Anti-immigrant attitudesOccupational skillsRelational skills

ings conrm our argument that workers with a higher possession of interpersonal skill

havese ricrs shcientring i

erce compadvance tactor endo

models: skilled workers will move from rich, skill-abundant countries to poor, labor-abundant countries, while unworkers will move from poor to rich countries, thereby driving unskilled workers in rich countries in vulnerable po(Scheve and Slaughter, 2001; Mayda, 2006; ORourke and Sinnott, 2006; Hainmueller and Hiscox, 2010). Based on this logic,

http://dx.doi.org/10.1016/j.ssresearch.2015.02.0040049-089X/ 2015 Elsevier Inc. All rights reserved.

Corresponding author at: 1126 E. 59th Street, Chicago, IL 60637, United States.E-mail addresses: naeyun@uchicago.edu (N. Lee), chslee@uchicago.edu (Cheol-Sung Lee).

Social Science Research 52 (2015) 270289

Contents lists available at ScienceDirect

Social Science Research

journal homepage: www.elsevier .com/locate /ssresearch(Borjas et al., 1997). In their perspective, the basis of anti-immigrant sentiment originates primarily from between low-skilled natives and immigrant workers. The recent theoretical developments and applicationsof labor-market competition theory by incorporating trade theories such as the Heckscher-Ohlin family of fetitionhis linewmentskilledsitionsSemyonov et al., 2002), perceived threats on group interests (Blumer, 1958; Bobo and Hutchings, 1996) or cross-nationalvariations in structural conditions such as population composition and unemployment (Quillian, 1995; Semyonov et al.,2006).

Meanwhile, on the other hand, scholars of labor market have applied their utilitarian ways of thinking to interpreting thegrowing anti-immigrant sentiments in Western societies, by emphasizing the role of skill-based labor-market competition1. Introduction

Growing numbers of immigrantscountries. Native populations in thecountries. While some native workeunderstand these responses, social sto immigrants, mostly by underscoassets relative to instrumental skill assets are exposed to less intense competitions withimmigrants, and are therefore less likely to express anti-immigrant sentiments. Our nd-ings suggest that occupational-level relational skill assets based on sociocultural differ-ences play an important role in shaping native workers attitudes toward immigrants.

2015 Elsevier Inc. All rights reserved.

transformed socio-economic and demographic structures in advanced industrialh countries have developed varied responses to the newcomers from developingow favorable attitudes toward migrants, others exhibit hostility. In an attempt toists, on one hand, have sought to identify the determinants of natives responsesndividuals social and demographic aspects (Espenshade and Hempstead, 1996;Relational skill assets and anti-immigrant sentiments

Naeyun Lee, Cheol-Sung Lee Department of Sociology, University of Chicago, United States

a r t i c l e i n f o

Article history:Received 24 January 2013Revised 11 January 2015Accepted 1 February 2015Available online 16 February 2015

Keywords:Anti-immigrant sentiment

a b s t r a c t

This study introduces the role of relational skill assets in accounting for attitudes towardimmigrants: relational skill assets. Drawing upon stratication researchers notion ofnon-cognitive skills, we build a theoretical framework highlighting the role of occupa-tional skill requirements in explaining anti-immigrant sentiment. Then, utilizing twooccupation-specic measures, interpersonal skill requirement and instrumental skillrequirement, we construct an explanatory factor, relational skill specicity. We test itseffect on anti-immigrant attitudes as well as on the concentration of foreign-born workersin occupations, using the 2004 national identity module of General Social Survey. The nd-

unskilled native workers in rich countries will be more discontented with the inux of unskilled immigrants, as it increasescompetition in the labor market already exacerbated by the rise of the service economy.

such measures as nurturant skill (Kilbourne et al., 1994). In recent years, scholars have started to use the Occupational

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 271Information Network (O NET) data (see Liu and Grusky, 2013; Kunovich, 2013a, 2013b). The previous studies on immigrantslabor market experience have also paid little attention to the role of relational skills. The conventional approaches havemostly attributed migrants earnings decit to the lack of human capital and the concentration of social capital within ethniccommunities (Reitz, 2001; Aguilera and Massey, 2003; Waldinger and Lichter, 2003; Li, 2004, 2008; Cranford, 2005; Ness,2005; Nakhaie, 2007).

Departing from these preexisting views, our study argues that relational skills play an important role in shapinganti-immigrant sentiments through two mechanisms. First, relational skills will limit the supply of not only domesticlabor-market competitors, but also immigrants or foreign workers who aspire to enter into or move up in the labor market.Subsequently, exposure to immigrants in a respective occupational labor market will affect the job qualities, wage levels, andthe risk of unemployment for native workers (at least their perceptions of them), eventually shaping their attitudes towardimmigrants. Second, relational skills required for occupational tasks tend to be correlated with personality traits (Heckmanand Kautz, 2012) that bring in openness to (new) experiences, tendency to act in a cooperative unselsh manner, and(sympathetic) tender-mindedness all of which may foster more favorable understanding of immigrants situations andtheir contributions to a host society. The following sections identify causal mechanisms between occupational skill assetsand native workers attitudes toward immigrants.

1 The following sections on interpersonal and instrumental skills further elaborate on the denition and the concept of occupational skills regarding peopleand tools dimensions.We nd that there are unexplored dimensions for explanatory variables in these studies of attitudes toward immigration,and introduce a new independent variable: the role of occupation-specic relational skills. Conventional approaches in thisarea mostly treat skills as a general construct of human capital (Becker, 1964). Students of attitudes toward immigrants paytheir primary attention to labor market competitions based on skill levels, structural conditions such as the size of out-grouppopulations and unemployment at regional level (Quillian, 1995; Kunovich, 2004; Semyonov et al., 2006). Scholars of immi-grant labor, on the other hand, have concentrated on the effect of human and social capital on immigrant earnings.

This study argues that two components of relational skills, interpersonal and instrumental skills, play a critical role inshaping anti-immigrant sentiments. We will propose two causal mechanisms: (1) rst, the notion of relational skill speci-city will capture the degree of transferability of skill assets across different cultures and societies, which will eventuallydetermine workers occupational-specic exposure to competitions with immigrants; (2) second, it will also represent per-sonality traits and facets (Heckman and Kautz, 2012) that involve intellectual and emotional openness, trust, and unselsh-ness, arising from pre- and post-labor market training. We assume that these personality traits, correlated withoccupational skill requirements, will also determine workers sentiments toward immigrants. In the subsequent sections,we propose three sets of hypotheses and causal explanations that link occupation-specic relational skills withanti-immigrant sentiments, and then test them with a General Social Survey module (National Identity, 2004).

2. Theoretical discussions

This study highlights the signicance of occupational-level relational skill standards in accounting for attitudes towardimmigrants. We dene relational skill standards as average skill requirements of an occupation needed for workers to per-form appropriate levels of job-specic tasks in their interactions with people and tools.1 Specically, they refer to occupation-al skill requirements of a workers functional tasks in relation to people such as co-workers, transaction partners, andcustomers, as well as things or objects such as products, devices, and tools.

Regarding people dimension, we note that any occupational tasks cannot be performed without simple or complex inter-actions among people. Workers of an occupation may have to help others or serve the requests by others. They may have toexchange information with others or closely communicate with others in order to produce products or provide services. Forsome occupations, the interpersonal skills constitute the core essence of their job functions. For lawyers, persuasion andnegotiation skills are the key components of different stages of legal procedures, not only for lawsuits in courts, but alsofor informal settlements outside courts (Heinz and Laumann, 1982). For sales representatives, explaining the functions ofproducts to their customers, and instructing subordinates and reporting to their supervisors are essential parts of their jobs.

With respect to things dimension, workers often deal with simple to complex tools and machines. For a carpenter, beingable to handle necessary tools comprises a core essence of his or her job performance. A scientist in a biology lab shouldknow how to operate and manage lab equipment, and how to raise and maintain living organisms on a regular basis.Note that, although manual laborers need to communicate with their colleagues for simple instructions or procedural pro-tocols, communication skills are less essential than their skill with tools or machines. However, a scientist working in alaboratory may be required to communicate with co-workers at very diverse, sophisticated, and complex levels while simul-taneously holding high levels of skills to handle laboratory facilities.

These aspects of skill assets have been largely neglected in stratication research since Cain and Treiman (1981)spioneering exploration of DOT (Dictionary of Occupational Titles) measures and a couple of subsequent investigations using

2.1. Interpersonal skills and anti-immigrant sentiments

We dene interpersonal skills as a type of mental and physical practices that involve different levels and repertoires ofinteractions and communications among human beings, needed to develop temporary or prolonged relationships in diverse

family background (Lareau, 2002) or neighborhood and friendship networks (Coleman, 1988). As an important part of

worlds eventually develop into the skills needed for adequate performance in higher education and the workplace.

272 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289This study assumes that there exist two primary venues by which native workers develop sentiments toward immigrantsthrough interpersonal skills: (1) labor market competition and labor supply; and (2) personality traits or soft skills. First,interpersonal skills may function as skill barriers, discouraging immigrants from entering certain occupations or industriesthat formally or informally require highly sophisticated, culturally shaped skills, thereby shielding native workers fromintense job competition. Interpersonal skills are generally less transferrable across different societies. Thus, immigrants lackof understanding of appropriate local norms, habits, conventions, routines, and practices may inhibit their adequate perfor-mance in particular job tasks that require highly nuanced skills for interpersonal interactions, and limit their chances ofmoving up to higher positions. With the presence of language barriers in certain occupations, language skills will also direct-ly determine immigrants aspirations in local job markets (Portes et al., 1978).2

The second mechanism operates through personality traits and soft skills formed by pre-labor market experiences andon-the-job training. Those with higher levels of interpersonal skills are more likely to possess certain personality traits thatfacilitate a better understanding of other peoples situations, and therefore give a sympathetic support for immigrants evenbefore their entries into occupational worlds primarily through earlier formal and informal educational experiences. Afterentering the workplace, workers with jobs requiring higher levels of interpersonal skills are more likely to be in situationsdealing with other peoples interests and concerns or cooperating with others, both of which require a certain level of trustof others and relatively unknown outsiders. Therefore, those who chose occupations requiring high levels of interpersonalskills will be more likely to develop more favorable views of immigrants.3

To summarize, interpersonal skills may not only serve as signicant skill barriers against the immigrant labor force, butalso capture certain personality traits or soft skills developed before or after entering the occupation. As a result, nativeworkers in jobs that require high levels of interpersonal skills will be insulated from direct competitions againstout-group members including immigrants, and also better able to understand the plight of immigrants. These theoreticalexpectations lead to the following hypotheses.

Hypothesis 1-1. Occupations requiring high levels of interpersonal skills will have fewer immigrant workers.

Hypothesis 1-2. Workers with high levels of occupation-specic interpersonal skills (pj) will show less anti-immigrantsentiments.

2.2. Instrumental skills and anti-immigrant sentiments

Instrumental skill standards capture skill requirements of each occupation in relation to tools, devices, machines, equip-ment, vehicles, and substance or raw materials. In contrast to interpersonal skill standards, instrumental skill standards donot carry social skill components. As it refers to workers ability to prepare, adjust, and control the operation of machines,tools, and materials on their job routines, it is impersonal, performance-oriented, and manual labor-related.

The two venues through which interpersonal skills affect anti-immigrant sentiments may be also applicable to instru-mental skills. First, the impersonal aspect of instrumental skill standards is likely to attract immigrant workers with or with-out high levels of instrumental skills, resulting in increased competition with native-born workers for jobs. Instrumental skill

2 Although it is unclear how much language skills and interpersonal skills shape each other, it is obvious that two skills are highly correlated with each other.We control for the effect of language skills through the variable, cognitive skill complexity in our analyses in Tables 2 and 3.

3 A reviewer pointed out that workers with higher interpersonal skills may be able to better relate to the situation of others and therefore those workersare less likely to develop anti-immigrant sentiments. We nd our rst mechanism (through personality traits) embraces this view.non-cognitive skills (Bowles and Gintis, 1976, 2002; Bowles et al., 2001), interpersonal skills also include work habits,self-condence, self-esteem, and calmness (Farkas, 2003). Interpersonal and communicative skills obtained in informal life-institutionalized work settings. Interpersonal skills range from mere helping or serving to more sophisticated skills, suchas mentoring (highest), negotiating, instructing, and supervising (US Department of Labor, 1991). Therefore, interpersonalskills carry social skill components within them and some scholars call this soft skills as opposed to formal or technicalknowledge (Moss and Tilly, 1996, 253).

Each occupation requires workers to demonstrate the ability to motivate cooperation in other actors (Fligstein, 1997,398) regarding its functional tasks. Employees in an occupation (or occupational eld) should be able to relate to the situa-tion of the other, where others include co-workers, transaction partners (suppliers and buyers), clients, and customers.They should be able to comprehend, and perform their roles and tasks based on their interpersonal relations or on theirorientations to each other or to shared goals (Martin, 2003, 29). A large share of these skills partly originates from or is rein-forced through formal educational training and on-the-job training (Becker, 1964), but also tends to be obtained through

standards can be directly convertible across cultures and societies without any discounts. Therefore, even immigrant work-ers who initially lack these instrumental skills may be willing to invest in them, whereas it is difcult to obtain interpersonalskills over a comparably short time span. In this sense, we expect that native skilled manual workers with higher invest-ments in instrumental skill assets, and concomitantly with higher incomes, will develop the strongest hostility toward immi-grants as they are likely to lose more in the event that they are replaced by immigrants.

Second, those with jobs requiring high levels of instrumental skills may be less likely to possess soft skills or personalitytraits (Heckman and Kautz, 2012; John and Srivastava, 1999) that enable a deeper understanding of others situations. Theirinitial entrance into jobs that require higher levels of instrumental rather than interpersonal skills may imply a lack of per-sonality traits that lead to sympathetic support for others. Furthermore, after entering the occupation, their isolated nature

ers capacity to understand outgroup members but also the degree of exposure to competition with immigrant workers,

from competitions with immigrants by skill barrier mechanisms based on the low convertibility of interpersonal skills. The

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 273rst characteristic leads them to support and favor the presence of more immigrants, but the second characteristic showsthat they are already economically sheltered from intense competitions.

However, workers in occupations requiring a lower possession of interpersonal skills relative to instrumental skillsmay face an opposite situation. They are less likely to possess personality traits that develop empathy for outgroup mem-bers, while the lower level of interpersonal skills requirement will facilitate the movements of workers across borders dueto the lack of social skill barriers. Therefore, incumbent native workers in this type of jobs will have to tolerate the highperceived risk of downward wage pressure and unemployment. We argue that two mechanisms, lack of soft skills orrelated personality traits, and intensied competition, may reinforce each other and end up producingstronger anti-immigrant sentiments among native workers with a higher possession of instrumental skills over interper-sonal skills.

4 We have a measure of the proportion of foreign-born workers/immigrants in each occupation imported from the Current Population Survey March 2000data, which allows us to test the labor market competition/labor supply causal venue, but do not have a direct measure of the pre- and post-labor markettraining mechanisms through personality traits or soft skills. In actual analyses, therefore, we will introduce our relational skill measures along with theproportion of immigrants in each occupation variable, and then surmise that the remaining signicant effects of relation skill measures will reect thepersonality traits aspects.

5 We develop this relational skill specicity as a summary measure of two dimensions of relational skills, and aim to introduce it in a regression equation asa single measure (as introducing both interpersonal and instrumental skills causes a collinearity problem due to a relatively high correlation between them).

6 The formula was directly borrowed from Iversen and Soskice (2001) skill asset specicity, s/(g + s), in which s denotes rm-specic skills, while grepresents general skills that are transferable across rms. For discussion on occupation or industry-specic skills (beyond rm-specic skills), see Mares(2003), Neal (1995, 2000), and Acemoglu and Pischke (1999)s works.which depends on the transferability of skill assets across different cultures and societies. Therefore, native workers in occu-pations requiring a higher possession of interpersonal skills in their total skill sets may be in dualistic situations. On onehand, they may be inclined to embrace immigrants based on their soft, communicative skills, while they are also protectedof work life may deprive their chance of developing these personality traits. Thus, we expect that these high-instrumentalskill-workers will be more likely to develop negative feelings toward members of outgroups.

In summary, native workers holding a high level of instrumental skill may not only have difculty understanding of theplight of immigrants and others, but also confront ercer competitions from immigrant workers with a similar level of skillsbut lower wage demands. Their lack of soft skills or personality traits as well as perceived or experienced vulnerability tocompetition with skilled immigrants in terms of high instrumental skills, ensuing downward pressures on their wages, andeventually higher risk of unemployment, lead them to develop stronger anti-immigrant sentiments.4 Based on the precedingdiscussion, we predict as follows:

Hypothesis 2-1. Occupations requiring high levels of instrumental skills will have more immigrant workers.

Hypothesis 2-2. Workers with high levels of occupation-specic instrumental skills (tj) will show more anti-immigrantsentiments.

2.3. Relational skill specicity and anti-immigrant sentiment: two dilemmas

Two types of aforementioned skill sets, interpersonal skills necessary for human interaction on job tasks and instrumentalskills required for dealing with things and objects on job routines and practices, constitute a notion of occupation-specicrelational skill specicity.5 The relative composition of the two types of skills captures the degree to which a worker possessesinterpersonal skill assets relative to the total skill assets (the sum of interpersonal and instrumental skill assets). In our frame-work, the relational skill specicity, pj/(tj + pj),6 reects two aspects. First, the higher the proportion of pj is, the less culturallyconvertible the entire skill set of a particular occupation will be. Therefore, native workers in occupations with higher interper-sonal skill requirements will be more insulated from competition with immigrants. Second, the higher the proportion of pj is,the greater the chance that a member of a particular occupation will have personality traits and soft skills that enable a betterunderstanding others situation and their contribution to the society.

In other words, this study assumes that this notion of relational skill specicity is likely to capture not only native work-

Hypothesis 3-1. Occupations with a higher possession of interpersonal skills over the entire skill assets have a lower level ofinux of immigrants.

Hypothesis 3-2. Workers employed in jobs requiring a greater possession of interpersonal skills over the entire skill assetswill show lower levels of anti-immigrant sentiments.

Fig. 1 summarizes the causal ows between relational skill specicity and anti-immigrant sentiment. Relational skillspecicity affects natives attitudes toward immigrants directly (A), but also indirectly through the percentage offoreign-born in an occupation (A-1 and A-2). The goal of this paper is to test these two channels: direct and indirect path-ways of relational skills on anti-immigrant attitudes. In the direct causal path (A), we assume that a high level of relationalskill specicity in ones occupation will lead to a lower level of anti-immigrant sentiments. For instance, those with highlevels of interpersonal skills may be better able to understand other peoples situation or those who succeed in building

274 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289Y4)/4. In addition, we also tested the simple sum of four raw scores, which achieves a fairly high level of reliability (Cronbachs a reliability coefcient = 0.78)and produces the same results.10 Detailed denitions and explanations of each stage are in Appendix B.interpersonal skills may intrinsically have higher levels of sympathy or empathy of others. In the indirect causal ows,low or high relational skill specicity for an occupation will affect the percentage of foreign-born workers in that occupation(A-1). The extent of exposure to foreign-born workers in ones occupation will induce incumbent native workers to develophigh or low anti-immigrant sentiment (A-2).

3. Data and measures

3.1. Anti-immigrant sentiment

This study tests the proposed hypotheses, using a module of the General Social Survey (2004) on national identity andcitizenship.7 Using four questions on respondents attitudes toward immigrations, this study constructs a composite, weightedaverage index of anti-immigrant sentiments based on the factor loadings identied by conrmatory factor analysis (CFA). Thequestion used to construct the composite index is: there are different opinions about immigrants from other countries living inAmerica (By immigrants we mean people who come to settle in America). How much do you agree or disagree with each ofthe following statements? The question subsequently display the following four statements: (Y1) Immigrants are generallygood for the economy; (Y2) Immigrants take jobs away from people who were born in America; (Y3) Immigrants improveAmerican society by bringing in new ideas and cultures; and (Y4) Immigrants increase crime rates.8 The response categoriesfollow the Likert scale: (1) strongly agree; (2) agree; (3) neither agree nor disagree; (4) disagree; (5) strongly disagree. The fourquestions deal with respondents sentiments of immigrants regarding the general economy, employment/unemployment, cog-nitive/cultural diversity, and everyday security, respectively.9 In constructing the composite index, response categories arerescaled to assign higher scores to stronger sentiments against immigrants.

3.2. Interpersonal and instrumental skill standards

In order to measure relational aspects of skill requirements for each occupation, we matched the occupation variable inthe GSS 2004 module with the DOT (Dictionary of Occupational Titles) (U.S. Department of Labor, Employment, and TrainingAdministration, 1991) classication system of occupational categories. The former based on the ISCO 88 (the InternationalStandard Classication of Occupations) contains about 400 occupational categories. We link these 400 occupations with theDOTs more detailed occupational groups. Through this matching process, we were able to import seven DOT measures ofoccupational skill requirements. Among them, two DOT measures, a jobs relationships to people and things, are used to cap-ture relational skill assets. First, a jobs functional requirement in relation to people (pj), interpersonal skill requirement, isbased on the following hierarchical ordering of interactive, communicative relationships among people: takinginstructions-helping (1), serving (2), speaking-signaling (3), persuading (4), diverting (5), supervising (6), instructing (7),negotiating (8), and mentoring (9). Second, a jobs skill requirements for dealing with things (tj) are based on the followingordering: handling (1), feeding-offbearing (2), tending (3), manipulating (4), driving-operating (5), operating-controlling (6),precision working (7), and setting up (8).10 Both measures are reverse-ordered from the original scale to denote higher skill

7 Currently, the GSS 2004 data is the most recent dataset in the U.S. with detailed information on native workers attitudes toward immigrants and theiroccupations.

8 Most studies using GSS (or ISSP, International Social Survey Programme) use another question, Do you think the number of immigrants to [country]nowadays should be. . . 1. Increased a lot; 2. Increased a little; 3. Remain the same; 4. Reduced a little; 5. Reduced a lot. I nd this question does not necessarilymeasure how respondents feel about immigrants in different areas of socio-economic matters. The question captures the degree to which respondents considerthe current and future numerical immigration level appropriate for the society. It is completely possible to answer remain the same for this question, whilebeing strongly positive or negative about the presence of immigrants.

9 The four questions load on one factor in exploratory factor analysis. Conrmatory factor analysis using STATA 12 gives the following factor loadingestimates: 1.000 for general economy (Y1); 1.034 for employment/unemployment (Y2); 1.048 for ideas/cultures (Y3); 1.073 for crime (Y4). Overall model tmeasures are satisfactory, when judged based on Bollen (1989)s recommendations, with the exception of RMSEA (p-value < 0.000; IFI = .976; NFI = .975;TLI = .929; RMSEA = 0.113). The nal composite score was calculated as: Anti-immigrant Sentiment Composite Score = (1.000Y1 + 1.002Y2 + .914Y3 + .842-

Relaonal Skill Specicity in Occupaon

An-Immigrant Senment

Causal Flow A

Beer able to relate to others situaons(due to so skills or personality traits)

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 275percentage of foreign-born in ones occupational category as a dependent variable and a control. First, we test whether highor low levels of relational skill specicity in occupations have an effect on the actual percentage of foreign-born workers inthese occupations. Secondly, we use the percentage of foreign-born workers in each occupational category to assess the indi-The measurement scheme of this relational skill specicity is inspired by Iversen and Soskices work and their Varieties ofCapitalism school (VoC hereafter) approach (Estevez-Abe et al., 2001; Hall and Soskice, 2001; Iversen and Soskice, 2001).They have recently highlighted the role of workers skill asset portability/specicity across rms in explaining demandsfor social protection. The VoC literature claims that workers who have invested in rm-specic skills relative to general skillsare more likely to support the public provision of social welfare, because they are exposed to greater risk of unemploymentand income loss due to their low transportability of skill assets across rms (Iversen and Soskice, 2001). Based on this logic,we expand the notion of skill specicity across rms to skill specicity across national borders and cultures. Relational skillspecicity is measured as the ratio, pj/(tj + pj), in which a respondents jobs interpersonal skill requirement divided by herjobs total relational skill requirements (the sum of instrumental skill requirement and interpersonal skill requirement). Itmeasures the necessary amount of interpersonal skills per one unit of total relational skills to demonstrate an adequateexecution of tasks specic to an occupation.

3.4. Percentage of foreign-born in respondents occupational category13

In order to ascertain the relationship between relational skill specicity and anti-immigrant sentiments, we introduce therequirements for higher values.11 These 1 to 9 and 1 to 8 ordered categories are directly used as scores for interpersonal skillasset and instrumental skill asset.12

3.3. Relational skill specicity

Then we constructed a measure of relational skill specicity with these two skill requirements at the occupational level.

A-1

Percentage of Foreign-Born in Occupaon

A-2

Fig. 1. Causal mechanism: the effect of relational skill specicity on anti-immigrant sentiment.rect effect of relational skill specicity on attitudes toward immigrants.Previous research by Kunovich (2013a, 2013b) also employs the percent of Mexicans and immigrants in occupations to

investigate the impact of labor market competition on attitudes toward immigrants and immigration policies by usingONET and GSS 2004 data. His ndings show that the effect of percent of Mexican in occupation on anti-immigrant atti-tudes is largely spurious, and is mostly due to the fact that occupations with high percentage of Mexicans tend to have work-ers with less education who are also more likely to perceive threat.

Our analysis differs from these previous studies in two important ways. First, our main focus is on examining both directand indirect effects of occupational skills (e.g. relational skill specicity) on anti-immigrant sentiments. Thus, we use thevariable percentage of foreign-born in ones occupation to test the indirect effect of relational skill specicity, i.e. exposureto competition with foreign-born workers within occupations. Second, we use regression (OLS) models with HuberWhite

11 Another dimension of relational skill requirements contained in the DOT is skill requirement in relation to data. This measure, based on factor analysis,turns out to be more related to cognitive skill dimension (Kilbourne et al., 1994). Therefore, it is included as a component of cognitive skill complexity.12 Obviously, we assume that ordinal scales may be treated as interval. Previous studies report that this treatment does not involve too much distortion(Labovitz, 1970; Kim, 1975).13 A reviewer has suggested that the effect of interpersonal skill requirements on anti-immigrant attitudes [may be] purely a function of their lack ofexposure to immigrant worker threat and that it is imperative to control for the percentage of foreign-born workers in respondents occupational category inorder to assess the true effect of ones level of relational skills on anti-immigrant sentiments. Therefore, we calculated the percentage of foreign-born in eachoccupational category (ISCO 88) from the Current Population Survey March 2000 data and included it in our analysis.

robust standard errors instead of HLM models. As Kunovich (2013a) has found that individual-level variables explain muchof the variance in attitudes toward immigrants, we focus on estimating the effects of individual-level variables.

In this paper, we assess the detailed skill requirements on occupations through the DOT measures. We calculate the per-centage of foreign-born in each occupational category (ISCO 88) from the Current Population Survey (CPS) March 2000 dataand merge with the General Social Survey (2004) data.14 Thus, for each job category (ISCO 88) in the General Social Survey

276 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 2702892004 module, we provide the percentage of foreign-born workers in that specic occupation.

3.5. Other control variables

3.5.1. Formal educationFormal education captures the individual-level human capital component (Becker, 1964), and has been the most sig-

nicant factor in explaining attitudes toward out-groups in previous studies (Quillian, 1995; Mayda, 2006; Hainmuellerand Hiscox, 2010). Highly educated individuals are more tolerant of the presence of ethnic or racial minorities, includingimmigrants, in the neighborhood and national economies. Formal education is measured as the highest year of school com-pleted for each respondent.

3.5.2. Cognitive skill complexityIn line with previous stratication research, this study includes occupation-specic cognitive skill complexity (Kilbourne

et al., 1994; Weeden, 2002) as one of the control variables. Following the previous studies (Kilbourne et al., 1994; Farkaset al., 1997), we use ve DOT measures as components of this cognitive skill complexity: three components of general educa-tional development in reasoning, math, and language (R, M, and L, respectively in Appendix A.2); specic vocational trainingrequirements (S); and skill requirement in relation to data (D).15 With the inclusion of the language-skill requirement in thecomposite measure, interpersonal skill standards are expected to capture only the effects of non-cognitive social skill elementson anti-immigrant sentiments.

3.5.3. Religious afliationsIn our attempt to control for the effects of religion on anti-immigrant sentiments, we consider not only denomination but

also cultural conicts between orthodox/conservative Protestants and mainline/liberal Protestants. We assume that culturaldifferences originating from respondents religious faiths (Hunter, 1994; Wuthnow, 1996) may shape their attitudes towardimmigrants in different ways.

On one hand, conservative Protestants, who tend to have a strong belief in the literal inerrancy of the Bible and an uncom-promising moral view based on it, may consider the greater presence of immigrants with different cultural and religiousbackgrounds as serious threats to their (uncompromising) religious doctrines and communities. They are more likely to pre-sume that immigrants might disrupt their traditional fabrics of religious and cultural communities, not only by increasingcrime and job loss, but also by bringing different cultural and religious ideas and practices along with them. Therefore, con-servative Protestants low tolerance for different religious and cultural practices is more likely to lead to greateranti-immigrant sentiment. On the other hand, liberal Protestants may think that moral views and religious doctrines canbe compromised and updated along with changing community norms (Hunter, 1991), which may lead to higher level ofopenness to immigrants presence and their values. Overall, we assume that, as both liberal Protestants and Catholics arecomposed of very heterogeneous groups, their overall differences in opinions toward immigrants from non-religious popula-tions may not be noticeable. Three indicator (0 or 1) variables for religious afliation, conservative Protestants, mainlineProtestants, and Catholics, are introduced with the reference category being no or other religions.

3.5.4. RaceWe expect that racial differences may exist in attitudes towards immigrants due to the racialization of immigration since

the 1960s. Three categories of individual racial identication, non-Hispanic whites, non-Hispanic blacks, and others areintroduced. We use three indicator (0 or 1) variables to compare group differences in attitudes toward immigrants withLatinos, the reference category.

3.5.5. Political afliationMany social scientists have recently highlighted the effect of political afliation on anti-immigrant sentiments especially

in advanced industrial countries (Semyonov et al., 2006), as (extremist) right-wing parties often resort to stirring up xeno-phobic fear of foreigners to consolidate their electoral base. In this study, we control for political party identity by introduc-ing an indicator variable (0 or 1) for Republican Party membership. Although some segments of Republicans may show

14 We deleted one occupational category (e.g. postmasters) in the General Social Survey (2004) data as this did not exist in the Current Population SurveyMarch 2000 data. Thus, the total number of occupational categories (ISCO 88) in the General Social Survey (2004) data changed from 161 to 160.15 We also conducted conrmatory factor analyses of ve measures and identied appropriate factor loadings for these sub-indicators; 1.000 for reasoning,1.063 for math, 1.112 for language, 1.438 for relation to data, 1.670 for specic vocational training (SVP). Fit statistics are as follows: p-value < 0.000; IFI = .987;NFI = .986; TLI = .973; RMSEA = 0.133).

stronger feelings against immigrants, others may not. Especially controlling for fundamentalist Protestants, we do not expectthat republican membership will have statistically meaningful positive effects on anti-immigrant sentiments.

3.5.6. Other individual-level demographic variablesSex is introduced as a component of the baseline model with an indicator (0 for men and 1 for women) variable, with the

expectation that men will show greater hostility toward out-group populations. With the expectation that older people areexpected to show more prejudice about out-group populations, age variable is also added as one of the controls. A group ofdummy variables are also introduced to control for individuals labor-market status: part-time employed, unemployed,

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 277retired and non-employed (students, housewives, and disabled) are controlled for, with full-time employed being the refer-ence category. Weaker status in the labor market is expected to be associated with stronger anti-immigrant sentiments. Wecontrol for union membership to see if workers memberships with unions lead them to show stronger anti-immigrant sen-timents, by using an indicator (0 or 1) variable for being a member of a union.

Finally, we control for respondents citizenship status. Respondents with citizenship are expected to show strongeranti-immigrant sentiments, compared to those without citizenship, mostly, immigrants.16 Obviously, there will be anunknown number of citizens who have recent immigration histories but also obtained citizenships after decades of residentialhistories. We assume that this segment of the population is likely to have similar attitudes toward the recent immigrants similarto those of native-born citizens, after having experienced a signicant process of assimilation.17

4. Method

We employ a linear regression (OLS) model using HuberWhite heteroskedasticity-consistent standard errors (or robuststandard errors) for 160 occupational-level clusters, adjusting for within-cluster correlations. In this model, we assume thatrespondents may be clustered into occupations and that observations may be correlated within each occupation, but wouldbe independent between occupational categories (Huber, 1967; White, 1980). In other words, the model allows off-diagonalelements (in the error term) from the same cluster to be non-zero, while still maintaining the assumption that there is nocorrelation among observations across clusters (Primo et al., 2007; 451). One may employ hierarchical linear modeling(HLM; Bryk and Raudenbush, 1992; Raudenbush and Bryk, 2002) to take into account this kind of error structure, but wechoose to employ the HuberWhite estimator for the following two reasons: rst, we are not interested in decomposing (-explained) variances into each level to know how much each level (of covariates) explains, which is the main concern of theHLM models. What we want to know in this study is simply the point estimation of our main covariates and their statisticalsignicance, with an appropriately specied error structure; second, the HuberWhite estimator is computationally lessintensive, as it does not require distributional assumptions for every level (Primo et al., 2007), which may impose heavydemands on theory and data (Steenbergen and Jones, 2002, 234).18

5. Results

5.1. Descriptive results

Table 1 presents the descriptive statistics of the main variables including measures of relational and cognitive skill assetsfor eight major occupational categories. The mean value of anti-immigrant sentiments is the highest for skilled machineoperators (2.99) and skilled craftsmen (2.93), while the lowest for professionals (2.35) and managers (2.53). Note that theelementary unskilled, those who are at the lowest (cognitive) skill level, do not show the highest level of anti-immigrantsentiment (2.80). The skilled craftsmen and machine operators, representative of manual working class occupations, haveone of the lowest interpersonal skill standards (1.76 and 2.01), and the highest instrumental skill standard (6.56 and6.23), thereby showing the lowest relational skill specicity (.21 and .25). These descriptive statistics aggregated by largeoccupational categories suggest that there may be a meaningfully strong negative association between relational skill speci-city and anti-immigrant sentiment, which may also coincide with variations along occupational lines.

5.2. Multivariate results

Table 2 displays the unstandardized coefcients from linear regression models testing the effects of relational skill speci-city, interpersonal skills, and instrumental skills on the percentage of foreign-born in occupations. Our ndings show that

16 As citizens with one or two immigrant parents (the second generation immigrants who hold citizenships) are more likely to support immigrants presence,we alternatively tested both parents citizenship status instead of respondents citizenship status, expecting that individuals whose parents are both citizens,will show stronger anti-immigrant sentiments, when compared to those with at least one parent without citizenship status. The results were largely similar tothe ones using simple citizenship. The results are available upon request.17 We tested whether the exclusion of this group with recent immigration histories from the sample affects the estimates of the relational skill measures. Theresults for this smaller sample (N = 929) are largely identical with the ones reported in the text. The results are available upon request.18 Actually, some of the models using cross-levels and interactions did not converge under the MLE (maximum likelihood estimation). Nevertheless, we alsotested some basic multi-level models treating the occupation-level as a higher level (but not specifying varying slopes across units), which is equivalent to ourcurrent models. The results were almost identical. Please see Appendix C.2 for the results using HLM models with 47 occupational groups.

Table 1Descriptive statistics of key variables, by major occupational categories (ISCO 88): General Social Survey, 2004.

Occupational Categories N Anti-immigrant Cognitive skill Interpersonal Instrumental Relational skill

278 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289sentiments complexity skill standard skill standard specicity

Mean Std dev. Mean Std dev. Mean Std dev. Mean Std dev. Mean Std dev.

Ofcials/Managers 176 2.51 0.64 7.05 0.32 8.00 0.00 1.00 0.00 0.89 0.00Professionals 223 2.35 0.72 7.47 0.61 5.35 2.46 3.17 2.71 0.65 0.25Technicians 126 2.63 0.81 6.43 0.58 3.51 1.51 2.96 2.67 0.60 0.24Clerks 146 2.76 0.64 4.29 0.82 2.16 0.99 5.62 1.32 0.28 0.11Service/Sales Workers 129 2.78 0.78 3.60 1.20 2.47 0.91 2.55 2.06 0.54 0.19Skilled Craftsmen & Agricultural and Fishery 100 2.94 0.75 5.48 0.85 1.78 0.98 6.58 1.20 0.21 0.09Machine Operators 81 2.98 0.84 4.18 1.16 2.01 1.01 6.25 1.44 0.25 0.12Elementary Unskilled 69 2.81 0.78 1.75 0.32 1.46 0.83 1.00 0.00 0.56 0.10

Table 2Regression results of the percentage of foreign-born in occupation against relational skill specicity, interpersonal skills, and instrumental skills in occupation.

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Relational skill specicity in occupation 6.719**(0.927)

7.264**(0.899)

6.656**(0.923)

** **Hypothesis 1-1, 2-1, and 3-1 hold across different model specications. Occupations with higher levels of relational skillspecicity and interpersonal skill requirements are more likely to have a lower percentage of foreign-born workers in theoccupation, controlling for cognitive skill complexity, average years of schooling, average log income, average percentageof union membership, and average level of unemployment in each occupation.19 In contrast, occupations with higher instru-mental skill standards are more likely to have a higher concentration of foreign-born workers in the occupation, ceteris paribus.Thus, interpersonal skill standards may function as skill barriers limiting the entry of immigrant workers to certainoccupations.

Table 3 presents unstandardized coefcients from linear regression models with HuberWhite standard errors testing theeffects of measures of relational skills on anti-immigrant sentiments. Model 1 reveals that interpersonal skill asset has anexpected, negative effect on anti-immigrant sentiment, in the presence of other individual-level covariates (Hypothesis1-2). Interpersonal skills may facilitate a greater understanding of others through personality traits and also operate as askill barrier, insulating native workers from competition with immigrants. Model 2 shows that instrumental skill assetshas a strong positive effect on anti-immigrant sentiment (Hypothesis 2-2).We predicted that workers with higher levelsof instrumental skill may not only lack soft skills or relevant personality traits, but also feel more vulnerable to the riskof unemployment and income loss, leading to strong anti-immigrant sentiments.

In Model 3, we test the effect of our main variable, relational skill specicity, on anti-immigrant sentiment, adjusting forall individual-level baseline controls. As we predicted, the coefcient is negative and highly signicant. The result also lends

Interpersonal skill in occupation 0.779(0.111)

0.700(0.115)

Instrumental skill in occupation 0.707**

(0.089)0.653**

(0.090)Cognitive skill complexity in occupation 1.198**

(0.179)1.404**(0.157)

1.161**(0.171)

1.082**(0.173)

0.875**(0.181)

1.719**(0.156)

1.416**

Avg. years of education in occupation 0.455*(0.194)

0.411*(0.184)

0.468*(0.185)

0.493**(0.180)

Avg. log income in occupation 0.351(0.505)

0.036(0.478)

0.233(0.479)

0.156(0.476)

Avg. union membership in occupation 1.790(1.693)

2.197(1.688)

1.644(1.680)

2.578(1.706)

2.005(1.699)

1.561(1.686)

1.015(1.675)

Avg.% of unemployed in occupation 2.526(2.495)

2.877(2.496)

2.489(2.494)

3.198(2.513)

2.776(2.510)

1.946(2.505)

1.560(2.499)

Constant 25.54** 24.53** 28.21** 23.87** 26.59** 21.11** 24.92**

(4.321) (4.308) (1.971) (4.367) (2.056) (4.392) (2.087)N 1076 1076 1076 1076 1076 1076 1076R2 0.212 0.208 0.212 0.197 0.201 0.207 0.212

Robust standard errors in parentheses.+ Signicant at 10%.

* Signicant at 5%.** Signicant at 1%.

19 Average years of schooling in occupation and average log income in occupation are highly correlated (r = .64), and, therefore are not included in the samemodel except for Model 1.

Table 3Regression results of anti-immigrant sentiments against relational skill specicity (using HuberWhite robust standard errors).

Model 1 Model 2 Model 3 Model 4

Sex (Female = 1) 0.048 0.048 0.046 0.041(0.043) (0.043) (0.043) (0.042)

Age 0.004* 0.004* 0.004* 0.004*(0.002) (0.002) (0.002) (0.002)

Education 0.083** 0.089** 0.083** 0.080**(0.008) (0.008) (0.008) (0.008)

Part-time Employed 0.009 0.013 0.003 0.004

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 279(0.066) (0.065) (0.066) (0.067)Unemployed 0.065 0.065 0.066 0.064

(0.095) (0.094) (0.093) (0.093)Retired 0.020 0.006 0.013 0.013

(0.075) (0.076) (0.076) (0.078)Non-employed 0.057 0.074 0.069 0.065

(0.071) (0.071) (0.071) (0.071)credibility to the central argument (Hypothesis 3-2) of this study that a greater possession of interpersonal skills relative toinstrumental skills will lead to lower anti-immigrant sentiments. Model 4 tests whether the signicant effect of relationalskill specicity remains robust against the percentage of foreign-born in respondents occupation and cognitive skill com-plexity.20 Relational skill specicity is again highly signicant at the 1% level (b = .288). With the introduction of additionalvariables, the effect of relational skill specicity has decreased moderately from Model 3 (b = .316), which implies that a part

Fundamental Protestant 0.158** 0.150* 0.152** 0.154**

(0.058) (0.059) (0.058) (0.058)Mainline Protestant 0.137* 0.131* 0.135* 0.135*

(0.065) (0.064) (0.064) (0.064)Catholic 0.084 0.084 0.083 0.082

(0.067) (0.068) (0.068) (0.067)Non-Hispanic White 0.512** 0.517** 0.512** 0.512**

(0.085) (0.085) (0.084) (0.084)Non-Hispanic Black 0.501** 0.509** 0.500** 0.494**

(0.093) (0.094) (0.092) (0.093)Other (Non-Hispanic) 0.004 0.013 0.007 0.003

(0.112) (0.107) (0.109) (0.109)Citizenship 0.463** 0.486** 0.484** 0.487**

(0.122) (0.117) (0.119) (0.118)Republican 0.022 0.034 0.033 0.035

(0.048) (0.047) (0.047) (0.046)Union Membership 0.054 0.070 0.059 0.059

(0.083) (0.082) (0.082) (0.082)Interpersonal Skill 0.024**

(0.007)Instrumental Skill 0.029**

(0.008)Relational Skill Specicity 0.316** 0.288**

(0.077) (0.088)Cognitive Skill Complexity 0.006

(0.014)% Foreign-born in Rs Occupation 0.228

(0.298)

Constant 3.104** 2.975** 3.159** 3.100**

(0.141) (0.153) (0.138) (0.146)R2 0.182 0.187 0.189 0.190

N = 1055; Robust standard errors in parentheses.+ Signicant at 10%.

* Signicant at 5%.** Signicant at 1%.

20 In Appendix C.1, we separately test the effect of exposure to foreign-born workers in 160 occupations on anti-immigrant sentiments. In Model 1, withoutrelational skill specicity, the percentage of foreign-born in ones occupation has a signicant positive effect on anti-immigrant attitudes at the 10% level. Thispositive effect disappears once relational skill specicity is introduced in Model 2. In Models 34, we combined adjacent occupations into ISCO 88s 28 largeroccupational categories. In Model 3, the percentage of foreign-born in ones broader occupational category has a signicant, positive effect on anti-immigrantfeelings at the 5% level, but this effect becomes non-signicant in Model 4 after adding relational skill specicity. Appendix C.2 shows the result from HLMmodels using 47 occupational groups. To collapse 160 occupations into 47 larger categories, we split the 28 larger occupational categories (especially thosewith more than 50 incumbents) and combined occupations in the similar category (according to ISCO88 categorization). In Model 1, the percentage offoreign-born workers in occupations has a signicant positive effect on anti-immigrant sentiment at the 5% level. However, this signicant effect disappearsonce relational skill specicity is introduced.

of relational skill specicity effects may have been mediated through exposure to foreign-born workers and cognitive skill com-plexity in ones occupation (but the effect is non-signicant).21 The nding shows that the effect of relational skill specicityoutweighs that of cognitive skill assets and the percentage of foreign-born in ones occupation, both of which do not have sig-nicant effects on anti-immigrant feelings. Based on these results, we conclude that personality traits mechanism prevails overlabor market competition mechanism.22

Model 3 and 4 show the effects of the baseline control variables on anti-immigrant sentiment. Respondents sex and

280 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289labor-market status variables are not associated with anti-immigrant sentiment. Conservative Protestants are more likelyto have negative feelings towards immigrants, compared to other religious adherents, which supports the argument thatthey will be less tolerant of out-groups different cultural and religious identities. Surprisingly, contrary to ourprediction, mainline Protestants are also more likely to exhibit stronger sentiments against immigrants, compared tothose with no or other religious afliations. The strong anti-immigrant sentiments held by mainline Protestants mayimply the division between the laity and the progressive clergy (Wuthnow, 1996). While the leadership of mainlineProtestants may be progressive and pro-immigrant, the followers may not exactly share the same views towardsimmigrants.

Non-Hispanic white respondents hold more negative views towards immigrants, compared to Latino respondents.Interestingly, non-Hispanic black respondents also show anti-immigrant sentiments as strong as white respondents. Thisresult suggests that black respondents consider themselves as incumbent majority-native insiders threatened by immi-grants, rather than a part of minority groups, especially regarding the immigration issue. African-Americans strongeranti-immigrant sentiment may be also attributable to their higher (perceived) group-level threats that originate from theirintense competition with immigrants in low-wage job markets. In addition, African-Americans may have a lower proportionof (recent) immigrants than Latinos, the omitted category, which also means that their friends and relatives are less likely tobe immigrants. Therefore, non-Hispanic blacks lower chances of contact (Pettigrew, 1998) with immigrants inside theirown group are likely to lead to stronger anti-immigrant sentiments.

As expected, the effect of education has a highly signicant, negative effect on anti-immigrant sentiment. This variable isthe single most dominant predictor of attitudes toward immigrants. The result simply signies that the more educated anindividual is, the less threatened or less prejudiced one is likely to be. The more educated are supportive of the presence ofimmigrants, because they have had the chance to absorb more valuable sources of information regarding immigrants posi-tive roles in industry and society. Respondents who are legal citizens are likely to have stronger anti-immigrant sentiments,compared to those without citizenship, as they may perceive themselves more as insiders against immigrants. Contrary toour expectations, older people are less likely to exhibit negative attitudes toward immigrants, net of economic self-interest.Being a Republican or a union member does not have a statistically signicant effect on anti-immigrant sentiment. Thenon-signicant effect of Republican Party membership is not surprising as mixed attitudes toward immigrants often coexistwithin the same political party. Unions stances on immigration also vary by industries, and thus it may be difcult to nd aconsistent relationship between union membership and sentiments toward immigrants.

The result overall offers sufcient support to our argument. Workers possessing greater interpersonal skills relative toinstrumental skills are more likely to possess soft skills or personality traits that enhance their capacity to relate to thesituation of others, and also less likely to suffer from erce competitions with immigrants (thanks to relatively high inter-personal skill barriers). Therefore, workers with higher relational skill specicity are less likely to express stronganti-immigrant sentiments. However, the fact that relational skill specicity is still highly signicant, after controlling forthe percentage of foreign-born in ones occupation (which has a non-signicant effect), implies that the direct effect of rela-tional skill specicity outweighs the indirect effect. In short, we conclude that the direct pathway through personality traitsprevails over the indirect pathway through the exposure to competition with immigrants in occupations.

Table 4 shows the results from ordered logit models for four separate dependent variables. In order to nd out whetherresults differ by each question, we analyzed the four outcome variables separately using random intercept proportional oddsmodels. There are no signicant differences in results between using the composite-index and using the individual out-comes. Relational skill specicity has a signicant, negative effect on anti-immigrant sentiments across three models(a = .01 level), except for the question, immigrants are good for economy, where it is only marginally signicant at the10% level. We also nd that, contrary to our prediction, fundamentalist Protestants focus more on the economic aspect ofimmigration while mainline Protestants express concern on the cultural aspect of immigration. Conservative Protestantsare more likely to think that immigrants are not good for the economy (a = .01 level), and that immigrants take jobs awayfrom natives (a = .05 level), compared to those with other religious beliefs. On the other hand, they are only marginally morelikely to view that immigrants improve society by bringing new ideas (a = .10 level). Mainline Protestants, in contrast, areless likely to think that immigrants improve society by bringing in new ideas (a = .01 level), and marginally more likelyto think that immigrants increase crime (a = .10 level).

21 Following Reviewers suggestion, we ran structural equation models (SEM) and Sobel tests. Both results show that the indirect effect of relational skillspecicity through the percentage of foreign-born in occupations is not signicant.22 As we do not have concrete measures of personality traits other than interpersonal and instrumental skills, we do not know exactly how much variationsare attributable to one or the other mechanism between personality traits and labor market competition. Personality traits effects may have absorbed labormarket competition effects, or there may, indeed, be no labor market competition effect. At this point, the SEM result lends credence to the latter scenario.

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 281Table 4Regression results from multi-level ordered logit models of four separate areas of anti-immigrant sentiments against relational skill assets (using HuberWhiterobust standard errors).

ImmigrantsIncrease Crimea

Immigrants Goodfor Economy

Immigrants TakeJobs Awaya

Immigrants ImproveSociety/New Ideas

Sex (Female = 1) 0.294* 0.136 0.131 0.037(0.125) (0.112) (0.119) (0.108)

Age 0.002 0.020** 0.008* 0.006(0.005) (0.005) (0.0042) (0.005)

Education 0.141** 0.156** 0.182** 0.198**(0.024) (0.024) (0.025) (0.023)

Part-time Employed 0.033 0.065 0.203 0.108(0.151) (0.173) (0.168) (0.155)

Unemployed 0.084 0.340 0.326 0.044(0.229) (0.225) (0.246) (0.234)

Retired 0.234 0.103 0.017 0.301(0.214) (0.197) (0.226) (0.206)Education has a negative effect on all four outcome variable, so that those who are more educated are less likely to holdnegative views on immigrants. Compared to Latino respondents, non-Hispanic Whites have higher degrees of negative atti-tudes toward immigrants in all four aspects, while non-Hispanic Blacks are more negative in all areas except for increasingcrime. Citizens are more likely to have unfavorable views towards immigrants regarding all three aspects (immigrants aregood for economy, immigrants take jobs away, and immigrants improve society), but not for increasing crime.Republicans are more likely to disagree that immigrants improve society culturally (a = .05 level). Females are less likelyto believe that immigrants increase crime rates (a = .05 level), while older people, contrary to our prediction, are less likelyto think that immigrants take away jobs and are not good for the economy.

6. Conclusion

This study attempted to highlight the role of relational dimensions of occupational skills in explaining anti-immigrantsentiment. We initially constructed two dimensions of relational skill assets: interpersonal and instrumental skill assetsat occupational level. We conceptualized the former as less culturally convertible and the latter as more portable across soci-eties. I inspired by the recent VoC schools emphasis on the role of skill asset portability/specicity (Hall and Soskice, 2001;

Non-employed 0.098 0.063 0.270 0.121(0.162) (0.172) (0.210) (0.178)

Fundamentalist Protestant 0.150 0.397** 0.341* 0.290+(0.141) (0.154) (0.147) (0.165)

Mainline Protestant 0.332+ 0.179 0.202 0.466**

(0.176) (0.176) (0.169) (0.178)Catholic 0.231 0.235 0.007 0.279

(0.164) (0.184) (0.171) (0.185)Non-Hispanic White 0.514* 1.209** 1.245** 0.989**

(0.236) (0.237) (0.229) (0.258)Non-Hispanic Black 0.303 1.213** 1.362** 1.151**

(0.273) (0.268) (0.262) (0.311)Other (non-Hispanic) 0.326 0.019 0.352 0.10

(0.344) (0.354) (0.381) (0.435)Citizenship 0.105 1.571** 1.424** 1.028*

(0.359) (0.387) (0.322) (0.478)Republican 0.136 0.029 0.085 0.299*

(0.127) (0.143) (0.122) (0.126)Union Membership 0.180 0.145 0.066 0.059

(0.240) (0.189) (0.228) (0.205)Relational Skill 0.703** 0.404+ 0.642** 0.746**Specicity (0.223) (0.212) (0.184) (0.229)

Cut1 Constant 4.488** 3.322** 3.686** 3.611**(0.613) (0.547) (0.435) (0.606)

Cut2 Constant 2.107** 0.606 1.400** 0.917(0.591) (0.481) (0.421) (0.579)

Cut3 Constant 0.763 0.869+ 0.346 0.446(0.587) (0.486) (0.431) (0.584)

Cut4 Constant 0.843 2.855** 1.496** 2.441**

(0.574) (0.514) (0.435) (0.602)

Log pseudo likelihood 1461.76 1365.44 1476.91 1324.66

N = 1055; Robust standard errors in parentheses.+ Signicant at 10%.

* Signicant at 5%.** Signicant at 1%.a Response categories are rescaled to assign higher scores to stronger sentiments against immigrants.

Iversen and Soskice, 2001), we developed a new variable, relational skill specicity, utilizing the above two measures. Ourcentral claim is that higher relational skill specicity (e.g. higher the ratio of interpersonal skills relative to the sum of instru-

282 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289mental and interpersonal skill assets) will be associated with less job market competition with immigrants and greater capa-city to relate to the situation of others, and, thus, produce less negative feelings toward immigrants.

In building the causal explanation between relational skill specicity and anti-immigrant sentiment, we primarily focusedon the long-termdevelopmental processes of non-cognitive skills (Bowles and Gintis, 1976) and their inconvertibility acrossdifferent societies. This paper has proposed two causal paths: (1) labor market competition and labor supply; and (2) person-ality traits or soft skills formulated before and through occupational training. First, relational skill specicity may indirectlyshapenativeworkers attitudes toward immigrants bydetermining the exposure to labormarket competitionwith immigrants(e.g. the percentage of foreign-born workers in occupations). Our hypothesis is that relational skill specicity will capture thedegree of convertibility of skills across different cultures and societies,whichwill eventually affect the intensity of competitionin the occupation-specic labor markets, leading to positive or negative feelings toward immigrants. Second, relational skillspecicitymayalsodirectly affect anti-immigrant sentiments. Relational skills for occupational tasks tend tobecorrelatedwithpersonality traits or soft skills (HeckmanandKautz, 2012; Johnand Srivastava, 1999), such as openness, cooperativeness, andtender-mindedness, that enable a favorable understanding of immigrants situations and their contribution to the hostsociety.

Overall, the ndings of this study impressively support our argument that relational skills play an important role inaccounting for citizens attitudes toward immigrants as well as the composition of foreign-born workers in occupations.Coefcients of relational skill specicity remained highly signicant and robust against different specications, whilecontrolling for predominantly strong factors such as individual-level education, citizenship status, cognitive skill variablesas well as demographic, religious and political variables. Relational skills specicity has a direct, negative effect onanti-immigrant sentiments, which persists after controlling for the extent of exposure to foreign-born workers in onesoccupations. In summary, we conclude that the direct causal pathway of relational skill specicity (throughpersonality traits) overweighs the indirect pathway (through labor market competition and labor supply).

These ndings advance our understanding of occupation-based social stratication processes, ethnic competition in thelabor market, and attitudinal studies of race and minorities. Although previous scholarship has underscored the impor-tance of cognitive skill assets in migration processes and attitudes toward immigration and out-group populations, theimportance of relational skills has not received much attention. We nd that the ndings of this study may provide a nov-el pathway of causal explanation of how native individual workers develop their anti-immigrant sentiments, along theline of group threat theory (Blalock, 1967; Blumer, 1958; Bobo, 1983; Quillian, 1995).

Group threat theory posits that each native individual worker should have a priorimembership with a dominant group (amajority ethnic/racial group) and be exposed to direct competitions with immigrants in their local residential settings orlabor markets. Our study has shown how perceived threats are formulated from a combinational process of individualsnon-cognitive skills (Bowles and Gintis, 1976) and economic vulnerability (Bonacich, 1972; Borjas and Tienda, 1987), oper-ating through occupational skill closures, as well as the capacity to understand others situation. Therefore, our theory andndings extend the applicability of the logic of group-threat theory to occupational boundaries beyond conventionalracial/ethnic boundaries. The notion of relational skill specicity enriches group-threat theory by illuminating how theorigin of prejudice toward out-group populations is not only based on racial/ethnic groupings, but also individuals skillassets formed along occupational lines.

This study has some limitations that need to be addressed. First, the heterogeneity of immigrants and their labor marketexperiences is not fully taken into consideration. Depending on the year of entry, country of origin, and destination, eachimmigrant may not only have a different composition of skill assets, but also face widely varying labor market situationsin the host society. Second, this paper does not fully explicate why the indirect effect of relational skill specicity throughthe proportion of immigrants in occupations is non-signicant. However, it is important to note that having a higher per-centage of foreign-born workers in ones occupation can lead to both positive and negative feelings toward immigrants.While native workers may feel threat due to increased competition, they may also develop solidarity with immigrantco-workers. Other factors, such as the geographic concentration of immigrants, the gap between actual and perceived groupsize, and the absence of measures of direct contact with immigrants (Kunovich, 2013a), further complicate the issue. Futureresearch may employ rich datasets on immigrant populations to investigate the relationship between occupational skills andlabor market outcomes.

Anti-immigrant sentimentmay be partly a combinational product of inadvertently evolved, macro-level occupational seg-mentation, global movements and reconguration of labor forces, and individuals strategic responses to changinglabor-market situations. In the recent debates of immigration reforms, the Obama administration has expressed an interestto open thedoors formore high-skilled immigrants, especially in the tech sectors.Whatwould this entail for thenative citizensattitudes toward immigrants? On one hand, the general public may view immigrants more favorably with the increase in thenumber of high-skilled professionals. On the other hand, professionals with high levels of relational skill specicity may feelthreatened by the inux of high-skilled foreign-born workers, but, at the same time, may exhibit positive attitudes towardimmigrants because they are better able to understand others situations.23

23 Thus, we may predict constant/stable anti-immigrant sentiments among professionals despite an increase in the number of high-skilled immigrants.

An increasing number of immigrants in a society shake the structure of nativity that has been formulated over a longcourse of time. Immigrants will try to nd cracks and rooms to carve out their own occupational niches in receiving soci-eties, despite being constrained by their cognitive and non-cognitive skill assets, while natives also build their own defen-sive mechanisms to maintain the inherited institutions. Those may include educational credentials, occupational skillcomplexity, and more importantly, interpersonal or social skills specic to the cultural sensitivities of the host country.This study suggests that occupation-specic compositions of relational skill assets may be one of those criteria which bothnative and immigrant workers utilize in order to maintain or secure their spaces in the labor market and socialhierarchies.

Acknowledgments

ea

Ap

A.

Latino .09 .28 0 1

Instrumental skill 3.45 2.68 1 8Relational skill asset specicity .54 .28 .11 .90

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 283% Foreign-born in Rs occupation .12 .09 0 .61Citizenship .96 .19 0 1Republican .40 .49 0 1Union membership .10 .31 0 1Cognitive skill complexity 5.53 1.90 1.26 8.50Interpersonal skill 3.92 2.65 1 9Non-Hispanic Others .03 .18 0 1Non-Hispanic WhiteNon-Hispanic Black.75

.13

.43.330011Catholics .24 .43 0 1

Mainline Protestants .14 .35 0 1

Conservative Protestants .26 .44 0 1

Non-employed .13 .34 0 1

Retired .14 .35 0 1

Unemployed .08 .26 0 1

Part time .11 .32 0 1

Education 13.97 2.75 0 20

Age 45.17 15.90 18 89

Sex (Female = 1) .56 .50 0 1

Anti-immigrant Sentiment Composite Index 2.65 .76 0.94 4.70Variables Mean SD Minimum Maximum1. Summary statisticspendix Arlier versions of this paper.Many thanks to Edward Laumann, Kazuo Yamaguchi and John Levi Martin for their insightful comments and critiques.We would also like to thank the anonymous reviewers of Social Science Research who have provided helpful comments on

A.2. Correlation coefcients among key independent and dependent variables

pin

for Cog. Complexity)Math Development (M) (Component for

Cog. Complexity)0.18 0.87 1.00

Languago

(Component for Cog. Complexity)Relation to Data (D) (Component for

Cog. C0.17 0.85 0.80

Cognitiv(Comp

rt

Instrumto Thi

da Z-scores m ),

was used to

Ap s

1 Taking instructions-helping: Attending to the work Handling: Using body members, handtools, and/or

appropriate tool, object, or materialso s o

v

e g i u gl a i h m hr i u c dp b s

sf ni

e S , g bc n p n s

d m l th g a ga a a ai w i n l m

d k s tf u t i , e

e j r. s la f m

e p n e ac l e

ll go d

r p : g ih e a

284 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 2702895 Diverting: Amusing others, usuamedium of stage, screen, televisi n, or ra io the action of mac ines or equipm nt for whichy throu h the

mD

anifestiving-o erating Startin , stopp ng, and controllingrto

gard tool, objerecisiot, or maattaineterial, ad and sthoughlectingthis is rppropadilyriate4 Persuading: Inuencing others inservice, or point of view d

m

vices toaterialswork, mInvolveove, gusomeide, or ptitudelace obor judgects oent withfavor o a prod

is

ct, M

involveanipulain maing: Using theng bodye adjusmembementsrs, tools or sp cialassistants

g

uges, tupping srning vitcheslves ton respollow ose to ligw of mhts. Littterialse judg, andentpeople to convey or exchange ingiving assignments and/or direct ons to helpers or a

c

justingangingateriauides,s or condjustinrols of ttimershe machand temine, superatuch asre3 Speaking-signaling: Talking with and/orormatioignaling. Includes

Tfu

nding:nctionintartingg of mastoppinhines a, and od equiservingment. Ithevolveor animals or the expressed or impeople. Immediate response is in olved o

o

equipmeratedent why otherch are aworkertomati or ten ed ornon-learning helpers2 Serving: Attending to the needs r reque

plicit w

ts of peishes ofple Fp

eding-oacing mffbearinterials: Insertn or remng, throoving twing, dem frompinmac, orinesassignment instructions or orders of supervisor. (Noimmediate response required unless clarication ofinstructions or orders is needed.) Helping applies to

special devices to work, move, or carry objects ormaterials. Involves little or no latitude for judgmentwith regard to attainment of standards or in selectingScores Interpersonal skill standards (relation to peopleb) Instrumental skill standards (relation to thingsc)pendix B. Detailed denitions of worker functions: interper onal and instrumental skill standards.aardized)

of P and T were constructed separately, and then 2.58 were added togenerate the specicity score (RSII).ake both scales bigger than 0. Then, the specicity formula, pj/(tj + pjRelation(Stanental Skill Standard (T) (Relationngs)

0.12 0.09 0.02

al Skill Specicity II (RSII)a 0.19 0.38 0.230.15 0.03 0.02 0.02 0.50 1.00

0.43 0.26 0.32 0.33 0.80 0.89 1.00Interpe(Relaomplexity)e Skill Complexity (Comp)osite Index)

0.19 0.95 0.91

sonal Skill Standard (P)ion to People)

0.20 0.62 0.450.85 1.00

0.83 0.83 1.00

0.93 0.96 0.93 1.00

0.60 0.53 0.58 0.59 1.00for CSpecice Development (L) (Componentg. Complexity)

0.22 0.93 0.86

Vocational Preparation (S) 0.16 0.89 0.841.00(ComReasonigrant Sentiments (AS)osite Index)

1.00

g Development (R) (Component 0.20 1.00AS R M L S D Comp P T RSII

Anti-Imm

Detailed denitions of worker functions: interpersonal and instrumental skill standards.a (continued)

Scores Interpersonal skill standards (relation to peopleb) Instrumental skill standards (relation to thingsc)

course must be steered or which must be guided tocontrol the movement of things or people for a varietyof purposes. Involves such activities as observinggauges and dials, estimating distances anddetermining speed and direction of other objects,turning cranks and wheels, and pushing or pulling

such as electric wheelbarrows and handtrucks6

responsibilities is involved in this function

Precision working: Using body members and/or toolsor work aids to work, move, guide, or place objects ormaterials in situations where ultimate responsibility

7 Instructing: Teaching subject matter to others, or

8

9

other professional principlesa The entire table was excerpted from Dictionary of Occupational Titles (1991), pb People: Human beings; also animals dealt with on an individual basis as if thec Things: Inanimate objects as distinguished from human beings, substances or A

thing is tangible and has shape, form, and other physical characteristics.

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 285p. 10051014.y were human.materials; and machines, tools, equipment, work aids, and products.Mentoring: Dealing with individuals in terms of theirtotal personality in order to advise, counsel, and/orguide them with regard to problems that may beresolved by legal, scientic, clinical, spiritual, and/orturning valves and other devices to regulate factorssuch as temperature, pressure, ow of liquids, speed ofpumps, and reactions of materialsNegotiating: Exchanging ideas, information, andopinions with others to formulate policies andprograms and/or arrive jointly at decisions,conclusions, or solutionsaccuracy of machine capabilities, properties ofmaterials, and shop practices. Uses tools, equipment,and work aids, such as precision gauges andmeasuring instruments. Workers who setup one or anumber of machines for other workers or who setupand personally operate a variety of machines areincluded hereOperating-Controlling: Starting, stopping, controlling,and adjusting the progress of machines or equipment.Operating machines involves setting up and adjustingthe machine or material(s) as the work progresses.Controlling involves observing gauges, dials, etc. andtraining others (including animals) throughexplanation, demonstration, and supervised practice;or making recommendations on the basis of technicaldisciplinesappropriate tools, objects, or materials, and theadjustment of the tool to the task require exercise ofconsiderable judgmentSetting up: Preparing machines (or equipment) foroperation by planning order of successive machineoperations, installing and adjusting tools and othermachine components, adjusting the position ofworkpiece or material, setting controls, and verifyingSupervising: Determining or interpreting workprocedures for a group of workers, assigning specicduties to them, maintaining harmonious relationsamong them, and promoting efciency. A variety of for the attainment of standards occurs and selection ofgear lifts or levers. Includes such machines as cranes,conveyor systems, tractors, furnace-chargingmachines, paving machines, and hoisting machines.Excludes manually powered machines, such ashandtrucks and dollies, and power-assisted machines,

Appendix C

C.1. Anti-immigrant sentiments against percentage of foreign-born and relational skill specicity (using HuberWhite robuststandard errors)

Age

Constant

N

Ro+ p < 0.1.*

**

286 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289p < 0.05.p < 0.01.bust standard errors in parentheses.R-squared 0.180 0.190 0.181 0.190(0.151)1055(0.148)1055(0.193)1055(0.189)10553.009** 3.087** 2.958** 3.065**(0.318) (0.323)

0.751* 0.358% Foreign-born in Rs 28 Larger Occ. Categories(0.280) (0.298)

0.528+ 0.259% Foreign-born in Rs Occupations(0.084) (0.067)

0.293** 0.287**Relational Skill Specicity(0.084) (0.082) (0.088) (0.084)

0.072 0.061 0.062 0.057Union Membership

(0.047) (0.046) (0.062) (0.060)

0.030 0.035 0.025 0.033Republican

(0.119) (0.118) (0.138) (0.137)

0.472** 0.489** 0.463** 0.484**Citizenship

(0.110) (0.109) (0.103) (0.101)

0.008 0.004 0.005 0.006Other (Non-Hispanic)

(0.095) (0.092) (0.088) (0.084)

0.500** 0.496** 0.498** 0.495**Non-Hispanic Black

(0.087) (0.084) (0.089) (0.085)

0.520** 0.513** 0.513** 0.510**Non-Hispanic White

(0.068) (0.067) (0.083) (0.084)

0.085 0.082 0.086 0.083Catholic

(0.065) (0.064) (0.068) (0.065)

0.132* 0.134* 0.136+ 0.136*Mainline Protestant

(0.058) (0.058) (0.082) (0.084)

0.158** 0.153** 0.161+ 0.154+Fundamental Protestant

(0.070) (0.072) (0.047) (0.049)

0.065 0.068 0.064 0.067Non-employed

(0.074) (0.077) (0.053) (0.053)

0.004 0.010 0.008 0.012Retired

(0.095) (0.093) (0.087) (0.086)

0.068 0.064 0.071 0.066Unemployed

(0.065) (0.066) (0.096) (0.093)

0.001 0.001 0.002 0.001Part-time Employed

(0.008) (0.008) (0.010) (0.011)

0.088** 0.082** 0.086** 0.081**

(0.002) (0.002) (0.002) (0.002)

Education0.004* 0.004* 0.004* 0.004*

(0.043) (0.042) (0.041) (0.042)Sex 0.045 0.040 0.043 0.040Clustered in 160occupations

Clustered in 28 largeoccupation categories

Model 1 Model 2 Model 3 Model 4

C.2. Anti-immigrant sentiments against percentage of foreign-born and relational skill specicity in 47 large occupational groups(using HLM models)

Part-time Employed

Unemployed

Retired

Relational Skill Specicity

% Foreign-born in 47 Occ groups

Constant

Level 1 variance

Intraclass correlation

N

Ro+

*

**

N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289 287p < 0.05.p < 0.01.bust standard errors in parentheses.p < 0.1.1055 1055

Chi-square 2204 21960.00658 2.38e07

Level 2 variance 0.00312 1.12e070.470 0.469

(0.196) (0.196)

2.961** 3.077**

(0.338) (0.334)

0.695* 0.342(0.0899)

0.267**Occupation group-level(0.0713) (0.0708)

0.0593 0.0577Union Membership

(0.0467) (0.0466)

0.0283 0.0292Republican

(0.118) (0.117)

0.468 0.476Citizenship Status

(0.139)

**(0.138)**0.00862 0.00138Other (Non-hispanic)

(0.0987) (0.0983)

0.504 0.500Black (Non-hispanic)

(0.0819)

**(0.0816)**0.516 0.511White (Non-hispanic)

(0.0593)

**(0.0591)**0.0861 0.0847Catholic

(0.0696) (0.0694)

0.136 0.135Mainline Protestant

(0.0576)

+

(0.0575)

+0.162 0.156Non-employed 0.0617 0.0593(0.0678) (0.0675)

Fundamental Protestant ** **(0.0792) (0.0786)

0.0103 0.0123

(0.0834) (0.0832)

0.0692 0.0634

(0.0697) (0.0693)

0.00504 0.00704

(0.0086) (0.0086)Education 0.086** 0.082**

(0.0017) (0.0017)Age 0.0039* 0.0038*

(0.0457) (0.0446)HLM models in 47 large occupational groups

Model 1 Model 2

Individual levelSex 0.0500 0.0393

288 N. Lee, C.-S. Lee / Social Science Research 52 (2015) 270289References

Acemoglu, Daron, Pischke, Jorn-Steffen, 1999. The structure of wages and investment in general training. J. Polit. Econ. 107 (3), 539572.Aguilera, Michael B., Massey, Douglas S., 2003. Social capital and the wages of Mexican migrants: new hypotheses and tests. Soc. Forces 82, 671701.Becker, Gary, 1964. Human Capital. University of Chicago Press, Chicago.Blalock, Hubert M., 1967. Toward a Theory of Minority-Group Relations. Wiley, New York.Blumer, Herbert, 1958. Race prejudice as a sense of group relation. Pac. Sociol. Rev. 1, 37.Bobo, Lawrence, 1983. Whites opposition to busing: symbolic racism or realistic group conict? J. Pers. Soc. Psychol. 45, 11961210.Bobo, Lawrence, Hutchings, Vicent L., 1996. Perceptions of racial group competition: extending Blumers theory of group position to a multiracial social

context. Am. Sociol. Rev. 61, 951972.Bollen, Kenneth A., 1989. Structural Equations with Latent Variables. Wiley, New York.Bonacich, Edna, 1972. A theory of ethnic antagonism: the split labor market. Am. Sociol. Rev. 37, 547559.Borjas, George J., Tienda, Marta, 1987. The economic consequences of immigration. Science 235, 645651.Borjas, G.J., Freeman, R.B., Katz, L.F., DiNardo, J., Abowd, J.M., 1997. Howmuch do immigration and trade affect labor market outcomes? Brookings Pap. Econ.

Activity 1997 (1), 190.Bowles, Samuel, Gintis, Herbert, 1976. Schooling in capitalist america: Educational reform and the contradictions of economic life. Basic Books, New York.Bowles, Samuel, Gintis, Herbert, 2002. Schooling in capitalist america revisited. Sociol. Educ. 75, 118.Bowles, Samuel, Gintis, Herbert, Osborne, Melissa, 2001. Incentive-enhancing preferences: personality, behavior and earnings. Am. Econ. Rev. 91, 155158.Bryk, Anthony S., Raudenbush, Stephan W., 1992. Hierarchical Linear Models in Social and Behavioral Research: Applications and Data Analysis Methods.

Sage Publications, Newbury Park, CA.Cain, Pamela S., Treiman, Donald J., 1981. The dictionary of occupational titles as a source of occupational data. Am. Sociol. Rev. 46, 253278.Coleman, James S., 1988. Social capital in the creation of human capital. Am. J. Sociol. 94, S95S120.Cranford, Cynthia J., 2005. Networks of exploitation: immigrant labor and the restructuring of the los angeles janitorial industry. Soc. Probl. 52, 379397.Espenshade, Thomas J., Hempstead, Katherine, 1996. Contemporary American attitudes toward U.S. immigration. Int. Migr. Rev. 30, 535570.Estevez-Abe, Margarita, Iversen, Torben, Soskice, David, 2001. Social protection and the formation of skills: a reinterpretation of the welfare state. In: Hall,

Peter, Soskice, David (Eds.), Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. Oxford University Press, New York.Farkas, George, 2003. Cognitive skills and non-cognitive traits and behaviors in stratication process. Annu. Rev. Sociol. 29, 541562.Farkas, George, England, Paula, Vicknair, Keven, Kilbourne, Barbara Stanek, 1997. Cognitive skill, skill demands of jobs, and earnings among young European

American, African American... more. Soc. Forces 75, 913938.Fligstein, Neil, 1997. Social skill and institutional theory. Am. Behav. Sci. 40, 397405.Hainmueller, Jens, Hiscox, Michael J., 2010. Attitudes toward highly skilled and low-skilled immigration: evidence from a survey experiment. Am. Polit. Sci.

Rev. 104, 6184.Hall, Peter, Soskice, David, 2001. An introduction to varieties of capitalism. In: Hall, P., Soskice, D. (Eds.), Varieties of Capitalism: The Institutional

Foundations of Comparative Advantage. Oxford University Press, New York.Heckman, James J., Kautz, Tim, 2012. Hard evidence on soft skills. Labour Econ. 19, 451464.Heinz, John P., Laumann, Edward O., 1982. Chicago Lawyers: The Social Structure of the Bar. Russell Sage Foundation/American Bar Foundation, New York.Huber, Peter J., 1967. The behavior of maximum likelihood estimates under nonstandard conditions. In: Proceedings of the Fifth Berkeley Symposium on

Mathematical Statistics and Probability. University of California Press, Berkeley, CA.Hunter, James D., 1991. Culture Wars: The Struggle to Dene America. BasicBooks, New York.Hunter, James D., 1994. Before The Shooting Begins: Searching for Democracy in Americas Culture War. Free Press, Maxwell Macmillan Canada, Maxwell

Macmillan International, New York, Toronto,