CONSENSUAL SOCIAL DOMINANCE ... - ScienceDirect.com

17 downloads 0 Views 1MB Size Report
Items 9 - 16 - space in which high- and low-status group members have the same mean ..... (14) Working as a lawyer for the state government prosecuting those.
fnt. J. Intrrculturcrl

Rel. Vol. 20. No. 3/4, pp. 385-408, 1996 Copyright @ 1996 Elsevier Science Ltd Printed in Great Bntain. All rights reserved Ol47-1767/96 $15.00 + 0.00

Pergamon

SO147-1767(96)00025-9

CONSENSUAL SOCIAL DOMINANCE ORIENTATION AND ITS CORRELATES WITHIN THE HIERARCHICAL STRUCTURE OF AMERICAN SOCIETY

JIM SIDANIUS

and SHANA

LEVIN

UCLA FELICIA

PRA TTO

Stanford

Univrrsity

The concepts of “false consciousness”, “the political formula”, and “ideological hegemony” have been around for some time, but have not been measurable. We present an operational definition of these related ideas, which we call “consensual” ideology. We measure consensual ideology with respect to social dominance orientation (SDO), which is dejned as a general positive orientation towards group dominance. We define consensual SD0 as that portion of SD0 that is shared across social groups, such as different ethnic groups. We also test social dominance theory’s ideological asymmetry hypothesis using Euro- and AfricanAmerican university students. This hypothesis states that the relationship between support for group inequality and social ideologies and policies supportive of hierarchical group relations will be significantly more positive within high-status groups than within low-status groups. Consistent with theoretical expectations, even though, by operational definition, Blacks and Whites have the same mean level of consensual social dominance orientation, there was consistent evidence for the type of asymmetry assumed. The theoretical implications are discussed. Copyright 6 1996 Elsevier Science Ltd. ABSTRACT.

After more than 40 years of active and intense effort to eliminate racism and ethnocentrism from American society and realize Martin Luther King’s noble Dream, one may be more impressed by how much race relations in the United States have remained the same rather than by how much they have changed. Casual inspection of the structure of American society reveals that it is still highly segmented along social class and ethnic group lines and that this group-based hierarchical structure has been highly stable over the last quarter of a century (see Smith, 1991). Requests for reprints should be addressed to Jim Sidanius, Department of Psychology, University of California at Los Angeles, 405 Hilgard Avenue, Los Angeles, CA 90095, U.S.A. Electronic mail may be sent via Internet to [email protected]

385

386

J. Sidanius et al.

According to social dominance (SD) theory, the stability of the American ethnic hierarchy should be seen as a special case of the more general tendency for large human social systems to form and maintain group-based systems of social hierarchy (see Sidanius, Devereux, & Pratto, 1992; Sidanius & Liu, 1992; Mitchell & Sidanius, 1993; Pratto, Sidanius, & Stallworth, 1993; Sidanius, 1993; Sidanius & Pratto, 1993a,b; Pratto, Sidanius, Stallworth, & Malle, 1994; Sidanius, Pratto, & Bobo, 1994; Sidanius, Pratto, & Mitchell, 1994; Sidanius, Pratto, & Rabinowitz, 1994). Although societies differ in the severity of their group-based, hierarchical organization, SD-theory assumes that the mechanisms which help to establish and maintain these group-based hierarchies are very similar across social systems. Three proximal mechanisms are assumed to be involved: (1) “Aggregated individual discrimination”, where members of high-status groups will tend to discriminate against individual members of low-status groups, While this discrimination can be very blatant and obvious within extremely hierarchical societies, this discrimination can also be extremely subtle in less hierarchical societies. For example, the widespread social practice of allowing wealth to be inherited helps to maintain race and class inequality across generations (see Pratto, Sidanius, & Stallworth, 1993 for a discussion). (2) “Institutional discrimination”, whereby positive social value (e.g., wealth, prestige, power) will be disproportionately allocated to members of dominant groups while negative social value (e.g., fines, beatings, prison terms, death sentences) will be disproportionately allocated to members of subordinate low-status groups. (3) “Behavioral asymmetry”, or the existence of systematic differences in the behavioral repertoires of members of the social system, depending upon the position of their primary group within the social hierarchy. (For a fuller discussion of these mechanisms, see Sidanius, 1993; see also Figure 1.) A number of theorists have argued that the structural integrity of hierarchically organized social systems is facilitated by the use of social attitudes variously known as “ideology” or “false consciousness” (see Marx & Engels, 1970) “the political formula” (see Mosca, 1939) or “ideological hegemony” (see Gramsci, 1971). Although there are differences in the definitions of these terms, they have in common the notion that ideologies and ideas are often used to convince both dominant and subordinate members of the social system that hierarchical relationships between socially constructed groups (e.g., social classes, “races”) are right and good. Social dominance theory adopts this basic notion and refers to such beliefs as ‘legitimizing myths’. Within the framework of social dominance theory, legitimizing myths (LMs) are defined as values, attitudes, beliefs, causal attributions and ideologies which provide moral and intellectual justification for increasing or decreasing levels of social inequality among social groups.

Consensual SD0

II

Group Status & Socialization Experiences

387

Indiviv. Discrim. Legitimizing Myths Racism Sexism Classism Caste-Maintenance Orientation Protestant Work Ethic Divine Right of Kings Manifest Destiny Political Conservatism Etc.

Hierarchy

Deference Asymmetrical Ingroup Bias Self-Handicapping Ideological Asymmetry FIGURE 1. The general social dominance model.

Remembering that ideologies can be used to argue for inequality or equality, social dominance theory distinguishes between two types of legitimizing myths: those that enhance hierarchy and those that attenuate hierarchy (see Sidanius, 1993). Hierarchy-enhancing myths are those social ideologies and beliefs which morally or intellectually justify greater group inequality. The more widespread or “consensual” these beliefs are, the easier it is to peacefully maintain the current state of group-based social inequality without the need for naked force (see also Sidanius, Pratto, Martin, & Stallworth, 1991). Hierarchy-attenuating myths, on the other hand, have the exact opposite effect, providing intellectual and moral justification for greater levels of social equality among socially constructed groups. By the term “myth”, we do not mean that such ideologies are inherently “wrong” or “incorrect”, but simply that when such ideologies are widely shared, they take on the quality of self-evident truths. Unlike the earlier classical models of social stratification, social dominance theory has also emphasized that individuals vary significantly in their general support for stratified social systems. We have defined social dominance orientation @DO) as a very broad orientation expressing one’s general endorsement of group-based inequality. Like many other individual difference variables, it is assumed to be normally distributed and to have its etiology in both cultural/socialization and biological factors (for a more detailed discussion of this construct, see Pratto, Sidanius, Stallworth, & Malle, 1994). SD0 has been both conceptually and empirically distinguished from related constructs, such as

388

J. Sidanius et al.

interpersonal dominance, authoritarianism, nationalism, sexism, political conservatism, and ethnocentrism (see Pratto, Sidanius, Stallworth, & Malle, 1994). Combining this individual-differences analysis with socialstructural analysis, SD-theory differs from most other classical models of social stratification and psychological models of prejudice in bridging the gap between the psychology of individual differences and the dynamic, structural and hierarchical interrelationships among groups. SD-theory asserts that individual political psychology cannot be fully understood unless one attends to individual psychological attachment to groups, and also to how these groups form hierarchically organized systems of groupbased relationships.

DISSENSUAL

AND CONSENSUAL

IDEOLOGIES

One of the implications of SD-theory is that legitimizing myths (LMs) or ideologies can be conceived of as consisting of three distinct components: dissensual components, consensual components, and random error. Dissensual LMs represent that portion of the total variance of legitimizing ideologies which distinguishes between people from different strata of the group-based hierarchy. In general, one should expect members of high-status groups to be more supportive of those ideologies justifying the unequal distribution of social value in favor of high-status groups (i.e., hierarchy-enhancing LMs) than members of low-status groups. The simple reason for this expectation is that people should generally favor those ideologies that benefit their primary reference groups. For example, within hierarchically structured societies organized on the basis of “race”, such as the United States, if we define “racism” as the belief in the inherent superiority of certain “races” over others (e.g., Whites over Blacks), one should then expect Whites to be more “racist” against Blacks than Blacks are racist against Blacks (see Sidanius, Pratto, Martin, & Stallworth, 1991). The consensual component of ideologies is that portion of the LM space in which high- and low-status group members have the same mean level of LM-endorsement. In other words, the consensual LMs are those legitimizing ideologies which are shared to the same degree by members of all status groups within the social system. For example, consensual racism is that portion of the racial attitude space that racial groups share in common. Within SD-theory, consensual racism is, by far, the most important portion of racial attitude space because it represents the degree to which different “races” share common evaluations of one another, common attributions of social events and maintain a common social and “racial” cosmology. The dissensual component of racism is often regarded as the most important part because it is the component of

Consensual SD0

389

racism that divides one group from the other. However, because of its relatively strong consensual component, racism can also be regarded as a belief system which actually unites people across the racial continuum, thereby enabling members of hierarchically organized racial groups to interact with one another in a relatively peaceful manner. One of the predictions of SD-theory is that when the hierarchical social structure is relatively stable, the consensual variance of a legitimizing myth should be considerably larger than the dissensual variance of the legitimizing myth. In other words: 2 ~Consensual~,~ > &ssensual~~

This implies that the amount of variance in legitimizing myths that unites high- and low-status groups will be considerably larger than the amount of variance which divides these groups. We expect that the greater the size of the consensual variance relative to the dissensual variance, the smaller the degree of ideological and political conflict we should find between the groups. Because we assume that social dominance orientation serves the interests of hegemonic groups more than oppressed groups, and because we expect that being in prestigious social strata will increase general support for group hierarchy, we expect people in hegemonic groups to have higher SD0 levels than people in oppressed groups. Using scales to measure SDO, this pattern has been found with respect to men vs. women (Sidanius, Pratto, & Bobo, 1994; Levin & Sidanius, 1995; Sidanius, Pratto, & Brief, 1995) White Americans vs. Black Americans (Choudhury & Pratto, 1995; Levin & Sidanius, 1995), middle-class vs. workingclass Americans (Levin & Sidanius, 1995), Jewish Israelis vs. Palestinians (Levin & Sidanius, 1995), and heterosexuals vs. gays, lesbians and bisexuals (Choudhury & Pratto, 1995). Because we can measure SD0 using a scale similar to an attitude or ideological scale, we can also speak of the dissensual and consensual components of SDO. We may define dissensual SD0 as that portion of this general anti-egalitarian ideology with respect to which different groups have different average attitudes. Likewise, we may also speak of consensual SD0 as that portion of this general anti-egalitarian ideology with respect to which different groups have the same mean level. Everything else being equal, the greater the relative size of this consensual SD0 space, the less conflict we should find between high- and low-status social groups regarding general principles of resource distribution. Ideological

Asymmetry

Not only should we expect that: (1) high-status groups should have significantly higher scores with respect to hierarchy-enhancing legitimiz-

390

J. Sidanius et al.

ing myths (e.g., racism) than low-status groups, but we should also expect that (2) at least among members of high-status groups, support of these hierarchy-enhancing LMs should be positively correlated with social dominance orientation. The basic logic underlying this second expectation is that members of high-status groups will hold hierarchy-enhancing views and support hierarchy-enhancing social policies to the extent that they desire to maintain the dominant position of their primary reference group. That is, high-status group members who are also high in SD0 are more likely to espouse hierarchy-enhancing LMs because these views and policies serve to increase and/or maintain social inequality among salient socially constructed groups. Using Katz’s (1960) functional theory of attitudes, one could say that these hierarchy-enhancing attitudes serve a value-expressive function for high SD0 individuals. The value being served in this case is the value that people place on generalized, groupbased anti-egalitarianism (i.e., SDO; for a similar argument, see Rokeach, 1979). Extending Katz’s (1960) functional reasoning, we would assert that ideologies serve functions for individuals not only as individuals, but as members of groups as well. While support for hierarchy-enhancing LMs is easily seen as being positively correlated with SD0 among members of high-status groups, SD-theory does not expect this to be the case to the same extent for members of low-status groups. While hierarchy-enhancing LMs will tend to serve group dominance values for members of high-status groups, these same hierarchy-enhancing LMs are less likely to serve group dominance values for members of low-status groups. In other words, we expect that the relationship between SD0 and support for hierarchy-enhancing LMs will not be symmetrical across the social status continuum; we call this thesis the “ideological asymmetry hypothesis” (Mitchell & Sidanius, 1993; Sidanius, 1993; Sidanius, Pratto, & Rabinowitz, 1994; Sidanius, Pratto, & Levin, 1995). The general ideological asymmetry hypothesis maintains that the correlation between SD0 and any hierarchy-enhancing LM among members of high-status groups should be more positive than the correlation between SD0 and the LM among members of low-status groups. Upon careful inspection, one can detect what appear to be instances of ideological asymmetry in other studies. If we simply consider the relationship between ideologies and policy attitudes regarding group relations or ingroup preference, we find empirical support for the same kinds of asymmetrical correlations predicted by the asymmetry hypothesis. For example, Mercer and Cairns (198 1) found that political conservatism was significantly correlated with anti-Catholic attitudes among Protestants in Northern Ireland (i.e., high-status group), but was not significantly correlated with anti-Protestant affect among Northern Irish Catholics (i.e., low-status group). Similarly, Bahr and Chadwick (1974)

Consensual SD0

391

found that general political conservatism was significantly associated with anti-Native American affect among European-Americans but was not associated with anti-White affect among Native Americans. Because the death penalty is systematically used against Black Americans more than Whites (see Mitchell & Sidanius, 1993, for a review), we would expect to find similar asymmetries in support for the death penalty. Combs and Comer (1982) found that the relationship between liberalism-conservatism and death penalty support was stronger among American Whites than among American Blacks. In addition, Young (1991) found that the relationship between crime attributions and support for the death penalty was higher for Whites than for Blacks. Following up on these ideas, Sidanius (1993) found that the strength of the association between ingroup bias and political conservatism varied as a function of the social status of one’s ethnic group. As expected, the relationship between ingroup bias and political conservatism was most positive among Whites and least positive and nonsignificant among Blacks. According to the logic of SD-theory, the more one favors one’s high-status ingroup over low-status outgroups, the more one will endorse conservative beliefs, because political conservatism operates to maintain status-quo intergroup status differences and thereby reinforces the dominant position of Whites in the social hierarchy. Social dominance theory assumes that ideological asymmetry occurs as SD0 drives greater support for hierarchy-enhancing LMs among highstatus group members. In line with this reasoning, previous research has found the relationship between SD0 and a series of LMs (e.g., redistributive social policy attitudes) to be stronger among high-status group members than among low-status group members (Mitchell & Sidanius, 1993; Sidanius, Liu, Shaw, & Pratto, 1994). In short, SD0 is assumed to be the driving force behind endorsement of hierarchy-enhancing LMs. (Of course, it is possible that one acquires one’s level of SD0 through adopting particular stances, like racism, towards group inequality. However, because SD0 has been shown to predict people’s support for or opposition to new hierarchy-enhancing ideologies, e.g., Pratt0 et al. (1994), we believe that it does serve as a predisposing orientation.) High SD0 members of high-status groups are expected to be especially likely to endorse these LMs because these views and policies fulfill desires for ingroup superiority. The empirical evidence of asymmetry involving SD0 has been found using various subject populations. For example, using student samples from two different universities (UCLA and San Jose State), Sidanius, Pratto, and Rabinowitz (1994) found asymmetry in the relationship between SD0 and ingroup affiliation. While SD0 was positively correlated with ingroup affiliation among European-Americans (i.e., highstatus group), SD0 was negatively associated with ingroup affiliation among Hispanicand African-Americans (i.e., low-status groups).

392

J. Sidanius et al.

Following up on the earlier death penalty research, Mitchell and Sidanius (1993) used a large random sample of adults from Los Angeles County and found evidence of ideological asymmetry in the relationship between SD0 and death penalty support. The SDO-death penalty support relationship was significantly more positive among high-status groups than among low-status groups. Similarly, using a large random sample of Los Angeles County residents, Sidanius, Pratt0 and Levin (1995) found that SD0 was positively associated with the ingroup preference of Whites vis-a-vis low-status groups, but was negatively associated with the ingroup preference of Blacks vis-a-vis high-status groups (see also Choudhury & Pratto, 1995). While the ideological asymmetry effect has been found to hold with respect to SD0 attitudes in general, thus far there is no research showing that it will also hold with respect to consensual SDO. Consensual SD0 is considered the most important component of the SD0 attitude space because desires for unequal status relations which are shared by members of all social strata ensure that the status hierarchy will be mutually reinforced by individuals at all social levels. The mechanisms used by high SD0 individuals to reinforce group status differences may also be used by individuals high in consensual SDO. Specifically, even though the mean level of consensual SD0 is, by definition, the same for all specified groups across the social status continuum, it may be that this consensual SD0 is still serving more hierarchy-enhancing functions for members of highstatus groups than for members of low-status groups. That is, beyond SD0 attitudes in general, even those anti-egalitarian beliefs which are shared across the status continuum may serve different functions for highand low-status group members. If this “differential function” idea is correct, then we should also expect consensual SD0 to be more positively and strongly correlated with hierarchy-enhancing attitudes among highstatus groups than among low-status groups. In other words, we should find some evidence of ideological asymmetry with respect to consensual SD0 as well. We will examine the relationship between consensual SD0 and several very important legitimizing myths with respect to two ethnic groups known to show substantial differences in their levels of social status: Euroand African-Americans (see Smith, 1991). Given the general reasoning above, we should expect to observe the following: (1) Consistent with previous findings, there should be a significant relationship between social dominance orientation and ethnicity. In particular, Euro-Americans (high-status group) should have significantly higher levels of social dominance orientation than African-Americans (low-status group). As mentioned above, this expectation is based on the general principle that high-status groups will be more supportive of general anti-egalitarian norms than low-status groups.

Consensual SD0

393

(2) The variance of consensual SD0 scores should be substantially greater than the variance of dissensual SD0 scores. This is expected to occur within hierarchically structured and relatively stable social systems because in order for harmonious intergroup interactions to occur, there must be more agreement than disagreement among different social groups as to the desirability of group-based status differences. (3) Even with respect to that portion of SD0 space where all ethnic groups have the same average level of SD0 (i.e., consensual SD0 space), among Euro-Americans (high-status group), there should be positive correlations between consensual SD0 and a long vector of legitimizing attitudes, ideologies and beliefs. (4) Again, even within consensual SD0 space, we should expect to find some evidence of ideological asymmetry. Specifically, among EuroAmericans (traditional high-status group), the relationship between consensual SD0 and an array of legitimizing myths should be significantly more positive than among African-Americans (traditional lowstatus group). METHOD

Participants The sample consisted of 823 UCLA undergraduate students from all educational levels (i.e., first-year students to senior students) surveyed in the fall of 1993. The respondents were enticed to participate in the study by the offer of four $50 prizes. Of the respondents, 42.1% were males and 56.9% were females. The sample was ethnically diverse. However, only the Euro-Americans (N = 161) and African-Americans (N = 127) were selected for analysis.

Measures’ Social Dominance Orientation. Social dominance orientation was measured by use of a new, balanced 16-item version of the SD0 scale (referred to as the “SD06 Scale”; see Appendix).2 The reliability of the scale within this sample was cy = .89.

‘Note that all measures of SD0 and legitimizing myths were coded so that a high score always indicates a high level of anti-egalitarian or hierarchy-enhancing values, beliefs, sociopolitical ideologies, social attributions, social policy attitudes or career preferences (e.g., a high SD0 score indicates a high level of anti-egalitarian sentiment). 2For a general discussion of the validity and reliability of the SD0 scales, see Pratto, Sidanius, Stallworth, & Malle (1994).

394

J. Sidanius et al.

Legitimizing Myths. We examined 21 legitimizing myths (e.g., attitudes, beliefs, attributions) which could be classified into seven general categories: (A) beliefs in system legitimacy, (B) general sociopolitical ideology, (C) social attributions for civil disturbance, (D) redistributive social policy attitudes, (E) ingroup identification, (F) attitudes toward hierarchy-enhancing careers and (G) attitudes toward hierarchy-attenuating careers. Unless otherwise indicated, all questions were answered on a seven-point scale ranging from “1 -Strongly disagree/disapprove” to “7 - Strongly agree/favor”.

(A) Beliefs in System Legitimacy This cluster of variables encompasses the degree to which respondents perceive American society to be just, fair and legitimate. The cluster consists of the following four variables: (1) Denial of discrimination. This measure is a composite of six variables and measures the degree to which respondents deny the existence of racial/ethnic discrimination within American society. The questions read: (a) “America is an open society where individuals of any ethnicity can achieve higher status”. (b) “Advancement in American society is possible for individuals of all ethnic groups”. (c) “Individual members of a low-status ethnic group have difficulty achieving higher status”. (d) “Individual members of certain ethnic groups are often unable to advance in American society”. (e) “African-Americans usually don’t get fair treatment (in the labor market, education, politics)“. (f) “People often discriminate against members of certain ethnic groups (e.g., AfricanAmericans and Latinos)“. The higher the score, the greater the degree of discrimination denial (a = .81). (2) Perceived social system legitimacy. This was assessed by asking whether the status differences among ethnic groups are perceived to be fair and justified. Five items comprise this scale: (a) “Differences in status between ethnic groups are the result of injustice”. (b) “America is a just society where differences in status between ethnic groups reflect actual group differences”. (c) “All ethnic groups are given the same treatment as other ethnic groups in the criminal justice system”. (d) “American society treats all ethnic groups equally”. (e) “Although there was discrimination in the past, today members of all ethnic groups have equal opportunity”. The higher the score, the greater the perceived system legitimacy (a = .77). (3) The Protestant work ethic. This was measured by three questions: (a) “If people work hard they almost always get what they want”. (b) “Most people who don’t get ahead should not blame the system; they really have only themselves to blame”. (c) “In America, getting ahead doesn’t always depend on hard work”. The higher the score, the greater one’s endorsement of the Protestant work ethic (a = .75).

Consensual SD0

395

(4) Support for L.A. police officers. This variable assesses whether respondents feel that the police officers accused of violating Rodney King’s civil rights should receive harsh or light sentences. There were seven allocation alternatives on a single item ranging from “be acquitted” to “15 years in jail”. The higher the score, the less severe the desired sanction. (B) Sociopolitical Ideology (5) Political conservatism. This was assessed by three items: (a) “How would you describe your political party preference?” (“Strong Democrat” to “Strong Republican”). (b) “In terms of economic issues, how would you describe your political attitudes and beliefs?” (c) “In terms of social issues, how would you describe your political attitudes and beliefs?” The higher the score, the greater the level of conservatism (CX= .83). (6) Classical racism. This was measured using a scale of six items: (a) “Blacks are inherently inferior”. (b) “Chicanos/Latinos are inherently inferior”. In addition, the respondents were asked to provide attributions for the economic and social status of African- and Hispanic-Americans. The stem question read: “Despite changes in social and economic policy, African- and Hispanic-Americans still suffer much lower living standards than other groups. Several explanations have been suggested for this poverty. Using the scale below, indicate the degree to which you agree or disagree with each of these explanations.” The alternatives used were: (c) “African-Americans are less intellectually able than other groups”. (d) “African-Americans are lazier than other groups”. (e) “Latinos are less intellectually able than other groups”. (f) “Latinos are lazier than other groups”. The higher the score, the greater the level of classical racism (N = .90). (C) Social Attributions for Civil Disturbance We also assessed the respondents’ attributions for the 1992 Los Angeles civil disturbance. (7) Criminal riot attributions. This variable assesses the degree to which respondents perceive the L.A. riots/uprising of 1992 as primarily caused by criminal elements. The question reads: “Do you think the L.A. riots/uprising was primarily caused by criminal elements?“. There was a seven-point response scale which ranged from “1 -Strongly disagree” to “7 - Strongly agree”; the higher the score, the greater the criminality attribution. (8) Social protest riot attributions. The respondents were asked to indicate whether or not they believe that the L.A. riots/uprising of 1992 was primarily a social protest. The question reads: “Do you think the L.A. riots/uprising was primarily a protest against social injustice?” The

396

J. Sidanius et al.

responses were scored so that the higher the score, the less the social protest attribution. (D) Redistributive Social Policy Attitudes This cluster of variables concerns the degree to which one supports government policies designed to redistribute positive or negative social value among groups. (9) Public spending on prisons vs. education. Using a “Tajfelian” (see Brewer, 1979) type of allocation matrix, the students were asked to allocate a given amount of money either to more prisons and police or to improved education in poor neighborhoods. The precise instructions were as follows: “Assume that the California State Legislature has decided to allocate $60 million for a long term plan to alleviate the State’s crime problem. Two kinds of long-term investment were discussed by the legislators, (a) more prisons and police and (b) more money to improve education in poor neighborhoods. Given the amount of money available, what investment mix do you feel is appropriate? Put a check in the space provided.” Money to more prisons and police

$55 million

$50 million

$40 million

$30 million

$20 million

$10 million

$5 million

Money to improved education in poor neighborhoods

$5 million

$10 million

$20 million

$30 million

$40 million

$50 million

$55 million

Respondents were asked to choose only one combination of money offered in one column of the matrix. The way the allocations were coded, the higher the score, the more desired spending on prisons and police. (10) Opposition to support for the poor. This was assessed by the respondents’ degree of positive or negative responses to the four questions: (a) “Greater assistance to the poor”. (b) “Government provided health care”. (c) “Reduced public support for the homeless”. (d) “Reduced benefits for the unemployed”. The greater the score, the greater the opposition to spending on the poor (a = .77). (11) Opposition to equal access policy. This variable measures respondents’ opposition to government effort to ensure that members of minority groups have equal access to housing, schools and jobs. The questions defining this dimension are: (a) “Government should see to it that minorities get fair treatment in jobs”. (b) “Government should not pass laws concerning the hiring of ethnic minorities”. (c) “Government should ensure that Whites and minorities go to the same schools”. (d) “Government should do what it can to improve the economic condition of poor ethnic minorities”. (e) “Affirmative action”. The higher

Consensual

SD0

397

the score, the greater the opposition to these equal access government policies (a = .80). (12) Opposition to public accommodations policy. This is a related, yet distinct policy dimension primarily concerned with equal access to public accommodations. The specific items are: (a) “Government should ensure that minorities can go to any hotel or restaurant they can afford”. (b) “Government has no business trying to ensure minority access to hotels and restaurants” (a = .76). The higher the score, the greater the opposition to public accommodations policy.

(E) Ingroup Identijkation (13) This variable assesses the degree to which the respondents identify with their own ethnic ingroup. There are four items: (a) “How strongly do you identify with other members of your ethnic group?” (b) “How important is your ethnicity to your identity?” (c) “How often do you think of yourself as a member of your ethnic group?” (d) “How close do you feel to other members of your ethnic group ?” The higher the score, the greater the degree of ingroup identification (cy = .89).

Attractiveness of Career Choice. Social dominance theory posits that, just as certain attitudes and ideologies can be regarded as hierarchy-enhancing or -attenuating, the same can be said for professional careers or institutional roles. Hierarchyenhancing careers are those which tend to contribute to the increase or maintenance of group-based hierarchical relationships within the social system. Hierarchy-enhancing careers can be identified as those which tend to favor the socially strong against the socially weak. Examples of such careers would be elements of the criminal justice system, police officers, internal security forces (FBI, CIA), and a number of careers with large private economic concerns (e.g., corporate lawyers). Hierarchy-attenuating careers are those which tend to aid the socially weak against the socially strong. Examples of such careers would be welfare officers, civiland human-rights organizers, and members of the public defenders office (see Sidanius, Liu, Shaw, & Pratto, 1994). The students were asked to indicate the attractiveness of eight career paths on a seven-point scale ranging from “1 -Not at al; attractive” to “7-Very attractive”. Four of these career paths are considered to be “hierarchy-enhancing” and four are considered to be “hierarchyattenuating” (see Sidanius, Pratto, Sinclair, & van Laar, 1996).

(F) Hierarchy-Enhancing Careers The four hierarchy-enhancing careers are: (14) Working as a lawyer for the state government prosecuting those charged with crimes.

398

J. Sidanius et al.

(15) Working to enforce the law and prevent crime within cities and towns of the United States. (16) Working for the FBI or other national police force. (17) Working to improve the profit of a certain business or organization. The greater the scores on these variables, the more attractive these careers are and the higher one’s degree of “hierarchy-enhancement”.

(G) Hierarchy-Attenuating

Careers

The four hierarchy-attenuating careers are: (18) Working as a civil rights lawyer for those who are discriminated against. (19) Working as a lawyer defending those unable to afford their own lawyer. (20) Working as an advocate for human rights, e.g., helping political prisoners or homeless people. (21) Working to improve the welfare of poor, ill and old individuals. The higher the scores on these variables, the more unattractive these careers are and the higher one’s degree of “hierarchy-enhancement”.

Analytic Procedure Combining social dominance theory with elementary classical measurement theory (e.g., Nunnally & Bernstein, 1994), we can separate the total variance of any legitimizing myth (e.g., racism, sexism, Protestant work ethic) into the following components: 2 ‘TTotal,_,

=

2 gDissensual~~

+

&zsidualLM.

(1)

This equation merely states that the total variance of the legitimizing myth (LM) can be divided into a “dissensual” component, or the component which is accountable for in terms of the socially constructed group differences (e.g., “race” differences) and a residual part which is independent of the socially constructed group distinctions. Furthermore, the residual LM scores can be broken down into two additional components:

One part of this residual variance is random error and the other part of the residual variance is the variance of the consensual LM scores. Applying classical measurement theory, we will recall that the reliability (r) of any variable is the proportion of that variable’s variance which is considered “true variance” (i.e., not stochastic, random error). This

Consensual SD0

399

implies that what we are calling “consensual” variance is merely the reliable portion of the residual variance of the LMs. This is to say:

rResidua1t.M

=

2 mConse”s”al&t 2 ITResidu&,,

Therefore, given the variance of the residual LM scores, and the reliability of these residual LM scores, the variance of the consensual LM scores can be simply defined as: a2CO~S~XISU~I~~ =

rResiduh_M

2 gResldual,~M

t

(4)

To summarize, the total variance of any legitimizing myth can be divided into the following components: 2 gTotal,,M

=

2 oDissensual~M

+ $Consensual~~

+ &mdom

errort,~~

(5)

Finally, and again applying the logic of classical measurement theory and expressing scores in standard form (see Nunnally & Bernstein, 1994, pp. 237-239) the estimated consensual LM score for each individual can be found as:

RESULTS Ethnicity and Social Dominance Orientation As our first hypothesis stipulates, SD0 was found to be significantly higher among high-status Euro-Americans (M = 2.32, SD = 1.02), than among low-status African-Americans (M = 1.74, SD = .54), F( 1,275) = 32.71, 7 = .33, p < lo-‘. As we will recall, the SD0 attitude space within which Blacks and Whites differ can be referred to as dissensual SD0 space. Operationalization of Consensual and Dissensual SD0 Scores Using the logic outlined above, we generated consensual SD0 scores for each individual in the following manner: (1) First, we regressed each of the 16 SD0 items upon ethnicity. We used all of the ethnic groups claimed by the respondents, including missing data cases and those who did not claim a specific ethnic group. This procedure provided us with ten ethnic categories: (1) EuroAmericans, N = 155, (2) Jews, N = 40, (3) Middle Easterners, N = 10,

400

J. Sidanius et al.

(4) Native Americans, N = 5, (5) African-Americans, N = 122, (6) White, European citizens, N = 11, (7) Hispanics/Latinos, N = 127, (8) Asians/ Asian-Americans, N = 194, (9) “Other”, N = 69, (10) Missing data, N = 54. In this manner, we exploited all of the information about ethnicity available.3 Ethnicity was dummy-variable coded and each SD0 item was regressed upon the nine dummy vectors. The residual and predicted SD0 scores were calculated for each SD0 item. (2) These predicted scores constitute the dissensual SD0 scores, which are simply defined as that portion of SD0 which could be predicted on the basis of ethnic group membership. (3) The reliability of the residual SD0 scores was estimated using the Cronbach a-coefficient. This reliability was found to be .88. This number indicates that, rather than largely consisting of random error, after the effect of ethnic group membership is removed, 88% of the variance of these residual scores is “true variance”, and, by definition, shared in common across all ethnic groups. (4) The consensual SD0 scores for each subject were computed by application of equation (6) above, adapted for SDO: ~Consensuals~

= YResidualsooZResidualso”

3

(7)

is the individual’s standardized = .88 and ZResidual,, residual SD0 score, computed as the average of the 16 residual SD0 items calculated in Step 1 above. where YResiduals,

Relative Sizes of Dissensual and Consensual SD0 The second hypothesis was explored by comparing the variances of the dissensual and consensual SD0 scores. The variance of the dissensual SD0 scores was computed to be 15.87, while the variance of the consensual SD0 scores was 175.64. This confirms our second hypothesis: the size of the common, consensual SD0 attitude space was substantially greater than that of the dissensual SD0 attitude space (i.e., more than nine times greater).

‘Note that we have defined the consensual SD0 space by use of all ten ethnic distinctions, even though we are only examining Black/White differences. We have done this deliberately because we would like to define consensual SD0 space as broadly as possible and controlling for as many ethnic distinctions as possible. In this manner, the resultant residual attitude space is reasonably independent of ethnicity as a factor and not simply independent of the Black/White distinction, and therefore, bears a stronger conceptual correspondence to what we mean by Consensual SDO. It should also be noted that defining consensual SD0 in this manner does not prejudice the results in our favor but, if anything, makes it more difficult to establish our point by eliminating even more of the true variance of the SD0 scores.

Consensual SD0

401

The Relationship Between Consensual SD0 and HierarchyEnhancing Legitimizing Myths among Euro-Americans Using consensual SD0 scores for each respondent, we tested our third hypothesis, namely that, among Euro-Americans, there should be positive and statistically significant correlations between consensual SD0 and the vector of 21 legitimizing beliefs and attitudes listed above (see Table 1). To examine this question, we performed 21 individual regression analyses for Whites, in which each of the LMs was regressed upon the consensual SD0 scores. The unstandardized regression coefficients are found in Table 1. As can be seen in Table 1, in general, the results were consistent with the third hypothesis. Among Euro-Americans, 19 of the 21 b-coefficients between consensual SD0 and the legitimizing myths were positive and statistically significant. Furthermore, not one of these relationships was significantly negative. For example, looking at the four variables assessing belief in system legitimacy, it was found that, even within consensual SD0 space, the higher the Whites’ consensual SD0 scores, the more likely they were to: (a) deny the existence of racial discrimination (b = .34, p < .OOl), (b) perceive the American social system as fair and just (b = .49, p < .OOl), (c) believe that success in America is almost exclusively determined by hard work and nothing else (h = .33, p < .OOl), and (d) show support for the police officers accused of the Rodney King beating (b = .53, p < ,001). Furthermore, among Whites, consensual SD0 was also consistently associated with the experienced unattractiveness of all four “hierarchyattenuating” career paths: (a) civil rights lawyer (b = .66, p < .OOl), (b) public defender (b = .55, p < .OOl), (c) human rights advocate (h = .93, p < .OOl) and (d) social worker (b = .77, p < ,001). The Ideological Asymmetry Hypothesis The fourth and last hypothesis concerns “ideologies asymmetry”. We will recall that, while the relationships between SD0 and group-relevant LMs are expected to be significant and positive among Euro-Americans, these relationships are expected to be significantly less positive among African-Americans. To examine this question, we also performed 21 individual regression analyses among the African-Americans, where each of the LMs was regressed upon the consensual SD0 scores. Inspection of Table 1 shows that the general trend in the differences between the unstandardized b-coefficients for Whites and Blacks appears to support the asymmetry hypothesis. The data show that, without a single exception, the b-coefficient between consensual SD0 and the legitimizing myths is more positive among Euro-Americans than among African-Americans. If we assume

402

J. Sidanius et al.

TABLE 1 Twenty-one Legltimlzlng Bellefa and Attitudes Regressed upon Consensual SD0 for Euro- and African-Americans (Entrles are Unstandardized Regresslon Coefficients) Ethnic group

Variables (A) Bellefs In system legltlmacy (1) Denial of discrimination (2) Perceived social system legitimacy (3) Protestant work ethic (4) Support for L.A. police officers

EuroAmericans b-coefficient

AfricanAmericans b-coefficient

Direction of coefficient difference

.34*** .49***

.lO .24*

+

.33*** .53***

.20 .04

+

(B) General soclopolltlcal Ideology (5) Political conservatism .67*** .48*** (6) Classical racism

.04 .14

(C) Social attrlbutlons for civil disturbance (7) Criminal riot attributions .45*** (8) Social protest riot .35** attributions (D) Redlstrlbutlve social policy attitudes (9) Public spending on prisons .30*** vs. education .71*** (10) Opposition to support for the poor .78*** (11) Opposition to equal access policy .59*** (12) Opposition to public accommodations policy (E) lngroup ldentlflcatlon (13) lngroup identification

.29***

*p < .05.

**p < .Ol. ***p