘explained’ and justified~’ (p.61). In otherwords,wedo not supportsocialstructuresunlessthey satisfy our cognitive needsfor validity andrationality. Sociologistsand organizationaltheoristshavedescribedthe excuses,accounts,andexplanationswe useto smoothoversocial interaction(e.g.,Scott & Lyman,1968), spreadorganizationalinfluence(e.g., Powell &DiMaggio, 1990),andpreservepublic reputations(e.g.,Elsbach& Sutton,1992).However, it is the disciplineofpsychologythat remindsus of the factthat peoplemustgeneratereasonsand justificationsprivately, for them-selves,aswell asfor others(e.g.,Aronson,1992;Festin-ger, 1957;Tetlock& Manstead,1985;Weick, 1993).

Psychologistsknow betterthan anyoneelsethat weare cognitivecreatures;we needreasonsandargumentsto justify bothaction and inaction. If we behavein aninconsistentor counterattitudinalway, we mustcomeupwith rationalizationsfor the departure.If we do some-thing that causessocial disapproval,we must defendourselveswith ideas. And sometimes,we just want toknow that thereare reasonsand explanationsfor theactionsof othersand theway thingsare.Theseprinciplesare assumedby such diverse theories and researchprogramsin social psychologyas balancetheory, cogni-tive dissonancetheory, equity theory, attribution theory,and social cognition. In otherwords,oneof the centralcontentionsof socialpsychologyfrom the 1950sto thepresenthas beenthat peoplejustify themselves,theirassociates,and the world aroundthemin thesensethattheyuseideasto providevalidationand legitimacyfor allof thesethings(Jost& Banaji, 1994).

Justificationsdiffer from explanations,in that theformer (but not necessarily the latter) render socialeventsright andappropriate.It is no accidentthatjusticeand justification derivefrom the sameLatin root. Theexistenceof classsystemsmay be justified — that is,madelegitimate— by postulatingwide individual differ-encesin effort or motivation;patriarchalsystemsmay bejustifiedby assertinginsurmountablegenderdifferencesin achievementor ability; andsystemsof ethnicor racialscgregationmaybejustifiedby claiming incommensura-

ble groupdifferencesin intelligenceor morality. In thisway, peopleuseideasaboutgroups and individuals toreinforceexistingsocial systemsandpreservethesensethat those systemsare fair, legitimate, and justifiable(Just& Banaji, 1994). This neednot be a consciousprocess,as Bern andBem (1970) pointed out in theiranalysis of gendersocializationand the subtle, non-consciousways in which girls are taughtto ‘know theirplace’ and to participatein a sexistworld.

Yet why would peopleperceivethe world aroundthemto be justifiedand legitimatewhensomany featuresof their environmentseemunfair and illegitimate? Oneanswerthat social psychologistshavegiven is that thereis a generalmotivation to “believe in ajust world” (e.g.,Lerner,1980;Olson& Hafer, 2001).Theguiding thesisis that living in circumstancesthat are unpredictable,uncontrollable,and capriciously unjust would be toopsychologically threatening,and so we cling to theillusion that peopleget what theydeserveand deservewhattheyget. The theoryof systemjustification buildson this essential insight (Jost & Banaji, 1994), de-emphasizingsomewhatthe universal, psychodynamicaspectsof theprocessandstressinginsteadtheimpact ofsociallearning, institutionalizednorms,andthe powerofideology (ef. Bem& Bem, 1970;Berger& Luckmann,1967;Major, 1994;Tyler & McGraw, 1986).

Stereotypesas IdeologicalJustifications

Oneof themostcommonways in which peopleuseideasand beliefs tojustify thesocial world aroundthemis bystereotypingmembersof disadvantagedgroups in waysthat rationalize the inequality. The stereotype thatAfrican Americans(orHispanicAmericansorblue-collarworkers)arenotasintelligent or hardworking asother,more successfulgroups, accordingto this perspective,servesas an ideologicaljustificationfor the substantialsocioeconomicdifferencesbetweenthesegroups andothers.Thisjustification or rationalizationfunction wasrecognizedby Katzand Braly (1933)andAllport (1954),

but it was not studieddirectly until relatively recently,when cognitive and ideological analyseswere againcombined.

In sociology,JackmanandSenter(1983)providedsurvey data leading to the conclusionthat ideologicalvaluesandstereotypeswere relatively consensualandfavoring of dominantgroups’ interestsacrossracialandgendercontexts.C. HoffmanandHurst(1990)conductedexperimentalresearchin which peopleformed stereo-types of fictional groups as a way of rationalizingun-equal divisions of labor and social roles; the authorsexplicitly calledinto questionthekernelof truth view of

6. OUTGROUPFAVORITISM AND SYSTEM JUSTIFICATION 91

stereotypes,arguing that stereotypecontents reflectrationalizationratherthan perceptualprocesses.Ridge-way (2001), too, hassummarizedfield andexperimentalstudies,leadingto theconclusionthat peopleform statusbeliefs in the courseof faceto faceinteractionin suchaway that structuralinequalitiestendto be legitimizedandperpetuated.In all of theseresearchprograms,we seethat stereotypesoperateas ideologicaldevicestojustifyorrationalizeinequality betweengroups.

The mostsurprisingand powerful casesof system-justifying stereotypesarisewhenmembersof low-statusgroupsinternalizeunfavorablestereotypesof themselvesandfavorablestereotypesof othersasa way tojustify theexistinghierarchy(Jost& Banaji, 1994).This processmay give rise to the attitudesandbeliefs that are out-groupfavoring. Such ideasandjustificationsaretaughtto us as children until they begin to operatenon-consciously(Bern & Bern, 1970). Ultimately, system-justifying ideologiesandstereotypesbecomeimpercepti-ble — like waterto the fish.

Outgroup-Favoring StereotypesUnder Extremeand Ordinary Circumstances

Fromapsychologicalstandpoint,it is especiallystrikingthat disadvantagedpeoplewould justify the very socialsystemthat placesthemat a disadvantage.Nevertheless,it seemsthat peoplesometimesdo favor thepreservationof the social order evenover their own personalandcollectiveinterests,as the opening quote from JamesBaldwin attests.Familiar examplesin the domain ofpublic opinion include women’s widespreadfailure tosupport the Equal Rights’ Amendment(Gurin, 1985;Mansbridge,1986) and the significant lack of supportamong the American working class for policies ofeconomicredistribution(Kluegel& Smith, 1986;Piven& Cloward,1977).

Evenmoredramaticexamplesarefoundin historicaland journalistic accountsof war camps and slavery.Psychoanalyticallyinspiredwork by AnnaFreud,BrunoBettelbeim,andotherson thephenomenonof “identifica-tion with theaggressor”suggestedthat,amongvictims ofextremeinjustice and deprivation,therewas an implicitrejectionof the ingroup and a preferencefor the out-group. A historical novel set during the Holocaustprovides a stirring literary examplein which a youngGypsy boy encountershis first Nazi officer. In ThePaintedBird, JerzyKosinski(1966)writes:

His entire personseemedto have somethingutterlysuperhumanaboutit... he seemedanexampleof neatperfectionthat could not besullied: the smooth,p01-ishedskin of his face, the bright goldenhairshowing

underhis peakedcap, his puremetal eyes...1 thoughthow good it would be to havesucha gleaming andhairlessskull insteadof my Gypsy face which was sofearedanddislikedby decentpeople(p. 100).

Similarly disturbingcasesof systemjustificationexist incontemporaryaccountsof the ongoingslave trade inAfrica and Latin America.A 1992 articlepublishedinNewsweekmagazine(Masland,Nordland,Liu, & Con-treras,1992),for example,quoteda 25-year-oldMaurita-nian slaveasfollows:

I am a slave,my wholefamily are slaves...Sometimesthey treat us well, sometimesthey treatus badly, butonly the childrenget beaten...A masteris a masteranda slaveis a slave.Mastersare white,slavesare black...Naturally, weblacksshouldbetheslavesof the whites(p. 32).

Theseanecdotalexamplesareof rareandextraordinarycircumstances.Theyare relayedto suggestthatpsycho-logical investmentin the statusquo may occur evenunderthemosthorrific systemsof inequalityandexploi-tation.If somedegreeof systemjustificationarisesundersuch dramaticconditions, system-justifying impulsesshouldbeevenmorelikely to arisein ordinarylife.

It was a widespreadassumption,in fact, of earlyresearchersof intergroup relations that membersofdisadvantagedgroupssuchas JewsandBlackscouldnothelp but internalizesociety’s biasesagainstthem andexhibit a kind of inferiority complexat the group level(e.g., Allport, 1954; Bettelbeim, 1960; Lewin, 1941).This wasalso theconclusionreachedby Clark andClark(1947) in their famous studiesof African-Americanchildren’s preferencesfor White dolls. A series oflaboratory studiesconductedby Sachdevand Bourhis(1985, 1987, 1991) demonstratedconvincingly thatassignmentto high-statusor powerful groups leadspeopletodisplayingroup favoritism,whereasassignmentto low-statusor powerlessgroups producesoutgroupfavoritism.

Becauseevidenceof outgroupfavoritism is oftendeemphasizedin the literatureon intergrouprelations,the data from tables reported in an influential meta-analyticstudy conductedby Mullen, Brown, andSmith(1992)arerearrangedandpresentedin Table6.1 accord-ing to thepercentageof experimentalgroupsshowingingroup favoritism, outgroup favoritism, and exactequality— brokendownaccordingto relativestatusof theingroup.Whatthis revealsis that membersof high-statusandequal-statusgroupsareindeedgoingoverwhelminglywith ingroup favoritism, but 85% of the low-statusgroupsaredisplayingoutgroupfaVoritism. That is, they

92 JOST

TABLE 6.1: Numberandpercentageof experimentalgroupsshowing ingroup favoritism, exact equality,favoritismby relativestatusof the ingroup.

and outgroup

Status

High Equal LowIngroupfavoritism

Number 20 27 3Percentage 100% 73% 15%

ExactequalityNumber 0 2 0Percentage 0% 5% 0%

OutgroupfavoritismNumber 0 8 17

Percentage 0% 22% 85%Note. Tableadaptedfrom a meta-analysisreportedby Mullen,Brown, and Smith (1992). Ingroup favoritism indicatesapositive Z-scorcreported by Mullen et al., exact equalityindicatesa Z-scoreof zero,outgroupfavoritism indicatesanegativeZ-score.

are saying that membersof the other group are moreintelligent, more industrious,andso on, than are mem-bersof their own group.This typeof evidenceled JostandBanaji (1994)to arguethatmanylow-statusgroupsacceptaslegitimatetheir allegedinferiority. Ratherthanattempting to raise self-esteemor enhanceingroupsolidarity, they usetheir evaluationsandjudgmentstoreinforceandjustify theexistingsystemof inequality. Insomeways, this interpretationimplies a rediscoveryoftheMarxianconceptof falseconsciousness(Jost, 1995;Jost& Banaji, 1994).

Contraryto someclaims (e.g.,Mullen et al., 1992),systematicevidence of outgroup favoritism does notseemto be restrictedto laboratorygroups.Fieldstudiesconductedby,amongothers,R. Brown (1978),Hewstoneand Ward (1985), Jost, Burgess,and Mosso (2001),Mlicki and Ellemers (1996), Skevington(1981), andSpearsand Manstead(1989) have turned up strongevidence of outgroup favoritism among membersofvariouslow-statusgroups.In fact, evidencecoming frommanyreal-worldgroupssupportsRogerBrown’s (1986)observationthat:

Subordinate groups like black Americans, SouthAfrican Bantus, the Mayansof Guatemala,and thelower castesof India either do, or until recentlydid,derogateor look down on the in-group and showpositive attitudestoward the depriving out-group(p.558).

of intergrouprelations,social identity theory, is notwell-equippedto handlethe phenomenonof outgroupfavorit-ism (e.g., Hewstone& Ward, 1985; Hinkle & Brown,1990; Jost & Banaji, 1994; Sidanius, 1993), althoughitis true that the subjecthasbeenaddressedin somedetailby social identity theoristssuch as Turner and Brown(1978),Tajfel andTurner (1986),SpearsandManstead(1989),andothers.In many ways, systemjustificationtheoryseeksto build on the foundation laid by socialidentity theory, much as social identity theoristssoughtto build on theoriesof social comparisonand realisticconflict (Tajfel & Turner, 1986).

Outgroup Favoritism and Social Identity Theory

Social identity theorywasdevelopedto accountfor theinitially unexpectedfinding that minimal laboratorygroupswith no history of interactiondisplayedingroupfavoritism with regard to social stereotyping,perfor-mance evaluation, and resourceallocation (Tajfel &Turner,1986).Drawingextensivelyon Festinger’ssocialcomparisontheory, it was argued that becausepeopleneedto evaluatethemselvesfavorably andbecausegroupmembershipis an importantconstituentof the self-concept,people tend to evaluate their ingroups morefavorably thantheyevaluateothergroups(e.g.,Tajfel &Turner, 1986;Turner, 1975).Thus, accordingto socialidentity theory, thereis a general drive to enhanceindividual andcollectiveself-esteemby making favor-able comparisonsbetween the ingroup and relevantoutgroups(e.g., Hogg& Abrams, 1988;Tajfel & Turner,1986).

Althoughit is truethat social identity theory empha-sizes the generalizabilityof ingroup favoritism amongmembersof manydifferenttypesof socialgroups (Hogg& Abrams, 1988;Mullen et al., 1992;Tajfel & Turner,1986),it hasalsodonemuchto framethesocial-psycho-logical understandingof how andwhengroupsthat arelow in socialstandingwill accepttheir allegedinferiorityandwhen theywill attemptto challengeit (e.g.,Ellemers,Wilke, & van Knippenberg, 1993; Murrtmendey &Schreiher,1984; Spears& Manstead,1989;Turner&Brown, 1978;van Knippenberg,1978;Wright, Taylor,& Moghaddam,1990).From this bodyof literature, it ispossibleto discernthreedistinct explanationsfor thephenomenonof outgroup favoritism among low-statusgroups.Oneaccounthas to do with self-categorizationprocessesof identification anddis-identification(Tajfel& Turner, 1986),anotheraccountdistinguishesbetweencomparativedimensionsthat arerelevantand irrelevantto thestatusdifferences(van Knippenberg,1978),andathird accountis relatedto perceptionsof the legitimacyIt hasoftenbeensuggestedthat theprevailing theory

6. OUTGROUPFAVORITISM AND SYSTEM JUSTIFICATION 93

and stability of statusdifferences(Turner & Brown,1978).

According to the first type of explanation,membersof low-statusgroupsexhibit outgroupfavoritismto theextent that they shun identification with their ownnegativelyvaluedgroup and identify insteadwith mem-bersof a positively valuedoutgroup(Tajfel & Turner,1986). It has beenarguedthat disidentification is thepreferredchoice amonglow-statusgroup membersingeneral(e.g., Ellemers,van Knippenberg,de Vries, &Wilke, 1988;Lewin, 1941; Tajfel, 1978),andresearchsuggeststhat whenthe option of individual mobility orexit is available,low-statusgroup memberstend to takeit (Lalonde & Silverman, 1994; Wright et al., 1990).Researchhas also demonstratedthat peoplewho aremadeto identify only weakly with a low-statusingroupare less committed to the group and more likely toexpressa desirefor individual mobility to anothergroupthan arepeoplewho aremadeto identify stronglywiththe ingroup(Ellemers,Spears,& Doosje,1997).

Although it seemsplausiblethat levels of ingroupidentification would predictthedirectionandmagnitudeof ingroup versusoutgroupfavoritismamongmembersof low-statusgroups, there are reasonsto think thatideological factors such as the perceptionof systemlegitimacyplay a more determiningrole. Forinstance,group consciousness-raisingamongwomenandminoritygroupsrequiresnot merelyan identification with one’sgroup, but a perception that the group’s low statusisillegitimate and unfair (e.g., Gurin, 1985; Kluegel &Smith, 1986).Thus, from our perspective,group identifi-cationis probablyanecessarybutnotsufficientconditionfor therejectionof outgroupfavoritism.

Accordingto the secondtypeof explanation,mem-hersof low-statusgroupsmayaccepttheir inferiority andengagein outgroup favoritism on dimensionsthat arehighly relevantto the statusdifferences,but they mayexhibit ingroup favoritismon irrelevantdimensionsas away to compensatefor an otherwisenegativesocialidentity(e.g., van Knippenberg,1978).In fact, thereis awealthof evidenceto supportthenotionthat membersoflow-statusgroups accept their inferiority and exhibitoutgroup favoritism on dimensions that are highlyrelevantto thestatusdifferences,but theyexhibit ingroupfavoritism on irrelevantdimensionsas a wayto compen-sate for an otherwise negativesocial identity (e.g.,Mullenetal., 1992;Mummendey& Schreiber,1984;vanKnippenberg, 1978). For instance,Skevington(1981)examined intergroup relations among professionalnursinggroupsthat differedin statusand foundthat low-statusgroup membersjudgedthe othergroupto be moreintelligent, ambitious, responsible, organized, and

confidentthan their own group,but theysawthemselvesas morecheerful,thoughtful,happy,andpractical thanthe outgroup. The strategyof compensatingfor theeffects of a low-statusposition by displaying strongingroup favoritism on dimensionsthat are unrelatedtothe statusdifference has been referred to as socialcreativity(Tajfel & Turner, 1986).

A third account— onethat is mostcloselyrelatedtothe concernsof systemjustification theory — hasto dowith perceptionsof the legitimacy and stability of thesocialsystem(e.g., Tajfel & Turner,1986).Specifically,it has beenfound that membersof a low-statusgroupaccepttheir inferiorityon dimensionsrelatedto thestatusdifferencesand displayoutgroupfavoritismunless theyperceivethestatusdifferencesto bebothillegitimateandlikely to change(Turner& Brown, 1978).Thus,whetherlow-statusgroup membersacceptor rejecttheir allegedinferiority is hypothesizedto dependon whethertheyperceive ‘cognitive alternatives”to the social system,which aresaid to be broughton by appraisalsof illegiti-macy and instability. Although thereis no publishedresearchto date linking perceptionsof legitimacy andstability to counterfactualthinking with regardto socialsystems,researchin socialidentity theoryhashighlightedthesevariablesas importantpredictorsof groupidentifi-cation and intergroupbehavior(e.g., Caddick, 1982;Ellemerset al., 1993;Turner & Brown, 1978).

The Influence of Social Identity Theory on theSystemJustification Perspective

Social identity theory is an importantprecursorto thetheoryof systemjustification inat leastthreeways.First,it brings a social-psychologicalperspectiveto bearonintergroup relations.Differences in statusor successbetweengroupsare predictedto affectgroup members’perceptionsof their own group andother groups,andtheseperceptionsare theorizedto affectthefuturecourseof relationsbetweenthe groupsas well astheviability ofthesocialsystem(e.g., Tajfel & Turner, 1986;Wrightetal., 1990). Second,the theory introducesideologicalfactorssuchasperceptionsof the legitimacyandstabilityof thestatussystemas relevantto ingroup and outgroupfavoritism(e.g.,Ellemerset al., 1993;Tajfel & Turner,1986; Turner & Brown, 1978). Third, the notion thatsocialgroupsinvent ideologiesthat justify their competi-tion againstothergroups,whencombinedwith perspec-tives emphasizingthe persuasivepower of dominantgroups’ ideologies,helpsexplain why stereotypesandother ideasjustifying social and material inequalitieseventuallycome to be endorsedevenby membersofsubordinategroups(e.g.,Hogg & Abrams, 1988;Jost&

94 JOST

Banaji, 1994).Nevertheless,therearesomelimitationsof thesocial

identity perspectivethat makeit moreof ajumpingoffpoint than a terminus for the theoryof systemjustifica-tion (cf. Jost& Banaji, 1994).Most important,theorizingaboutwhenlow-statusgroupsacceptthestatusquo andwhen they reject it has tendedto be relativelyundevel-oped in thesocial identity tradition. Turner and Brown(1978)arguedthat “subordinategroupswill seekpositivedistinctivenessfrom dominantgroupsto thedegreethattheir inferiority is not perceivedas inherent,immutableor fully legitimate” (p. 207), but the theory fails tospecify when the system is perceived as ‘inherent,immutable, or fully legitimate” and when it is not.Perceptionsof legitimacy and stability havebeen ad-dressedby social identity researchers,but they haveenteredinto the theoryas independentvariables(e.g.,Caddick, 1982;Ellemerset al., 1993;Turner& Brown,1978). As a result, not much is known from socialidentity theoryaboutthecausesof perceivedlegitimacyor aboutwhy membersof low statusgroupswould everfind the systemto be legitimate,whensucha perceptionclearly conflicts with group-servingmotivations. Theguiding assumptionof social identity theory is thatpeopleare motivatedto favor theirown groupoverothergroups,but thismotivation seemsconspicuouslylackingin any display of outgroup favoritism, evenoutgroupfavoritism underconditionsof legitimacy and stability(Hinkle & Brown, 1990).

Systemjustificationtheory,by contrast,drawson thevast literature on the toleranceof injustice (see Jost,1995)andpositsa motiveto investin andrationalizethestatusquo,and this motive is thoughtto bepresentevenamong membersof disadvantagedgroups, althoughtypically to a lesserdegree.Furthermore,researchonsystemjustification theoryhasbegunto treatlegitimacyand stability as moderatingand dependentvariables,demonstrating,for instance,that theact of stereotypingincreasesperceptionsof themagnitude,legitimacy,andstability of status differences(see lost, Burgess,&Mosso,in press).Thus,systemjustificationtheoryseeksto elaboratefurthersome of thesociostructuralvariablesidentified by social identity theorists.

In fact, Tajfel (1984)seemedto realizesomeof thelimitations of social identity theorywhenhe wrote that:

This disymmetrybetweenthe ‘superior’ and ‘inferior’groupshasbeenrecognizedto someextent in the socialidentity approach to intergrouprelations which hasspecifiedthe different strategiesfor achievingdistinc-tivenessthat can be adoptedby membersof groupswhich differ in status...But this is notenough(p. 700).

Elsewhere,he noted the importanceof justice percep-tions in particularfor an understandingof when groupmemberswill acceptandwhen they will reject the socialsystem:

[ani importantrequirementof researchon socialjusticewould consistof establishingin detail the links betweensocial myths and thegeneralacceptanceof injustice,andresearchwhich would attempt to specify thesocio-psychologicalconditionswhich could be expectedtocontributeto thedissolutionof thesepatternsof accep-tance(1982,p. 164).

Here,Tajfel seemedto be appealingto justiceresearch-ers to determinewhen people will engagein systemjustification andwhenthey will not. Healluded,it seems,to the needfor a theory of false consciousness(Jost,1995). Obviously, theseare top priorities of thesystemjustification approach,the implicationsof which are stillbeing developedand testedin emergingresearchpara-digms.After abriefoverviewof the theory, wedescribean experimentalparadigmthat has beenused to shedfurther light on the dynamicsof ingroup andoutgroupfavoritism and therole of severalvariablesidentified bysocial identity theory, including group identification,attributerelevance,andperceivedlegitimacy.

THE THEORY OF SYSTEMJUSTIFICATION

Although phenomenasuchas outgroupfavoritism andinternalizationof inferiority may bepuzzling to social,political, and psychologicaltheorists,who assumethatattitudesand behaviorsaredrivenlargely by self-interest,group-interest,or needsfor personalor collective self-esteem,thereis a rich traditionof Marxist and feministscholarshipon the problemof falseconsciousness(seelost, 1995, for a review). This work especiallyempha-sizesthe cognitivedimensionsof oppressionandsystempreservation,building on Marx andEngels’ (1846/1970)observationthat, “The class which has the means ofmaterial production at its disposal,has control at thesametimeoverthe meansof mentalproduction” (p. 64;emphasisadded). This theme is carried on in 20thcenturyMarxism, most especiallyin Gyiirgy Lukdcs’historical analysisof classconsciousnessand AntonioGramsci’s cultural theory of hegemonyand consent.Contemporarysociologistsworking underthe bannerofdominantideologytheory havecontinuedto exploretheextentto which subordinategroupsare persuadedto holdbeliefsthat are at oddswith their objectivesocial inter-ests (Abercrombie,Hill, & Turner, 1990; Kluegel &Smith, 1986).

6. OUTGROUPFAVORITISM AND SYSTEM JUSTIFICATION 95

Systemjustification theory,then,is a theoryof socialcognitionthat takesits impetusfrom the Marxian ideo-logical tradition, with its focus on thejustification ofinequalityandexploitation(Jost& Banaji, 1994).Oneofthe main theoreticalassumptionsof this perspectiveisthat, all otherthingsbeingequal,peopletendto useideasaboutgroupsand individuals tojustify the way thingsare, so thatexistingsocial arrangementsareperceivedasfair and legitimate,perhapsevennaturalandinevita-ble. Thereare at leastsevenestablishedsocial-psycho-logical phenomenathat we drawon and taketo supportthegeneralsystemjustificationperspective.Theymay besummarizedas follows:

I. Membersof groupslow in social standingexhibit“outgroup favoritism” by internalizing unfavorablestereotypesof their own group and subscribingtofavorablestereotypesof successfuloutgroups(Hinkle&Brown,1990;Jost& Banaji, 1994; Sidanius,1993).

2. Peopleform stereotypesasa way of “rationalizing”unequaldivisionsof roles,especiallyin terms ofessen-tialistic biological categories(Eagly & Steffen, 1984;Hoffman & Hurst, 1990;Jackman& Senter,1983).

3. Membersof disadvantagedgroupstendto drawintra-personalandintragroupsocial comparisonsratherthanintergroupcomparisons,as when women judge thelegitimacyof their own incomeagainststandardsof theincomeof otherwomenand of their own incomein thepast (Jost, 1997;Major, 1994).

4. Peopleperceiveexistinginstitutions,procedures,andoutcomesas fair and legitimate,even when therearereasonsto suspectthat they are not (Lerner, 1980;Martin, 1986;Tyler& N’lcGraw, 1986).

5. Peopleexhibit decision-makingbiasesin favor ofwhateveroption is perceivedas the ‘status quo” andavoid choices that are perceived to entail change(Samuelson& Zeckhauser,1988;Silver & Mitchell,1990).

6. Peoplestickdisproportionatelywith pastbehavioralpracticessimply becausethey are familiar or habitualand fail to considerinnovativealternatives(Hackman&Oldham,1980;Silver& Mitchell, 1990).

7. Peopledisplay“outcomebiases’ in their evaluationsof groupsand individuals,so that peopledescribedas‘winners” are selectively perceived as possessingenduringattributesthat areconsistentwith their successand peopledescribedas ‘losers” are seenas alwayshavingpossessedattributesthat areconsistentwith theirfailure (Allison, Mackie, & Messick,1996).

is thenotionthat whatis tendsto beexperiencedaswhatoughtto be; Although someof thesephenomena,mostespeciallystatusquo and outcomebiases,are usuallyexplainedin purely cognitivetermsby socialpsycholo-gists, thereis an ideological tenorto them that adds alayerof political significanceandmotivation to thebasicinformation processingfunctions. According to thepresentview, cognitionis deployedin the serviceofthesocialsystem.

An Experimental Paradigm

On occasion,thephenomenonof outgroupfavoritismhasbeendismissedas somethingof anexperimentalartifactthat doesnot occur in real-worldgroups.For instance,Mullen et al. (1992) wrote that “a concentrationontransitory, task-specific conceptualizationsof statuswould lead to the misguidedconclusionthat ingroupbiasoccurspredominantlyin higherstatusgroups”(p. 118).One of the goalsof the researchparadigmsummarizedhereis to examineingroupandoutgroupfavoritism byusing an experimentalmanipulationof status that isneithertransitorynor task-specific.Instead,wesoughttodevisean experimentalparadigmin which ingroup andoutgroupfavoritism couldbeinvestigatedin thecontextof real-worldgroup memberships,whereasrelativesocialstatus could be manipulated experimentallyso thatdifferencesdueto social statusof the ingroup could beattributedsolely to variationsin statusandnot to otherfactorsassociatedwith particularreal-world groups.

Additionally, themanipulationof statuswasdefinedin terms of socioeconomicsuccess.This wasdone toimprove on the ecological validity of previousexperi-mentalmanipulationsof statusin termsof performancefeedbackon testsof creativity (Sachdev& Bourhis,1987, 1991),reasoningskills (Turner& Brown, 1978),or other “transitory, task-specificconceptualizations”(Mullen et al., 1992,p. 118).Becausesystemjustifica-tion theory is especially relevant for understandingcognitive responsesto wealth and poverty (e.g., lost,1995;Lane, 1962;Major, 1994), statusof the ingroupwasoperationalizedas relativesocioeconomicsuccess.

Thebasicprocedurefor varyingperceivedsocioeco-nomic successis as follows. Shortly afterarriving for anexperimentbilled as “The Inter-CollegiateStudy ofAbstractThought,”universitystudentsare told that theyare aboutto participatein a researchprojectaimedatunderstanding“why differencesin social andeconomicsuccessexistbetweengraduatesof differentcollegesanduniversities.”Half of the participantsare presentedwithstatisticsindicating that alumni membersof their ownuniversitygroup aresignificantly lesssuccessfulin terms

Whatthesedistinctbodiesof evidencehave in common

96 JOST

of socioeconomicachievementthan are membersof arival outgroup,and theotherhalf areled to believethatthe ingroup is moresuccessfulthan theoutgroup.Thesestatisticsinclude information concerningaveragefinan-cial income,careeradvancementandpromotions,statusof professionsentered,ratesof admissiontograduateandprofessionalschools,andyearsof postgraduateeducationcompleted.The materialsused for a study involvingUniversity of Marylandstudents(low-successcondition)are presentedin Table 6.2 (see also Jost & Burgess,2000,for more information).

Thisprocedurehasbeenusedsuccessfullyin a seriesof experimentsinvolving studentsat YaleUniversity, theUniversity of Maryland,andthe Universityof Californiaat SantaBarbara(U.C.S.B.),with comparisonoutgroupsof StanfordUniversity, the University of Virginia, andthe Universityof Californiaat Los Angeles(U.C.L.A.),respectively.Thesethreeexperimentsare describedinabbreviatedform in this chapter;two of them arepre-sentedmoresystematicallyelsewhere(see Jost, 1996;Jost & Burgess,2000). Manipulationchecks indicatethat, in all studiesconducted thus far, the statisticsconcerningsocioeconomicsuccessdifferences(ascribedto sourcessuchas U.S. NewsandWorldReportandtheChronicle of Higher Education) were acceptedascredibleandconvincing.Thesestudiesalsodemonstratethat relativesocioeconomicsuccesshasa major impacton thestereotypesandevaluationsthatpeoplehaveaboutingroup and outgroup members.As a general rule,randomassignmentto the high-successcondition leadspeopleto displayingroupfavoritism — that is, to expressbeliefsthat their own group is superioron a numberofstereotypicalcharacteristics,whereasassignmentto thelow-successconditionleadspeopleto displayoutgroupfavoritism — that is, to expressbeliefs that the moresuccessfuloutgroupis superior.

Thispatternhasbeenobservedon qualitative,open-endedmeasuresof attribution in which respondentsarenot constrainedby the expectanciesor questionsof theresearchers.In theYalestudy,following the experimen-tal inductionof relativesuccess,participantsweregiventhe following instructions:“Think aboutthedifferencesin social and economic successbetween Yale andStanfordalumnilae.Canyou think of anyexplanationsorjustificationsfor why Yale and Stanfordgraduateswouldhavedifferentratesof socioeconomicsuccess?”Partici-pantswere thenaskedto spend2 to 3 minuteslisting upto five responsesin clearly numberedspaces.Twoindependentjudgescodedthe open-endedresponsesasfocusing either on the ingroup (Yale) or the outgroup(Stanford) and as expressing something favorable,unfavorable,or neutralaboutthat ingroupor outgroup.

TABLE 6.2: Sample materialsfor manipulatingperceivedsocioeconomicsuccess.

Virginia MarylandAlumni Alumni

MeanFinancialIncomeafter5 yearsafter 10 yearsafter20 yearsat retirement

CareerAdvancementMeannumberof pro-motionsafter 5 yearsMeannumberof pro-motionsafter 10 yearsMeannumberof pro-motionsafter20 yearsNumberof CEOsofmajorcorporations

$38,500$53,200$69,700$78,300

2.4

5.3

9.5

41

$24,700$39,500$54,100$62,500

1.3

3.0

6.2

17

PosteraduateEducationMeanyearsof post-graduateeducation% of applicantsadmit-ted to medicalschool% of applicantsadmit-ted to law school% of applicantsadmit-ted to businessschool% of applicantsadmit-ted to graduateschool% receivingpost-bac-calaureatedegrees

1.3

44%

43%

57%

60%

49%

0.4

21%

19%

30%

41%

23%

Sources: U.S. News& World Report,1993;TheChronicleof HigherEducation,1994

When Yale studentswere assignedto the high-successcondition, explanationsmaking referencetocharacteristicsof theingrouptendedto beveryfavorable(e.g., “Maybe Yaliesaresmarter”; “Yale admitsstudentswith better records who are innately more driven”).According to the independentjudges,81.5%of ingroup-related statementswere favorablein content,and only2.4%of thesewere unfavorable.Whenmembersof high-successgroupsgeneratedexplanationspertainingto theoutgroup,42.1%of thesewerejudgedto beunfavorable(e.g., “BecauseStanfordis a sportscholarship-grantingschool, they are going to get athletesthat are not asintelligent as the studentswho get in regularly”; “Stan-ford studentsare somehowsuperficial.They simply glide

6. OUTGROUPFAVORITISM AND SYSTEM JUSTIFICATION 97

on the surfacewithout seekingthe deepreasons”).Only15.8% of theexplanationshaving to do with featuresoftheoutgroupwere favorable.

WhenYale studentswereassignedto a position oflow socioeconomicsuccess,the results wereverydiffer-ent.Undertheseconditions,only 12.3%of theexplana-tions involving theingroup werejudgedto befavorablein nature,whereas42.5%wereunfavorable(e.g., “Yaleis full of your bookworms and your dorks”; “Yalestudentsare too idealistic,andusuallyhaveimpracticalor falseimaginationsaboutrealworld life”). Ofexplana-tions involving the outgroup, 62.2% were judged asfavorable(e.g., “Stanfordoffers a bettereducationthanYale”; “Stanford is a more selectiveschool, so it hassmarterpeople”),andonly 4.2%wereunfavorable.Thus,membersof low-successgroupsdisplayoutgroupfavorit-ism in making open-endedattributionsfor thesocioeco-nomic successdifferences.

As with stereotypes,evaluations,resourcealloca-tions,andothertypesof socialjudgments,the researchliteratureon intergrouprelationshasstressedthe ethno-centric natureof attributionsaboutingroup andoutgroupmembers(Cooper & Fazio, 1986; Hewstone, 1990;Pettigrew,1979).Thenotion that peoplegenerategroup-servingattributionsfor outcomesis also highly consistentwith social identity theory. Our researchsuggests,however,thatlow-statusgroup membersdo notattributetheir inferiorposition to situationalfactorsor extenuatingcircumstances,hut ratherseemto internalizetheinequal-ity in the form of internalattributionsabouttheunfavor-able characteristicsof the ingroup and the favorablecharacteristicsof theoutgroup.

According to a systemjustificationanalysis,mem-bersof groupsthat arelow in social or materialstandingshouldexhibit ingroup derogationandoutgroupfavorit-ism to the extentthattheyperceivetheoverarchingsocialsystemto be fair, legitimate,andjustifiable. Thus, it ishypothesizedthat perceived legitimacy is negativelyrelated to ingroup favoritism amongmembersof low-statusgroups,insofaras peoplewho acceptideologicaljustifications for a statusquo that placesthem at adisadvantageshould be more likely to consentto theirown inferiority. However, perceived legitimacy ishypothesizedto relatepositively to ingroup favoritismamong high-statusgroups, insofar as they gain confi-denceandesteemfrom the sensethat their advantageislegitimized;their senseof superiority is increasedby theperceptionthat the systemis fair, legitimate, andjusti-fied.

This interactionhypothesisis slightly differentfromthat which hasbeenpredictedby social identity theorists.Turner andBrown (1978)hypothesizedthat “Ilgiroups

with illegitimate status relations would display moreingroup bias than thosewith legitimatestatusrelations~~(p. 210), regardlessof the statusof the ingroup.Theirreasoningwas that perceivedillegitimacyshould renderthe systemof statusdifferencesunstableand insecure,leadingboth groupsto vie for a position of superiority(see also Caddick, 1982). This implies a main effecthypothesissuchthat ingroupfavoritismshouldbegreateramongpeoplewho perceivethestatusdifferencesto below in legitimacythan amongpeoplewho perceivethemto be high in legitimacy.Thus,although social identityand systemjustificationtheoriesmakethe samepredic-tion with regard to the behaviorof low-status groupmembers,the two perspectivesdiffer whenit comestopredictionsabouthigh-statusgroup members.

In an experimentinvolving studentsat theUniversityof Maryland, theproceduredescribedearlierwas usedtomanipulateperceivedsocioeconomicsuccess.Followingthis induction,participantswereaskedhow fair or unfair,how justifiable or unjustifiable,and how legitimate orillegitimate thesocioeconomicsuccessdifferenceswerebetweenthe ingroup (University of Maryland) and theoutgroup(Universityof Virginia), andtheir responsestothesethreeitemswere averagedto createa generalindexof perceivedlegitimacy. Ingroupandoutgroupratingsonstatusrelevant(intelligent, hardworking,andskilled atverbalreasoning)andstatusirrelevant(friendly, honest,and interesting)attributesweresolicited.

As illustrated in Fig. 6.1, perceived legitimacyincreasedingroup favoritism amongmembersof high-successgroups,butit decreasedingroup favoritism(andincreasedoutgroupfavoritism)amongmembersof low-successgroups. This interaction patternwas also ob-servedin thpYalestudy (Jost, 1996),suggestingthat thefocus on legitimation in intergroup relations is wellplaced. Perceivedlegitimacy seemsto have oppositeeffectson high-statusand low-=’tatusgroup members,assystemjustification theorypredicts,andnot as Turnerand Brown (1978)suggested.

In addition, Jost and Burgess(2000) found thatMaryland studentsassignedto the position of lowsocioeconomicsuccessshowed significantly greaterattitudinalambivalencedirectedat their own group thandid studentsassignedto the high-successcondition.Thiswas explained in terms of a psychologicalconflictbetweenopposingtendenciestowardgroupjustficationandsystemjustification — a conflict that facesmembersof low-statusbutnothigh-statusgroups(seealso Justetal., 2001). We also reasonedthat membersof a psy-chologicallymeaningfulgroup suchas this (for whom atleast moderatelevels of group justification motiveswould be present),ambivalencetowardthe ingroup

98 JOST

Fig. 6.1. Meanson ingroupfavoritism(stereotyping)by ingroupsuccessandperceivedlegitimacy

1.00

IngroupFavoritism

0.50

0.00

OutgroupFavoritism

-0.50

-1.00 Low High

PerceivedLegitimacy

would be increasedas levelsof systemjustificationwereincreasedfor membersof the low-status group, butambivalencewould be decreasedas levels of systemjustification were increasedfor peopleassignedto thehigh-statuscondition.And this is whatwefound.Percep-tions of the legitimacy of the statusdifferenceswereassociatedwith ik.creasedambivalenceamonglow-statusgroupmembersanddecreasedambivalenceamonghigh-status group members (Jost & Burgess, 2000).

Onepotential limitation of theMarylandstudy is thatperceivedlegitimacy wasmeasuredratherthan manipu-lated, and so there was no random assignment to condi-tions of legitimacy or illegitimacy. A follow-up studyinvolving studentsatU.C.S.B.did employ an experimen-tal manipulationof perceivedlegitimacy.After learningthat their own groupwas less socioeconomicallysuccess-ful thanthe comparisonoutgroupof U.C.L.A. students,participantswereexposedto a pairof persuasiveessaysthat wereallegedlywritten by membersof the ingroup(as part of a cover story concerning abstract verbalreasoning); these essays were in actuality designed toalterperceptionsof the legitimacy of thesocioeconomicdifferences.In thehigh-legitimacycondition,for exam-ple, oneof theessaysread:

Therearetwo good reasonswhy UCSB studentsarelesseconomically successfulthan UCLA students: (I)UCLA admits studentswith more varied and more

experiencedbackgrounds,and these peoplehave abettersenseof what theywant todo laterin life; and (2)thereis a perceptionout there(and it’s probablyright!)that UCSB studentsare partyers who do not takeacademicsseriously enough.Both of these reasonswouldeasilyexplain thedisparities.With regardto thefirst, everyoneknowsthatpeoplewhoareambitiousaz~dknowledgeableare in a better position to succeedeconomically, and they deservethat success.Withregard to the secondreason,potential employersareprobably sensitive to legitimate differences in thequalificationsof studentsat the two schools.

The correspondingessayfor the low-legitimacycondi-tion was asfollows:

Therearetwo main reasonswhy UCSB studentsare lesseconomicallysuccessfulthanUCLA students:(1) UCLAadmitsmore studentswith privilegedbackgrounds,andthesepeoplehave more advantagesto begin with andmore connectionslater in life; and (2) thereis a mis-perceptionthat UCSB studentsarejust partyerswho donot takeacademicsseriously.Neitheroneof thesereasonsis fair. With regard to the first, everyoneknows that‘wealthbegetswealth’and that it is far easierfor peopleofhigher social classesto succeedeconomically,whetherthey deservethat successor not. With regard to thesecondreason,potential employersareprobablyrelyingon falseperceptions,without payingenoughattentiontothe meritsof qualified individualsat UCSB.

—S

Success

— + .89 High

Low-.60 Success

6. OUTGROUPFAVORITISM AND SYSTEMJUSTIFICATION 99

Fig. 6.2. Effectsof legitimacymanipulationon ingroupandoutgroupfavoritismon thepart of low statusgroupmembers

Although a subsetof participants(12%) werenotpersuadedby theessays,thosewho werepersuadedthatthe systemwaseitherlegitimateor illegitimate showedchangesin their stereotypes(see Fig. 6.2). Comparedwith studentswho were assignedto the low- legitimaccondition, thosewho were assignedto the high-legiti-macy conditionexhibitedstrongeroutgroupfavoritismon status-relevantattributes(intelligent, hard working,skilled at verbalreasoning)andlesseringroupfavoritismon status-irrelevant attributes (honest, friendly, interest-ing). In addition to clarifying the importantrole ofperceived legitimacy in determining the degree ofingroup or outgroupfavoritism exhibitedby low-statusgroup members,the experimentalparadigmwe devel-oped alsoprovidesuswith someinsightconcerningtheissuesof disidentificationandattributerelevance.

Does Disidentification Account for OutgroupFavoritism in Low-status Groups?

Accordingto oneprominentaccountderived from socialidentity theory, membersof low-statusgroupsexhibitoutgroup favoritism to the extentthat they disidentifywith their own group. However, it may not always befeasible for peopleto avoid perceiving themselvesin

termsof ascribedgroup membershipsand to persistinthinking of themselvesaspart of a group to which theydo not belong. Research indicates, in fact, that ingroupidentification tends to be stronger among members ofsome low-status groups (e.g., African Americans, His-panic Americans) than among members of high-statusgroups(e.g.,EuropeanAmericans),insofaras the formergroup membershipsare more numerically distinctive(rarer) than the latter (MeGuire& McGuire, 1988).Anevenbiggerchallengeto the disidentificationthesisis thefact that correlations between ingroup identification andingroupfavoritismare weakand inconsistent(Hinkle &Brown, 1990).A studyby Mlicki and Ellemers(1996),forexample,finds that Polishcitizensidentify especiallystronglywith their own national image,but the imagethat they hold is a predominantlynegativeone,presum-ably becauseof recentfailuresof theireconomicsystem.Therefore, it is doubtful that decreased ingroup identifi-cation amongmembersof low-statusgroups,evenif itcanbedemonstrated,would be sufficientto producetheextentof outgroupfavoritism that hasbeenobserved.

Although ingroup identificationhasnotbeena focalpoint of the researchprogramreportedon here, it wasincludedasa measuredvariablein theYale studyand asa manipulatedvariable in theU.C.S.B.study. In the Yalestudy, no evidencewas obtainedfor the disidentification

100 JOST

Fig. 6.3. Stereotypic evaluations of high status and low status target groups by an observer group

Favorability

of Ratings

hypothesis that members of low-success groups wouldidentify less with the ingroup than would membersofhigh-success groups. Ingroup identification was in factnonsignificantlyhigheramongmembersof low-successgroups than among members of high-successgroups,possibly because cognitive dissonance is aroused bybelonging to a group that is low in social standing, andthis dissonance may be reducedby redoubling one’sinvestment in the group (cf. Turner, Hogg, Oakes, &Smith, 1984). In the U.C.S.B. study, a boguspipelineprocedure was used to convince members of low-statusgroups they were either especially high or low in ingroupidentification (cf. Doosje,Ellemers,& Spears,1995).Nosignificant main effect of ingroup identification oningroup favoritism was obtained, although this shouldhave been expected on the basisof social identity theory(Tajfel & Turner, 1986).

Taken as a whole, these findings deepen existingworries that social identity theory insufficiently explainsstatus differences on ingroup favoritism (Hinkle &Brown, 1990; Jost & Banaji, 1994; Sidanius, 1993).None of this is to say that disidentificationprocessesplayno role in the phenomenonof outgroupfavoritismor thatissues of groupjustification are unrelated to intergroupevaluations. What our results suggest, however, is thatdisidentification with the ingroupdoesnotseemto bea

StatusIntlevant 7IThits StatusRelevantTmits(honest,friendly. (intelligent,hard-interestilw) working,skilled)

Target Group

I UC Santa Cruz

llJStanfoid

necessaryprerequisitefor thesort of system justifyingoutgroupfavoritismobservedamong membersof low-statusgroups.

Does the Moderating Role of Attribute Rele-vanceSupport Social Identity Theory?

Studies employing our experimental paradigm formanipulatingperceivedsocioeconomicsuccessreplicatethe finding that attributerelevancemoderatesthe displayof ingroup and outgroup favoritism on the part of low-status group members (e.g., Mullen et al., 1992;Mummendey & Schreiber, 1984; Skevington, 1981;Spears& Manstead,1989;van Knippenberg,1978).Inthe Maryland study (inst & Burgess, 2000), for example,membersof low-statusgroupsexhibitedstrongoutgroupfavoritism on status-relevantattributesof intelligence,industriousness,and verbal reasoningability, but theyexhibited strong ingroup favoritism on status-irrelevantattributesof honesty,friendliness,and interestingness,afinding that was replicated in the U.C.S.B. study. Mem-bersof high-statusgroups,by contrast,exhibited ingroupfavoritism on relevantandirrelevantattributes.

Although we havereplicatedthepatternof resultsobtainedby social identity theoristsfor relevantversus

6. OUTGROUP FAVORITISM AND SYSTEM JUSTIFICATION 101

irrelevant attributes,anotherone of our studiescastsdoubt on the theoretical interpretation that has beenoffered repeatedlyfor this pattern. U.C.S.B. studentswereaskedto ratetwo outgroups— Stanfordand U.C.-SantaCruz— on exactlythe samestereotypingmeasuresused in all of the prior studies. Although U.C.S.B.studentsdid not belong to either of the groupsbeingrated, the stereotypepatternthey showed was almostidentical to other studies in which the ingroup wasimplicated in the comparison:The higher statusgroup(Stanford) was rated as more intelligent, more hardworking, andmoreskilledat verbalreasoning,whereasthe lower statusgroup (U.C.-SantaCruz) wasratedasmorehonest,friendly, andinteresting(seeFig. 6.3). Infact, themeandifferencescorrespondedalmostexactlyto meanlevels of ingroupandoutgroupfavoritism in ourother studies.This suggeststhat the tendencyamongmembers of low-status groups to favor the ingroup onstatus-irrelevantdimensionsof comparisonmay notactuallybedriven by group-justifyingneedsto compen-satefor threatenedsocial identification,as hasalwaysbeen assumed. The possibility that people subscribegenerally to lay or folk theories in which a negativecorrelation exists between social or economic status andfavorable socioemotional characteristics (such as honestyand friendliness) is being addressed in ongoing research.In fact, it may be that such beliefs servesystem-justifyingends and that high- and low-status group members mightall feel that the system is more legitimate if they cansustain stereotypes of the ‘poor but honest” or “poor buthappy variety.

FUTURE DIRECTIONS IN THE THEORYOF SYSTEM JUSTIFICATION

Onequestionthat has notbeenaddressedsufficiently inthe researchreportedhereinis “why do people engage insystemjustification?” Our responseis that it is probablyoverdetermined.There are many forces, internal andexternalto the individual, thatpull for systemjustifica-tion sortsof attitudesandbehaviors,and theseshouldbeinvestigatedin future research.

Thereare cognitivefactors,suchas genuineattemptsto explainandunderstandtheworld, as well as tenden-cies to preserveexisting attitudesand beliefs and toachievecertaintyand closure.Thus,we havearguedthatpolitical conservatism and other system-justifyingattitudesserveto reduceuncertaintyandsatisfy the ‘needfor cognitive closure” (e.g.,Jost,Kruglanski,& Simon,1999). In addition, there are motivations to staveoffexistentialterrorby preservingthe sensethat theworld isa fair andmanageableplacein which people“get what

they deserveand deservewhat they get” (e.g.,Lerner,1980).Situationalfactorsalso determinethestrengthofsystem-justifying responses.In recent researchwithYephatKivetz, we have found that the presenceof anideological threatdirected against the national systemincreasesstereotypicdifferentiationbetweenhigh-statusAhskenaziJewsandlow-statusSepharadicJewsin Israel.In general, too, people face structural constraints, such asrewards for obeying authority and tolerating unequaloutcomes, and punishments for challengingthe legiti-macyof thesystem.

Therefore, we argue that one major (and over-determined) function of attitudes, stereotypes, and socialjudgments is to justify existing social arrangements.System justification refers to a set of social, cognitive,and motivational tendenciesto preservethe statusquothrough ideological means, especiallywhenit involvesrationalizing inequality among social groups. Ourresearch (in collaboration with Grazia Guermandi,Monica Rubini, and Cristina Mosso) has begun toexplore the ways in which people use stereotypes tojustify socioeconomic differences between northernersand southerners in the United States, England, and Italy.

There are other issues, too, that future researchwould do well to address. For one thing, it would beimportant to know the extent to which system justifica-tion and outgroup favoritism are truly internalized on thepart of low-status group members, as opposed to beingstrategic, self-presentations displays that do not reflectprivately held beliefs and attitudes. Jost and Banaji(1994) argued that outgroup favoritism might be evenstronger at an unconscious or unexamined level ofawarenessinsofarassuch impulseswould beless subjectto controlledprocessingand to consciousactivation ofego and group justification motives. Our researchhasbegun to explore these issues, using reaction timeparadigms and other unobtrusive methodologiestoestimate the extent of outgroup favoritism on non-consciouscognitive, affective,andbehavioralmeasures(Jost,Pelham,& Carvallo,2000).

Finally, the futureof system-justificationtheorywillhaveto accommodateexceptionsto the rule by explain-ing when and why peoplefail to provide ideologicalsupportfor theexistingsocialsystem.In otherwords,thetheory should be useful also for identifying opponentprocessesthat govern group consciousness-raisingandinspire social andorganizationalchange.Someof thesearelikely to be associatedwith processesof egojustifica-tion andgroupjustification, which are hypothesizedtostand in an inverse relation to systemjustificationprocessesfor membersof low-statusgroups(Jostet al.,2001).Individual differencevariablesmight alsoidentify

102 JOST

peoplewho areespeciallylikely or unlikely to engageinsystem-justifying outgroup favoritism. Candidatesinclude the ‘belief in ajustworld” scaleand the ‘socialdominanceorientation”scale(seeJost& Burgess,2000;Jost& Thompson.2000),as well aspolitical orientation(Jost,Burgess,& Mosso,in press).In collaborationwithGrazia Guermandiand Erik Thompson,we havealsobeendevelopinga scaleof ‘economicsystemjustifica-tion” (seeJost& Thompson,2000).Evidencepresentedhere— thatperceptionsof illegitimacyare associatedwiththe rejectionof outgroupfavoritism — is alsoa step inthe right direction, but more researchis neededtodeterminewhen andwhy peoplewill shedthe layersoffalseconsciousnessandbegin to substantiallychallengethe existenceof social inequality in all of its variousguises.

ACKNOWLEDGMENT

Portionsof this chapterarebasedon writings submittedin partial fulfillment for thePh.D.degreegrantedby theDepartmentof Psychologyat Yale University underthesupervisionof William J. McGuire. Funding for theresearchdescribedhereinwasprovided by a RobertM.Leylan Fellowshipin SocialSciencefrom Yale Univer-sity, a grantfromtheFaculty ResearchAssistantProgramat U.C.-SantaBarbara(both awardedto the author),NIMH Grant#5R01-MH32588awardedto William J.McGuire,andNIMH Grant#ROI-MH52578awardedtoArie W. Kruglanski.I am grateful to JoyceLiu, CristinaMosso, and Oliver Sheldonfor assistancewith manu-script preparationand to Sonya Grier and MichaelMorris for helpful suggestionsconcerningrevisionsofthis chapter.