Racial Diversity and White Political Views

Racial Diversity and White Political Views:

How Latino Context Transforms White Political Orientations

Zoltan Hajnal, University of California, San Diego

Marisa Abrajano, University of California, San Diego

Abstract

Immigration is profoundly changing the racial makeup of America. In this article, we seek to understand if and how increasing racial diversity in the form of larger Latino and Asian

American populations structures the politics of individual white Americans. We show that local

Latino context has broad effects on the policy positions and political ideology of white

Americans. Using large and geographically diverse samples from the 2000 and 2004 National

Annenberg Election Studies, we find that, all else equal, whites in zip codes with larger Latino populations are more conservative on a range of policy issues linked to immigrants and more apt to identify as conservative. This rightward shift suggests that racial diversity associated with immigration is re-shaping American politics.

Zoltan Hajnal and Marisa Abrajano can be reached at the Department of Political Science,

UCSD, La Jolla, CA 92093-0521. Please direct all correspondence to Zoltan Hajnal [email protected]. (858)822-5015. Marisa Abrajano is at (858)534-7201 [email protected] The authors wish to thank Nicholas Warner for his exemplary research assistance and James

Gimpel, Rodney Hero, and David Leal for their thoughtful comments.

Immigration has transformed the racial demographics of America. Asian Americas are the fastest growing racial group and Latinos have surpassed African Americans as the largest minority. White Americans have declined from 89 percent of the population in 1960 to 66 percent today. And the future is likely to bring even more change. The Census projects that by sometime in the middle of this century, white America’s majority status will be in jeopardy.

The impact of immigration is evident not just in demographic numbers. Immigration has stirred a wide array of fears related to economic losses and a cultural demise (Huntington 2005,

Borjas 2001). In the political arena, immigration has infused debates over welfare, crime, and partisan politics. Proposition 187 in California and the welfare reform act of 1996 are only two of the most prominent examples. The debate is, of course, not all one-sided. Many highlight the economic benefits and the social contributions of the immigrant population (Bean and Stevens

2005, Portes and Rumbaut 2006). Others point to the near complete assimilation of immigrants and their children into the American way of life (Alba and Nee 2005). Regardless of which side of the debate one is on, there seems little doubt that America’s changing demographics are influencing the politics of this nation.

In this article, we investigate the extent of this influence on the political orientations of individual white Americans. One of the most inescapable features of immigration today is its uneven nature. Some Americans live in areas where there is almost no perceptible evidence of immigration and others live in neighborhoods, cities, and states that have been dramatically reshaped. We take advantage of this geographic variation to assess the impact of larger Latino and Asian American populations on individual attitudes about a range of political issues related to immigration. 2 We are, by no means, the first to examine racial or immigrant context. We are, however, among the first to examine the broader political implications of these contexts. The vast array of scholarship on immigrant and racial context has been narrowly focused on how context affects attitudes about, behavior towards, and policy on the minority group itself (but see Johnson 2001,

Hero 1998, Alesina et al 1999). That research has at times shown that larger minority or immigrant populations are associated with more negative attitudes about minorities and the immigrant population, violence against these groups, and heightened opposition to expanded immigration (eg Ha 2009, Hood and Morris 1998, Stein et al 2000, Green et al 1998).

We believe, however, that the effects of minority context could be much broader. There are two reasons for this belief. First, given that immigration has permeated debates about a wide range of policy questions, there is reason to suspect that it could have more far-reaching implications for policy views. The fact that the Democratic and Republican parties have, at least at times, taken sharply divergent tacks on immigration also suggest that there may also be a strong partisan component to the effects of immigration. Second, such broad ranging contextual effects are far from unprecedented in American political history. An array of white political activity including candidate choice, policy views, and violence has, in the past, been linked to the size of the local black community (Black and Black 1973,Fosset and Kiecolt 1989, Corzine et al

1983). Given that Latinos have replaced blacks as the largest ethnic/racial minority population, it is at least plausible that Latinos and other immigrant groups have become more central in the political thinking of white America.

We also hope to advance upon existing studies by dealing with selection bias problems in a novel and hopefully instructive way, by distinguishing between the effects of Latinos and

Asian Americans – two groups that have very different socioeconomic standing and quite distinct stereotypes in American society, by relying on newer data that can capture the growing 3 significance of immigration in America, and by employing a unique data set that offers greater geographic variation and thus a potentially more telling look at how wide-ranging the effects of immigration have reached.

How Context Might Work

Theoretically, how might white Americans react to living in contexts with large minority or immigrant populations? The literature offers three competing theories about the impact of context: 1) a racial threat perspective, 2) a racial contact theory, and 3) an individualistic account. Probably the most prominent view is that proximity tends to increase the threat posed by out-groups. As developed by Key (1949) and Blalock (1967), this racial group threat perspective maintains that proximity enhances real (or perceived) competition over jobs, homes, political offices, and other scarce resources. Individuals in contexts with larger minority populations should, therefore, express greater racial animosity and be especially supportive of a host of policies aimed at maintaining the in-group’s political and economic privilege.

A second view contends that racial animosity is more often the result of inaccurate perceptions about the minority out-group than it is about real competition over resources. From this racial contact view, personal interaction with minority out-groups exposes majority group members to new, more accurate information about minority group members that should disconfirm negative stereotypes and allow for the development of more favorable views (Allport

1954, Jackman and Crane 1986, Kinder and Mendelberg 1995). Understanding and approval should, thus, increase with the size of the local minority population.1

Finally, a third possibility is that context has no independent effect on majority-minority relations. Either because racial views are predicated on a rigid type of prejudice that is impervious to change or because other individual characteristics like education or economic

1 It is important to note, that in the analysis that follows, we do not directly test the contact hypotheses. Our measures of racial are probably correlated with contact but only weakly 4 status are the primary factors shaping racial views, geographic context is largely irrelevant for understanding inter-group conflict and cooperation. If true, racial views should be unrelated to geographic context once we control for individual characteristics.

Past Patterns, Current Studies

Existing research provides an array of evidence relating to the impact of context on white

Americans. A wide range of studies has demonstrated a strong relationship between black context and white political behavior during different periods in American politics.

Proximity to blacks has been linked to greater racial antagonism in a variety of forms including more negative views of blacks themselves (Dixon 2006, Taylor 1998, Quillian 1996 but see Oliver and Wong 2003, Kinder and Mendelberg 1995) violence against blacks including riots and lynching (Corzine et al 1983), support for racist candidates (Black and Black 1973 but see Voss 1996), and greater opposition to policies designed to aid blacks (Giles and Evans 1986,

Fosset and Kiecolt 1989, Key 1949).2 Notably, the confirmed effects of black context are even broader. Larger black populations have also been linked to more conservative views on a range of implicitly racial policies (Hero and Preuhs 2006, Fellowes and Rowe 2004), to diminished public goods (Hero 1998, Alesina et al 1999) and to large scale defection from the Democratic

Party (Huckfeldt and Kohfeld 1989). While several more recent studies have found that the effects of black context are contingent on socioeconomic status (Oliver and Mendelberg 2000,

Branton and Jones 2005), there seems little doubt that black racial context has often been a important force shaping white political choices and actions.

A number of studies that have shifted the focus to immigrant context have also found that more immigrants can represent a threat and drive a white backlash. In particular, larger local immigrant populations have been associated with more hostile views toward the immigrant

2 These findings say little about the potential of certain forms of inter-racial contact. Studies that measure actual contact tend to find that inter-racial contact can reduce racial enmity (Pettigrew 1997, Dixon and Rosenbaum 2004). 5 population, violence against immigrants, and heightened opposition to expanded immigration

(Ha 2010, Campbell et al 2006, Stein et al 2000). Similarly, a range of comparative research has found that larger immigrant populations at the national level are associated with more negative views of immigrants and more support for restrictionist immigration policy (Quillian 1995,

McLaren 2003, Dustman and Preston 2001).

Although the bulk of the research points to an immigrant backlash, there are at least a few studies that reach very different conclusions about immigrant context. Some researchers have found that immigrant context has few significant implications (Dixon 2006, Taylor 1998, Burns and Gimpel 2000). And still others have revealed a positive relationship between immigrant context and white views – especially when that context is primarily composed of Asian

Americans (Ha 2010, Hood and Morris 1998, 2000, Fox 2004). Finally, several more recent studies have suggested that the impact of ethnic or immigrant context is contingent on either the socioeconomic status of the neighborhood, the skill level of the immigrants, or the national political debate (Hopkins 2010, Scheve and Slaughter 2001, Oliver and Mendelberg 2000,

Branton and Jones 2005).

One read of these mixed findings is that immigrant context has a real but limited and contingent impact on white political behavior. There are, however, reasons to hesitate before concluding that immigrant context is not a central feature in the minds of white Americans.

Understanding a New Racial Threat and its Political Implications

Perhaps, the most important reason to re-assess the role of race in shaping white views is that the world is changing. Latinos have recently replaced African Americans as the largest minority group. Asian Americans are now the fastest growing racial and ethnic group. By contrast, the number of African Americans living in this country is close to stagnant. One core result of these 6 demographic changes is that immigration, in general, and Latinos and Asian Americans, in particular, may be becoming more central in the minds of white Americans. The growing reach of immigration suggests that existing studies may be too narrow. Almost every study that has focused on the contextual impact of immigrants has limited its purview to minority specific outcomes (but see Hopkins 2009, Hero and Preuhs 2007, Hero 1998). They study how context affects attitudes toward immigrants (Scheve and Slaughter 2001, Fox 2006), how context affects violence toward minorities (Green et al 1998), or how context affects racial policy and immigration policy (Hood and Morris 1998, 2000, Stein et al 2000, Burns and Gimpel 2000,

Branton and Jones 2005). Although these studies are extremely informative, we maintain that their focus is too limited and we suggest that the effects of immigrant context could reach more broadly into a range of policy debates that are central to the politics of the nation.

Why Context Could Affect Policy Views

If immigrants are becoming more central in the thinking of white Americans and if immigrants can and do pose a threat as many previous studies suggest, then there is reason to believe that minority context could have broader political implications. The first is that immigrant based groups and especially Latinos are visibly gaining political power. It would be difficult for most white Americans to have missed the growth of the Latino vote (up 64% between 2000 and 2008), dramatic increases in the number of Latino elected officials (from almost none to over 5000 nationwide), and regular references to the sleeping Latino giant. All of this clearly politicizes the Latino population and increases the potential that they are seen as a threat.

Second, debates about core policy concerns are increasingly being infused with references to illegal immigrants and the broader Latino population. Take health care, for example. After much of the recent debate on health care reform focused on whether or not the 7 Democratic reform package would cover undocumented immigrants, a recent PEW poll found that 66% of those opposed to the plan reported that they were opposed because the plan might cover illegal immigrants. Welfare reform since the 1990s has similarly been permeated with discussions of Latinos and illegal immigrants. California’s Prop 187 which sought to restrict public services to illegal immigrants is only the most prominent example. The fact that just under 70 percent of whites view Latinos as particularly prone to be on welfare suggests that the connection between Latinos and welfare is now firmly in place (Bobo 2001). Latinos and crime is another readily apparent script throughout the nation. The story on crime is analogous. Given that crime, terrorism, and illegal immigration account for fully 66 percent of network news coverage of Latinos, it should not be surprising to find that white Americans often hold stereotypes of Latinos as being particularly violent (Bobo 2001).

Third, evidence of race impacting core policy views would not be new. Research has already shown that non-racial policy can be racially coded. In particular, there is evidence that individual policy preferences on welfare, education, crime, and a host of other core issue arenas have, at least at some points in the past, been shaped by attitudes toward blacks (Gilens 1999,

Kinder and Sanders 1996).

Given the increasingly central role played by immigration and Latinos, it is crucial that we expand the scope of our research. We need to consider the effects of racial context not just on how we think about minorities themselves but on how we think about the broader array of policies that are at different times at least implicitly linked to immigration. Only through a broader study can we begin to understand the extent of immigration’s impact on American politics.

Latinos vs Asian Americans

8 Another concern is that most existing studies of immigrant context lump all immigrants together.

We argue that Latinos and Asian Americans should be examined separately because the two groups are likely to be viewed very differently by white Americans. Asian Americans and

Latinos tend to have noticeably different socioeconomic standing and the two groups tend to be subject to very different stereotyping. In terms of economic status, there is a fairly clear racial hierarchy. Asian Americans and whites are, on average, well off while Latinos and African

Americans are, on average, significantly worse off. The median household income of Asian

Americans, for example, was almost $60,000 in 2005, roughly $10,000 more than the figure for whites. At the opposite end of the spectrum, median black and Latino households only earned about $30,000. Also, stereotypes of Latinos, Asian Americans, and African Americans differ dramatically (Lee 2000). Whereas large numbers of white Americans tend to view Latinos as less intelligent, more prone to welfare, and not especially hard working, the bulk of white

Americans tend to view Asian Americans in roughly the opposite fashion (Bobo 2001). These distinct stereotypes are also consistent with survey findings on inter-group attitudes. In a poll asking whites how well they generally get along with other racial groups, 92 percent say they get along with Asians, while only 67 percent felt the same way about Latinos (NCCJ 2005). Exactly what these two patterns imply in terms of a threat to the white community is not clear

Selection Bias

Finally, most studies of contextual effects have been impacted by concerns about selection. We know, for example, that whites with more negative views of racial and ethnic minorities have in the past migrated in large numbers away from neighborhoods and cities where black population growth was dramatic (Clark 1992, Massey and Hajnal 1995). If relocation decisions are driven partly by attitudinal factors, then simple, uncorrected estimates of the connection between context and attitudes will be biased. In our analysis, we employ an 9 instrumental variables procedure that may help us to get around the selection problem. As we will explain below, we use exogenous state level demographics to instrument for neighborhood level demographics that are the likely to be affected by endogenous selection effects.

To assess the effects of context on white views we turn to the 2000 and 2004 National

Annenberg Election Surveys (NAES). The NAES is an ideal tool because it has a large sample

(over 50,000 respondents per survey) and extensive geographic variation (respondents from over

14,000 different zip codes). In addition, the surveys contain questions on a range of policy issues, basic ideology, and measures of party identification. Since the NAES identifies the zip code of each respondent, contextual data can be merged into the sample. Data on local demographics, including immigrant population size broken down by country of origin come from the 2000 U.S. Census of Population and Housing. In subsequent analyses we present results from the 2000 NAES – largely because it has a wider array of policy questions – but analysis of the 2004 NAES leads to a nearly identical set of conclusions.

Our analysis consists of a series a regression models that simultaneously incorporate immigrant context, individual characteristics, and economic context to try to explain the individual political orientations of white Americans. We focus only on policy areas that have been clearly and regularly linked to immigrants and the issue of immigration. Specifically, we examine views on immigration (How serious of a problem is immigration into the United

States?), poverty (How serious of a problem is poverty?), inequality (Should the federal government try to reduce income differences between rich and poor Americans?), four different

10 health-care related questions,3 four different questions on criminal justice,4 and one question on education policy.5 To determine if racial context has influenced more fundamental political identities, we also incorporate ideology. Liberal-conservative ideology is measured with a standard self-identified question (Generally speaking, do you consider your political views very conservative…very liberal?).

To measure context, we use zip code demographics provided by the US Census. There are a number of different ways to measure immigrant context. The most obvious is to simply record the percent foreign born of the area. But as we have already mentioned, there is reason to suspect that white Americans will react differently to the two main immigrant populations from

Asia and Latin America. Thus, another logical set of alternatives might be percent foreign born

Asian and percent foreign born Latin American. But here there is a real question as to whether white Americans distinguish between the foreign born and the native born of these two groups or whether they tend to lump all Latinos and all Asian Americans together. Patterns in racial hate crimes and patterns in past discriminatory actions by the government also suggest that foreign nativity can be irrelevant to white actions (Kim 1999, Almageur 1994). Systematic analysis of racial stereotypes also indicates that at least for Asian Americans, one of the main stereotypes of the group is as ‘perpetual foreigners’ unable or unwilling to assimilate (Lee 2000, Wu 2003).

Thus, it is possible that regardless of nativity, all Asian Americans and Latinos who look

‘different’ will be lumped together in the eyes of white Americans. For this reason, we focus on

3 The questions on health care policy are: 1) Do you favor or oppose using government funds to make sure that every child in the US is covered by health insurance? 2) How serious of a problem is having Americans without health insurance? 3) Should the federal government spend more or less money on health care for the poor? 4) Should the federal government spend more or less money on health care for the elderly? And 5) Should the federal government spend more or less money for health care to cover the uninsured? 4 The questions on criminal justice are: 1) Do you favor or oppose the death penalty? 2) How serious a problem is the number of criminals who are not punished enough? 3) Do you personally favor or oppose requiring a license for a person to buy a handgun? 4) Should the federal government do more to restrict gun purchases? 5 The question is: should the federal government provide more financial assistance to public elementary and secondary schools? 11 percent Latino and percent Asian American as our primary measures of context. As such our analysis should be considered first and foremost as a test of racial context rather than immigrant context. Since it is not clear in advance what the best measure of context is, we repeat our analysis with each of these different measures to try to determine which best explains white views. The results of these tests are described below.

Given that the Annenberg survey identifies the zip code, county, city, metropolitan area, and state of each respondent, we could have chosen to measure context at a number of different levels. However, since we seek to assess the effects of living near a large minority population, we believe that the smallest available unit – a zip code – is most appropriate. In larger geographic units such as counties or cities, residential segregation may mean that there is little correlation between diversity at the city or metropolitan level and the personal environment of a respondent who lives in that city or metropolitan area. Since zip codes are relatively small, there is more likely to be a correspondence between the demographics of the zip code and the experiences of any given respondent in that zip code. Although zip codes are our best available option, we readily admit that they represent an imperfect measure of local context in that they are not always drawn around well-defined neighborhoods in which individuals from the neighborhood regularly interact.6

Finally, to ensure that the relationships we see between context and white views are not spurious, we include several measures of individual characteristics and several different measures of neighborhood context that have also been tied to political views. In terms of individual demographic characteristics, we control for education (the last grade of school completed sorted into nine categories), household income (divided into nine categories), gender, age in years, whether the respondent is unemployed or not, and whether anyone in the household

6 It is also not clear how aware individual whites are about the racial makeup of their neighborhoods (Wong 2007). The “noise” in our contextual measure should, however, only serve to reduce significance levels. 12 is a union member or not, and whether or not there are any children in the household Also, since policy views are often driven by political ideology and party identification, we include both measures in our model. Party identification is measured with a standard self-identified question

(Generally speaking, do you usually think of yourself as a Republican, a Democrat, an

Independent or something else?).

To help ensure that other features of the local neighborhood are not driving white attitudes, we include controls for the socioeconomic status of the zip code. Specifically, we incorporate median household income and percent college educated in the respondent’s zip code. Since

Branton and Jones (2005), Fetzer (2000), and others have highlighted the impact that economic downturns can play in shaping white views, we also incorporated a measure of local unemployment in our analysis. To help discount the role of whites selecting into and out of neighborhoods with more or fewer minorities, we controlled for mobility by adding a variable that measures the number of years the respondent has lived in their current address. Also, to ensure that our analysis is not picking up regional effects, we include dummy variables for region. As additional controls, we distinguish between respondents who live in urban areas, suburban areas, and rural areas. Our sample is restricted to those respondents who self-identify as both white and non-Hispanic.

Dealing with Selection Bias

As we have already noted, one concern with much of the existing literature is selection.

Individuals with a particular set of political views may select into or out of different types of neighborhoods. The most likely possibility is that white Americans with more negative views of minorities will leave neighborhoods and cities that have disproportionately large numbers of minorities. Indeed, there is clear evidence of wide scale white flight from neighborhoods and

13 cities with large black populations (Clark 1992, Massey and Hajnal 1995).7 And if past patterns are being repeated today, we may be seeing similar movement away from predominantly Latino or Asian American locales. If this kind of selection is occurring then any simple correlation between immigrant context and white views would likely understate the role of racial threat and might even produce a spurious positive relationship between immigrant context and white views.

Indeed, post-estimation diagnostics (Hausman and Durban-Watson) of our two-stage least squares estimation procedure reveals substantial endogeneity between each respondent’s zip code demographics (percent Latino and percent Asian American) and their policy views. As such, the issue of endogeneity is a concern that must be accounted for in our analysis.

Most studies recognize this problem but few are able to deal with it effectively.

Fortunately, we can use larger geographic units to help model away the endogeneity of neighborhood selection. All of our respondents live not only in neighborhoods but also in states.

The states where individuals reside in no small part shape the diversity of the neighborhood they live in. It is hard, if not impossible, to live in a primarily black neighborhood in Idaho or

Wyoming or several other states. At the other end of the spectrum, it is often hard to find an all- white neighborhood in cities in California or Texas. In fact, for the Annenberg respondents, the three state racial contexts that we examine - percent Hispanic, percent Asian American, and percent African American - are correlated between .43 and .59 with the same variables at the zip code level.

The other key factor with state context is that movement and selection across states is much less common than movement and selection across neighborhoods. Just as we know that one’s neighborhood choice can be significantly impacted by one’s racial views (Bobo and

Zubrinsky 1996, Clark 1992, Oliver and Wong 2003), we also have fairly strong evidence

7 It is also quite possible that minorities choose not to move into areas where they expect to face high levels of racial intolerance. 14 indicating that choice of state is largely unrelated to racial views. The first piece of evidence is that very few people move across states. According to the Census about one half of one percent of all Americans move across state borders in any given five-year period (Census Bureau 2003).

If few people move from state to state, there cannot be a significant amount of selection occurring at the state level. Second, studies of inter-state migration – unlike studies of neighborhoods or municipalities – have not found that race plays a significant role in migration decisions (Greenwood 2000). Mobility across states is relatively costly and is thus driven almost exclusively by employment and family (Gimpel 1999). Concerns about the race and ethnicity of one’s neighbors, may be enough to move one out of the neighborhood or even to next municipality but they are seldom strong enough to move one out of the state.

As such, state context represents a relatively exogenous context that can predict local context and serve as an instrument to deal with selection at the local level (see Dustmann and

Preston 2001 for a more formal test of this proposition). We take advantage of this set of state level instruments and in all subsequent regressions we use a two-stage least squares estimation procedure with percent Latino, percent Asian American, percent African American at the state level as instruments for percent Latino, percent Asian American, percent African American at the zip code level. While this instrument is far from perfect, it should substantially reduce concerns about selection affecting our results. We report the results from the first-stage estimation process in supplementary documentation.

Context and White Policy Views

Is living near large minority populations associated with distinct political views? And more specifically are white Americans reacting to significant numbers of Latinos and Asian

Americans in their local neighborhoods with a racial backlash and a shift to the right politically?

Our first answers to these questions emerge in Table 1. The table displays the second stage 15 estimates of the two-stage least squares regression that models individual white views on immigration.8 [TABLE 1 HERE]

Immigration

We start with views on immigration in column one because it is the policy area that can be most logically tied to immigrant context. If Latino and Asian American context matters anywhere, it should be here. The estimates indicate that local context is, in fact, a significant factor shaping views on immigration. In a pattern that we will see repeated, white Americans who live in neighborhoods with larger Latino populations have more conservative views and are, in this case, significantly more likely to view immigration as a serious problem. The magnitude of the effect, as Figure One shows, is substantial. All else equal, an individual white American living in a zip code that is comprised almost totally of Latinos is predicted to be one half full category more conservative on a four category immigration scale than a white American living with no Latinos in the area. In other words, living in a Latino context as opposed to a white context might mean moving from viewing immigration as a ‘not too serious’ problem to a seeing it as a ‘serious’ problem. This pattern comports well with a racial threat view of the world.

As we expected, Asian American context does not have the same effect. Whites who live in areas with high concentrations of Asian Americans are no more or less likely to view immigration as a serious problem. These divergent effects for Latinos and Asian Americans mirror the divergent views that white Americans have of these two pan-ethnic groups. The estimates presented in column one also begins to inform us about the role of African American context in American politics. The results, as one would expect, indicate that whites living with larger concentrations of African Americans are no more or less concerned about immigration.

8 We use robust errors and cluster by zip code to account for the non-independence of observations within the same zip code. 16 Our results on attitudes towards immigration policy essentially reaffirm the existing work of previous scholars (eg Brader et al 2008). In the remaining estimates presented in Table 1,

Figure 1 and in subsequent analyses, we examine policy areas that scholars of racial context have largely ignored. We start with the policy questions that we believe are most closely and most logically linked with concerns about immigrants and racial minorities and then slowly shift to away to less immigration focused but more core political decisions.

Health Care and Social Welfare

The results in the remaining columns of Table 1 indicate that the impact of minority context does, in fact, extend beyond the confines of immigration policy. The table reveals the effect of context on two policy areas often tied to the immigration debate – social welfare policy and issues related to health care. In each case, subsets of the American population have expressed concern that immigrants are a fiscal burden on public resources (Tichenor 2002). And in each of these two policy areas, attempts have been made to limit the resources expended on immigrants (eg Proposition 187). The results suggest that Latino context is affecting these policy debates. In both policy areas, white Americans who live in proximity to large numbers of

Latinos tend to have more conservative views. All else equal, whites living in zip codes with larger Latino populations are less likely to support the federal government’s efforts to reduce income inequality, less likely to seek increased spending on health care for the poor, less likely to want to do more to cover the uninsured, and almost significantly less likely to view poverty as a serious problem. Moreover, Figure One indicates that these effects are substantial. Latino context has a similar impact for two other questions in the Annenberg survey that address health care. Whites living in areas with larger Latino populations are significantly less apt to want to spend on Medicare and significantly less eager to expand resources to ensure that all children are covered by health care. 17 The implications from these findings is that Latino context now appears to be shaping core policy views across the American public. And it is doing so in a way that mirrors the negative reactions that have often faced the African American community in the past (Gilens

2001, Kinder and Sanders 1994). In contexts where Latinos are prominent (and perhaps threatening), whites tend to support a reduction in services and expenditures that would benefit the bottom rungs of society.

Across the two policy areas of welfare and health, we see again that Asian American context has a different and largely negligible relationship with white views. In three of the four cases, proximity to Asian Americans is unrelated to white policy preferences. The only exception is with respect to white attitudes towards poverty. And in this case, Asian American context is associated with more liberal rather than more conservative views. One reading of the divergent contextual effects for Latinos and Asian Americans is that the two pan-ethnic groups do, in fact, represent very different threats and possibilities to members of the white community.

The other interesting conclusion that emerges from the results presented in Table 1 concerns the relevance of black context. In only one case do the effects of black context even approach statistical significance. In this particular situation, there is a hint that whites living in areas with large black populations are more opposed to expanding services to the needy. But for the other three policy questions, black context seems irrelevant.

Criminal Justice

The estimates from Table 2 further the same themes by illustrating a range of links between context and criminal justice policy. Once again, we see the importance of Latino context and the backlash that a large Latino population seems to engender. Whites who live with higher concentrations of Latinos are significantly more likely to approve of the death penalty.

As before, we also see a very different effect for Asian American context. Proximity to Asian 18 Americans is either unrelated to white views or is associated with greater leniency on the part of white respondents. The contrast between the punitive nature of the response to Latino context and the forgiving nature of the response to Asian American context suggests that white perceptions vary widely about how threatening or deserving the two pan-ethnic groups are

(Brader et al 2008). Echoing past research, whites faced with a large local black population are also apt to want to punish criminals more harshly – although the effect is not robust across the two questions (Gilliam and Iyengar 2000).

Education

The results for educational policy are similar. On the one question in the 2000 NAES on school spending, higher concentrations of Latinos are associated with more conservative white views, higher numbers of Asian Americans have no observable effect, and more blacks once again means less liberal views.

Ideology

An important test of how far Latino context is re-shaping white views is to determine if proximity to Latinos changes how Americans identify politically. Are the right-ward shifts that we see on health, welfare, criminal justice, trade, and education policy accompanied by a broader shift to the right on the core liberal-conservative ideology scale that underlies much of the politics of this nation? In the last column of Table Two, we attempt to answer this question. The results are striking. Minority context does play a role in the core political identities of white

Americans. The effects mirror the patterns we have already seen. Larger Latino populations push white Americans to the right politically and the effects are substantial. In terms of self- described ideology, whites living in zip codes with no Latinos are, all else equal, predicted to place themselves almost one full category to the left of whites living in zip codes that are largely

19 Latino. On the five point ideology scale, that could signal a shift from ‘very liberal’ to ‘liberal or from ‘liberal to ‘moderate. See also Figure One.

Proximity to Asian Americans in this final case has a significant effect and the direction is a liberalizing one. Whites in zip codes with greater concentrations of Asian Americans are substantially more likely to identify as liberal than whites residing in areas with few Asian

Americans. Finally, our analysis of ideology suggests that black context plays no direct role in shaping white perceptions of where they fit politically. Whites who live in zip codes with larger black populations are no more or less likely to identify as conservative.

Results from 2004

There is additional evidence in the 2004 NAES that context is altering core beliefs and partisan attitudes. The 2004 NAES includes views on different groups in American society and approval of the Bush presidency. In both cases, Latino context plays a significant role.

Proximity to large Latino populations is associated with less favorable views of labor unions and more favorable views of corporations. More Latinos also means stronger approval of the Bush administration. The results suggest again that a growing Latino population is pushing white

Americans to the right. The results for Asian American context on these two sets of questions mirror our earlier findings. Whites living in zip codes with larger Asian American populations are no more or less likely to favor unions or corporations but are less apt to favor the Bush

Administration. Similarly, black context had no significant impact on white views of unions but proximity to a larger black population was linked to more positive views of Bush and corporations. [analysis in supplementary documentation].

All told, these results point to the centrality of the Latino population in the minds and political calculations of white Americans. When many whites think about core policy questions and their own ideological identity, their views appear to be at least partly driven by their local 20 Latino context. Where Latinos are numerous, concerns and fears appear to lead to more conservative or punitive views. Where Latinos are yet to arrive in large numbers, the absence of these concerns and fears appears to engender a more liberal or lenient political viewpoint. The fact that a large Latino population can lead to fear and resentment is far from novel (eg Brader et al 2008). The fact that a large Latino population helps to shape white public opinion on core areas of American politics is both novel and potentially alarming.

Asian American context clearly plays a different and less central role. In most cases, white views are unrelated to proximity to Asian Americans. Either due to the different stereotypes that whites have of each group, differences in the socioeconomic attainment of each group, or differences in the extent to which politicians and the media focus on each group, few white Americans seem threatened enough by large Asian American populations to shift their political views in any noticeable way. Indeed, to the extent that Asian American context matters, it has a liberalizing influence. Although likely surprising to most readers, the intermittent liberalizing effect of the Asian American population on white views is not entirely new and has been echoed in past research. In a number of existing studies, there is some indication that reactions to the Asian American population are more positive. Hood and Morris

(2000, 1998), Hero and Preuhs (2006), and Tolbert et al (1999) in different ways all find that proximity to larger Asian Americans communities breeds more positive views about Asian

Americans or policies related to Asian Americans.

Our findings also suggest that black context plays an uneven role. When proximity to

African Americans did matter, the effect was clear. Whites were less generous and more punitive. But in most of the areas we examine, living near blacks had no obvious implications for white policy opinions. These mixed results should not, however, lead to the conclusion that the black-white divide is irrelevant in politics today or indeed is less central than immigrant context 21 in any of the policy arenas we examine. It is entirely possible that views of African Americans have become so deeply entrenched in the minds of white Americans that perceptions of blacks affect white views regardless of whether individual whites live near to or far from large concentrations of blacks. We know, for example, that across all contexts perceptions of the black community influence white views on crime, welfare, and other policies (Gilens 2001,

Kinder and Sanders 1994). And concerns about policies designed to help the African American community have, at times, arguably been the core factor driving white partisan choices

(Carmines and Stimson 1989 but see Abramowitz and Saunders 1998). We believe that our results warrant making relatively strong claims about the increasingly relevance of Latino contexts in the political calculus of white Americans but they do not merit dismissing the centrality of a black-white divide.

Robustness Checks

To help ensure that the results in Tables 1 and 2 measure underlying relationships between context and white political views, we performed a serious of additional tests. First, we repeated as much of the analysis as possible on the 2004 NAES. The 2004 survey has fewer policy questions but does have the same large sample and wide geographic sampling and includes questions on the same set of individual characteristics that we incorporated into the regression models in all of the tables in this paper. Thus, for at least a few of the policy areas as well as for the liberal-conservative ideology question we can replicate our earlier tests. This replication corroborates the results from the 2000 survey. In 2004, as in 2000, proximity to large concentrations of Latinos is associated with significant shifts to the political right. Whites living near large numbers of Latinos were once again more conservative on social welfare and health care policy and they tended to be more likely to self-identify as conservative. The effects of

Asian American context were as we see here either insignificant or led in some cases to more 22 left-leaning policy choices. Black context once again had an uneven but at least partially conservative impact [analysis in supplementary appendix].9

Second, we repeated the analysis using a Hierarchical Linear Model (HLM). We started with the two-stage least squares regressions for two reasons. First, because of the selection concerns that we have outlined, running an HLM model that employs simple uncorrected measures of local context is problematic. Second, existing studies indicate that clustering standard errors can provide a better and more straightforward estimate of contextual effects than can multi-level modeling (Primo et al 2007). Nevertheless, since an HLM model does take into account the different levels of units and minimizes the correlation in error terms among respondents in the same geographic unit, it provides a chance to assess the robustness of our results under different methodological constraints.10 Fortunately, the results of this HLM analysis generally mirror the findings that we present here.

One other potential issue is multicollinearity between the key contextual variables.

Fortunately, at the zip code level, the proportions of residents who are Latino, Asian American, and African American are only minimally correlated with each other (r<. 13). Nevertheless, to help ensure that correlation between different contexts was not affecting our results, we repeated the analysis while including only one of the three racial contexts and dropped the other two.

This did nothing to alter the basic conclusions that we present.

Finally, we also performed a series of tests in which we added a range of different individual characteristics into our model. Specifically, we accounted for one’s religious denomination (Catholic, Protestant, Jewish, Muslim), a measure for whether the respondent was

9 Black context also played roughly the same role in 2004 as in 2000. 10 HLM, for example, enables us to simultaneously control for state and local context – a test that helps ensure that the contextual effects detailed in Tables 1-2 are not solely due to state level effects. 23 born again or not, another for households with a member in the armed forces, and a variable that gauged occupational status. These additional controls did little to affect the results.

Selection Revisited

Our use of state level context in a two-stage least-squares model may not allay all concerns about selection. To help discount these concerns, we performed a series of additional tests. First, we sought to verify our explanation by examining Latino contextual effects in areas where we should not find them - on views of groups that are orthogonal to discussions of Latinos and Latino immigration. Accordingly, we found no significant ties between Latino context and attitudes towards feminist groups, gay and lesbian groups, and Muslims. This suggests that we are not seeing effects for Latino context merely because Latinos move into areas where whites are especially intolerant or conservative. Instead, Latino context tends only to be associated with white views on those issues where the policy debate often centers on immigrants or Latinos.

One might also wonder much these results are driven by a particular state. To ensure that one state did not account for the pattern of results, we repeated the analysis dropping each large state one-by-one. Basic conclusions about the impact of Latino, Asian American, and African

American context remained the same whether states like California, Texas, or New York are included or dropped.

But perhaps the best defense of our results is that the general patterns we present here do not fit the typical selection story. If the selection of whites were driving the results (e.g. large numbers of whites with more negative views of Latinos leave areas with high concentrations of

Latinos) then we would expect a positive link between Latino context and white views rather than the negative one that we find.

A different possibility is that Latinos and Asian Americans are selecting into distinct kinds of neighborhoods. Latinos, due to their low socioeconomic status, could be forced to 24 choose poorer and less well educated locales – the kinds of neighborhoods that are typically more racially intolerant (Schuman et al 1997). By contrast, Asian Americans, because of their greater economic resources, could move into more desirable neighborhoods that contain residents with higher levels of income and education and thus greater levels of racial tolerance.

There are, however, two problems with this account. First, the zip codes that Latinos and Asian

Americans move into are not all that different on these measures. In our data set, Latinos live in zip codes that have on average a 10.9 percent white poverty rate. Asian Americans live in zip codes that have on average an 8.3 percent white poverty rate. Similarly, the average Latino lives in a zip code where the proportion of whites with less than a high-school education averages 5.3 percent. The comparable figure for the average Asian American respondent is 3.3 percent.

Second, our analysis controls for individual socioeconomic status, individual political orientation, and the overall socioeconomic well-being of each zip code. This makes it difficult to explain away the contextual effects due to the kind of selection story we have just presented.

The conservative Latino contextual effects cannot simply be the result of less well off Latinos moving into less well-off neighborhoods where whites might tend to be conservative and unwelcoming. Since we control for the education, ideology, and party identification of each individual white respondent and the overall socioeconomic status of the zip code (including the median income, the average education level, and the overall poverty rate), the only way selection could account for the results is if Latinos move into neighborhoods with individuals whose policy views are more conservative than their party, ideology, and individual and neighborhood socioeconomic status would suggest– a possibility but a remote one.

Is it Race or Foreignness?

Up to this point, we have focused on two ‘racial’ measures of context – percent Latino and percent Asian American. We have separated out the two groups because we suspect that 25 white Americans view the two pan-ethnic groups very differently. We have also combined foreign born and native born members of each group because we suspect that white Americans tend not to distinguish between the native and foreign born of each pan-ethnic group when reacting to the local population. But we could be wrong on both accounts. To determine if white

Americans most strongly react to race or foreignness, we repeated our analysis with a serious of different measures of context including percent foreign born overall, percent foreign born Latino and percent foreign born Asian American.

The overall percent foreign born results are the most telling. The reason for these null findings is likely quite simple. We believe that because white Americans tend to react to immigrants from Latin America by shifting to the right politically and to immigrants from Asia either not at all or by shifting to the left politically, the two effects tend to offset each other and to reduce the significance of any findings. Thus, if we want to uncover how immigrant context is shaping white views, we need to separate out these two divergent effects [Results in

Supplementary Documentation].

The results for percent foreign born Latin American and percent foreign born Asian

American are more complicated. When we substituted these two contextual measures into our model, they tended to work in the same manner as the percent Latino and percent Asian

American measures. This is because the two sets of measures are very highly correlated.

Percent foreign born Asian is correlated at .84 with percent Asian American at the zip code level.

The correlation for percent Latino and percent foreign born Latin American is .89. Thus, from an empirical standpoint, very little exists that distinguishes these two measures. If we do, however, include all four measures of context in the same model, the two racial measures perform slightly better than the two foreign born measures.11 From this meager evidence it

11 Given the high correlations amongst these variables, this specification generates a significant amount of collinearity, thus making firm conclusions problematic. 26 appears that whites are reacting more to the larger pan-ethnic group and less to the foreign born members of each group. At least at first glance, it appears to be more about race than about foreignness. This is, however, a finding that requires more extended analysis in the future.12

Finally, we also looked to see if the socioeconomic standing of the local Latino population affected white reactions. It is possible that white reactions are shaped by social distance and that white Americans are reacting to the relatively low status of Latinos rather than to any racial differences. To address this possibility, we asked if white Americans react more punitively when the local Latino population is worse off socioeconomically? Specifically, we re- ran the analyses adding measures for the local Latino poverty rate and an interaction between the

Latino poverty rate and Latino population size. With only one exception, the economic standing of the local Latino population did not interact with the size of the Latino population to affect white reactions. White Americans tended to be more regressive when they lived in close proximity to large Latino populations whether or not the Latino population was largely poor or not [Analysis in Supplementary Documentation].13As far as we could tell, white Americans appear to be not all that discerning in their reactions to the Latino population.14

Partisanship and the Contingent Effects of Minority Context

The analysis to this point has assumed that all white Americans react to minority context in the same way. There is, however, reason to expect something less than a uniform reaction across the white population. Given the divergent stances of the Democratic and Republican parties on the issue of immigration, given that Republican candidates have more than occasionally portrayed immigrants as illegal and dangerous (e.g. former governor Pete Wilson),

12 We also considered separating out the foreign born population by national origin group but variation in the size of most national origin groups at the zip code level was too limited to assess empirically. 13 We also found no obvious patterns relating to the social distance between local whites and local Latinos. White reactions were not exaggerated in areas where the gap between white and Latino income (or education) was higher. 14 Work by Hood and Morris (1998) suggests that responses to illegal immigrants will be more severe. Unfortunately, data on the legal status of immigrants by zipcode is not available. 27 and given that racial prejudice and stereotyping both tend to be more pronounced among

Republicans than they are among Democrats, white Republicans might react more viscerally to large immigrant or Latino populations (Kinder and Sanders 1996, Carmines and Stimson 1989,

Chavez 2008, Hajnal 2006). Democrats, by contrast, might be more receptive to the Latino community and less threatened by proximity to large numbers of Latinos.

To test this possibility, we added a series of interaction terms between Republican identification and minority context to the models in Tables 1 and 2. The results are displayed in

Table 3 and in the Supplementary Documentation.15 There is some evidence –albeit highly inconsistent -that reactions to the Latino context are more pronounced amongst Republicans.

Only one of the five interactions are significant, but in this case, Republicans appear to shift even more severely to the right in contexts with large Latino populations than do other whites. In particular, white Republicans are especially apt to favor stiffer criminal punishment when living in close proximity to large Latino populations. The fact that most of the interactions with Latino context are insignificant does, however, give us pause. [TABLE THREE HERE]

There is more consistent and stronger evidence that white reactions to Asian American communities are contingent on partisanship. All but one of the interactions is significant and all of the coefficients point in the same direction. The significant interactions between Republicans and Asian American context suggest that white Republicans are more prone than other whites to shift to the right when faced with a large Asian American population. This is true on social welfare policy, health care, and crime.16 This all suggests that white Republicans may be less discerning about the minority population than white Democrats and nonpartisans. For white

Republicans, it appears that both Latinos and Asian Americans may represent a threat that does not warrant increased governmental assistance.

15 Given a similar pattern of results and limited space, we moved three of the regressions to the Online Appendix. 16 The results in the Supplementary Documentation displays the results for education spending. 28 Although we view these interactions with partisanship as telling, we also believe that this is far from the last word on variations in white reactions to minority context. There are any number of factors that could influence who responds negatively to minority context. Scheve and

Slaughter (2001), for example, have predicted that the consequences of immigrant context will be more severe when natives and immigrants have similar skills levels and are in direct economic competition. Others might point to the interaction of minority context and local partisan or local media context. Still others contend that the national framing of immigration is a critical mediator (Hopkins 2010). Our tests are merely an initial foray into this line of inquiry.

It is clear that more work needs to be done before the full contours of white reactions are known.

Discussion

The patterns illustrated in this paper suggest that the nation’s increasingly diverse population is having a profound impact on the politics of white America. This may not be particularly surprising to many observers. What is novel here is the focus on Latinos rather than blacks and the broad nature of the reaction to minority context. It is not simply that larger concentrations of

Latinos do or do not cause fear and concern. It is that larger concentrations of Latinos appear to be leading to a fundamental shift in the political orientation of many members of the white population. On several of the major policy debates that we face and on the core liberal- conservative ideological line that delineates much of our politics, Latino context is a key contributor. The degree to which white Americans live in areas with large numbers of Latinos seems to shape who they are politically. Importantly, politics, or at least partisanship, also shapes whites reactions to minority context. In particular, there is some indication that

Republicans react more sharply to Asian American context than do Democrats.

Our findings also suggest that reactions to all immigrant-based groups are far from uniform. Whites residing in close proximity to Latinos hold distinct opinions from whites living 29 in close proximity to Asian Americans. Residing in areas with high numbers of Latinos is associated with a consistent move to the right politically. Whites residing in zip codes with high concentrations of Latinos are more concerned about immigration, more conservative on social welfare policy, health care and educational issues, more punitive on criminal justice questions, and more likely to see themselves as conservative. By contrast, living in close proximity to

Asian Americans often has little to no impact on white attitudes. And in the few cases where

Asian American context is significant, it is associated with a shift to the left.

Just why white Americans react so differently to these two pan-ethnic groups is less clear.

As with most contextual analysis, the results we have presented here are somewhat of a black box. We can demonstrate that different sets of demographic numbers translate into distinct sets of views but we have not shown how that translation works.17 Does a larger Latino population, as we have sometimes implied, generate a greater sense of threat among local white residents?

And if so, are whites afraid of Latinos taking jobs and lowering wages, are they concerned about

Latinos using up precious public resources, or do they simply not like the Latino community?

Likewise, why is proximity to a larger Asian American not associated with more conservative views? Is it that white stereotypes of Asian Americans are more empathetic, is it that Asian

Americans tend be better educated and have higher incomes and thus tend to represent less of a threat to public resources, or is it that politicians and the media on Latinos more?

And how does proximity actually ‘work’ to change individual white views? Are whites responding to personal interactions with Latinos, to local immigrant political activism, to local white political campaigns, to portrayals of immigrants in local media outlets, or to something else altogether? We are doubtful that individual interactions play a big role – in large part because studies that actually test for the effects of personal interactions find that they generally

17 Unfortunately, the survey we use does not include questions on racial threat, stereotypes, or ethnocentrism. 30 have a positive effect on white views of minorities (Dixon and Rosenbaum 2004). But the other explanations seem more plausible. We know, for example, that media portrayals of Latinos are more negative in areas with higher concentrations of Latinos (Branton et al 2009).

Future research will also have to consider other aspects of this reaction. We have demonstrated the broad political consequences of minority context at one point in time. Others will have to determine if and how reactions vary over time? Are white reactions, for example, more heated when national politics focuses on immigration (Hopkins 2010) or when the nation faces severe economic stress (Fetzer 2000, Branton and Jones 2005)? Do whites respond more severely in cases of rapid expansion of the local Latino population than when that population is relatively stagnant (Hopkins 2010)? In short, how dynamic is this reaction?

While there is certainly more work to be done, we provide an important contribution by demonstrating that Latino context helps shape the political calculus of white Americans. In

American history, when race mattered, it was often driven by a black-white dynamic. That may no longer true today. Along with dramatic growth in the immigrant population, major changes in the racial makeup of the nation and the increasing visibility of Latinos in the economic, social, cultural, and political spheres of this nation have brought forth a real change in the racial dynamics of our politics. Blacks still matter but when we talk about the role of race in American politics, we have to talk about the fears and concerns that a growing Latino population provokes.

31 BIBLIOGRAPHY

Abramowitz, Alan I., and Kyle L Saunders. 1998. "Ideological Realignment in the U.S. Electorate." Journal of Politics 60 (3):634-52. Alba, Richard, and Victor Nee. 2005. Remaking the American Mainstream: Assimilation and Contemporary Immigration. Cambridge: Harvard University Press. Alesina, Alberto, R. Baqir and William Easterly. 1999. “Public Goods and Ethnic Divisions.” Quarterly Journal of Economics 114(4):1243–84. Allport, G. W. 1954. The Nature of Prejudice. Menlo Park, CA: Addison-Wesley. Bean, Frank D., and Gillian Stevens. 2003. America's Newcomers and the Dynamics of Diversity. New York: Russell Sage. Black, Earl, and Merle Black. 1973. "The Wallace Vote in Alabama: A Multiple Regression Analysis." Journal of Politics 35:730-6. Blalock, Hubert M. 1967. Toward a Theory of Minority-Group Relations. New York: Wiley. Bobo, Lawrence D. 2001. "Racial Attitudes and Relations at the Close of the Twentieth Century." In America Becoming: Racial Trends and Their Consequences, ed. N. Smelser, W. J. Wilson and F. Mitchell. Washington DC: National Academy Press. Borjas, George J. 2001. Heaven's Door: Immigration Policy and the American Economy. Princeton: Princeton University Press. Branton, Regina, and Bradford S. Jones. 2005. "Re-Examining Racial Attitudes: The Conditional Relationship Between Diversity and Socio-Economic Environment." American Journal of Political Science 49 (2):359-72. Branton, Regina and Johanna Dunaway. 2009. “Spatial Proximity to the U.S.-Mexico Border and Newspaper Coverage of Immigration Issues” Political Research Quarterly 62(2): 289- 302. Burns, Peter, and James G. Gimpel. 2000. "Economic Insecurity, Prejudicial Stereotypes, and Public Opinion on Immigration Policy." Political Science Quarterly 115:201-25. Carmines, Edward G., and James A. Stimson. 1989. Issue Evolution: Race and the Transformation of American Politics. Princeton, NJ: Princeton University Press. Clark, W.A.V. 1992. "Residential Preferences and Residential Choices in a Multiethnic Context." Demography 29 (3):451-66. Corzine, Jay, James Creech, and Lin Corzine. 1983. "Black Concentration and Lynchings in the South: Testing Blalock's Power-Threat Hypothesis." Social Forces 61:774-96. Dustmann, C., and I. Preston. 2001. "Attitudes to ethnic minorities, ethnic context, and location decisions." Economic Journal 111:353-73. Fellowes, Matthew C. and Gretchen Rowe. 2004. "Politics and the New American Welfare States." American Journal of Political Science 48 (2 (April)):362-73. Fetzer, J. 2000. Public Attitudes Toward Immigration in the United States, France, and Germany. Cambridge: Cambridge University Press. Fossett, Mark A., and K. Jill Kiecolt. 1989. "The Relative Size of Minority Populations and White Racial Attitudes." Social Science Quarterly 70 (4):820-35. Fox, Cybelle. 2004. "The Changing Color of Welfare? How White's Attitudes toward Latinos Influence Support for Welfare." American Journal of Sociology 110 (3):580-625. Gilens, Martin. 2001. Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy. Chicago: University of Chicago Press. Giles, Michael W., and Arthur Evans. 1986. "The Power Approach to Intergroup Hostility." Journal of Conflict Resolution 30:469-86. 32 Gilliam, Frank D. and Shanto Iyengar. 2000. “Prime Suspects: The Influence of Local Television News on the Viewing Public” American Journal of Political Science 44(3):560-573. Gimpel, James. 1999. Separate Destinations: Migration, Immigration and the Politics of Place. Ann Arbor, MI: University of Michigan Press. Green, Donald P., Dara Z. Strolovitch, and Janelle S. Wong. 1998. "Defended Neighborhoods, Integration, and Racially Motivated Crime." American Journal of Sociology 104 (2):372- 403. Ha, S. E. (2010). The Consequences of Multiracial Contexts on Public Attitudes Toward Immigration. Political Research Quarterly, 63(1), 29-42. Hero, Rodney E. and Robert R. Preuhs. 2006. "From Civil Rights to Multiculturalism and Welfare for Immigrants: An Egalitarian Tradition Across American States?" Du Bois Review 3 (2):317-40. Hero, Rodney. 1998. Faces of Inequality: Social Diversity in American Politics. New York: Oxford University Press. Hood III, M. V. and Irwin L. Morris. 2000. “Brother, Can You Spare a Dime? Racial/Ethnic Context and the Anglo Vote on Proposition 187.” Social Science Quarterly 81 (1):194- 206. Hood III, M.V. and Irwin L. Morris. 1998. “Give Us Your Tired, Your Poor...But Make Sure They Have a Green Card: The Effects of Documented and Undocumented Migrant Context on Anglo Opinion Toward Immigration.” Political Behavior 20 (1):1-15. Hopkins, D. J. (2009). The Diversity Discount: When Increasing Ethnic and Racial Diversity Prevents Tax Increases. The Journal of Politics, 71(1), 160-177. Hopkins, D. J. (2010). Politicized Places: Explaining Where and When Immigrants Provoke Local Opposition. American Political Science Review. Huckfeldt, Robert, and Carol Weitzel Kohfeld. 1989. Race and the Decline of Class in American Politics. Urbana, IL: University of Illinois Press. Huntington, Samuel. 2005. Who Are We?: The Challenges to America's National Identity. New York: Simon and Schuster. Jackman, Mary R., and Marie Crane. 1986. “`Some of my best friends are black...’: Interracial Friendship and Whites' Racial Attitudes.” Public Opinion Quarterly 50:459-86. Johnson, Martin. 2001. "The Impact of Social Diversity and Racial Attitudes on Social Welfare Policy." State Politics and Policy Quarterly 1 (1):27-47. Key, V. O. 1984. Southern Politics in State and Nation. Knoxville, TN: University of Tennessee Press. Kim, Claire Jean. 2003. Bitter Fruit: The Politics of Black-Korean Conflict in New York City. New Haven, CT: Yale University Press. Kinder, Donald R and Tali Mendelberg. 1995. "Cracks in American Apartheid: The Political Impact of Prejudice among Desegregated Whites." Journal of Politics 57 (2):402-24. Kinder, Donald R. and Lynn Sanders. 1996. Divided by Color: Racial Politics and Democratic Ideals. Chicago: University of Chicago Press. Klinkner, Philip A. and Rogers M. Smith. 1999. The Unsteady March: The Rise and Decline of Racial Equality in America. Chicago: University of Chicago Press. Kousser, J. Morgan. 1999. Colorblind injustice: minority voting rights and the undoing of the Second Reconstruction. Chapel Hill: The University of North Carolina Press. Massey, Douglas, and Zoltan Hajnal. 1995. "The Changing Geographic Structure of Black-White Segregation in the United States." Social Science Quarterly 76 (3):527-42.

33 McLaren, Lauren M. 2003. "Anti-Immigrant Prejudice in Europe: Contact, Threat Perception, and Preferences for the Exclusion of Migrants." Social Forces 81 (3):909-36. Oliver, J. Eric and Tali Mendelberg. 2000. "Reconsidering the Environmental Determinants of White Racial Attitudes." American Journal of Political Science 44 (3 (July)):574-87. Oliver, J. Eric and Janelle Wong. 2003. "Intergroup Prejudice in Multiethnic Settings." American Journal of Political Science 47 (4):567-82. Parker, Frank R. 1990. Black Votes Count: Political Empowerment in Mississippi after 1965. Chapel Hill: University of North Carolina Press. PEW. 2006. "America's Immigration Quandry." Washington: PEW Hispanic Center. Portes, Alejandro, and Ruban G. Rumbaut. 2006. Immigrant America: A Portrait. Berkeley: University of California Press. Quillian, Lincoln. 1995. "Prejudice As a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial Prejudice in Europe." American Sociological Review 60 (4):586-611. Scheve, K.F. and JM.J. Slaughter. 2001. "Labor Market Competition and Individual Preferences over Immigration Policy." Review of Economics and Statistics 83 (1):133-45. Stein, Robert M., Stephanie Shirley Post, and Allison L. Rinden. 2000. "Reconciling Context and Contact Effects on Racial Attitudes." Political Research Quarterly 53 (2 (June)):285-303. Taylor, Marylee C. 1998. "How White Attitudes Vary With The Racial Composition of Local Populations: Numbers Count." American Sociological Review 63 (August):512-35. Tichenor, Daniel J. 2002. Dividing Lines: The Politics of Immigration Control in America. Princeton: Princeton University Press. Tolbert, Caroline and Rodney Hero. 2001. “Dealing with Diversity: Racial/Ethnic Context and Social Policy Change” Political Research Quarterly 54(3): 571-604.

34 Table One. Context and Views on Immigration, Poverty, and Health Care1 Social Welfare Policy Health Care Policy Immigration a Poverty a Serious Serious Reduce More for Help Cover Problem Problem Inequality Medicaid Uninsured NEIGHBORHOOD CONTEXT Percent Latino2 .58 (.25)* -.20 (.13) -.34 (.14)* -.71 (.21)** -.49 (.14)** Percent Asian American2 -.35 (.78) .93 (.41)* .58 (.43) .83 (.60) .60 (.43) Percent African American2 .29 (.32) -.10 (.17) -.06 (.18) -.51 (.26) -.08 (.18) Median Income .08 (.10) -.00 (.06) -.06 (.06) -.01 (.01) -.06 (.07) Percent College Educated -.61 (.14)** -.13 (.07) -.15 (.07)** .15 (.10) .10 (.07) Percent Unemployed -1.6 (.76)* .42 (.39) .32 (.42) 1.2 (.62) .45 (.40) POLITICAL ORIENTATION Party Identification (hi=Rep) .04 (.01)** -.08 (.01)** -.07 (.01)** -.12 (.01)** -.14 (.01)** Ideology (hi=Conservative) .06 (.01)** -.12 (.01)** -.09 (.01)** -.13 (.01)** -.16 (.01)** SOCIOECONOMIC STATUS Education -.09 (.00)** -.01 (.00)** -.02 (.00)** -.01 (.00)** -.01 (.00)** Income -.01 (.01) -.02 (.00)** -.03 (.00)** -.02 (.00)** -.02 (.00)** Unemployed -.04 (.08) .06 (.04) -.06 (.04) .03 (.06) .03 (.06) Age .00 (.01) -.01 (.00)* .01 (.00)** -.00 (.01) -.01 (.01) Gender (1=female) .10 (.02)** .27 (.01)** .06 (.01)** -.01 (.02) .09 (.01)** Union Member .02 (.03) .04 (.01)** .01 (.01) -.01 (.02) .02 (.01) Have Children -.02 (.02) .04 (.01)** .06 (.01)** -.02 (.02) .04 (.01)** CONTROLS Years in home .02 (.00)** -.03 (00)** -.02 (.00)** -.01 (.04) -.09 (.05) Urban -.12 (.04)** -.04 (.02) .02 (.02) .06 (.03)* .02 (.02) Suburban -.06 (.03)* .02 (.02) .02 (.02) .04 (.02) .01 (.02) West -.03 (.05) -.03 (.03) .01 (.03) -.01 (.04) -.00 (.03) Midwest -.11 (.04)* -.04 (.02)* .01 (.02) -.08 (.03)** -.03 (.02) Northeast -.10 (.04)* .05 (.02)* .08 (.02)** -.01 (.03) .07 (.02)** Constant 2.95 (.07)** 1.62 (.04)** .81 (.04)** .62 (.06) .59 (.04)** N 7640 18599 7523 7644 18314 R Squared .09 .10 .12 .08 .11 1Two-stage least squares regression. 2 Endogenous variables-instruments are percent Latino, percent Asian American, and percent African American at the state level. **P<.01 *P<.05

35 Table Two. Context and Criminal Justice and Education Policy and Basic Ideology1 Criminal Justice Punish Increase Liberal- Favor Death Criminals Education Conservative Penalty More Spending Ideology NEIGHBORHOOD CONTEXT Percent Latino2 .29 (.11)** .08 (.14) -.42 (.21)* .47 (.16)** Percent Asian American2 -1.16 (.34)** -1.16 (.45)* .86 (.66) -3.51 (.56)** Percent African American2 .01 (.09) .53 (.17)** -.60 (.28)* .03 (.21) Median Income .01 (.00)** .02 (.00)** -.02 (.00)** .04 (.01)** Percent College Educated -.18 (.06)** -.56 (.07)** -.08 (.11) -.39 (.08)** Percent Unemployed -.33 (.34) -.99 (.41)* 1.4 (.63)* -.55 (.51) POLITICAL ORIENTATION Party Identification (hi=Rep) .06 (.01)** .07 (.01)** -.10 (.01)** ---- Ideology (hi=Conservative) .05 (.01)** .11 (.01)** -.16 (.01)** ---- SOCIOECONOMIC STATUS Education -.02 (.00)** -.07 (.00)** -.00 (.00) -.03 (.00)** Income .00 (.01) -.01 (.00)** -.02 (.00) .04 (.00)** Unemployed -.00 (.04) .02 (.04) -.04 (.06) -.13 (.05)** Age -.00 (.00) -.00 (.00) -.07 (.02)** .01 (.00)** Gender (1=female) -.06 (.01)** .15 (.01)** .11 (.02)** -.14 (.01)** Union Member -.01 (.01) .03 (.01)* .04 (.02) -.12 (.02)** Have Children .01 (.01) .05 (.01)** .13 (.02)** .02 (.01) CONTROLS Years in Home -.01 (.00)** -.01 (.01) -.05 (.01)** .06 (.01)** Urban -.07 (.02)** -.05 (.02)* .05 (.04) -.17 (.02)** Suburban -.02 (.01) .01 (.02) .03 (.03) -.10 (.02)** West -.02 (.02) -.03 (.03) -.11 (.04)* -.07 (.03)* Midwest -.06 (.02)** -.09 (.02)** -.13 (.03)** -.09 (.02)** Northeast -.04 (.02)* -.02 (.02) -.06 (.04) -.23 (.03)** Constant 1.62 (.03)** 2.76 (.04)** .62 (.06) 3.29 (.04)** N 8089 18419 7667 22709 R Squared .07 .10 .10 .04 1Two-stage least squares regression. 2 Endogenous variables-instruments are percent Latino, percent Asian American, and percent African American at the state level. **P<.01 *P<.05

36 Table Three. Context, Partisanship and White Policy Views Immigration Poverty a Help Cover Punish Increase a Serious Serious Uninsured Criminals School Problem Problem More Spending PARTISANSHIP * CONTEXT Republican*%Latino1 -.41 (.37) -.03 (.24) -.01 (.27) .50 (.25)* .06 (.43) Republican*% Asian1 5.1 (1.1)** 1.7 (.69)* -2.4 (.76)** 2.6 (.73)** .89 (1.2) Republican*% Black1 .44 (.39) -.02 (.25) .26 (.28) .23 (.24) .87 (.44) NEIGHBORHOOD CONTEXT Percent Latino2 .82 (.32)** .29 (.22) -.64 (.25)** -.16 (.22) -.52 (.35) Percent Asian American2 -2.7 (1.1)* -2.3 (.74)** 2.2 (.84)** -2.8 (.76)** 1.6 (1.3) Percent African American2 .19 (.35) .15 (.23) -.30 (.28) .63 (.23)** -1.1 (.43)* Median Income .02 (.02) .01 (.01) -.00 (.01) .04 (.01)** -.01 (.00)** Percent College Educated -.71 (.18)** .21 (.11) -.09 (.13) -.82 (.11)** -.05 (.19) Percent Unemployed -.03 (.11) -.29 (.37) .21 (.42) -.31 (.37) .73 (.65) POLITICAL ORIENTATION Republican -.03 (.06) .22 (.04)** -.49 (.04)** .08 (.03)* -.47 (.06)** Independent .07 (.03)* .08 (.02)** -.25 (.03)** .10 (.02)** -.22 (.04)** Ideology (hi=Conservative) .07 (.02)** .17 (.01)** -.28 (.01)** .16 (.01)** -.28 (.02)** SOCIOECONOMIC STATUS Education -.11 (.01)** .02 (.00)** -.02 (.00)** -.10 (.00)** .00 (.01) Income -.01 (.01) .04 (.00)** -.04 (.01)** -.01 (.00)** -.03 (.01)** Unemployed -.03 (.11) -.09 (.07) .07 (.09) .04 (.07) -.10 (.11) Age -.00 (.01) .00 (.00) .01 (.01) -.00 (.00) -.01 (.00)** Gender (1=female) .12 (.03)** -.44 (.02)** .17 (.02)** .23 (.02)** .19 (.03)** Union Member .04 (.03) -.07 (.02)** .02 (.03) .06 (.02)** .08 (.04)* Have Children -.01 (.03) -.05 (.02) .06 (.02)** .07 (.02)** .24 (.03)** CONTROLS Years in home .01 (.00)** .04 (.00)** -.02 (.00)** -.00 (.01) -.01 (.00)** Urban -.08 (.05) .05 (.03) .04 (.04) -.01 (.03) .05 (.06) Suburban -.03 (.04) .04 (.03) -.01 (.03) .03 (.03) .04 (.04) West -.02 (.06) .04 (.04) -.02 (.04) -.07 (.04) -.16 (.06)* Midwest -.15 (.04)** .07 (.03)* -.06 (.03)* -.17 (.03)** -.18 (.04)** Northeast -.13 (.04)** -.08 (.03)** .14 (.03)** -.06 (.03)* -.07 (.05) N 7636 18599 18314 18409 7667 R Squared .05 .05 .07 .06 .07 1Endogenous variables-instruments are percent Latino, percent Asian American, and percent African American at the state level. **P<.01 *P<.05

37 Figure 1: The Impact of Latino Context on White Policy Attitudes