SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Policies and programs designed to challenge the effects of racial discrimination (such as affirmative action) are hotly contested. Factors which have been proposed to explain opposition to these policies include racial prejudice, group threat and self-interest, and perceptions of intergroup justice. We report the results of two random national telephone surveys which tested a theoretically based model of the predictors of policy support in post-apartheid South Africa. The results provided limited support for Blumer's group position model. Compensatory and preferential treatment policies had different underlying predictors: Violated entitlement featured in the models of compensatory policy attitudes, but not preferential treatment policy attitudes, where threat was the strongest predictor. In addition to threat and violated entitlement, policy attitudes among the black sample were related to ingroup identification but those of the white sample were related to prejudice. The effects of these variables were in the opposite directions for the two samples: Policy support was associated with strong ingroup identification and high levels of threat among the black sample (i.e. prospective beneficiaries of the transformation policies), but with low levels of prejudice and threat among the white sample. We conclude by considering the implications that these findings have for social change programs. Copyright © 2009 John Wiley & Sons, Ltd.

Racial transformation policies include laws, policies, programs and initiatives that are designed to improve the socio-economic opportunities of groups who have been disadvantaged by racism. Because these policies are perceived as supporting one or more social group at the expense of others, they have provoked strong reactions. Predictors of support for racial transformation have been studied in terms of attitudes towards desegregation (Greeley & Sheatsley, 1971), busing (Bobo, 1983; McClendon, 1985), affirmative action (Kravitz, 1995; Kravitz, Klineberg, Avery, Nguyen, Lund, & Fu, 2000), welfare (Peffley, Hurwitz, & Sniderman, 1997; Virtanen & Huddy, 1998) and immigration policy (Citrin, Green, Muste, & Wong, 1997).

This paper aims to identify and model the predictors of support for transformation policies in South Africa. During the apartheid era social stratification was entrenched by means of laws and policies prescribing extensive racial segregation, job reservation and political disenfranchisement (see Ashforth, 1990; Crankshaw, 1997). Black South Africans were barred from many categories of employment, from better resourced educational institutions and even from national sports teams.

Since 1994, the African National Congress (ANC) government has scrapped apartheid legislation and implemented racial transformation policies in an effort to redress inequality (cf. Habib & Bentley, 2008). These race-targeted interventions have been far more extensive than those employed elsewhere. They have included affirmative action in education, employment and sport.1 A far reaching program of black economic empowerment (BEE) has set targets for the transformation of medium and large business enterprises in term of black ownership and management, skills development and the procurement of goods and services. In addition, an extensive land reform program has aimed to secure land ownership for black populations by means of land restitutions (returning land to communities that were forcibly displaced), land tenure reform (securing tenure and preventing arbitrary evictions) and land redistribution (increasing black land ownership). Together, these policies have challenged the racial hierarchy in South Africa and their impact can be felt in all aspects of life, including the programs and languages that are available on public media, the composition of national sports teams, and the prospects for employment and acceptance to university study.

Public opinion about these policies is complex, multi-layered and fiercely divided. Whereas the policies and laws attract strong support, based on the perceived need to eradicate the effects of apartheid, they also attract strong opposition. Echoing reactions in the US (Pincus, 2001), affirmative action has been represented by some as reverse racism, and concerns have been raised about a drop in standards, with the beneficiaries of affirmative action being portrayed as less competent than other students, job applicants or sports players. Moreover, some of the younger generation of whites complain that they are unfairly prejudiced by the legislation since they did not personally benefit from apartheid.

Building on the wider literature on attitudes towards race-targeted policies, the present research was designed to explore some social and psychological factors that explain variation in support for transformation policies in post-apartheid South Africa. Although there is some controversy about the predictors of racial policy opinion, there is also consistent evidence of the importance of the following predictors (1) perceptions of personal and group-based threat, (2) racial prejudice and (3) a range of ideological factors related to sense of violated group entitlement, including stratification beliefs, conservatism, and perceptions of fairness (Krysan, 2000).

Since nearly all of the previous research has been conducted on white survey respondents in the US, a first aim of our research was to determine whether or not these findings are applicable in the South African context. Given the political, geographic, historical and economic context, we expected all three factors to play an even stronger role in shaping transformation policy attitudes in South Africa. As a numerical minority in a context where black people have both political power and control over the implementation of transformation policies, we expected whites to manifest a high sense of threat. With the recent history of apartheid, we expected racial prejudice to be an important factor in shaping whites' policy opinions. Finally, because apartheid ideology left many whites with a lingering sense that their privileged status was justly deserved (Steyn, 2001), we expected that their sense of violated entitlement would predict their attitudes towards transformation policies.

A second aim of this study was to expand the focus of the research to include an analysis of black people's attitudes towards transformation policies. As in other areas in social psychology (Crosby, Ferdman, & Wingate, 2001; Shelton, 2000), this field of research has been marked by ‘an unfortunate bias towards examining only the views of the dominant group’ (Bobo & Tuan, 2006, p. 21). Since racial change policies affect intergroup relations we anticipated that similar factors would be implicated in policy support among black and white respondents; however, we also anticipated that the nature and direction of their influence might vary for white and black South Africans.

In addition to identifying predictors of policy attitudes, we aimed to model these relationships. To date, research in this field has seldom explored the way in which the different variables combine to predict policy attitudes. In their recent groundbreaking study, Bobo and Tuan (2006) showed that the predictors of policy attitudes can be arranged in an order of causal priority which conforms to the predictions of Blumer's group position model. Group threat and sense of violated entitlement are immediate proximal predictors, whereas prejudice is a ‘distal’ variable. Our third aim was to determine whether the South African data for black and white respondents conformed to these predictions.

In the remainder of the introduction we first review evidence for racial prejudice, threat and sense of violated entitlement as predictors of policy attitudes, and then discuss the group position model.

RACIAL PREJUDICE

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Symbolic racism researchers were first to identify the significance of policy attitudes as a forum for the expression of prejudice in the US, a context where crude racism had gone out of fashion (Kinder & Sears, 1981; McConahay & Hough, 1976). They argued that deep-seated anti-black affect, acquired in early childhood socialization—alongside the values of equality and democracy—was expressed in subtle and indirect ways, including by way of opposition to policies such as busing, affirmative action and social welfare (Sears, 1988; Sears, van Laar, Carrillo, & Kosterman, 1997). Pettigrew and Meertens (1995) argue that a lack of positive emotions towards an outgroup is also a manifestation of subtle prejudice. In their European research, they found that a lack of sympathy for immigrants, rather than overt hatred or hostility, was related to racial policy attitudes.

Other research has shown that more traditional expressions of prejudice may also play a role in shaping policy attitudes. Sniderman and Piazza (1993) showed that a sizeable minority of white Americans agreed that most blacks are lazy, complaining, and have a chip on their shoulder; and experimental research has shown that whites who embraced negative stereotypes about blacks, displayed a ‘discriminatory double standard’ by offering negative judgements of policies that assist blacks but not policies that assist whites (Peffley et al., 1997).

There is empirical support for both the symbolic racism and traditional racism theses. Surveys in the US (Sears & Henry, 2005), Europe (Pettigrew & Meertens, 1995), Australia (Pedersen & Walker, 1997) and South Africa (Pillay-Singh, & Collings, 2004) have demonstrated that symbolic racism and old-fashioned racism form two distinct but correlated factors, and much of this research shows that symbolic racism is a stronger predictor of racial policy opinions than old-fashioned racism. However, the work of Sniderman and his colleagues provides a valuable corrective to the emphasis on new subtle forms of modern racism. Certainly, it is both ‘premature and misleading to jettison measures of old-fashioned racism from studies predicting white opposition to racial policies’. (Virtanen & Huddy, 1998, p. 327).

The role of old-fashioned racism may even be more prominent in South Africa, where the majority of the white adult population were socialized in the crudely racist context of apartheid. One of the few studies that investigated the relationship between racism and policy opinions in that context found that old-fashioned racism was a stronger predictor of opposition to transformation policies than symbolic racism among a supposedly liberal student sample (Durrheim, 2003). Although there is no equivalent data, opposition to transformation may be related to racial attitudes for the black population. Black people may oppose the policies out of sympathy for whites who are disadvantaged by such policies.

THREAT

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Since transformation policies are often perceived to threaten the interests of one group relative to another, perceptions of threat are expected to correlate closely with opposition to the policies. Early research conceptualized and measured threat in terms of self-interest. White families directly affected by busing in the US were expected to oppose the policy more than those whose interests were not at stake. However, McConahay (1982) found that none of nine self-interest measures—including such factors as having a child at school—were significantly related to anti-busing sentiment (see Sears & Allen, 1984). On the basis of his review of this work Kinder (1986) concluded that ‘self-interest …packs much less political whallop (sic) than our assumptions and theories have typically and rather uncritically assigned to it’ (p. 164).

Over the years, these initially perplexing findings have been qualified by research that has introduced conceptual and methodological refinements. Perceptions of threats to self-interest have been shown to be associated with willingness to protest against busing (Green & Cowden, 1992; McClendon, 1985) as well as opposition to social welfare programs (Virtanen & Huddy, 1998), immigration (Citrin et al., 1997) and affirmative action (Jacobson, 1985; Kluegel & Smith, 1983; Kravitz, 1995; Summers, 1995; Tougas, Brown, Beaton, & Joly, 1995). In overall terms, however, these effects have not been especially strong and indices of racism generally have stronger relationships with policy attitudes (Sears & Funk, 1991). Thus although self-interest has some ‘political wallop’, this varies across policies, and might be overshadowed by the effects of racial prejudice (Sears, 2001).

Bobo has argued that much of this research has focused on the wrong level of analysis. Rather than threats to the interests of individuals and their families, racial policy opinions primarily reflect threats to collective interests (Bobo, 1983; cf. Funk, 2000). Bobo and Hutchings (1996) used Blumer's group position theory to argue that opposition to racial transformation policies emerges from perceived threats to the positional arrangement of social groups in society.

In their survey concerning the controversy about Chippewa Indian fishing rights in Wisconsin, USA, Bobo and Tuan (2006) measured threat by asking whether respondents felt that ‘many Indians have been getting ahead economically at the expense of many non-Indians’. Such items operationalize threat in group position terms, as ‘perceived group competition or zero-sum access to important social resources’ (Bobo, 1999, p. 458). Bobo and Tuan (2006) showed that the effects of personal self-interest were reduced to non-significance when group threat was entered into the regression model predicting opposition to Indian treaty rights. Likewise, Citrin et al. (1997) found that personal economic motives (such as living in states with high concentrations of recent immigrants) played a restricted role in shaping opinion towards immigration reform in the US, but that concern about resources of the dominant group (e.g. beliefs about the state of the economy, anxiety over taxes) was a stronger determinant of restrictionist sentiment.

In the present research, we expected that group threat would be strongly related to policy opinions; we also wanted to explore whether self-interest would predict support for transformation policies. After the transition to black majority rule many whites in South Africa have felt increasingly marginalized as a small minority with much to lose from race-targeted policies of transformation. In addition to this group threat, individuals and families may oppose the affirmative action policies in the belief that they have been (or will be) directly affected by them. In contrast, we expected to find support for racial change policies among black respondents who believe that whites continue to be advantaged by the legacy of racism and thus pose a threat to their group position in society. It is also possible that black respondents would support the policies out of self-interest, believing that they would benefit personally from them.

SENSE OF VIOLATED ENTITLEMENT

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Sense of violated entitlement is closely related to perceptions of threat in Blumer's group position theory (Bobo, 1999). Perceptions of entitlement arise historically as dominant groups forge ideologies that justify their relative social and material advantages; in contrast, groups can be said to be racially alienated when they perceive themselves to be unfairly disenfranchised. Sense of violated entitlement is a product of ideological depictions of intergroup justice and just deserts. It is a sense of the relative injustice of social and economic rewards of the ingroup in comparison with the outgroup.

Although sense of violated entitlement has not been an explicit focus in the racial policy attitude literature, a number of closely related constructs have been proposed as explanations of opposition to racial change policies. These have included conservatism (Sniderman & Tetlock, 1986), individualism (Lipset & Schneider, 1978), equality and fairness (Sears, Henry, & Kosterman, 2000), stratification ideology (Kluegel & Smith, 1983), social dominance orientation (Sidanius, Pratto, & Bobo, 1996) and racial alienation (Bobo, 1999). As racial policy in the US shifted from Jim Crow segregation to desegregation, busing and affirmative action, so the prime concerns about race changed fundamentally: ‘Instead of debating blacks’ innate ability or intelligence, racial policy elicited concerns about black deservingness and explanations for persistent poverty'. (Virtanen & Huddy, 1998, p. 313).

Especially important are beliefs about the causes and persistence of discrimination, which are rooted in perceptions of whether unequal social outcomes are just or not. The more racial inequality is attributed to the history of discrimination the more entitled disadvantaged groups are perceived to be, leading to greater support for transformation policies. In their work on ‘stratification ideology’, Kluegel and Smith (1983) showed that opposition to affirmative action among whites was strongly related to (1) the belief that there is very little discrimination against blacks, and (2) the related belief that the causes of inequality are individual factors such as laziness or lack of motivation, not structural factors such as unequal education or discrimination (Bobo & Kluegel, 1993; Kluegel, 1985, 1990; Kluegel & Smith, 1982, 1983, 1986). These results have been confirmed by Kravitz and his colleagues (Kravitz, 1995; Kravitz et al., 2000), who showed that positive attitudes towards affirmative action were associated with the belief that the policies were fair and that the beneficiaries of the policies were in need of assistance.

We anticipated that white respondents who believed that they were entitled to more than they were getting would be opposed to the transformation policies; but that black respondents who believed that they were entitled to more than they were getting would support the transformation policies.

The Model: Group Position Theory

The literature reviewed above has often taken the form of a debate about the relative importance of the different predictors of policy attitudes, which have often been ‘framed as antithetical lines of analysis’ (Bobo & Tuan, 2006, p. 216). Recently, Bobo and Tuan (2006) have adapted Blumer's (1958) group position theory to provide an integrative model of policy attitudes. Rather than giving priority to one factor over others—e.g. is it prejudice or non-racial interest that underlie opposition?—they seek to develop a theory that ‘blends core ideas from orthodox prejudice models with a sociological understanding of group relations’ (p. 31). They use theoretically informed hierarchical regression analysis to develop a multivariate model of predictors, showing how the effects of some (distal) factors are superseded by the effects of more immediate (proximal) factors. This analytic strategy seems very appropriate to a field of investigation which has accumulated a good deal of evidence to support all three sets of predictors, but where none of the effects are particularly large and the variables appear to be intercorrelated. Prejudiced whites, for example, are expected to perceive transformation policies as a threat to their deserved status and oppose them. Thus, analysis needs to proceed on the basis of a theoretical model that allows us to appreciate the structure of the intercorrelations and the causal priority of the predictors.

Group position theory provides a model of proximal and distal influences of policy opinions. The factor of prime interest in this theory is sense of group position, which is a socially shared and psychologically potent sense of one's group position in the racialized social order (Bobo & Tuan, 2006). Sense of group position has two central components; namely, sense of group entitlement and sense of intergroup threat (see Bobo, 1999). Sense of entitlement is a prescriptive view of the status that the ingroup is entitled to vis-à-vis outgroups; and sense of intergroup threat is a perception of outgroup challenge to those entitlements. Racial transformation policies are opposed by members of the dominant group who experience a sense of violated entitlement and outgroup threat. They feel the policies would undermine their rightful status or privilege in society.

Sense of group position is a social force because it shapes opinion and behaviour directed at preserving or challenging the existing order of social stratification. Accordingly, these two aspects of sense of group position—i.e. entitlement and threat—are expected to be directly implicated in shaping opinions towards policies that are designed to alter the existing order of racial stratification. According to the group position model, prejudice remains an important factor. However, threat and entitlement are hypothesized to be proximate determinants of policy opinion whereas prejudice is a distal force: ‘Prejudice translates into an active, adaptive political force via a sense of proper entitlement or prior rights to enjoy certain resources and a sense of threat or challenge to these entitlements from a subordinate group’. (Bobo & Tuan, 2006, p. 151).

In the only previous test of the group position explanation of policy attitudes, Bobo and Tuan (2006) found that outgroup prejudice was correlated with policy attitudes, but that the effects of prejudice became non-significant once threat and entitlement were entered into the regression model. This suggested that the impact of prejudice on policy opinions was ‘entirely indirect’ (p. 169). Thus, although we expected prejudice to be correlated with policy attitudes, we hypothesized that prejudice would affect policy attitudes indirectly—through sense of threat and violated entitlement—and that the direct effect would not be significant in the structural models.

Although Blumer was primarily interested in how the prejudices and opinions of dominant group members served to defend their sense of group position, Bobo and Tuan (2006) suggest that the theory can be applied to explain subordinate group members' attitudes to dominant groups. To our knowledge such an application of group position theory to the policy attitudes of prospective beneficiaries of the policies has yet to be made, and thus we cannot make firm predictions about the model of policy attitudes for our black samples. We proposed that sense of group position would influence the policy attitudes of both white and black respondents and we thus expected the same predictors of policy attitudes. However, because black people are the targeted beneficiaries of the policies, we expected the signs of the relationships between policy attitudes and prejudice, sense of threat and violated entitlement to be in opposite directions for the black and white samples. We predicted that transformation policies would receive higher levels of support from black respondents who believe that black people are unfairly disadvantaged and who believe that whites pose a threat to their group status. In contrast, the policies would receive lower levels of support from white respondents who believe that whites are unfairly disadvantaged and collectively threatened. In line with Bobo & Tuan's (2006) application of group position theory, in the structural equation model we expected that these group position effects would reduce the anticipated zero-order correlation between prejudice and policy attitudes to non-significance.

The Surveys

We tested the predictions of the group position model of policy attitudes by conducting two opinion surveys in South Africa, each using different measures of the central constructs. Sampling was done by means of random digit dialling of cellular telephone numbers. Cell phone numbers were randomly generated and then screened with the aid of auto-dialing software to eliminate non-existent numbers. Cell phone surveys are a relatively inexpensive way of generating a diverse sample in South Africa. In 2007, 56.3% of the adult population (>16 years) either owned or rented a cellphone, including 52% of the black population and 82% of the white population, with equal proportions of males and females (South African Advertising Research Foundation, 2007). Not only was this twice the proportion of landline ownership (23%), but cell phone ownership is also reasonably frequent in non-urban settlements (44%), among the non-working population (47%) and among the poor (37% of people living in households earning less than $140 per month).

Survey 1 was conducted between February and April 2006, and Survey 2 was conducted between March and July 2007. Both samples were composed of black African and white South African citizens who were aged 18 years and older. Table 1 shows that the sample composition of Survey 1 corresponded closely with national demographics, except that black and white respondents were overrepresented (at the expense of other population groups, Indians and coloureds); and the poor were overrepresented. To achieve comparable levels of point estimate precision across the black and white samples, Survey 2 was designed to include an over representation of white respondents. We did this by over representing cell phone numbers of contract subscribers.

Table 1. Comparison of survey demographics with cellphone ownership and national demographics
 National demographics (>16 years old)% Cellphone ownershipSurvey 1Survey 2
N31 109 000 19172484
Mean age  34.62 (SD = 12)36.7 (SD = 13)
Race
Black23 446 000 (75.4%)51.7%1556 (81.2%)1684 (68%)
White4 102 000 (13.2%)82.3%361 (18.8%)793 (32%)
Gender
Male15 501 000 (49.8%)55.8%921 (48%)1227 (49.4%)
Female15 608 000 (50.2%)56.9%996 (52%)1257 (50.6%)
Monthly household income
<$1408 041 000 (25.8%)36.6%637 (36.4%)447 (21.9%)
>$13334 912 000 (15.8%)86.9%161 (11.2%)531 (26%)

We expected high response rates because in South Africa (1) there are no costs to persons receiving the call, and (2) cell phone surveys are not frequent and so there should be little fatigue from unsolicited calls. This expectation was confirmed by our data. Of the 3687 people who were contacted, 2930 agreed to participate, and 2859 completed the questionnaire (including 200 coloured and 173 Indian respondents who were not included in the analyses reported in this paper). The final response rate was thus 77.5% (95%CI: 0.76 < RR < 0.791). The 71 people who did not complete the survey did not differ significantly from the sample in terms of age, gender, race, income or education. Although we did not record the response rate for Survey 1, there is no reason to assume that it was any different. Both surveys used similar methods and questions, and were of similar length (requiring between 15 and 30 minutes to complete).

The survey questionnaires were developed in English and translated into five other commonly spoken languages in South Africa (isiZulu, isiXhosa, seSotho, sePedi and Afrikaans). Each translation was done independently by two fully bilingual mother-tongue speakers of the African language who met to resolve any discrepancies. Microsoft Access was used to construct an electronic questionnaire that allowed researchers to administer the questionnaire in any of the six languages, and record responses onto a common database. A team of five interviewers who collectively spoke all of the six languages administered the questionnaire in the home language of the respondent. To deal with potential race of interviewer effects, we matched the language of the respondent to that of the interviewer. Although this did not eliminate racially unmatched interviewers and respondents, it did substantially reduce its incidence. At the start of the interviews, all respondents were told that their phone numbers were randomly selected and they were invited to participate anonymously in the interview about ‘race and transformation in South Africa’.

STUDY 1

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Measures

Racial Policy Attitudes

The nature and construction of racial transformation policies determine how they are evaluated (Fine, 1992). Policies framed or justified in a non-racial and universalistic manner (e.g. to help schools in poor neighbourhoods) gain more support than policies framed in racial terms (e.g. to help schools in black neighbourhoods) (Bobo & Kluegel, 1993; Sniderman, Carmines, Laymen, & Carter, 1996). Also, different kinds of interventions attract different levels of support (Kravitz, 1995). Tuch and Hughes (1996) showed that white respondents favour compensatory programs which enable members of disadvantaged groups to compete more successfully, but they oppose preferential treatment affirmative action programs. However, they found similar predictors of the two kinds of policy attitudes, and concluded that ‘the process by which support for these disparate policies is created does not vary by policy type’ (p. 738).

We compiled a set of nine items to measure attitudes towards diverse racial change policies. These included both compensatory and preferential treatment policies, some specifically targeting black people but others more generally targeting ‘historically disadvantaged’ people. Three-point rating scales were used to record whether the respondent was opposed to (scored 0), unsure (scored 1), or in favour of (scored 2) the policy. The items included:

  • (1)
    Providing special scholarships for black children who maintain good grades. (compensatory)
  • (2)
    Spending more money on schools in largely black neighbourhoods, especially for early education programs. (compensatory)
  • (3)
    Setting up quota systems to ensure racial integration at universities and schools. (preferential)
  • (4)
    Redistributing land by helping emerging black farmers get loans to buy land. (compensatory)
  • (5)
    Creating laws that stop farmers evicting farm labourers. (preferential)
  • (6)
    Redistributing land, by settling emerging black farmers on white owned farms. (preferential)
  • (7)
    Programmes that help to skill historically disadvantaged people so that they can compete fairly with whites for jobs and promotion. (compensatory)
  • (8)
    Affirmation action in hiring and promoting people who have been historically disadvantaged. (preferential)
  • (9)
    BEE [black economic empowerment] policies, giving preferential contracts and tax breaks to black businesses. (preferential)

The compensatory item scores were too skewed to be used as an independent measure, with the majority of the white sample and almost the entire black sample favouring the policies. Summing the scores for all nine compensatory and preferential treatment items yielded an overall score of policy support with a possible range from 0 (no support) to 18 (high support). The scores were skewed for the black sample (−1.158) but less so for the white sample (−.289). The internal consistency of the scale was adequate for the white sample (α = .72), but somewhat low (α = .55) for the black sample. We suspect that the low internal consistency and the skewness are partly attributable to the ‘ceiling effect’ that resulted from the use of a crude 3-point rating scale in this survey.

Threat
Self-interest

In two separate indicators of threat, respondents were asked to indicate the likelihood—response options: unlikely (scored 0), unsure (scored 1), likely (scored 2)—that the following events would happen: (1) You or a family member lose your job, or are prevented from getting a job or a promotion because of your race? (2) You will have to relocate or that your house or land value will depreciate because of racial integration? Scores had a potential range from 0 to 2, with high scores indicated perceptions of a high likelihood of a negative event occurring, and thus indexed high levels of race-based threat to personal self-interest.

Group Threat

Respondents were asked to rate the social and economic life circumstances of blacks and whites on Cantril's (1965) self-anchoring scale. This measure consists of a ladder with steps from 1 to 10 on which respondents indicated the socio-economic status of different social groups. By asking respondents to make group status comparisons, this measure directly assesses a sense of relative group position, and has previously been used to measure fraternal relative deprivation (Bobo, 1983). We operationalized threat in terms of contrasting perceptions of status change. Was the ingroup gaining or losing status in comparison with the outgroup? To this end, we asked respondents to indicate where they thought blacks and whites stood on the ladder 5 years ago, and where they expected them to stand 5 years in the future. The threat index was computed by, firstly, determining perceptions of ingroup and outgroup status change, by subtracting past from future status of each group. This index reflected how much each group had increased in socio-economic status. Group threat was then operationalized in terms of relative perceptions of ingroup versus outgroup status improvement. This was computed by subtracting the ingroup status change from the outgroup status change. Scores had a possible range from −18 to 18, where high scores represent perceptions that the outgroup has enjoyed a greater increase in status than the ingroup, and thus indicate higher levels of group threat.

Prejudice

In order to capture the diverse ways in which racial prejudice has been measured in the literature, we employed a range of measures of racial attitudes including old-fashioned racism, stereotypes and lack of sympathy. All of the racial attitudes indices were scored so that high scores indicated high levels of prejudice towards the outgroup.

Sympathy

Following Pettigrew and Meertens (1995), we measured prejudice subtly in terms of lack of positive emotions towards the outgroup. Respondents rated how often they felt sympathy for black and white people on a scale ranging from 1 to 10. A sympathy score was computed by subtracting outgroup sympathy from ingroup sympathy, yielding an index of the degree to which respondents showed less sympathy for the outgroup than their own group. Scores had a possible range from -9 to 9.

Stereotypes

In two items, respondents were asked to rate (on a scales ranging from 1 to 10) how trustworthy (versus untrustworthy) and how law-abiding (versus criminal) black and white people are in general. Relative scores of the ingroup being more trustworthy and law-abiding than the outgroup were computed by subtracting the outgroup from the ingroup ratings. The scores on the two scales were correlated for the black (r = .35) and white (r = .49) samples, and were summed to create an overall measure of outgroup stereotypes. Scores had a possible range from −18 to 18.

For both sympathy and stereotypes we computed relative scores by subtracting ratings of the outgroup from ratings of the ingroup ratings. This effectively standardized these indices by eliminating individual differences in generalized negativity that is reflected in an overall tendency to stereotype and lack sympathy for both ingroup and outgroup members. Exploratory analysis showed that these relative scores were stronger predictors of policy attitudes than the raw scores.

Old-fashioned Racism

We included one widely used measure of old-fashioned racism for the white sample: ‘Do you think the differences [in social status] between blacks and whites are because most black people have less in-born ability to learn?’ (Yes, Unsure, No). Scores had a possible range from 0 to 2 with high scores index high levels of old-fashioned racism.

Sense of Violated Entitlement

Respondents were asked to rate the social and economic life circumstances of blacks and whites on the Cantril self-anchoring scale (steps from 1 to 10). For each group they indicated where they thought blacks and whites stand today, and where they should be placed if they had their ‘fair and rightful share of wealth in the country’. First, perceptions of ingroup and outgroup inequity were calculated by subtracting the present status from the justly entitled status for each group. This inequity index reflected the extent to which each group currently received less than they rightfully deserved. Following Duckitt and Mphuthing (1998), we computed sense of violated entitlement by subtracting the outgroup inequity score from the ingroup inequity score. Scores had a possible range from −18 to 18, where high scores indicate perceptions of more ingroup inequity relative to outgroup inequity.

Results

Table 2 shows a pattern of expected group mean scores on the measures. There were high levels of support for the policies, with the black sample showing stronger support (with less variance) than the white sample (t = 19.40, df = 1915, p < .0001, η2 = .26). The white respondents reported higher levels of threat to job-related self-interest (t = −3.15, df = 1516, p < .003, η2 = .007) and group threat (t = −53.97, df = 1872, p < .0001, η2 = .60) than the black respondents. Underscoring the value of sympathy as a subtle measure of prejudice, whites on average expressed more sympathy for blacks than for whites (t(mean ≠ 0) = −4.75; df = 337; p < .0001), but they stereotyped blacks more negatively than whites (t(mean ≠ 0) = 9.64; df = 298; p < .0001). Black respondents also expressed more sympathy for blacks than for whites (t(mean ≠ 0) = 40.72; df = 1368, p < .0001), and they stereotyped blacks more negatively than whites (t(mean ≠ 0) = −8.07; df = 1368; p < .0001). As expected, blacks respondents expressed higher levels of violated entitlement than whites (t = 34.42, df = 1864, p < .0001, η2 = .33).

Table 2. Descriptive scale statistics for white and black samples, Study 1
 Observed rangeWhite sampleBlack sample
MeanSDMeanSD
Policy attitude0–1810.844.2815.402.60
Self-interest (job)0–21.04.95.85.92
Self-interest (housing)0–2.41.73.45.69
Group threat−18–184.903.21−5.463.31
Sympathy−9–9−.622.383.182.95
Stereotype−18–181.833.29−1.165.35
Old-fashioned racism0–2.35.73
Violated entitlement−8–18−1.332.604.313.23

The bivariate correlations reported in Table 3 show that all the predictors—except job-related self-interest for the black sample—were significantly correlated with policy support for both samples, and were of a comparable magnitude to the effects generally found in the literature. The two self-interest indices had the strongest relationships with policy attitudes, whereas the index of group threat had much weaker relationships. For both samples, higher threat was associated with lower levels of policy support. Interestingly, it was primarily job-related threat that was associated with policy attitudes among whites but housing-related threat that predicted policy attitudes in the black sample. This reflects popular concerns among whites about affirmative action, and concerns among blacks about adequate housing and the redistribution of land. We anticipated that whites would be concerned that affirmative action policies might threaten their employment, and that this would lead to opposition to the policies. However, we did not anticipate a relationship of the same direction for the black sample. Black respondents who scored high on the property threat item showed more opposition to the policies. We hypothesized that social class might explain these anomalous findings. Perhaps the wealthier black respondents were politically conservative and threatened by anti-free market policies such as land redistribution. Regression analysis showed that the relationship between policy support and housing threat (β = −.2, p < .0001) and group status (β = −.09, p < .002) remained significant after controlling for income and education, which suggests that the class hypothesis was wrong.

Table 3. Correlations between measures for black sample (above diagonal) and white sample (below diagonal), Study 1
 Policy attitudesSelf-interest (job)Self-interest (housing)Group threatSympathyStereotypeViolated entitlement
  • *

    p < .05;

  • **

    p < .0001.

Policy attitudes1−.03−.24**−.14**.19**.08*.10**
Self-interest (job)−.32**1.12**.07*.01.04.09*
Self-interest (housing)−.19**.24**1.14**−.09*−.03−.07*
Group threat−.12*.11*.11*1−.13−.06*−.03
Sympathy−.28**.08.18*.13*1.20**.19**
Stereotype−.22**.13*.20**.09.28**1.02
Violated entitlement−.22**.08.20**.12*.26**.14*1
Old-fashioned racism−.28**.19*.17*.18*.27**.21**.32**

In contrast, as hypothesized, the indices of prejudice were associated with policy support in opposite directions for the black and white samples. The lower the anti-black prejudice, the greater the degree of support for transformation policies among whites, but higher levels of anti-white prejudice were associated with greater policy support among blacks. These results suggest that both groups understand that the policies benefit blacks at the expense of whites. They are thus opposed by blacks who are less ethnocentric in their sympathies (they sympathize with whites relatively strongly in comparison with their own group sympathy) and they are opposed by whites who are more ethnocentric in their sympathies (they sympathize with blacks relatively less than with whites).

The correlation between policy attitudes and violated entitlement was more than twice as strong for the white than the black sample. In both cases, these results supported our hypotheses: A high sense of violated entitlement among whites was associated with low levels of support for transformation policies, whereas a high sense of violated entitlement among blacks was associated with high levels of support for transformation policies.

We then used AMOS to assess structural equation models that represented the predictions of group position theory. We had originally intended to use all three threat scores as observed indices of the latent threat variable in our structural equation model. As suggested by the correlations reported in Table 3, our measurement model indicated that group threat did not perform as expected for the white sample (see Figure 1) and that job related self-interest did not perform as expected for the black sample (see Figure 2). These indices did not correlate strongly with the other threat indices or with policy attitudes and were thus eliminated in later iterations of the model.

thumbnail image

Figure 1. Model of predictors of policy attitudes, white sample, Study 1. Note: Standardized regression coefficients are reported

Download figure to PowerPoint

thumbnail image

Figure 2. Model of predictors of policy attitudes, black sample, Study 1. Note: Standardized regression coefficients are reported

Download figure to PowerPoint

Our central aim was to determine whether prejudice had direct effects on levels of support for transformation policies, over and above the indirect effects it might exert through the two measures of sense of group position, namely, sense of threat and violated entitlement. The model for the white sample provided a good fit to the data (χ2 = 21.74, df = 11, p = .026; GFI = .98; RMSEA = .052). Manifest indicators had acceptable associations with the latent variables, threat and prejudice (all p < .01). The results of the structural model only partially fitted the predictions of group position. As expected, there was a significant indirect effect between prejudice and policy attitudes, mediated by threat (β = −.20, Sobel z = −1.96, p < .05). However, prejudice continued to have a strong direct relationship with policy attitudes (β = −.32, p < .05). In contrast to Bobo and Tuan's (2006) results, prejudice was not simply a distal variable whose effects were mediated by the more proximal group position variables. In addition, the association between sense of violated entitlement and policy attitudes was not significant (CR = .485, p < .63).

The model for the black sample also provided very good fit to the data (χ2 = 13.70, df = 6, p = .033; GFI = .997; RMSEA = .029). As was the case with the model for the white sample, (1) the observed indices of prejudice and threat were acceptable measures of the latent variables (p < .01), (2) the standardized regression weights were in the same direction as the bivariate correlations and (3) the direct effect between sense of violated entitlement and policy attitudes was not significant (CR = 1.22; p < .23). This model provided stronger support for the predictions of group position theory than the model for the white sample: The significant positive zero-order relationship between prejudice and policy attitudes was reduced to a non-significant effect (β = .07, p < .23) in the model, but the indirect effect, through threat, remained significant (β = .18, p < .0001). The Sobel test confirmed that the relationship between prejudice and policy attitudes was mediated by threat (z = 3.14, p < .002).

Discussion

Our first telephone survey showed that the predictors of policy attitudes are much the same in South Africa as elsewhere. Among the white sample, correlation analysis showed that opposition to transformation policies was associated with high levels of threat, prejudice and with a high sense of violated entitlement. These same variables served as predictors among the black sample, but that the direction of the effects for prejudice and sense of violated entitlement were in the opposite direction. This reflects the fact that these policies were designed to advantage blacks in comparison with whites. They were thus opposed by black respondents who felt high levels of sympathy for whites in comparison with blacks, who were unprejudiced towards whites, and who had a low sense of violated entitlement. The structural equation models provided limited support for the predictions of group position theory. Although, as hypothesized, the effect of prejudice on policy attitudes was mediated by threat for the white sample, contrary to our expectations, this mediation was only partial, as prejudice continued to have a significant independent relationship with policy attitudes. In addition, counter to our hypotheses, sense of violated entitlement did not significantly predict policy attitudes in the model. The data for the black sample provided somewhat stronger support for the group position model: The prejudice effect was reduced to non-significance once the effects of the ‘proximal’ variables (threat and entitlement) were included in the model. However, here too, sense of violated entitlement did not predict policy attitudes.

In Study 1 we did not anticipate that threat would be correlated with policy attitudes in the same direction for the black and the white samples. It is possible that these anomalous findings were the result of the way in which policy attitudes and threat were measured. The vast majority of black respondents supported nearly all the policy items, and most of the variation in responses revolved around the single land redistribution item favouring expropriation of white farms. At the same time, threat was measured in terms of realistic threat to personal self-interest. Perhaps expropriation was equally threatening to black and white respondents' housing security regardless of the race-based intent of the policy.

STUDY 2

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

This second survey aimed to replicate, refine and extend the findings of the first survey using better measures. We aimed to test an extended version of group position theory by (1) using reliable measures of compensatory and preferential treatment racial policy attitudes, (2) extending our model to include group identification as an additional ‘distal’ predictor variable and (3) investigating the different effects that realistic and symbolic threats play in the model.

To reduce degenerate variance and improve the internal consistency of the racial transformation policy attitude scale we substituted 5-point Likert response scales for the three point scales used in Study 1. This allowed us to reliably measure distinct attitudes towards compensatory and preferential treatment policies, and to determine whether the predictive models differed across these two kinds of policies. Since preferential treatment policies foster intergroup competition, we expected that threat may be a stronger predictor of attitudes towards these rather than the compensatory policies.

Second, we sought to extend our model by including a measure of group identification. Very little research has investigated the relationship between group identification and policy attitudes. The evidence shows that policy attitudes are aligned with group identities, and that identification with one's group is associated with support for programs which recognize that their group is ‘unjustly treated and which [are] designed to improve the situation’ (Conover, 1984; Schmermund, Sellers, Mueller, & Crosby, 2001; Tougas & Veilleux, 1988, p. 25). In addition, ingroup identification has also been shown to be related to prejudice, sense of relative deprivation and concerns about intergroup justice (Smith, Spears, & Hamstra, 1999; Tropp & Wright, 1999; Wenzel, 2004), and is thus closely related to a sense of group position (cf. Bobo, 1999). Since racial policies target intergroup relations, we hypothesized that strong ingroup identification would predict opposition to the policies among whites and support for the policies among blacks. However, as with prejudice, we expected that the effects of ingroup identification on policy attitudes would be entirely mediated by the group position variables. In our model, group identification is a distal factor whose effects are superseded by proximal effects of sense of threat and violated entitlement.

Much of the policy attitudes research and work in the Blumerian tradition has focused on realistic threat to a person's material well being. Drawing on work by McLaren (e.g. McLaren, 2002, 2003; McLaren & Johnson, 2007) and on Integrated Threat Theory (Stephan & Renfro, 2002; Stephan & Stephan, 2000), we proposed that threat to a sense of group position may, additionally, be related to symbolic threats. Realistic threats centre on conflicts over collective possessions and rights which can mostly be defined in material terms (e.g. competition for jobs, housing, access to social services, political power, etc.). Symbolic threat stems from perceived challenges to the ingroup's values and is manifest when there are perceived incompatibilities of the cultural norms and worldview of groups. In this second survey we measured both kinds of threat so that we could determine whether they are differentially related to policy attitudes.

There is a considerable body of empirical research linking both of these types of threat to intergroup attitudes. A recent meta-analysis of 95 research articles indicated that symbolic and realistic threats each contribute unique effects in predicting prejudice (Riek, Mania, & Gaertner, 2006). Realistic threat, but not symbolic threat, was a better predictor of outgroup attitudes towards low-status groups than towards high-status groups. Riek et al. (2006) suggest that competition over resources may be more aversive for those who have the most to lose (i.e. the high-status group); thus, ‘realistic threats posed by low status outgroups have a larger impact on attitudes than the [realistic] threats from high status outgroups’ (p. 345). Symbolic threat, however, does not seem to show this kind of status-based asymmetry.

Although Stephan and colleagues suggest that integrated threat theory can be applied to understanding ‘opposition to policies favouring the outgroup’ (Stephan, Renfro, & Davis, 2008, p. 62), very little research has investigated the differential impact of symbolic and realistic threat on racial policy attitudes. McLaren showed that symbolic and realistic threat predicted support for the European Union (McLaren, 2002) and anti-immigrant prejudice in Europe (McLaren, 2003; McLaren & Johnson, 2007). Renfro, Duran, Stephan, & Clason (2006) investigated the attitudes of two samples of American college students towards affirmative action. Their first study—with a white male student sample—found that only realistic threats predicted attitudes towards affirmative action; but their second study—with a racially diverse student sample—found that realistic and symbolic threat independently predicted attitudes towards affirmative action policies.

We expected to find that both types of threat would contribute to the prediction of policy attitudes. Although we had no a priori reason to predict an interaction between the participants' race and types of threat in the prediction of policy attitudes, we were curious as to whether this would be the case. It seemed plausible that we might find something akin to the asymmetry observed by Riek et al. (2006): That prejudice was more strongly predicted by realistic threat in high status rather than and low status groups; but that symbolic threat was equally relevant to high status and low status groups.

Measures

Racial Policy Attitudes

The following 11 items were used to measure support for racial policies:

  • (1)
    Using some of the national education budget for special scholarships for black children who do well in school. (compensatory)
  • (2)
    Spending more of your province's education budget on schools in largely black neighbourhoods. (compensatory)
  • (3)
    Setting up quota systems to ensure racial integration at universities and schools. (preferential)
  • (4)
    Making it easier for emerging black farmers get loans to buy land. (compensatory)
  • (5)
    Creating laws that stop farmers evicting black farm labourers. (preferential)
  • (6)
    Forcing farmers to sell land for less than it is worth to settle emerging black farmers on farms. (preferential)
  • (7)
    Special training programmes for black people so that they can compete fairly for jobs and promotion. (compensatory)
  • (8)
    Affirmation action in hiring and promoting black employees. (preferential)
  • (9)
    BEE policies, giving preferential contracts and tax breaks to black business people. (preferential)
  • (10)
    Using tax money to support emerging black artists and performers. (compensatory)
  • (11)
    Ensuring that the SABC (South African Broadcasting Corporation) gives much more TV and radio time to programmes in local black languages. (preferential)

Although these were similar to the items used in Study 1, there were three important differences. First, in Study 1 the targeted beneficiaries of the policies were labelled as either ‘blacks’ or ‘historically disadvantaged’, creating potential referential ambiguity. In order to eliminate possible measurement error arising from such ambiguity all the racial policy items in Study 2 targeted black people. Second, we substituted a 5-point Likert scale (with response options ranging from strongly agree to strongly disagree) for the 3-point scale we had used previously. Third, we made minor changes to the wording of some items and included two additional items (10 and 11) about change in the broadcasting media and the arts.

These changes improved the internal consistency of the scale, which was now acceptable for both the compensatory policies and the preferential treatment policies respectively for the white (αcomp = .81.; αpref = .76) and the black samples (αcomp = .77; αpref = .80). The means of the items, with a possible range from 0 to 4, were used as indices of support for the two policy attitudes. High scores indicate strong support.

Threat

We created an 11-item scale to measure realistic and symbolic threats. With respect to realistic threat, we used the four items devised by Bobo (1999) to measure threat in the Blumerian sense. These items, referring to competition for jobs, housing, political influence and economic resources generally, were adapted to refer to black or white South Africans. Two additional items were added. An item referring to unequal leniency in the legal system was adapted from the Racial Attitude Questionnaire devised by Walter Stephan (http://www-psych.nmsu.edu/faculty/wstephan.html). A sixth item referring to race-based exclusion from positions of power was written by the current authors.

Five items measured symbolic threat. Due to the survey length constraints, we were not able to adapt all 12 of the symbolic threat items used by Stephan and colleagues. We chose three items from the Racial Attitude Questionnaire (dealing with differences in values, and a lack of mutual understanding between groups). Two additional items (items 10 and 11 below) were written for the present survey. We felt it was important to assess perceptions of competition for status and hegemony in the domain of group culture. In the South African context, the black–white divide is commonly framed in terms of cultural differences (European/Western versus African), and both blacks and whites might potentially feel threatened by the prospect of cultural assimilation or domination by the other group.

In the survey, each item was adapted to the race of the respondent, so that black respondents were asked about threats from whites and white respondents were asked about threats from blacks. The items of the threat scale used by the white sample are listed below. Responses were given on 5-point Likert scales, with response options ranging from strongly agree to strongly disagree. Items 1–6 measure realistic threat, while items 7–11 measure symbolic threat.

  • (1)
    More good jobs for black people mean fewer good jobs for members of other groups.
  • (2)
    The more influence that black people have in South African politics the less influence other groups will have.
  • (3)
    As more good houses and neighbourhoods go to black people, the fewer good houses and neighbourhoods there will be for members of other groups.
  • (4)
    Many black people have been trying to get ahead economically at the expense of other groups.
  • (5)
    The legal system is more lenient on blacks than on whites.
  • (6)
    Blacks tend to exclude whites from positions of power and responsibility.
  • (7)
    Whites and blacks have very different values.
  • (8)
    Blacks don't understand the way that whites view the world.
  • (9)
    Most blacks will never understand what whites are like.
  • (10)
    There is pressure on white people to change their ways and fit in with black people's way of life.
  • (11)
    The traditions and values that are important to white South Africans are under threat, because of the influence of black people's values.

Principal components analysis with varimax rotation clearly separated the realistic threat and the symbolic threat items into two factors. The internal consistency of the symbolic threat scale (black α = .70; white α = .64) and the realistic threat scale (black α = .73; white α = .72) were acceptable. The means of each set of items, with a possible range from 0 to 4, were used as indices of threat, where high scores indicated high levels of threat.

Prejudice

Prejudice was measured by means of a five-item semantic differential scale (Zanna, 1994) on which respondents rated (on 10 point scales) how they felt about members of the outgroup: (1) Negative–Positive, (2) Cold–Warm, (3) Hostile–Friendly, (4) Suspicious–Trusting, (5) Disrespect–Respect. The scales were unidimensional and internally consistent (black α = .81; white α = .90). Mean scores were computed and then linearly transformed to create an index of outgroup prejudice ranging from low (0) to high (4) prejudice.

Sense of Violated Entitlement

As in Study 1, sense of violated entitlement was operationalized in terms of perceptions of relative inequity (cf. Duckitt and Mphuthing, 1998). Respondents rated the social and economic life circumstances of blacks and whites on the Cantril (1965) self-anchoring scale (steps from 1 to 10). For each group they indicated where they thought blacks and whites stand today, and where they should be placed if they had their ‘fair and rightful share of wealth in the country’. First, perceptions of ingroup and outgroup inequity were calculated by subtracting the present status from the justly entitled status for each group. Sense of violated entitlement was then calculated by subtracting the outgroup inequity score from the ingroup inequity score. High scores thus indicated perceptions of more ingroup inequity relative to outgroup inequity, that is, a high sense of violated entitlement. A linear transformation was applied so that the scores had a possible range from 0 to 4.

Strength of Identification

A three-item scale with a 5-point Likert response format (options ranging from strongly agree to strongly disagree) was used to measure strength ingroup identification: (1) Being [black/white] is an important part of who I am, as a person. (2) I feel good about being [black/white]. (3) I feel strong ties with [blacks/whites] as a group. These items were drawn from commonly used existing measures (e.g. Ellemers, Spears, & Doosje, 1997; Jetten, Postmes, & McAuliffe, 2002; Luhtanen & Crocker, 1992; Sellers, Rowley, Chavous, Shelton, & Smith, 1997), and have been piloted in South Africa (Eaton, 2006). Alphas were acceptable (black α = .81; white α = .71). Mean scores were computed and then linearly transformed to create an index of strength of ingroup identification ranging from weak (0) to strong ingroup identification (4).

Results

The pattern of group mean differences reported in Table 4 replicated the findings of Study 1. Support for transformation policies was much stronger among the black than the white sample. As expected these differences were especially strong for the preferential treatment policies, whose effect size was .53. The black respondents also reported higher levels of threat—especially symbolic threat—prejudice and violated entitlement than the white respondents. On average, whites perceived no difference in the current levels of black and white inequity whereas blacks believed their position was more inequitable than whites and hence had a higher sense of violated entitlement. Black respondents also reported higher levels of ingroup identification than white respondents.

Table 4. Descriptive scale statistics for white and black samples, Study 2
 Observed rangeWhite sampleBlack sampletdfp<η2
MeanSDMeanSD
Policy attitude (compensatory)0–42.58.783.31.6424.592479.0001.20
Policy attitude (preferential)0–41.49.843.15.6752.702478.0001.53
Threat (symbolic)0–42.12.763.03.7328.462478.0001.25
Threat (realistic)0–42.26.782.72.8312.822478.0001.06
Prejudice0–41.18.711.77.8516.602482.0001.10
Violated entitlement.89–41.93.322.50.3935.172417.0001.34
Ingroup identification0–43.03.843.65.5122.552473.0001.17

The correlations between policy support and the predictor variables threat, prejudice, violated entitlement and identification are reported in Table 5. For the white sample these correlations (reported below the diagonal) replicated and extended the findings of Study 1: Racial transformation policies were supported by whites who experienced low levels of intergroup threat, were less prejudiced, had a low sense of violated entitlement, and who reported lower levels of ingroup identification.

Table 5. Correlations between measures for the black sample (above diagonal) and white sample (below diagonal), Study 2
 Policy attitudes (compensatory)Policy attitudes (preferential)Threat (symbolic)Threat (realistic)PrejudiceViolated entitlementIngroup identification
  • *

    p < .01;

  • **

    p < .0001.

Policy attitudes (compensatory)1.70**.34**.30**.06−.04.46**
Policy attitudes (preferential).51**1.43**.39**.14**.01.39**
Threat (symbolic)−.21**−.31**1.51**.20**.13**.30**
Threat (realistic)−.33**−.38**.51**1.27**.11**.19**
Prejudice−.30**−.36**.43**.40**1.15**.07*
Violated entitlement−.26**−.20**.31**.35**.25**1−.02
Ingroup identification−.12*−.26**.27**.23**.26**.11*1

The results for the black sample present a contrasting picture to the results for the white sample, but largely replicated the findings of Study 1. Sense of violated entitlement did not predict policy support. Prejudice against whites was positively correlated with support of preferential treatment policies, but prejudice was not correlated with support of compensatory policies. However, threat and ingroup identification were strongly correlated with support of both policies among the black sample. In contrast to the results of Study 1, the correlations between the threat and policy support indices were positive, indicating that the policies attracted greatest support from black participants who experienced most threat. This is as we had originally hypothesized: Transformation policies are supported by black respondents who experience high levels of group threat, who are prejudiced against whites, and who strongly identified with the black ingroup.

The compensatory and preferential treatment policy attitudes were strongly correlated with each other for both the black and the white samples, and they had very similar correlations with all the predictor variables. The exception was ingroup identification, which was more strongly correlated with preferential treatment than compensatory policy attitudes for the white sample. The results for the two threat indices showed similar consistency. For both samples, the measures of symbolic and realistic threat were strongly correlated with each other, and they had similar correlations with other predictor variables.

We conducted path analysis to determine whether the pattern of intercorrelations conformed to the expectations of Blumer's group position model. We conducted a series of four separate analyses for each group, testing the differential impact of symbolic versus realistic threat, along with the other predictors, on either compensatory or preferential treatment policy attitudes. The results of the analyses are reported in Figures 3 and 4. The results show noteworthy differences across the models for the black and white samples, and they suggest that the distinction between preferential treatment and compensatory policy attitudes does matter.

thumbnail image

Figure 4. Model of predictors of policy attitudes, black sample, Study 2. Note: The non-standardized regression coefficients are reported for four different models in which the index of threat and policy attitudes varied. The paths that are represented with one or two effects have identical effects in the other models

Download figure to PowerPoint

thumbnail image

Figure 3. Model of predictors of policy attitudes, white sample, Study 2. Note: The non-standardized regression coefficients are reported for four different models in which the index of threat and policy attitudes varied. The paths that are represented with one or two effects have identical effects in the other models

Download figure to PowerPoint

As was the case with the zero order correlations, the regression coefficients for the predictors and policy attitudes are negative for the white sample but positive for the black sample. Black respondents who supported the policies showed higher levels of threat and ingroup identification but lower levels of violated entitlement, whereas whites who supported the policies showed lower levels of threat, ingroup identification, prejudice and violated entitlement. For the black sample, the threat effects for preferential treatment policies were stronger than the effects for compensatory policies (p < .01), but there were no differences between the coefficients for symbolic and material threat for the models of the two types of policy attitudes. For the white sample, the regression coefficients for realistic threat were lower than the regression coefficients for symbolic threat (p < .0001), and in fact the path between realistic threat and compensatory policy attitudes was not significant. For this model it was violated entitlement rather than threat that predicted compensatory policy attitudes. In fact, for both samples, the violated entitlement effects are significant for the models of compensatory policy attitudes, but—with the one exception—are not significant for models of preferential treatment policy attitudes. Also the regression coefficients for the paths between violated entitlement and policy attitudes are lower for the black than the white samples (p < .05). In sum, it appears as though both intergroup threat and violated entitlement—the proximal variables of group position theory—are important features of the models. However, threat appears to be more important in the model for the competitive and threatening preferential policy attitudes, whereas violated entitlement features significantly in the models for compensatory policy attitudes. Also, the effects for violated entitlement are larger for the white than the black sample.

No direct effects between prejudice and policy attitudes were significant for any of the models for the black sample, but they were all significant of the white sample. For both samples, prejudice had a significant indirect effect on preferential treatment policy attitudes, through symbolic (Black: Sobel z = 6.75, p < .0001; White: Sobel z = −5.83, p < .0001) and realistic threat (Black: Sobel z = 8.90, p < .0001; White: Sobel z = −3.78, p < .001). Although realistic threat did not mediate prejudice in the white model for compensatory policy attitudes, prejudice had a significant indirect effect on compensatory policy attitudes, through symbolic threat (Black: Sobel z = 6.11, p < .0001; White: Sobel z = −5.16, p < .0001) and realistic threat (Black: Sobel z = 7.76, p < .0001). Although violated entitlement did not serve as a mediator of the prejudice effect for preferential treatment attitudes among the black or white samples, it did mediate prejudice for both the symbolic threat (Black: Sobel z = −2.82, p < .005; White: Sobel z = −3.63, p < .001) and realistic threat (Black: Sobel z = −2.61, p < .01; White: Sobel z = −4.28, p < .0001) models of compensatory policy attitude models. Overall, the prejudice effects for the black sample did largely conform to the expectations of group position theory, which considers prejudice a ‘distal variable’: The significant zero order correlation between prejudice and policy attitudes became a non-significant direct effect in the two models: The effect of prejudice was entirely mediated by the threat indices and violated entitlement for the model of compensatory policy attitudes, and was entirely mediated by the threat indices only for preferential treatment policy attitudes. This replicated the findings of Study 1. In contrast, as was the case too in Study 1, the prejudice effects did not conform to the expectations of group position theory for the white sample. Prejudice continued to have a substantial direct effect on both compensatory and preferential treatment policy attitudes in the models. In addition to these direct effects, violated entitlement mediated the prejudice effects for the compensatory policy attitude model and both threat indices mediated the effects of prejudice for preferential treatment policy attitudes.

Whereas prejudice played a stronger role than ingroup identification in white compensatory (standardized total effect: prejudice β = −.29; identification β = −.12) and preferential treatment policy attitudes (prejudice β = −.31; identification β = −.25), ingroup identification had a greater impact on black attitudes towards both compensatory (prejudice β = .03; identification β = .46) and preferential treatment policies (prejudice β = .11; identification β = .39). For the black sample, ingroup identification had direct effects with all models of policy attitudes. It also had significant indirect effects on policy attitudes (all p < .0001), through symbolic (compensatory: Sobel z = 8.11; preferential: Sobel z = 9.84) and realistic threat (compensatory: Sobel z = 6.16; preferential: Sobel z = 6.70), but not though prejudice or violated entitlement. For the white sample, ingroup identification affected compensatory policy attitudes indirectly both through prejudice (symbolic: Sobel z = −4.03, p < .0001; realistic: Sobel z = −4.56, p < .0001) and symbolic threat (Sobel z = −3.23, p < .01). Ingroup identification affected preferential treatment policy attitudes indirectly both through prejudice (symbolic: Sobel z = −4.5, p < .0001; realistic: Sobel z = −4.81, p < .0001) and threat (symbolic: Sobel z = −3.36, p < .001; realistic: Sobel z = −3.08, p < .01).

The fit statistics for the group position models represented in Figures 3 and 4 are reported in Table 6. Although the goodness of fit statistics indicated good fit for all models, these are influenced by sample size and are misleading. The other model fit statistics all suggest that fit was poor for all models except the models for realistic threat for the black sample, where the RMSEA statistic of .077 suggests a barely adequate fit.

Table 6. Group position model fit statistics
 White sampleBlack sample
 χ2GFIRMSEAAICχ2GFIRMSEAAIC
  1. Note: AIC statistics for saturated models are reported below the statistics for the default model.

Preferential-symbolicχ2 = 63.04.970.28091.0χ2 = 22.05.995.11250.1
df = 1, p < .0001  30.0df = 1, p < .0001  30.0
Preferential-realisticχ2 = 44.07.979.23372.1χ2 = 10.92.997.07738.9
df = 1, p < .0001  30.0df = 1, p < .001  30.0
Compensatory-symbolicχ2 = 63.04.970.28091.0χ2 = 22.05.995.11250.1
df = 1, p < .0001  30.0df = 1, p < .0001  30.0
Compensatory-realisticχ2 = 44.07.979.23372.1χ2 = 10.92.997.07738.9
df = 1, p < .0001  30.0df = 1, p < .001  30.0

Since the data did not support the proposed group position model of policy attitudes, we ran a series of path analyses to identify best fit models. These are represented in Figure 5 and the model fit statistics are reported in Table 7. Aside from the RMSEA statistic for the preferential treatment models for the white sample, the statistics show that the models provide very good fit to the data, and all represent substantial improvements over the group position models.

thumbnail image

Figure 5. Best fit models policy attitudes. Note: With the exception of the model for white compensatory policy attitudes, the path coefficients are identical to those reported in Figures 3 and 4

Download figure to PowerPoint

Table 7. Best model fit statistics
 White sampleBlack sample
 χ2GFIRMSEAAICχ2GFIRMSEAAIC
  1. Note: AIC statistics for saturated models are reported below the statistics for the default model.

Preferential-symbolicχ2 = 16.62.990.14034.6χ2 = 5.39.998.05123.4
df = 1, p < .0001  20.0df = 1, p < .02  20.0
Preferential-realisticχ2 = 17.4.989.14435.4χ2 = 1.301.00.01319.3
df = 1, p < .0001  20.0df = 1, p < .225  20.0
Compensatory-symbolic Compensatory-realisticχ2 = 2.31 df = 2, p < .307 χ2 = 2.31 df = 2, p < .307.999.01518.4 20.0χ2 = .250 df = 1, p < .6171.00.00018.2 20.0
.999 .01518.4 20.0χ2 = 2.18 df = 1, p < .139.999.00018.2 20.0

The best fitting models for the black sample can be labelled threat models. Dropping violated entitlement and removing the direct effect from prejudice to policy attitudes improved the fit of all the models. The data for the white sample supported different models for compensatory and preferential treatment policy attitudes. The best fitting model for preferential treatment policy attitudes is similar to the threat model that described the black sample's policy attitudes, with the addition of a direct effect between prejudice and policy attitudes and the removal of the direct effect between identification and policy attitude. In contrast, the threat model provided poor fit for compensatory policy attitudes, which were better described by (what we called) a violated entitlement model.

Discussion

The results of Study 2 confirmed and extended the findings of Study 1. We were able to reliably measure policy attitudes and its predictors in a general survey of the South African population. The improved measures used in Study 2 had the effect of increasing our effect sizes, and the results confirmed that the predictors of policy attitudes in the South African context are similar to those reported in the literature elsewhere. Threat and prejudice were the strongest predictors of policy attitudes among the white sample, and threat and ingroup identification were the strongest predictors among the black sample. As we hypothesized, although the same set of variables predicted the policy attitudes of the black and white samples, the effects were in the opposite directions.

Path analysis showed a different structure of predictors of compensatory and preferential treatment policy attitudes. Threat was an especially strong predictor of preferential treatment policy attitudes. Sense of violated entitlement featured more prominently in the models for compensatory than preferential policy attitudes, and surpassed the effects for threat in the compensatory policy attitudes model for the white sample. By extrapolating from the findings of Riek et al. (2006), we had suggested that realistic threat may have stronger effects than symbolic threat for the high status (white) sample, but that the effects of symbolic and realistic threat should not differ significantly for the low status (black) sample. Our data did not support this tentative hypothesis. Although there were no differences between the effects for symbolic and realistic threat for the black sample; for the white sample, the difference between the regression coefficients between the threat variables and policy attitudes was in the opposite direction: Symbolic threat was a stronger predictor of both kinds of policy attitudes.

The results of Study 2 suggested that both symbolic and realistic threat play significant roles in all of the models, except for realistic threat in the white compensatory policy attitude model. However, the results do not provide strong grounds for distinguishing between the two kinds of threat, which appear to have similar effects on policy attitudes. Not only were the two threat variables strongly correlated, but distinguishing between them made little substantive difference to the predictive models. To further explore the differential impact of symbolic and realistic threat it may be worthwhile in future research to include individual difference variables such as right-wing authoritarianism or social dominance orientation. According to Duckitt's (2001) dual process model of prejudice, people high in right-wing authoritarianism ought to be particularly influenced by perceptions of symbolic threat, while people high in social dominance orientation would be expected to be more strongly influenced by perceived competition over resources (i.e. realistic threat). How these individual-level variables interact with group membership in the South African context remains to be seen.

Our data provided very little support for group position theory. Although the group position variables—threat and violated entitlement—did predict transformation policies, they did not entirely mediate the effects of the ‘distal’ variables in the models. Prejudice and ingroup identification continued to have strong independent effects on policy attitudes for the white and black samples respectively. Path analysis showed that the group position models fitted the data poorly. Model fit was substantially improved by dropping violated entitlement from the models of the black sample. For the white sample, a similar threat and prejudice based model best fit the data for preferential treatment policy attitudes, but the model for compensatory policy attitudes fitted best when threat was dropped and violated entitlement was included. Thus, the two central variables of group position theory appear to have a differential impact on opinions about social transformation policies, depending on the kind of policies under consideration, and the group identity of the person doing the evaluation.

CONCLUSION

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

Like many other places in the world, South Africa is in the process of profound socio-political transformation. A comprehensive set of policies has been implemented in an attempt to eradicate the legacy of apartheid and race-based privilege. Since these policies are often viewed as promoting the interests of one group at the expense of others, they are predictably highly contested. However, in addition to expected group differences in support for the policies, there are also substantial individual differences in support for the policies within each group. The purpose of this paper was to apply social psychology theory to understand the sources of support for or opposition to racial transformation policies in South Africa.

Our first aim was to determine whether variables that had been tested elsewhere were related to policy attitudes in the South African context. We expected these social psychological factors to play a role in South Africa because of the recent history of apartheid and the racism, threat and conflict that continues to characterize intergroup relations in the context where the racial order of social stratification is being challenged. The surveys reported in this paper show that ingroup identification, prejudice, and a sense of group threat and violated entitlement all play a role in shaping policy attitudes in South Africa.

Policy attitude research has largely been confined to the study of majority groups who are potentially threatened by the policies, investigating, for example, white people's attitudes to affirmative action. In general, this research has sought to understand sources of white people's resistance to change, highlighting the ongoing role of racism. In South Africa, in contrast, although whites are a historically dominant group, they are a numerical minority, and are thus potentially more strongly threatened by these policies than white samples in the US or Europe. At the same time, despite political change, economic change has been very slow and the legacy of apartheid persists in the form of extreme material inequalities between blacks and whites. Although a small proportion of the black population has moved into the middle class (Seekings & Natrass, 2006) the black majority continue to live in extreme poverty characterized by unemployment, lack of formal housing, and poor health care and education. South Africa's GINI index of inequality is among the highest in the world and continues to grow despite political change. In this context, in addition to whites' attitudes, it is essential to understand the sources of resistance to transformation among blacks, who make up the ruling class and have the power to implement change.

A second aim of our research, therefore, was to expand our focus to include an analysis of the policy attitudes of a historically disadvantaged group who are prospective beneficiaries of the policies, and to compare the predictors of policy attitudes among black and white samples. The results of both of our surveys confirmed that a similar set of variables is related to the policy attitudes of historically low and high status groups. Predictably, however, the effects were in the opposite direction. The results for our white sample were essentially the same as the results of the US research: The correlations reported in Tables 3 and 5 show that transformation policies were opposed by prejudiced whites who experienced a threat from blacks and who felt that in contrast to white losses, blacks had made more gains than they really deserved. These results conform to the expectation that outgroup prejudice underpins opposition to transformation. In contrast, among black respondents who are prospective beneficiaries of the policies, high levels of these same variables—threat (in Study 2), ingroup identification, (anti-white) prejudice and violated entitlement—were associated with support for the transformation policies. These findings suggest that the nature of the threat or benefit that these policies potentially hold for one's group is critical to understanding the sources and structure of attitudes towards the policies.

For most of its history the literature on racial policy attitudes has pitted one explanation against others. Early work by McConahay (1982) and Kinder and Sears (1981) showed that prejudice was a stronger predictor of opposition to desegregation policies than self-interest. Social psychological factors seemed to play a greater role than structural factors in shaping responses to changing intergroup relations. Since that time evidence has accumulated to show that group based-threats to self-interest, prejudice as well as a number of beliefs about violated group entitlement and social stratification are related to policy attitudes (see Krysan, 2000). More recently Bobo and Tuan (2006) have attempted to test multivariate theoretical models of the predictors of policy attitudes. In particular, they used Blumer's group position theory to differentiate distal and proximal predictors. In this view, the group position variables of threat and violated entitlement are immediate proximal predictors of policy attitudes whereas ingroup identification, like prejudice, is a distal factor, a ‘subsumed element’ or a minor tributary of ‘the larger flow of the great river that is the struggle for group position’ (Bobo & Tuan, 2006, p. 134).

Our third aim was to determine whether our data conformed to the expectations of group position theory. Certainly, there were aspects of our findings that where anticipated by group position theory: Intergroup threat was centrally important in the various models in both Study 1 and Study 2, and did function as a ‘proximal’ variable that entirely mediated the effects of the ‘distal’ variables, namely, prejudice (for the black sample) and ingroup identification (for the white sample).

Overall, however, our results did not confirm the model. There were two striking results which were not anticipated by group position theory. First, the ‘distal’ variables—prejudice for the white sample (Study 1 and 2) and ingroup identification for the black sample (Study 2)—continued to have strong direct effects on policy attitudes over and above the mediated effects through the group position variables. These results may very well reflect the history and dynamics of intergroup relations in South Africa. The extreme white racism of the apartheid years may be having a lingering effect, accounting for the independent role that prejudice plays in shaping opposition to transformation policies among whites. Also, since the perception (and reality) of inequitable social stratification is so pervasive, ingroup identification among black South Africans could shape an overwhelming sense of need for transformation that impacts on policy attitudes regardless of other factors.

The second result which was not anticipated by group position theory was the differential effect that threat and violated entitlement had for the different models. Sense of violated entitlement did not significantly predict policy attitudes for both the white or black models in Study 1, and the group position models did not provide a good fit to the data in Study 2. Dropping violated entitlement improved the fit of models, suggesting that it is a sense of intergroup threat rather than a fuller sense of intergroup position that shapes policy attitudes in the South African context. As we stressed in the introduction to this paper, the history of apartheid and ongoing race-based inequality make for a very threatening intergroup scenario both for whites who are a numerical minority and for blacks who continue to bear the brunt of socio-economic inequality despite being the ruling political class. So perhaps our findings are attributable to the particular constellation of threats that characterize the South African context.

On the other hand, our data indicate that sense of violated entitlement remains an important social psychological element in policy evaluation. For both the black and white samples in Study 2, violated entitlement significantly predicted compensatory policy attitudes. Indeed, the best fitting model for compensatory attitudes for the white sample included violated entitlement but not threat. It thus appears as though threat is more relevant to the consideration of competitive policies that are constructed as favouring one group at the expense of the other. However, violated entitlement appears to be more relevant to the evaluation of policies that are constructed in non-competitive intergroup terms, as helping to support historically disadvantaged groups. In other words, the nature and construction of social transformation policies may not only have an impact on the level of support for the policies as has been shown (see e.g. Fine, 1992; Kravitz, 1995, Kravitz et al., 2000) but also have an impact on the underlying social psychological factors that influence such support. This result is not anticipated in the literature. Not only is there a dearth of research data addressing this question, but also the current wisdom suggests that ‘the process by which support for these disparate policies is created does not vary by policy type’ (Tuch & Hughes, 1996, p. 738).

The results of our research have a number of implications for social change. Among historically dominant groups, such as our white sample, gaining support for policies will depend not only on the construction of the policies as involving compensation or preferential treatment, but will also involve concomitant work of presenting of compensatory policies in terms of just entitlements, and preferential treatment policies as non-threatening. Ongoing work to challenge racial prejudice is also a necessary prerequisite for gaining acceptance for social transformation policies.

The implications of our data for fostering social change among historically disadvantaged groups are perplexing. Among the black sample, high levels of intergroup threat and ingroup identification were associated with support for the transformation policies. Thus, while policies should be presented as non-threatening to whites, to gain support, they should be presented in ways that underscore a sense of intergroup threat among blacks.

Although Blumer was primarily interested in the prejudice of dominant groups, his model allows us to draw important distinctions between the intergroup threat experienced by dominant and subjugated groups. Intergroup threat among dominant groups is supported by racial prejudice as was evident in our data. However, the threat that predicted policy support among our black respondents was not associated with prejudice, but with a sense of ingroup identification. Thus attempts to challenge the order of social stratification in South Africa should target group-specific forms of threat, reducing threats associated with racial prejudice among whites, and increasing threats associated with identifying with a historically oppressed group among blacks. The kinds of threats we are dealing with here are not primarily abstract psychological states and emotions. Rather, they are historically and structurally specific concerns, which in South Africa are associated with a sense of entitlement among whites and of historical oppression, brutality and social injustice among blacks.

Acknowledgements

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES

This research was funded by grants awarded to the first author by the NRF and the Oppenheimer Foundation in South Africa, and to the second author by the ESRC in the United Kingdom. We thank Ask Afrika for supplying and screening random cell phone numbers with their autodialler.

  • 1

    Affirmative action policies in South Africa were effected though the following legislation: The Prevention of Unfair Discrimination Act (Act 4 of 2000), the Employment Equity Act (Act 55 of 1998) and the Broad Based Black Economic Empowerment Act (Act 53 of 2003). The land reform program was supported by the Restitution of Land Rights Act (Act 22 of 1994) and the Land Reform (Labour Tenants) Act (Act 3 of 1996).

REFERENCES

  1. Top of page
  2. Abstract
  3. RACIAL PREJUDICE
  4. THREAT
  5. SENSE OF VIOLATED ENTITLEMENT
  6. STUDY 1
  7. STUDY 2
  8. CONCLUSION
  9. Acknowledgements
  10. REFERENCES
  • Ashforth, A. (1990). The politics of official discourse in twentieth century South Africa. Oxford: Clarendon.
  • Blumer, H. (1958). Race prejudice as sense of group position. The Pacific Sociological Review, 1, 37.
  • Bobo, L. (1983). Whites' opposition to busing: Symbolic racism or realistic group conflict? Journal of Personality and Social Psychology, 45, 11961210.
  • Bobo, L. (1999). Prejudice as group position: Microfoundations of a sociological approach to racism and race relations. Journal of Social Issues, 55, 445472.
  • Bobo, L., & Hutchings, V. L. (1996). Perceptions of racial group competition: Extending Blumer's theory of group position to a multiracial social context. American Sociological Review, 51, 950962.
  • Bobo, L., & Kluegel, J. R. (1993). Opposition to race-targeting: Self-interest, stratification ideology or racial attitudes? American Sociological Review, 58, 443464.
  • Bobo, L., & Tuan, M. (2006). Prejudice in politics: Group position, public opinion, and the Wisconsin treaty rights dispute. Cambridge: Harvard University Press.
  • Cantril, H. (1965). The pattern of human concerns. New Brunswick, NJ: Rutgers University Press.
  • Citrin, J., Green, D. P., Muste, C., & Wong, C. (1997). Public opinion toward immigration reform: The role of economic motivations. The Journal of Politics, 59, 858881.
  • Conover, P. J. (1984). The influence of group identifications on political perception and evaluation. Journal of Politics, 46, 760785.
  • Crankshaw, O. (1997). Race, class and the changing division of labour under apartheid. London: Routledge.
  • Crosby, F. J., Ferdman, B. M., & Wingate, B. R. (2001). Addressing and redressing discrimination: Affirmative action in social psychological perspective. In R.Brown, & S. L.Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 495513). Malden, MA: Blackwell.
  • Duckitt, J. (2001). A dual-process cognitive-motivational theory of ideology and prejudice. In M. P.Zanna (Ed.), Advances in experimental social psychology (Vol. 33, pp. 41113). New York: Academic Press.
  • Duckitt, J., & Mphuthing, T. (1998). Political power and race relations in South Africa: African attitudes before and after the transition. Political Psychology, 19, 809832.
  • Durrheim, K. (2003). White opposition to racial transformation: Is it racism? South African Journal of Psychology, 33, 241249.
  • Eaton, L. (2006). National identity in the ‘new’ South Africa: A social-psychological investigation. Unpublished Doctoral Dissertation, University of Cape Town, Cape Town, South Africa.
  • Ellemers, N., Spears, R., & Doosje, B. (1997). Sticking together of falling apart: In-group identification as a psychological determinant of group commitment versus individual mobility. Journal of Personality and Social Psychology, 72, 617628.
  • Fine, T. S. (1992). The impact of issue framing on public opinion: Toward affirmative action programs. The Social Science Journal, 29, 323334.
  • Funk, C. L. (2000). The dual influence of self-interest and societal interest in public opinion. Political Research Quarterly, 53, 3762.
  • Greeley, A. M., & Sheatsley, P. B. (1971). Attitudes towards racial integration. Scientific American, 225, 1319.
  • Green, D. P., & Cowden, J. A. (1992). Who protests: Self-interest and white opposition to busing. The Journal of Politics, 54, 471496.
  • Habib, A. & BentleyK. (Eds.), (2008). Racial redress and citizenship in South Africa. Cape Town: HSRC Press.
  • Jacobson, C. K. (1985). Resistance to affirmative action: Self-interest or racism? Journal of Conflict Resolution, 29, 306329.
  • Jetten, J., Postmes, T., & McAuliffe, B. J. (2002). ‘We're all individuals’: Group norms of individualism and collectivism, levels of identification and identity threat. European Journal of Social Psychology, 32, 189207.
  • Kinder, D. R. (1986). The continuing American dilemma: White resistance to racial change 40 years after Myrdal. Journal of Social Issues, 42, 151171.
  • Kinder, D. R., & Sears, D. O. (1981). Prejudice and politics: Symbolic racism versus racial threats to the good life. Journal of Personality and Social Psychology, 40, 414431.
  • Kluegel, J. R. (1985). ‘If there isn't a problem you don't need a solution’: The bases of contemporary affirmative action attitudes. American Behavioral Scientist, 28, 761784.
  • Kluegel, J. R. (1990). Trends in whites' explanations of the black-white gap in socio-economic status, 1977–1989. American Sociological Review, 55, 512525.
  • Kluegel, J. R., & Smith, E. R. (1982). Whites beliefs about blacks' opportunity. American Sociological Review, 47, 518532.
  • Kluegel, J. R., & Smith, E. R. (1983). Affirmative action attitudes: Effects of self-interest, racial affect and stratification beliefs on whites' views. Social Forces, 61, 797824.
  • Kluegel, J. R., & Smith, E. R. (1986). Beliefs about inequality: Americans' views about what is and what ought to be. New York: Aldine de Gruyter.
  • Kravitz, D. A. (1995). Attitudes toward affirmative action plans directed at blacks: Effects of plan and individual differences. Journal of Applied Social Psychology, 24, 21922220.
  • Kravitz, D. A., Klineberg, S. L., Avery, D. R., Nguyen, A. K., Lund, C., & Fu, E. J. (2000). Attitudes toward affirmative action: Correlations with demographic variables and with beliefs about targets, actions and economic effects. Journal of Applied Social Psychology, 30, 11091136.
  • Krysan, M. (2000). Prejuduce, politics, and public opinion: Understanding the sources of racial policy attitudes. Annual Review of Sociology, 26, 135168.
  • Lipset, S. M., & Schneider, W. (1978). The Bakke case: How would it be decided at the bar of public opinion? Public Opinion, March/April, 3844.
  • Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self evaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302318.
  • McClendon, M. J. (1985). Racism, rational choice, and white opposition to racial change: A case study of busing. Public Opinion Quarterly, 49, 214233.
  • McConahay, J. B. (1982). Self-interest versus racial attitudes as correlates of anti-busing attitudes in Louisville: Is it the busses or the blacks? Journal of Politics, 40, 692729.
  • McConahay, J. B., & Hough, J. C. (1976). Symbolic racism. Journal of Social Issues, 32, 2345.
  • McLaren, L. M. (2002). Public support for the European Union: Cost/benefit analysis or perceived cultural threat. Journal of Politics, 64, 551566.
  • McLaren, L. M. (2003). Anti-immigrant prejudice in Europe: Contact, threat perception, and preferences for the exclusion of migrants. Social Forces, 81, 909936.
  • McLaren, L. M., & Johnson, M. (2007). Resources, group conflict, and symbols: Explaining anti-immigration hostility in Britain. Political Studies, 55, 709732.
  • Pedersen, A., & Walker, I. (1997). Prejudice against Australian aborigines: Old-fashioned and modern forms. European Journal of Social Psychology, 27, 561587.
  • Peffley, M., Hurwitz, J., & Sniderman, P. M. (1997). Racial stereotypes and whites' political views of blacks in the context of welfare and crime. American Journal of Political Science, 41, 3060.
  • Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in Western Europe. European Journal of Social Psychology, 25, 5775.
  • Pillay-Singh, N., & Collings, S. (2004). Racism on a South African campus: A survey of students' experiences and attitudes. Social Behaviour and Personality: An International Journal, 32, 607618.
  • Pincus, F. L. (2001). The social construction of reverse discrimination: The impact of affirmative action on whites. Journal of Intergroup Relations, 38, 3344.
  • Renfro, C. L., Duran, A., Stephan, W. G., & Clason, D. L. (2006). The role of threat in attitudes toward affirmative action and its beneficiaries. Journal of Applied Social Psychology, 36, 4174.
  • Riek, B. M., Mania, E. W., & Gaertner, S. L. (2006). Intergroup threat and outgroup attitudes: A meta-analytic review. Personality and Social Psychology Review, 10, 336353.
  • Schmermund, A., Sellers, R., Mueller, B., & Crosby, F. (2001). Attitudes toward affirmative action as a function of racial identity among African American college students. Political Psychology, 22, 759774.
  • Sears, D. O. (1988). Symbolic Racism. In P. A.Katz, & D. A.Taylor (Eds.), Eliminating racism: Profiles in controversy. New York: Plenum.
  • Sears, D. O. (2001). The role of affect in symbolic politics. In J. H.Kuklinski (Ed.), Citizens and politics: Perspectives from political psychology. Cambridge: Cambridge University Press.
  • Sears, D. O., & Allen, H. M. Jr., (1984). The trajectory of local desegregation controversies and whites' opposition to busing. In N.Miller, & M. B.Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 123151). New York: Academic Press.
  • Sears, D. O., & Funk, C. L. (1991). The role of self-interest in social and political attitudes. In M. P.Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 291). New York: Academic Press.
  • Sears, D. O., & Henry, P. J. (2005). Over thirty years later: A contemporary look at symbolic racism. Advances in Experimental Social Psychology, 37, 95150.
  • Sears, D. O., Henry, P. J., & Kosterman, R. (2000). Egalitarian values and contemporary racial politics. In D. O.Sears, J.Sidanius, & L.Bobo (Eds.), Racialized politics: The debate about racism in America (pp. 75117). Chicago: Chicago University Press.
  • Sears, D. O., van Laar, C., Carrillo, M., & Kosterman, R. (1997). Is it really racism? The origins of white Americans' opposition to race targeted policies. Public Opinion Quarterly, 61, 1653.
  • Seekings, J., & Natrass, N. (2006). Class, race and inequality in South Africa. Pietermaritzburg: University of kwaZulu-Natal Press.
  • Sellers, R. M., Rowley, S. A. J., Chavous, T. M., Shelton, J. N., & Smith, M. A. (1997). Multidimensional Inventory of Black Identity: A preliminary investigation of reliability and construct validity. Journal of Personality and Social Psychology, 73, 805815.
  • Shelton, N. J. (2000). A reconceptualisation of how we study issues of racial prejudice. Personality and Social Psychological Review, 4, 374390.
  • Sidanius, J., Pratto, F., & Bobo, L. (1996). Racism, conservatism, affirmative action, and intellectual sophistication: A matter of principled conservatism or group dominance? Journal of Personality and Social Psychology, 70, 476490.
  • Smith, H. J., Spears, R., & Hamstra, I. J. (1999). Social identity and the context of relative deprivation. In N.Ellemers, R.Spears, & B.Doosje (Eds.), Social identity: Context, commitment, content (pp. 205229). Oxford: Blackwell.
  • Sniderman, P. M., Carmines, E. G., Layman, G. C., & Carter, M. (1996). Beyond race: Social justice as a race neutral idea. American Journal of Political Science, 40, 3355.
  • Sniderman, P. M., & Piazza, T. (1993). The scar of race. Cambridge, MA: Harvard University Press.
  • Sniderman, P. M., & Tetlock, P. E. (1986). Symbolic racism: Problems of motive attribution in political debate. Journal of Social Issues, 42, 129150.
  • South African Advertising Research Foundation. (2007). http://www.saarf.co.za/
  • Stephan, W. G., & Renfro, C. L. (2002). The role of threat in intergroup relations. In D. M.Mackie, & E. R.Smith (Eds.), From prejudice to intergroup emotions: Differentiated reactions to social groups++ (pp. 191207). New York: Psychology Press.
  • Stephan, W. G., Renfro, C. L., & Davis, M. (2008). The role of threat in intergroup relations. In U.Wagner, L. R.Tropp, G.Finchilescu, & C.Tredoux (Eds.), Improving intergroup relations: Building on the legacy of Thomas F. Pettigrew (pp. 5672). Malden, Massachusetts: Blackwell.
  • Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S.Oskamp (Ed.), Reducing prejudice and discrimination: The Claremont symposium on applied social psychology (pp. 2345). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
  • Steyn, M. (2001). ‘Whiteness just isn't what it used to be’: White identity in a changing South Africa. Albany, NY: State University of New York Press.
  • Summers, R. J. (1995). Attitudes toward different methods of affirmative action. Journal of Applied Social Psychology, 25, 10901104.
  • Tougas, F., Brown, R., Beaton, A. M., & Joly, S. (1995). Neosexism: Plus Ça Change, plus C'est Pareil. Personality and Social Psychology Bulletin, 21, 842849.
  • Tougas, F., & Veilleux, F. (1988). The influence of identification, collective relative deprivation, and procedure of implementation on women's response to affirmative action: A causal modeling approach. Canadian Journal of Behavioral Science, 20, 528.
  • Tropp, L. R., & Wright, S. C. (1999). Ingroup identification and relative deprivation: An examination across multiple social comparisons. European Journal of Social Psychology, 29, 707724.
  • Tuch, S. A., & Hughes, M. H. (1996). Whites' racial policy attitudes. Social Science Quarterly, 77, 723774.
  • Virtanen, S. V., & Huddy, L. (1998). Old-fashioned racism and new forms of racial prejudice. The Journal of Politics, 60, 311332.
  • Wenzel, M. (2004). Social identification as a determinant of concerns about individual-, group-, and inclusive-level justice. Social Psychology Quarterly, 67, 7087.
  • Zanna, M. P. (1994). On the nature of prejudice. Canadian Psychology, 35, 1123.