Read Campbell, Andrea, Cara Wong, and Jack Citrin. 1999. "V.O. Key Visits California: Context, Racial Threat and Voting o Propositions 187, 209, and 227." Paper presented to the Annual Meetings of the American Political Science Association, Atlanta, Georgia. text version

Inter-Group Prejudice in Multi-Ethnic Settings

Eric Oliver Department of Politics and Woodrow Wilson School of Public and International Affairs, Princeton University Robertson Hall Princeton, NJ 08544 [email protected]

This paper examines how out-group perceptions among Asian Americans, blacks, Latinos and whites vary with the racial composition of their surroundings. Previous research on the contextual determinants of racial attitudes offers mixed expectations: some studies indicate that larger percentages of proximate out-groups generate inter-group conflict and hostility, others suggest that such environments promote interracial contact and understanding. As most of this research has been directed at black-white relations, the applicability of these theories to a multi-ethnic context remains unclear. Using data that merge the 1992-1994 Multi-City Study of Urban Inequality and 1990 Census, we find that in neighborhood contexts, inter-ethnic propinquity corresponds with lower levels of out-group prejudice and competition, although intergroup hostility is higher in metropolitan areas with greater minority populations. Further tests reveal that these results do not occur from individual self-selection. Rather, our findings suggest that ethnic spatial and social isolation reinforces negative out-group perceptions. From these results, we offer a new formulation to theories of racial threat and contact and suggest the value of residential integration for alleviating ethnic antagonism.

Under Revise and Resubmit for the American Sociological Review July 2001

1 Recent data from the 2000 Census are revealing what demographers have been heralding for decades: new waves of immigration from Asia and Latin America are transforming the United States from a country monochromatically divided between blacks and whites into a "prismatic" nation composed of a polychromatic range of ethnic and racial groups (Bobo et al., 2000). By 2050, whites will comprise only 50 percent of the total population, a pattern already evident in many regions (Frey 1999). But even as America moves toward this multiracial plurality, it remains a nation "divided by color." Whites and blacks and, to a lesser extent, Asians and Latinos, continue to be highly segregated by neighborhood and municipal boundaries (Massey and Denton 1993). Despite the growing suburbanization of minority and immigrant populations (Frey 1993), most ethnic groups continue to live more apart than together. The implications of this multi-racial segregation for American race relations are unclear. Previous research on the relationship between social contexts and racial attitudes are seemingly contradictory and focused mostly on white attitudes. Some studies find that larger proximate populations of African Americans correspond with greater white racial animosity (Fossett and Kiecolt 1989, Frisbie and Niedert 1977, Giles and Buckner 1993, Glaser 1984, Quillian 1996, Taylor 1998, Wright 1977). Other studies report that inter-racial propinquity promotes interracial contact and lessens racial antagonisms (Brewer and Miller 1988, Ellison and Powers 1994, Fitzpatrick and Hwang 1992, Sigelman and Welch 1993, Sigelman et al., 1995). The few studies to include Asian Americans and Latinos (Cummings and Lambert 1997, Hood and Morris 1998, Lee 2000, Stein et al., 2000, Taylor 1998, Welch and Sigelman 2000) have focused largely on white attitudes and are inconclusive because of limited sample sizes or poor measures of racial context. Almost no research has focused on how racial environments shape the attitudes of minority groups, particularly Latinos and Asian Americans. Despite a large body of related research, the theoretical and empirical relationship between racial contexts and attitudes, particularly in a multi-ethnic context, remains undetermined. Part of the uncertainty also is due to the problematic way in which theories about racial environments and race relations have been operationalized in past research. Early theories on group conflict and contact (Key 1949, Allport 1954, Blalock 1968, Blumer 1958) stipulated a number of environmental conditions aside from the size of out-group populations that might influence racial relations, such as economic distress and historical norms, yet most research on the environmental determinants of racial relations focuses only on superordinate group hostility as a single, linear function of a single out-group size (e.g., Fossett and Kiecolt 1989, Frisbie and Niedert 1977, Giles

2 1977, Giles and Evans 1985, Glaser 1984, Quillian 1996, Taylor 1998, Wright 1977; although see Quillian 1995 and Oliver and Mendelberg 2000). In addition, high levels of neighborhood and municipal racial segregation call into question whether counties or metropolitan areas, the typical contextual units of analysis, are appropriate as either arenas of interracial competition or as predictors of inter-group contact. Finally, previous research has focused almost entirely on superordinate racial groups. It is not self-evident that many theories, such as ones on racial threat, are applicable for relations among minority groups. Bearing these problems in mind, this paper examines the relationship between racial contexts and racial attitudes in multi-ethnic settings. We argue that the dominant hypotheses for explaining racial attitudes and racial contexts need revision when examining multi-racial contexts, particularly with regards to size of in-groups and the environmental unit of analysis. Using data from the 1992-94 Multi-City Study of Urban Inequality (MCSUI) and the 1990 Census, we examine how racial environments (at both the metropolitan and neighborhood level) affect racial stereotypes and perceptions of threat among Asian Americans, blacks, Latinos, and whites. The effects of racial context depend upon which contextual unit is examined. At the neighborhood level, we find consistent patterns: with the exception of Asian Americans, those who live amongst more out-groups have more positive attitudes toward those groups; those who live amongst more of their own racial group hold more negative views of out-groups and perceive more threat and competition from out-groups. These differences are heightened in metropolitan contexts with larger minority populations. Further tests reveal limited evidence of intergroup threat or self-selection as the source of these results; rather, we suggest that it is simple exposure to out-groups through residential proximity that is a powerful force for reducing inter-group antagonism in a multi-ethnic setting. These results both challenge previous assumptions about the environmental determinants of racial attitudes and reveal the dynamics of race and place in multi-ethnic settings.

Perceptions of Ethnic Threat and Conflict The most widely accepted theory regarding the contextual determinants of racial attitudes is the "powerthreat" or "real conflict" hypothesis (Key 1949, Blumer 1958, Blalock 1967, Bobo 1988). This argument is typically articulated with a very simple formula: superordinate groups become more racially hostile as the size of a proximate subordinate group increases, which putatively threatens the former's economic and social privilege. In the context of black-white relations' in the United States, several studies provide convincing evidence in support of

3 this claim. Across many different datasets and time periods, there is a consistent increase in white racial antagonism with larger black populations in counties and metropolitan areas (Fossett and Kiecolt 1989, Giles 1977, Giles and Buckner 1993, Giles and Evans 1985, Glaser 1994, Quillian 1996, Taylor 1998, Wright 1977). The suitability of the "power-threat" hypothesis for explaining white attitudes towards non-black minorities, however, is unclear. If whites view all ethnic out-groups as similarly alien or threatening, then presumably they will respond to increasing populations of other minority groups with the same degree of animosity (for possible evidence, see Hood and Morris 1997, Quillian 1995, and Stein et al. 2000). But given the involuntary immigration of most African Americans and the differing cultural and economic position of Asian Americans and Latinos, perceptions of racial threat by whites may not be the same towards other groups (Bobo and Hutchings 1996). The stereotype of Asian Americans as "model minorities" may valorize them relative to blacks (Kim 1997), the racial status of Latinos is difficult to assess because the category Latino includes a broad range of racial groups (Oboler 1995; Skerry 1997), and the large immigrant proportions of both Asian American and Latino populations raise the question of whether attitudes towards these groups arise from racial or nationalist sentiments, or both. Indeed, there is little consistency in white racial attitudes towards all minority groups. Link and Oldendick (1996) find that whites are successively most hostile to blacks, less hostile to Latinos and least prejudicial towards Asian Americans. If this dynamic also relates to the greater racial context, then larger percentages of Asian Americans and Latinos may not generate as much racial animus among white populations. The applicability of the threat hypothesis for explaining the environmental determinants of attitudes between minority groups is even less certain. Within the American racial hierarchy, the relative locations of Asian Americans, blacks, and Latinos are in flux and the economic or political threat that any one group may pose to another is not self-evident (Bobo 1999). Popular images in the media and some research suggests that competition for jobs, housing, and political power among minority groups concentrated in urban areas contribute to greater racial hostility (Johnson and Oliver 1989, McClain and Karnig 1990), especially when this coincides with perceptions of economic distress or increasing levels of immigration (Campbell et al., 2000). Among minority groups, high levels of racial stereotyping and animosity continue (Bobo and Hutchings 1996, Cummings and Lambert 1997). For instance, Johnson et al., (1997) find that a majority of Asian Americans and a large percentage of Latinos view blacks as less intelligent and more welfare dependent than their own groups; Asian Americans are also likely to view Latinos negatively in terms of intelligence and welfare dependency; and, over two-thirds of black respondents

4 and nearly half of all Latino respondents rated Asian Americans as "difficult to get along with." But whether or how these inter-minority antagonisms relate to patterns of social segregation or group size is unclear. To understand how the dynamics of inter-group competition might shape racial attitudes in multi-ethnic settings, it is important to reconsider the fundamentals of the real conflict hypotheses. Early theories regarding racial environments and racial attitudes (Key 1949, Blalock 1968, Blumer 1958) viewed out-group prejudice as a product of feelings of threat to social status that comes with racial group membership. This status is sensitive to the size of out-groups, as is often operationalized in existing research. However, racial status is also related to the salience and vulnerability of the in-group identification. Such feelings of vulnerability may have nothing to do with the size or objective positions of out-groups and be related largely to the status of the in-group. For example, Key (1949) argued that differences in white racial antagonism across the South were less the consequence of the size of black populations (which were equally large) and more the product of white economic position, a finding echoed in other research (Oliver and Mendelberg 2000, Quillian 1995). Therefore, to better understand the impact of racial environments in multi-ethnic settings, several components of inter-racial competition need to be considered. First, as the threat hypothesis is traditionally formulated, feelings of vulnerability and group position will be a function of the size of subordinate out-groups. Clearly, the challenge that such groups pose to either superordinate groups or other subordinate groups will depend partly on their size. Second, in a multi-ethnic context, perceptions of racial vulnerability will also be a function of in-group size. The importance of in-group size depends upon the position of the group in a racial hierarchy. As the size of a subordinate group decreases, its members will have incentives to downplay their own group membership as a determinant of their own social identity (Brown 1985). Indeed, this seems to be the case ­ blacks who live amongst more whites tend to feel less efficacious, while blacks who live in predominantly black areas are more likely to express group solidarity and be more politically active (Bledsoe et al., Lau 1989, Bobo and Gilliam 1990). Conversely, as the size of a superordinate group decreases, its members may feel greater racial vulnerability and exacerbate racial differences (Blalock 1968). Finally, out-group animosity will depend on the relative economic and political position of both in-groups and out-groups and the perceptions of individuals that group membership is important as a determinant of certain rights or privileges (Blumer 1958, Bobo 1988). Such perceptions of group competition may be influenced by culturally and historically defined relationships (Blumer 1958), the stability of political institutions reinforcing racial

5 privileges, and conditions of economic distress (Oliver and Mendelberg 2000, Quillian 1995). They will also depend upon the porousness of group boundaries and the coherence of internal group membership. For example, as a group, Latinos may pose less of a threat to whites than other racial groups because their own linguistic and national diversity and because many may be viewed or identify as white. When considering racial environments and racial attitudes in multi-ethnic settings, both the size of out-groups and in-groups as well as their relative economic and cultural position needs to be taken into account.

Inter-racial Contact, Self-Selection, and the Importance of Contextual Units In direct contrast to the threat hypothesis, the contact hypothesis predicts that larger populations of outgroups can improve inter-group relations as individuals correct negative racial stereotypes with first-hand social experience (Allport 1954). Although past scholars have routinely questioned the validity of the contact hypothesis (Jackman and Crane 1986, Pettigrew 1998), numerous recent studies purport to find strong correlations between racial contact and positive out-group sentiments (Bledsoe, Welch et al., 1995, Ellison and Powers 1994, Sigelman and Welch 1993, Stein et al., 2000, Welch and Sigelman 2000, Yancy 1999). Yet the findings produce a difficult puzzle: how can the proximity of out-groups, a necessary prerequisite for interracial contact, simultaneously promote inter-racial understanding yet also correspond with greater levels of inter-racial competition? In seeking to reconcile the research that supports both the threat and contact hypotheses, three possible explanations are evident. The first relates to the unit of analysis. Most evidence of racial threat comes from larger minority populations in relatively expansive environmental units, such as counties or metropolitan areas, but racial contact is more likely to be affected by smaller geographic units such as towns or neighborhoods. High degrees of racial segregation along municipal or neighborhood boundaries will profoundly influence how both racial conflict and contact may occur. For example, residence in a predominantly white suburb may shield residents from sharing public resources with nearby minority groups and also limit the possibilities of meaningful contact with out-groups. When comparing the impact of contact and conflict across racial environments, the racial composition of both micro and macro contextual units needs to be considered. Second, the positive effects of racial diversity in micro-contexts may not arise from interracial contact but from individual self-selection (Pettigrew 1998, Powers and Ellison 1995). In other words, people with more interracial contact patterns or who live in more racially heterogeneous settings may report more positive attitudes toward

6 out-groups not because inter-racial contact causes them to be less racist but because prejudice is a strong determinant of residential and friendship choice: people with strong animosity toward out-groups will seek to live amongst in-groups, limit contact to people of their own race, and discourage other races from moving into their neighborhood. But while self-selection is undoubtedly a factor that will need to be considered, other research suggests its effects may not be necessarily so large. Powers and Ellison (1995) report that contact is still an important determinant of racial attitudes, even when self-selection of the respondents is controlled. Other studies find that in-group racial preferences are less important than economic or other factors in driving residential location (Bobo and Zubrinsky 1996, Clark 1992, Frey 1979, Zubrinsky and Bobo 1996). Self-selection also may be less of a factor in explaining the relationship between residential location and racial attitudes for blacks and other minorities who still face limited housing choices due to discriminatory real estate practices (Farley et al. 1994). Finally, inter-racial proximity may also promote racial tolerance through means other than direct interpersonal contact. The sharpest criticism of the contact hypothesis is that the conditions necessary for the effects of contact to occur (equal status, cooperation towards mutual goals, opportunities for socializing) are difficult to attain, particularly in highly segregated racial settings (Pettigrew 1998, Jackman and Crane 1986). But while direct interpersonal contact may not always occur in integrated neighborhoods, casual exposure to minority groups on an everyday basis may counteract negative stereotypes. Negative images promoted through the mass media, as well as those triggered by unfamiliar languages or traditions associated with particular ethnic groups, may reinforce the belief that a group is foreign or does not "belong." Such negative attitudes can be countered through increased exposure to different communities. For example, Lee (2000) finds that whites with knowledge of Asian American history are less likely to hold Anti-Asian attitudes. Residents of highly segregated neighborhoods who are never exposed to out-groups save through images in the mass media may have few opportunities to correct or disconfirm erroneous stereotypes (Dyer et al., 1989, Lee 2000, Hum and Zonata 2000). Inter-racial social experiences or simple exposure may alleviate racial animosity, but minority groups need to have the opportunity before any effects will be felt. In sum, although the contact hypothesis is subject to numerous criticisms, it does not mean that interracial proximity is not without potential benefits. Understanding the positive effects of racial integration while retaining an appreciation of inter-group competition requires analyzing different units of analysis, controlling for self-

7 selection, and conceiving alternatives to the mechanisms of interracial contact outlined in past research. In a multiethnic context, simple social exposure may be sufficient for diminishing negative out-group stereotypes.

DATA & DISTRIBUTION OF SAMPLE To test these ideas, we utilize data from the 1992-1994 Multi-City Study of Urban Inequality (MCSUI).1 The MCSUI is a stratified area probability household survey that generated over 8,900 face-to-face interviews with over-samples of blacks, Latinos, and Asian Americans in the metropolitan areas of Atlanta, Boston, and Detroit and in Los Angeles County.2 Adults 21 years and older were interviewed in English, Spanish, Korean, Mandarin, or Cantonese. Because a large number of survey questions were not utilized in the Detroit sample, we excluded them from the sample. The Investigators of the MCSUI also provided block-group level census data from the 1990 Census Summary Tape File (STF3) on the racial and educational composition of the respondents' neighborhoods.3 With these multi-level data, we are able to examine the effects of both individual-level and contextual variables. INSERT TABLE 1 HERE As illustrated in Table 1, the respondents in the MCSUI sample display patterns of racial segregation comparable to the at-large population. On the metropolitan level, the sample is divided into three groups. Atlanta is largely bifurcated between whites and a relatively large black population. Los Angeles is composed of large white and Latino pluralities of approximate size with smaller but equivalently sized black and Asian populations. The greater Boston Metropolitan Area (the area sampled in the MCSUI) is overwhelmingly white, although Suffolk county which contains the city of Boston has higher percentages of both blacks and Latinos.


The MCSUI data is the product of more than 40 researchers at 15 colleges and universities. Beginning in 1992 a household survey was undertaken of adults over 21 years of age, oversampling in census tracts with high proportions of poor and minority residents. Although primarily designed as a study of labor market outcomes, the extensive surveys also had questions regarding racial attitudes and neighborhood choice. 2 Among the 1,054 Asian Americans surveyed, the majority (nearly 80 percent) were of Chinese and Korean origin, and the remainder of the sample was made up of respondents of Japanese and South Asian decent. The majority (87.7%) of the Asian American sample were foreign-born. Over half of the Los Angeles Latino sample (68.2%) were of Mexican origin. Central Americans, primarily from El Salvador and Guatemala, comprised the remainder of the sample. Similar to the Asian American sample, the majority of Los Angeles Latinos included in the study (80%) were foreign-born. The overwhelming majority of Latinos in the Boston sample were either Puerto Rican or Dominican in descent.


The block groups range in size from 46 to 10,988 in size although the average size is 1,858 (with a standard deviation of 1,700). Most block groups are relatively close in size to neighborhoods and we use the two terms interchangably.

8 Despite the aggregate racial differences across the metropolitan areas, most respondents' neighborhoods are similarly stratified along racial lines. Of all racial groups, African Americans are the most racially isolated. Despite only being 11 percent of the Los Angeles County population and only 6 percent of the Boston Metropolitan Area population, black respondents, on average, live in neighborhoods where at least half of the residents are black. In Atlanta, the black respondents' neighborhoods are, on average, nearly 80 percent black. Black respondents in Boston and Los Angeles are more likely to live among Latinos than either Asian Americans and whites. In contrast, Asian Americans, also roughly 11 percent of the Los Angeles County population, are much less isolated ­ the average percent of whites, Latinos, and Asians is near 30 percent each, although the black percentage is just 4 percent. In Boston and Los Angeles, whites and Latinos, as larger ethnic groups, are likely to have larger proportions of their own group in their neighborhood but are less isolated from non-black minority groups than in Atlanta. White respondents live in neighborhoods where, on average, more half of the residents are white, are roughly 20 percent Latino, and contain a smaller percentage of Asian Americans and blacks. Latinos tend to be in predominantly Latino neighborhoods with larger portions of whites than either Asian Americans or blacks. Neighborhoods for white respondents in Atlanta are, on average, 18 percent black. Although most groups are segregated by neighborhoods, not all neighborhoods are racially homogeneous. The large standard deviations associated with the means presented in Table 1 indicate that the actual percent of each racial group varies a great deal across different census block groups. STEREOTYPES AND SOCIAL CONTEXTS How do these patterns of metropolitan and neighborhood racial segregation shape racial attitudes? We start our analysis by examining how perceptions of ethnic out-groups vary with the racial composition of the neighborhood. Table 2 lists the coefficients predicting changes in summary negative stereotype scores with the percent of that particular out-group in the neighborhood across the three different metropolitan areas.4 In other


The MCSUI asked respondents to evaluate on a seven-point scale the extent to which certain "characteristics" fit the description of Asians, Blacks, Latinos and Whites. For example, for the characteristics of intelligence, a score of 1 would mean the respondent thought all members of the ethnic group were intelligent, 7 would mean that all members of the group were unintelligent and a score of 4 is "that the group is not towards one end or the other." Although different stereotypes exist for various racial and ethnic groups, for the sake of consistency, we chose four measures for all four groups: intelligence, self-sufficiency versus welfare dependence, easy to get along with, and treats other groups equally. Each of the seven point scale was re-coded so that positive and neutral perceptions were counted as zero and any negative perception as one. The four two-point negative stereotype scores were added to produce a zero to four point summary negative stereotype score.

9 words, it measures what effect living amongst more of a particular out-group has on a person's attitudes towards that group. To control for individual- level characteristics while estimating the distinct effect of the social context, we utilized an Ordinary Least Squares Regression.5 Each group of coefficients depicted in Table 2 comes from a separate equation in which the negative stereotype scale was regressed on several individual-level variables (education, age, home-ownership, sex, and length of residence) as well as two block group-level variables, the percent of residents of a distinct ethnic group and the percent of adults with a high school diploma.6 Table 2 displays just the coefficients for the percent of the out-group from each equation while the full equations are listed in Appendix B.7 INSERT TABLE 2 HERE With few exceptions, people who have more out-group members in their neighborhood also hold less negative stereotypes about that group. In Atlanta and Los Angeles, blacks, Latinos and whites all report lower negative stereotype scores about out-groups as the percent of that out-group in their neighborhood increases. Although the differences are not always statistically significant (particularly in Los Angeles with the case of blacks' views towards Latinos and whites, Latinos' views towards whites, and whites' attitudes towards Asian Americans), the direction of the effect in these two metropolitan areas in generally the same ­ propinquity corresponds with less negative out-group stereotyping. The exceptions to this trend are among Asian-Americans' attitudes towards blacks in Los Angeles and among blacks' and whites' attitudes towards Latinos in the Boston metropolitan area. Asian Americans who live in Los Angeles neighborhoods with higher percentages of blacks also hold much more negative stereotypes of blacks. Similarly, our findings from the Boston sample are distinct. Blacks and whites in the Boston


Ideally a hierarchical linear model should be used in a multi-level analysis to minimize the correlation in error terms among respondents in the same geographic unit. However, because the respondents in the MCSUI data were distributed across so many different block groups and because most block groups had fewer than 30 respondents per group, we could not generate estimates using HLM. Given the small number of cases per block group, however, the in-context error correlation should be relatively small.


Although individual income is often used as a predictor of negative out-group stereotypes, and is related the place of residence, a significant portion of the MCSUI sample, as with most surveys, refused to answer the household income item. In order to minimize the number of missing cases, we did not included individual income in the model. This action seems justified in two respects. First, the individual education and home-ownership items as well as the block group education items capture a good deal the individual's own income and social status level. Second, when the regressions were estimated with income included, the results were generally the same. 7 Because of the tiny Asian American and Latino populations in Atlanta, there is not enough variance to estimate the effects of the size of these groups in the neighborhood. The same holds for Asian Americans in Boston. Hence the effects of neighborhood context were estimated separately for each metropolitan area.

10 area who live in neighborhoods with more Latinos hold negative stereotypes of Latinos, although with blacks this difference is not statistically significant. Furthermore, among residents of Boston, outside of whites' views towards Latinos, none of the differences across neighborhoods in out-group attitudes are statistically significant. INSERT TABLE 3 HERE Table 3 replicates the analyses conducted in Table 2, only substituting the out-group percentage in the neighborhood with a measure of in-group percentage, while still controlling for the standard set of predictors. In short, it measures what effect living amongst one's own racial group has on the perception of other groups. Because Table 3 is examining in-group contexts on out-group sentiments, the samples were divided only by individual race of respondent rather than by metropolitan area. To measure the distinct effects of the metropolitan area, dummy variables for Atlanta and Boston and interaction terms between the dummy variables and the term measuring percent white in the neighborhood were included in the equation. As would be expected, the results in Table 3 largely mirror the findings in Table 2. Just as non-Asian American respondents in Atlanta and Los Angeles were all less likely to harbor negative out-group stereotypes as the percent of the out-group in their neighborhood increases, they are more likely to express such sentiments as the percent of their own in-group increases in their neighborhood. For example, whites in predominantly white neighborhoods score much higher in their negative stereotypes of Asians, blacks, and Latinos. Blacks in predominantly black neighborhoods also hold more negative impressions of Asian Americans and Latinos (and whites in Atlanta), a sentiment reciprocated by Latinos in predominantly Los Angeles Latino neighborhoods. Consistent with the results shown in Table 2, Asian Americans in predominantly Asian neighborhoods in Los Angeles are less negative in their attitudes towards out-groups than those in more racially mixed settings. And, once again, the trends in Boston are different - the effects of neighborhood racial homogeneity are attenuated, particularly for Latinos and whites. In other words, the differences in attitudes by the racial composition of neighborhoods in Boston are simply not as great as those in Atlanta or Los Angeles.8


Interestingly, as illustrated in the full equations in the Appendix, Asians who live in more high status neighborhoods (as measured by the percent with a high school degree) also hold more negative views towards blacks and Latinos. Otherwise there is no general trend across all races and all metropolitan areas between the economic status of the neighborhood and the negative stereotypes towards out-groups. In some cases, such as whites' attitudes toward Latinos in Boston or towards Asians in Los Angeles or blacks' attitudes towards Latinos in Atlanta, residents of higher status neighborhoods are less likely to hold negative stereotypes. In other cases, such as Latinos' attitudes blacks in Boston, the opposite is the case. The impact of neighborhood economic status seems highly dependent on its racial composition and on particular group dynamics.

11 These results demonstrate that racial environments are important predictors of individual racial attitudes. In many equations the size of the coefficients for the neighborhood racial percentage are larger than those for individual-level predictors of racial attitudes such as education and age.9 As previous research on white attitudes has illustrated, people's surroundings do correspond with their racial beliefs. But perhaps most interesting is the remarkable consistency of the findings. For blacks, Latinos, and whites, the trend is the almost always the same: the greater the percentage of minority out-group neighbors, the lower the racial animosity toward out-group minorities; the higher the percentage of in-group neighbors, the greater the animosity. The only exceptions to this trend are with Asian Americans, whose attitudes toward blacks, Latinos, and whites become more negative with greater proximity to these groups and among Boston area residents, whose views vary less across neighborhoods. The attitudes of Asian Americans may be very specific to the time and locale being sampled. The surveys in Los Angeles were conducted soon after the 1992 riots in which hostility between Asian Americans and other minority groups was quite high. The trends across Boston highlight the importance of metropolitan racial contexts for shaping neighborhood racial attitudes. Racial contexts are generally less salient among respondents in the Boston metropolitan area which also has much smaller minority proportions than either Los Angeles or Atlanta. Similarly, whites and blacks in Atlanta exhibit much lower rates of hostility toward Latinos than their counterparts in Los Angeles which has a much larger Latino population. If we review the results from both units of analysis, we can conclude that interracial proximity corresponds with less negative attitudes towards those out-groups on a neighborhood level (or conversely racial isolation fosters more negative attitudes) but only in metropolitan areas with large percentages of minority groups. In areas with smaller percentages of minorities, the differences in racial attitudes across neighborhoods are less significant.

PERCEPTIONS OF INTER-RACIAL COMPETITION The most striking aspect of these findings is how, at one level, they seem to contradict the predictions of the threat hypothesis and so many findings from past research. Unlike earlier studies which consistently report that larger percentages of proximate minority members contribute to higher white racial animosity (Fossett and Kiecolt 1989, Quillian 1996, Taylor 1998, Wright 1977), the results in Tables 2 and 3 indicate just the opposite - on a


When the different scaling of the variables are taken into account, the predicted differences in racial attitudes between residents of the highest and lowest ends of the neighborhood racial composition are usually greater than

12 neighborhood level, whites who live amongst more minorities hold less negative views of those groups. Moreover, living amongst other out-group minorities does not seem to fuel inter-ethnic hostility among most minority groups either. Does this mean the threat hypothesis is invalid, or just invalid for this level of context? One way to answer this question is to directly measure the perception of threat across contexts. Starting with black and white respondents, Table 4 lists coefficients from OLS equations that regress several measures of threat from minority outgroups by the percent of the in-group in the neighborhood with the same set of controls used above. Once again, samples from all three metropolitan areas were pooled with dummy variables measuring residence in Atlanta and Boston and interaction terms between the percent in-group in the neighborhood and the metropolitan dummy variables. These equations predict perceptions of zero-sum competition with specific minority groups, threats to political and economic power by immigrants, and opposition to affirmative action for specific minority groups.10 INSERT TABLE 4 HERE Among blacks and whites, the findings mimic the results about group stereotypes: the greater the percentage of in-group members within the neighborhood, the greater the sense of zero-sum competition with minority out-groups, the greater the perception of threats from immigration, and the higher the opposition for affirmative action programs for minority out-group members, although this effect varies across metropolitan regions. Whites in predominantly white Los Angeles neighborhoods feel the greatest competition from Asian Americans while whites in predominantly white Atlanta neighborhoods feel more competition with blacks. Whites in Boston generally have lower feelings of inter-group competition and the effects of living in predominantly white neighborhoods are lower, on average, than in Atlanta or Los Angeles. For whites, the economic status of their neighborhoods is also a consistent predictor of their sense of competition with minority groups: those in higher

between the highest and lowest values for the individual level variables. 10 The zero-sum variables are constructed from a combination of two five-point scales asking how much the respondent agrees or disagrees with statements on whether "more good jobs for [out-group] means fewer good jobs for [the respondent's in-group]" and whether "the more influence [out-group] have in local politics, the less influence [the respondent's in-group] will have in local politics." The sample was divided into separate groups who were asked about specific out- groups, i.e., blacks were asked about competition with Latinos and Asians, Asians with blacks and Latinos, whites with all three minority groups. The immigrant threat variable was constructed from two five point scales asking how much "political influence" and "economic opportunity" people of the respondent's race would have if immigration continues at its current rate. The opposition to affirmative action variables were constructed from two five-point Likert scales asking how much respondents support programs to give special "job training and assistance" and "preferences in hiring and promotion" for groups that have previously discriminated. To control for non-race specific opposition to affirmative action, we included a measure of opposition to affirmative action for women in these later equations. The full wording of the questions are in the Appendix.

13 status neighborhoods perceive less zero-sum competition with all minority groups, less threat from immigration, and are less opposed to affirmative action. The racial composition of African Americans' neighborhoods is an important determinant of their sense of competition with minority groups in Los Angeles but not so much in Boston.11 Black residents of predominantly black Los Angeles neighborhoods report a greater sense of zero-sum competition with Asians and Latinos, greater threat from immigration, and are more opposed to affirmative action benefits for Asians and Latinos (although more supportive of affirmative action for blacks). Among blacks in both Atlanta and Boston, there are lower overall levels of such inter-minority competition and the effects of neighborhood racial composition are attenuated. The highest levels of perceived threat from other minority groups occurs among blacks in predominantly black neighborhoods but that also live in a metropolitan area with higher numbers of minority out-groups (i.e., Los Angeles). INSERT TABLE 5 ABOUT HERE These same results, interestingly enough, do not generally occur among the Asian American or Latino respondents. Table 5 lists the results from OLS equations predicting feelings of zero-sum competition and opposition to affirmative action among Asian Americans in Los Angeles and Latinos in Los Angeles and Boston by the percent of in-group members within the neighborhood.12 Asian Americans who live in predominantly Asian neighborhoods express a greater sense of zero-sum competition with and are more opposed to affirmative action programs that benefit blacks and Latinos, although these differences are not statistically significant. Unlike with blacks, Asian Americans who live amongst more of their own ethnic groups are not more supportive of affirmative action programs that benefit in-group members. Latinos also exhibit similar preferences. Latinos who live in predominantly Latino neighborhoods perceive more zero-sum competition with Asian Americans and blacks and are more opposed to affirmative action programs for these other groups although, once again, these differences are not statistically significant. Latinos in Boston are generally less opposed to affirmative action except for, interestingly enough, those that live in predominantly Latino neighborhoods who are actually more opposed to programs that benefit Latinos.

11 12

Black respondents in Atlanta were not asked questions about zero-sum competition with other groups. Because these groups are composed of large immigrant populations, the variables on immigrant threat were not analyzed.

14 At one level of contextual analysis, these findings provide a strong contradiction to the threat hypothesis. Most research posits a relatively linear relationship between the size of out-group populations and the level of racial hostility, largely because of a putative threat mechanism (Blalock 1967, Key 1949, Taylor 1998): the larger the outgroup, the greater the threat, and the higher the racial animosity. It should be noted, however, that much of this past research only measures the variance of racial animosity or violence across racial contexts rather than direct perceptions of economic or political threat. Our results show no evidence that being proximate to minority groups at a neighborhood level promotes any perceptions of zero-sum conflict. In fact, we find just the opposite. Blacks and whites who live in neighborhoods with higher percentages of their own group are more likely to express feelings of threat from other minority groups. This pattern is roughly similar among Asian Americans and Latinos. But on the metropolitan level, inter-group competition is clearly a factor driving the relationship between racial environments and racial attitudes. Like past research, we find that residents of metropolitan areas with relatively large out-group populations express a greater sense of competition with those groups. For example, black and white Los Angelenos feel more competition with Asian Americans and Latinos than their counterparts in either Atlanta or Boston, while Los Angeles Latinos feel more competition with blacks and Asians than those in Boston. Whites in lower status neighborhoods also feel greater threats from out-groups. Proximity to out-groups is an important determinant of racial conflict but only when that proximity is defined relative to a larger geographic area.

RACIAL SELF-SELECTION Given past research, the most likely explanation for the findings above is one of individual self-selection. As scholars of residential segregation have pointed out, individuals with strong prejudicial feelings toward outgroups and preferences toward in-groups are most likely to locate in segregated neighborhoods (Bobo and Zubrinsky 1996, Clark 1992). Consequently, the higher instances of racial prejudice among people who live amongst more of their own in-group may simply be the geographic distribution of racial attitudes rather than any causal factor from the environment. The MCSUI data allow for this possibility to be tested. Respondents were given a series of five cards each depicting a neighborhood comprised of houses with different collections of racial groups. These ranged from neighborhoods composed of all their race to those with few of members of their own race. Respondents were asked to prioritize the neighborhoods from most to least desirable and, from these choices, an index of in-group

15 neighborhood racial preference was constructed.13 Table 6 lists the coefficients of OLS equations that regressed the percent of in-group members in the neighborhood on the neighborhood racial preference measure for each specific group with the standard set of controls. INSERT TABLE 6 HERE With the exception of Asian Americans, respondents who live amongst more in-group members are also more likely to express a preference for a predominantly in-group neighborhood. The biggest differences occur for blacks and Latinos in Los Angeles and whites in Atlanta. Compared to black residents of largely non-black neighborhoods, blacks in predominantly black Los Angeles neighborhoods score .7 points higher on the in-group neighbor preference scale a difference that is much less across neighborhoods in Boston and Atlanta. Conversely, the relationship between neighborhood racial composition and preference for whites in Atlanta is nearly three times as great as for whites in either Boston or Los Angeles. For Latinos, the correspondence between neighborhood racial preference and actual neighborhood composition is positive in Los Angeles but attenuated to no difference in Boston. Among Asian Americans there is no statistical relationship between the in-group neighborhood racial preferences and actual neighborhood racial composition, which is partly the consequence of Asians preferring predominantly white neighborhoods to predominantly Asian ones.14 But while, at first glance, these results would seem to provide strong support for the self-selection hypothesis, further tests show that self-selection does not account for the positive relationship between in-group neighborhood composition and negative attitudes towards out-groups as demonstrated above. Table 7 lists the results of OLS regressions of the measures of out-group negative stereotypes on the same set of measures used in Table 3 only adding the indicator of in-group neighborhood preference. INSERT TABLE 7 ABOUT HERE First, a preference for in-group neighbors does not always correspond with hostile attitudes towards outgroups, particularly among non-whites. Although in all equations, whites who prefer only white neighbors are more


This index was constructed on a five-point scale. Respondents who listed the two neighborhoods with either all or most residents as in-groups were coded 5, who listed one as a top preference were coded 4, as neither as a top two preference as 3, those who listed one of the two neighborhoods with the least number of in-group members as a first or second choice as a two, and those who listed both predominantly out-group neighborhoods as top choices as a 1. 14 The MCSUI data also provide other mechanisms for testing neighborhood racial preferences including a measure of neighborhoods the respondent would not move into and a measure where respondents fill in the racial composition of their ideal neighborhood using a blank card. Similar results are derived with these alternative

16 likely to have negative views towards all three minority out-groups, this pattern is only replicated in black views towards Asian Americans and in Asian American views towards whites. In no other instance does an in-group neighborhood preference correspond with an out-group prejudice. To a large extent, self-selection occurs only for whites. Beyond this, however, even when controlling for neighborhood racial preference, the positive effects of ingroup neighborhood composition on negative out-group attitudes remain. Although the effects of neighborhood racial composition on negative out-group attitudes diminish slightly for whites, they do not lose most of their size nor statistical significance. In the case of non-whites, the previous coefficients for neighborhood racial composition are virtually unchanged with the new control. While residential location does correspond with neighborhood racial preferences, this fact does not explain why blacks, Latinos, and whites who live amongst more of their own race hold more negative views of other groups. To whatever extent self-selection is behind the contextual differences in racial attitudes, it is largely a white phenomenon.

DISCUSSION This is the first study to examine the effects of racial contexts on out-group attitudes across a variety of racial groups in different multi-ethnic settings. As we have seen, determining the relationship between racial environments and attitudes in a multi-ethnic context requires looking beyond simple formulations. Perceptions of threat from out-groups will depend not simply on the size of that out-group, but on their relative economic position, the historical period, the size of the in-group population, and the contextual unit of analysis to be measured. Assessing the likelihood of contact (or at least the negative consequences of social isolation) depends upon controlling for geographic self-selection as well as utilizing smaller environmental units. Taking these factors into account, we find that past theories derived largely from the interaction between blacks and whites are applicable in multi-ethnic contexts, as long as the considerations above are heeded. We also find that some new ways for thinking about the relationship between racial environments and attitudes are in order. Like past research on inter-racial threat and competition, we find strong evidence that, in multi-ethnic settings, racial attitudes towards other ethnic groups are influenced by racial environments and that these results are remarkably consistent in terms of direction. Yet, unlike most of this research, we find that close proximity to out-

measures too: among non-Asian Americans, those who prefer in-group members as neighbors (or resist having outgroup members as neighbors) are more likely to live in predominantly in-group neighborhoods.

17 groups corresponds with less racial antagonism. Among blacks, Latinos, and whites, as the neighborhood percentage of a minority out-group increases, the likelihood of holding negative out-group perceptions or feeling inter-group competition generally declines. Negative perceptions of out-groups are higher for those who live in neighborhoods with more of their own racial group. These results do not contradict the threat hypothesis, however, because they are predicated on the racial composition of the greater metropolitan area as well as the economic and cultural position of distinct groups. The effects of racial isolation are greatest for those individuals who live amongst more minority out-groups in macro contexts. The effects of neighborhood racial composition on out-group attitudes are the lowest among respondents in the overwhelming white Boston metropolitan area. Among respondents in both Atlanta and Los Angeles, where minority populations are much larger, the differences across neighborhood racial contexts are much greater. These effects are also evident in the neighborhood choices of residents. Respondents who prefer same race neighbors generally live in more racially segregated neighborhoods, particularly if they live in more racially mixed metropolitan areas such as Atlanta and Los Angeles. In short, neighborhood racial isolation has a much more profound effect in the context of regional racial diversity. The higher levels of racial antagonism among those in racially isolated neighborhoods are not the consequence of self-selection of these residents. Neighborhood racial preferences do not explain the neighborhood differences in racial attitudes, particularly for minorities. Self-selection is primarily a white phenomenon - whites are the only group whose preference for same race neighborhoods corresponds with racial antagonisms toward outgroups. Neighborhood racial preferences generally do not correspond with negative out-group perceptions among non-whites. Moreover, these geographic preferences do not explain the higher racial antagonism for any group, white or non-white, living in same-race neighborhoods. Despite the general consistency in the size and direction of these effects, several exceptions reflect the complexities of racial threat in multi-ethnic contexts. First, economic as well as racial contexts are important to note when considering the environmental determinants of racial attitudes. Both blacks and whites in low status neighborhoods have more negative attitudes and perceive more competition with minorities than those in high status neighborhoods. Interestingly, these effects are just the opposite for Asian Americans and Latinos. Among these groups, residence in a high status neighborhood corresponds with greater animosity toward minority out-groups (although not toward whites). These findings reflect the interaction between race, class, and immigration in

18 American society. Groups with fewer immigrant members (i.e., blacks and whites) feel greater inter-group competition with greater economic vulnerability; groups with more immigrant members (i.e., Asian Americans and Latinos) express out-group animosity relative to a position of economic well-being. Second, these results also highlight an Asian American exceptionalism with respect to racial attitudes. Unlike with other groups, higher levels of education among Asian Americans in Los Angeles corresponds with more negative stereotypes toward out-groups and highly educated Asian Americans do not feel less zero-sum competition with other groups. In addition, the effects of neighborhood contexts run completely opposite to those of other groups - Asian Americans who live in largely Asian neighborhoods actually have more positive perceptions of outgroups while those who live in more integrated settings express more negative out-group attitudes, particularly toward African Americans. It may be that the development of racial attitudes among Asian Americans, especially those who are foreign-born, is distinct from other groups. Intra-ethnic group differences in settlement patterns (Hum and Zonta 1999) and racial attitudes (Johnson, Farrell and Guinn 1997) among Asian Americans may also be at play. And, since length of residence in the U.S. among Asian American immigrants tends to be shorter than length of residence among Latino immigrants, Asian American immigrants may not be experiencing the long-term effects of living in a particular racial environment to the same degree as other groups.15 Most importantly, these surveys were conducted in Los Angeles immediately following the urban unrest when violence targeting Asian American small business owners was high. It is unclear whether these results are largely a product of a very distinct place and time. The remaining question is why do non Asian Americans who live among more out-groups express fewer negative stereotypes towards those groups? We believe that the explanation lies in the level of social exposure. The MCSUI data do not allow us to adequately test the contact hypothesis, particularly given the particular conditions under which contact is supposed to have its salutary effects. But this does not mean that racial integration is without its benefits. Rather than direct interpersonal contact, it may simply be exposure to people from outside one's own racial group (i.e., casual contact with people who live in comparable circumstances) that disconfirms unfavorable views of out-groups. For example, in multi-ethnic Los Angeles county, negative racial stereotypes may be countered in racially diverse residential settings such as Monterey Park but not countered in the hyper-segregated

19 white beach communities or city neighborhoods that consist almost entirely of Latino or black residents. Clearly, this is not always the case - Asian Americans in Los Angeles and the attitudes of whites toward Latinos in Boston does not benefit from such proximity. Nevertheless, for most groups, racial integration can have very high benefits for reducing racial animosity, particularly in a society with a growing number new minority groups from Asia and Latin America. Our findings indicate that the increasing racial diversity of America's metropolitan areas is likely to fuel greater racial conflict. This tension can be alleviated, however, through greater residential integration. In an increasingly multi-racial nation, neighborhood integration and social exposure will be the key for reducing intergroup conflict and hostility.


Although we do control for length of residence in the models above, the effects of length of residence are examined within a particular sample (Asian American, Latino, black, white), not in terms of between-group differences.

20 TABLE 1 Average Percent of Racial Group in Neighborhood by Race in Atlanta, Boston and Los Angeles Per. Asian Atlanta Blacks Whites 1990 Atlanta MSA* Boston Blacks Latinos Whites 1990 Boston CMSA* Per. Black Per. Latino Per. White

.01 (.02) .79 (.25) .01 (.03) .18 (.24) .02 (.02) .19 (.24) .02 (.02) .76 (.24) .02 .26 .02


.02 (.03) .64 (.29) .17 (.17) .15 (.19) .02 (.05) .15 (.24) .52 (.26) .29 (.21) .04 (.06) .08 (.15) .20 (.25) .59 (.25) .03 .06 .05


Los Angeles Asian Americans .35 (.17) .04 (.07) .27 (.16) .33 (.23) Blacks .05 (.08) .58 (.25) .27 (.18) .09 (.15) Latinos .09 (.12) .13 (.19) .60 (.26) .17 (.20) Whites .13 (.12) .08 (.14) .23 (.17) .55 (.23) 1990 Los Angeles County* .11 .11 .37


Source: Multi-City Study of Urban Inequality 1992-1994 (Standard deviations in parentheses). * Source: 1990 U.S. Census

21 TABLE 2 Negative Stereotypes of Out-Groups by Percent of that Out-group in Neighborhood ATLANTA Respondent is Black Percent Black in Neighborhood Percent White in Neighborhood -.535 (.150)** Respondent is White -.839 (.188)**

BOSTON Respondent is Black Percent Black in Neighborhood Percent Latino in Neighborhood Percent White in Neighborhood .228 (.351) -.336 (.250) -.183 (.205) Respondent is Latino .080 (.193) Respondent is White -.324 (.307) .690 (.331)**

LOS ANGELES Respondent is Asian Percent Asian in Neighborhood Percent Black in Neighborhood Percent Latino in Neighborhood Percent White in Neighborhood 2.16 (.537)** .230 (.287) -.050 (.186) -.155 (.264) -.184 (.215) -.345 (.234) Respondent is Black -1.62 (.325)** Respondent is Latino -1.01 (.256)** -.333 (.199)* Respondent is White -.303 (.243) -1.36 (.284)** -.781 (.403)*

Source: MCSUI. Cell entries are coefficients from OLS equations. Full Equations listed in Appendix B. ** p < .05, * p < .1.


Table 3 Negative Out-group Stereotypes by the Percent of In-Group in Neighborhood

Blacks (ncases = 2388) Percent Black Atlanta Atl. x Per. Blk. Boston Bos. x Per. Blk. Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared

Anti-Asian .242 (.119)** -.416 (.139)** .057 (.182) -.206 (.134) -.074 (.196) -.025 (.129) -.006 (.001)** .000 (.022) .073 (.053) -.015 (.020) .031 (.083) -.037 (.048) 1.52 (.117)** .033

Anti-Latino .269 (.133)** -.457 (.156)** -.055 (.205) -.062 (.150) -.200 (.220) -.164 (.144) -.004 (.002)** -.064 (.025)** -.043 (.059) -.019 (.022) .110 (.093) .142 (.054)** 1.66 (.131)** .050 Anti-Black .628 (.225)** -.129 (.142) -.938 (.228)** .152 (.270) -.007 (.002)** -.045 (.034) .006 (.085) -.009 (.030) .247 (.081)** .101 (.062) 1.75 (.270)** .091

Anti-White -.083 (.117) -.494 (.137)** .487 (.179)** -.259 (.131)** .169 (.193) -.061 (.126) -.007 (.001)** -.008 (.022) .054 (.052) -.002 (.019) -.028 (.081) .050 (.047) 1.76 (.115)** .021 Anti-White .349 (.178)** .212 (.113)* -.280 (.181) .093 (.214) -.003 (.002) -.016 (.027) -.127 (.068)* .033 (.024) .060 (.064) -.022 (.049) 1.03 (.214)** .011

Whites (ncases = 2117) Percent White Atlanta Atl. x Per. Wht. Boston Bos x Per. Wht. Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared Asians (ncases = 1049) Percent Asian Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared

Anti-Asian .126 (.145) -.439 (.133)** .329 (.191)* -.010 (.118) -.034 (.177) -.266 (.149)* -.003 (.001)** -.048 (.017)** -.070 (.046) -.014 (.019) -.085 (.062) .151 (.040)** 1.33 (.122)** .039 Anti-Black -.611 (.232)** .436 (.247)* -.006 (.003)** .125 (.034)** -.086 (.091) -.005 (.042) .064 (.123) .261 (.081)** 1.54 (.278)** .060

Anti-Black .627 (.196)** -.558 (.180)** .222 (.257) .112 (.160) -.395 (.240)* -.022 (.201) -.001 (.002) -.091 (.023)** -.035 (.062) -.009 (.026) -.024 (.083) .285 (.054)** 1.25 (.165)** .041 Anti-Latino -.196 (.206) .271 (.219) -.006 (.002)** .131 (.030)** -.115 (.081) .005 (.037) -.000 (.109) .181 (.072)** 1.36 (.247)** .052

Anti-Latino .521 (.192)** -.441 (.176)** -.044 (.252) .481 (.157)** -.538 (.235)** -.214 (.197) -.004 (.002)** -.106 (.022)** -.050 (.061) .024 (.025) -.092 (.082) .230 (.053)** 1.45 (.162)** .057 Anti-White -.422 (.168)** -.091 (.179) -.005 (.002)** .050 (.025)** .022 (.066) -.090 (.030)** .058 (.089) .062 (.059) 1.27 (.201)** .038

Latinos (ncases = 1691) Anti-Asian Percent Latino .487 (.183)** Boston .146 (.116) Bos. x Per. Lat. -.375 (.186)** Percent w/ H.S. Deg. .333 (.220) Age -.001 (.002) Education .001 (.028) Homeowner -.097 (.070) Length of Residence .029 (.024) Immigrant .027 (.066) Conservative .062 (.050) Constant .679 (.220)** r-squared .009


Source: MCSUI, excluded category is Los Angeles residence, standard error in parentheses. ** p < .05, * p < .1.

24 TABLE 4: Perceptions of Inter-Group Threat by Blacks and Whites in Atlanta, Boston, and Los Angeles Areas WHITES Percent White Atlanta Atl. x Per. Wht. Boston Bos. x Per. Wht. Per. w/ H.S. Deg. Af. Act. Women Age Education Homeowner Length of Resid. Female Immigrant Conservative Constant r-squared ncases Zerosum w/ Asians .754 (.363)** ------.094 (.272) -.360 (.405) -.818 (.463)* ---.007 (.004)* -.140 (.045)** -.005 (.129) -.042 (.053) .011 (.103) .038 (.156) .069 (.112) .987 (.337)** .056 472 Zerosum w/ Blacks .168 (.303) -.345 (.218) .754 (.338)** -.136 (.266) -.115 (.400) -.548 (.236)** ---.005 (.002)** -.192 (.030)** -.090 (.080) -.022 (.033) -.085 (.067) -.037 (.119) .059 (.069) 1.22 (.234)** .099 1145 Zerosum w/ Latinos .188 (.331) -------.202 (.251) .219 (.369) -.781 (.445)* ---.002 (.003) -.129 (.043)** .089 (.116) .012 (.049) .048 (.098) .065 (.141) .212 (.104) 1.23 (.324)** .051 497 Immig. Threat .606 (.207)** -.256 (.190) .202 (.272) .257 (.169) -.621 (.253)** -.445 (.213)** ---.003 (.002)* -.127 (.024)** -.125 (.066)* .021 (.027) -.013 (.055) -.423 (.088)** .177 (.057)** 1.38 (.178)** .051 2117 Oppose Aff. Act. for Asns. .469 (.214)** .217 (.197) -.264 (.282) .032 (.175) -.228 (.262) -.595 (.220)** .820 (.013)** .003 (.002)* -.093 (.025)** -.002 (.068) -.050 (.028)* .104 (.057)* -.119 (.091) .106 (.060)* 1.66 (.197)** .660 2117 Oppose Aff. Act. for Blks. .453 (.162)** .229 (.149) -.218 (.214) -.001 (.133) -.004 (.199) -.440 (.167)** .897 (.010)** .001 (.001) -.107 (.019)** -.021 (.052) -.017 (.021) .151 (.043)** -.003 (.069) .178 (.045)** .818 (.149)** .801 2117 Oppose Aff. Act for Lats. .562 (.196)** .392 (.180)** -.382 (.258) .115 (.160) -.241 (.240) -.428 (.201)** .852 (.012)** .004 (.002)** -.127 (.023)** -.064 (.062) -.036 (.026) .170 (.052)** -.041 (.084) .156 (.055)** 1.07 (.180) .719 2117

BLACKS Percent Black Atlanta Atl. x Per. Blk. Boston Bos. x Per. Blk. Per. w/ H.S. Deg. Af. Act. Women Age Education Homeowner Length of Resid. Female Immigrant Conservative Constant r-squared

Zerosum w/ Asians .757 (.263)** -------.549 (.298)* -.654 (.405) -.470 (.350) ---.004 (.004) -.135 (.058)** -.053 (.139) .025 (.051) .280 (.114)** -.006 (.177) .148 (.119) 1.83 (.306)** .108

Zerosum w/ Latinos .951 (.256)** -------.720 (.273)** -.215 (.406)** -1.29 (.359)** ---.007 (.004)* -.152 (.058)** -.103 (.142) -.004 (.050) .111 (.112) -.074 (.194) .169 (.121) 1.82 (.300)** .127

Immig. Threat .771 (.163)** -.351 (.191)* -.341 (.250) -.197 (.183) -.632 (.269)** -.028 (.177) ---.003 (.002) -.050 (.030)* .176 (.072)** -.017 (.027) -.037 (.059) -.382 (.114)** .108 (.065)* 1.06 (.165)** .064

Oppose Aff. Act. for Asns. .604 (.219)** .069 (.256) -.398 (.334) -.152 (.245) -.940 (.360)** .154 (.236) .624 (.020)** -.006 (.002)** -.083 (.040)** -.032 (.097)* -.006 (.036) .275 (.080)** -.126 (.152) -.107 (.087) 2.61 (.236)** .310

Oppose Aff. Act. for Blks. -.307 (.108)** -.071 (.126) .212 (.165) -.135 (.121) .456 (.178)** .038 (.117) .871 (.010)** .001 (.001) -.029 (.020) .049 (.048) .014 (.018) .228 (.040)** .243 (.075)** .033 (.043) .202 (.116)* .783

Oppose Aff. Act for Lats. .565 (.180)** .451 (.211)** -.203 (.275) .071 (.202) -.666 (.297)** -.009 (.195) .710 (.016)** .003 (.002) -.097 (.033)** -.086 (.080) .007 (.030) .186 (.066)** -.039 (.125) -.005 (.072) 1.27 (.194)** .469

25 ncases 808 770 2388 2388 2388 2388

28 Table 7 Negative Out-group Stereotypes by the Percent of In-Group in Neighborhood Controlling for Neighborhood Preference

Blacks (ncases = 2388) Prefer Black Neigh. Percent Black Atlanta Atl. x Per. Blk. Boston Bos. x Per. Blk. Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared

Anti-Asian .028 (.015)* .222 (.119)** -.425 (.139)** .069 (.182) -.199 (.134) -.074 (.196) -.014 (.129) -.006 (.001)** .001 (.022) .076 (.053) -.014 (.020) .032 (.083) -.037 (.048) 1.44 (.123)** .035

Anti-Latino -.003 (.017) .271 (.133)** -.456 (.156)** -.056 (.205) -.063 (.150) -.200 (.220) -.165 (.144) -.004 (.002)** -.064 (.025)** -.043 (.059) -.019 (.022) .110 (.093) .142 (.054)** 1.67 (.139)** .055 Anti-Black -.001 (.021) .629 (.225)** -.129 (.142) -.939 (.228)** .151 (.270) -.007 (.002)** -.046 (.034) .006 (.085) -.009 (.030) .247 (.081)** .101 (.062) 1.75 (.270)** .094

Anti-White .013 (.015) -.092 (.117) -.498 (.137)** .492 (.179)** -.256 (.131)** .172 (.193) -.056 (.126) -.006 (.001)** -.007 (.022) .055 (.052) -.001 (.019) -.027 (.081) .053 (.047) 1.73 (.121)** .021 Anti-White -.008 (.017) .355 (.178)** .213 (.113)* -.287 (.181) .089 (.214) -.003 (.002) -.017 (.027) -.128 (.068)* .034 (.024) .062 (.064) -.022 (.049) 1.05 (.219)** .011

Whites (ncases = 2117) Prefer White Neigh. Percent White Atlanta Atl. x Per. Wht. Boston Bos x Per. Wht. Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared Asians (ncases = 1049) Prefer Asian Neigh. Percent Asian Percent w/ H.S. Deg. Age Education Homeowner Length of Residence Immigrant Conservative Constant r-squared

Anti-Asian .082 (.016)** .168 (.144) -.401 (.133)** .252 (.191) -.022 (.118) -.039 (.177) -.228 (.149) -.003 (.001)** -.038 (.017)** -.072 (.046) -.016 (.019) -.093 (.062) .135 (.040)** 1.33 (.122)** .051 Anti-Black -.013 (.029) -.614 (.232)** .439 (.247)* -.006 (.003)** .125 (.034)** -.086 (.091) -.005 (.042) .062 (.123) .262 (.081)** 1.51 (.291)** .060

Anti-Black .287 (.021)** .479 (.187)** -.422 (.172)** -.050 (.247) .071 (.153) -.411 (.229)* .111 (.192) -.003 (.002)* -.057 (.022)** -.043 (.060) -.018 (.025) -.055 (.080) .230 (.052)** 1.11 (.158)** .123 Anti-Latino -.000 (.026) -.196 (.206) .271 (.219) -.006 (.002)** .131 (.030)** -.115 (.081) .005 (.037) -.000 (.109) .181 (.072)** 1.36 (.258)** .052

Anti-Latino .216 (.021)** .410 (.187)** -.330 (.172)** -.249 (.252) .387 (.153)** -.549 (.235)** -.113 (.192) -.006 (.002)** -.079 (.022)** -.056 (.061) .017 (.025) -.115 (.082) .188 (.053)** 1.34 (.158)** .105 Anti-White .046 (.021)** -.430 (.168)** -.080 (.179) -.005 (.002)** .053 (.025)** .023 (.066) -.089 (.030)** .052 (.089) .062 (.059) 1.13 (.210)** .043

Latinos (ncases = 1691) Anti-Asian Prefer Latino Neigh. .003 (.017) Percent Latino .485 (.183)** Boston .146 (.116) Bos. x Per. Lat. -.373 (.186)** Percent w/ H.S. Deg. .335 (.220) Age -.001 (.002) Education .002 (.028) Homeowner -.096 (.070) Length of Residence .029 (.024) Immigrant .027 (.066) Conservative .062 (.050) Constant .669 (.220)** r-squared .009

Source: MCSUI, excluded category is Los Angeles residence, standard error in parentheses. ** p < .05, * p < .1.

32 Appendix B Continued Los Angeles

Asian Americans (n = 1049) Percent Asian American Percent Black Percent Latino Percent White Percent w/ H.S. Degree Age Education Homeowner Length of Residence Female Immigrant Conservative Constant r-squared Blacks (n = 1116) Percent Asian American Percent Black Percent Latino Percent White Percent w/ H.S. Degree Age Education Homeowner Length of Residence Female Immigrant Conservative Constant r-squared

Anti-Black -.651 (.232)** ---------.511 (.248)** -.010 (.003)** .100 (.036)** -.092 (.091) .003 (.042) -.194 (.079)** .089 (.123) .260 (.081)** 1.65 (.281)** .066 Anti-Asian ---.233 (.118)** -------.024 (.181) -.010 (.002)** .004 (.032) .154 (.075)** -.007 (.027) -.025 (.060) .071 (.144) -.133 (.063)** 1.57 (.151)** .020

Anti-Black ---2.16 (.537)** ------.765 (.243)** -.007 (.003)** .102 (.035)** -.002 (.091) .002 (.042) -.173 (.079)** .105 (.123) .241 (.081)** 1.12 (.250)** .073

Anti-Latino -.217 (.206) ---------.312 (.221) -.006 (.002)** .118 (.032)** -.119 (.081) .009 (.037) -.104 (.070) .013 (.109) .180 (.072) 1.42 (.250)** .054

Anti-Latino ------.230 (.287) ---.510 (.285)* -.006 (.002)** .119 (.032)** -.116 (.081) .008 (.037) -.098 (.070) .013 (.110) .186 (.072)** 1.12 (.310)** .053 Anti-Latino -------.155 (.264) ----.312 (.219) -.001 (.002) -.042 (.039) -.004 (.091) -.033 (.033) -.071 (.072) -.042 (.174) .082 (.076) 1.86 (.273)** .006

Anti-White -.445 (.168)** ----------.046 (.180) -.005 (.002)** .036 (.026) .018 (.066) -.085 (.030)** -.115 (.057)** .073 (.089) .060 (.059) 1.33 (.204) .042 Anti-White ----.024 (.120) -------.295 (.184) -.005 (.002)** -.002 (.028) .041 (.076) .005 (.028) -.078 (.061) .015 (.146) -.065 (.064) 1.87 (.153) .011

Anti-White ----------.050 (.186) .102 (.246) -.005 (.002)** .039 (.026) .038 (.066) -.087 (.030)** -.103 (.058)* .067 (.090) .066 (.059) 1.05 (.196) .036 Anti-White ----------.184 (.215) -.235 (.196) -.005 (.002)** .001 (.032) .035 (.076) .000 (.028) -.087 (.061) .023 (.146) -.068 (.064) 1.84 (.151) .011

Anti-Asian Anti-Latino -1.619 (.325)** ------.320 (.142)** ------------.117 (.174) -.311 (.219) -.006 (.002)** -.000 (.002) .017 (.032) -.032 (.039) .119 (.075) -.019 (.091) -.009 (.027) -.045 (.033) -.033 (.059) -.091 (.072) .087 (.143) -.019 (.174) -.139 (.063)** .073 (.076) 1.12 (.250)** 1.64 (.182)** .038 .010


APPENDIX B Continued Latinos (n = 1116) Percent Asian American Percent Black Percent Latino Percent White Percent w/ H.S. Degree Age Education Homeowner Length of Residence Female Immigrant Conservative Constant r-squared Anti-Asian ------.437 (.212)** ---.271 (.282) .000 (.002) -.001 (.035) -.104 (.082) .025 (.031) .057 (.059) .033 (.080) .022 (.065) .683 (.277)** .011 Anti-Asian -1.01 (.256)** ---------.048 (.169) .001 (.002) -.001 (.034) -.156 (.075) .003 (.031) .056 (.059) .033 (.079) .009 (.079) 1.14 (.142)** .023 Anti-Black ------.389 (.270) ----.106 (.359) -.007 (.003)** -.094 (.044)** -.019 (.105) .015 (.039) -.018 (.075) .334 (.101)** .082 (.082) 1.96 (.351)** .048 Anti-Black ----.333 (.199)* -------.493 (.200)** -.008 (.003)** -.101 (.045)** .000 (.105) .016 (.039) -.015 (.075) .351 (.102)** .088 (.082) 2.43 (.182)** .049 Anti-White ------.334 (.206) ---.075 (.274) -.005 (.002)** -.006 (.034) -.149 (.080)* .042 (.030) -.042 (.057) .034 (.078) -.143 (.063)** 1.20 (.269)** .027 Anti-White ----------.345 (.234) -.052 (.225) -.004 (.002)* .001 (.034) -.159 (.080)** .042 (.030) -.045 (.057) .001 (.079) -.144 (.063)** 1.52 (.144)** .026

Whites (n = 859) Percent Asian American Percent Black Percent Latino Percent White Percent w/ H.S. Degree Age Education Homeowner Length of Residence Female Immigrant Conservative Constant r-squared

Anti-Asian ---------.116 (.172) -.768 (.274)** -.004 (.002)* -.070 (.025)** -.102 (.067) .016 (.028) -.065 (.056) -.031 (.078) .127 (.059)** 1.68 (.189)** .046

Anti-Asian -.303 (.243) ----------.622 (.199)** -.004 (.002)** -.068 (.025)** -.102 (.067) .019 (.028) -.067 (.056) -.023 (.079) .126 (.059)** 1.66 (.181)** .047

Anti-Black ---------.653 (.250)** -.271 (.397) .002 (.003) -.050 (.036) .040 (.098) .029 (.041) -.031 (.082) .100 (.114) .343 (.086) 1.11 (.274)** .036

Anti-Black Anti-Latino -------1.36 (.284)** ------------.349 (.241) .238 (.289) .271 (.383) .002 (.003) .001 (.002) -.064 (.036)* -.095 (.035)** .021 (.097) -.126 (.094) .009 (.041) .017 (.040) -.014 (.081) .296 (.079)** .073 (.113) .009 (.110) .366 (.085)** .267 (.083)** 1.21 (.268)** .810 (.264)** .054 .044

Anti-Latino -------.781 (.403)* ----.184 (.489) -.001 (.003) -.092 (.035)** -.110 (.094) .020 (.040) .296 (.079)** .009 (.110) .256 (.083)** 1.48 (.477)** .046

Source: MCSUI Data/1990 U.S. Census, standard errors in parentheses, ** p < .05, * p < .1.



Campbell, Andrea, Cara Wong, and Jack Citrin. 1999. "V.O. Key Visits California: Context, Racial Threat and Voting o Propositions 187, 209, and 227." Paper presented to the Annual Meetings of the American Political Science Association, Atlanta, Georgia.

39 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in

Self-esteem: The costs and causes of low self-worth
Biased-Based Policing in Nashville: Police and Community Perceptions
Microsoft Word - Research Roundup Parenting.doc
Microsoft Word - FiddlerStudyGuide.doc