Proximal contact with Latino immigrants and immigration attitudes

Abstract

The growth of the Latino community has increased levels of contact between this minority group and the predominately white majority. How does exposure to immigrants impact attitudes towards immigrants and immigration held by white Americans? We argue that previous work has not adequately tested the relationship posited by inter-group contact theory, whereby contact should shape policy attitudes. We test our theory drawing on the 2018 Cooperative Congressional Election Study (CCES). We find that among non-Hispanic whites, having a loved one who is a Latino immigrant improves attitudes towards immigrants as a group, which in turn is associated with declining support for punitive immigration policy. We likewise find that attitudes towards immigrants as a group mediates the relationship between proximal contact and policy attitudes. We draw on a variety of strategies to assess the robustness of the findings and to tease out evidence for the causal pathway we theorize is at work. As the demographics of the nation continue to change, growing closer to minority-majority status, understanding the factors that shape how white Americans regard their minority neighbors is of heightened importance.

Introduction

Latinos are the largest ethnic minority group in the United States, accounting for 17.8% of the population in 2018 (American Community Survey 2018). While states like California, Texas, and Florida boast the most significant concentrations of Latinos, between 2000 and 2010 Latino communities established a presence in other parts of the country, with states like South Carolina, Alabama, and Tennessee seeing significant rates of growth. In many places, immigration drives this growth, and many Latino communities are characterized by mixed citizenship status and a relative closeness to the immigrant experience.¹ The growth of these communities likewise increases levels of contact between this minority group and white Americans.

How does exposure to immigrants impact attitudes towards immigrants and immigration held by white Americans? Previous research is mixed. Inter-group contact theory establishes that having relationships with out-group members can reduce prejudice towards that group. But, this body of work struggles to connect out-group contact to attitudinal changes towards immigration policy. Another stream of work developed around what scholars refer to as proximal contact – having a loved one threatened by punitive immigration policy – finds that individuals learn civic lessons about how a given set of policies work by watching their loved ones negotiate the system. While individuals update their attitudes towards government as a result, researchers have yet to explore whether they likewise change their policy attitudes. Instead, scholars find that other factors, like one’s geographic context and partisanship, determine policy positions.

In this paper, we argue that previous work has not adequately tested the relationship posited by inter-group contact theory, whereby contact should shape policy attitudes. Inter-group contact theory suggests that attitudes towards policies that target out-groups should shift when one has a relationship with an out-group member, the nature of that relationship is positive, and individuals extend their positive experiences to perceptions of the out-group writ large. That is, individuals may become less supportive of punitive immigration policy when they have a positive and meaningful relationship with an immigrant, and as a consequence of that positive relationship, they view immigrants as a group in more favorable terms.

While the indirect nature of this relationship is intuitive, previous research suffers from imprecise measures of contact, relying on geographic context for example, and/or has not examined the indirect effects of contact on policy attitudes, instead looking for a direct link. This project therefore contributes to the group contact literature by homing in on an ego-centric measure of contact, bypassing the methodological pitfalls and mixed results associated with geographic measures (Tolbert and Grummel 2003; Glaser 1994; Oliver and Mendelberg 2000; Rocha and Espino 2009). Wong et al. (2012) highlight that self-definitions of concepts like community and inter-group contact may be preferable because to the extent that one’s geographic context impacts policy attitudes, “it must matter via perceptions” (1157). Thus, following the proximal contact literature, to measure social contact we inquire about relationships individuals may have with immigrants. Work on proximal contact also helps clarify the mechanism by which inter-group contact should impact policy attitudes – via political learning, where knowing someone threatened by a given policy can educate one to the consequences of said policy.

Still, research on proximal contact with immigration policy focuses on Latinos who are embedded in mixed-status communities, offering little insight into how proximal contact may impact white Americans. The inter-group contact literature speaks to this question by pointing to a similar set of processes identified as political learning that demonstrably reduce prejudice towards out-group members. Thus, these two bodies of work together provide the theoretical building blocks to understand how exposure to immigrants can impact white Americans’ attitudes towards immigrants and immigration policy. This paper contributes to both bodies of work by specifying and testing an indirect relationship between contact and policy attitudes. Even when the group contact literature examines the impact of relationships with immigrants on policy attitudes, researchers focus on direct effects and find none (e.g., Gravelle 2016; Pearson-Merkowitz, Filindra, and Dyck 2016). Likewise, work on proximal contact has not yet examined policy outcomes.

In the forthcoming analysis, we test our theory drawing on the 2018 Cooperative Congressional Election Study (CCES). We find that having a loved one who is a Latino immigrant improves attitudes towards immigrants as a group, which in turn is associated with heightened support for less punitive, more permissive immigration policy. Our findings hold across a variety of robustness checks. While preliminary, these findings are practically, as well as intellectually, important. These findings shed light on potential advocacy strategies, including those that might prime personal connections to immigrants, that may yield improved support for permissive policies. In an increasingly hostile environment, immigrants and their native-born loved ones are a driving force shaping the nature and growth of the U.S. population. It is thus essential to understand how personal relationships with people different from one’s self shape policy preferences.

In what follows, we begin by reviewing the relevant literature around proximal contact and inter-group contact theory. We outline our theory for how proximal contact may impact policy attitudes, deriving a set of testable hypotheses. We then detail our data, methods, and extant findings. Implications are discussed.

The impact of personal relationships on public opinion

Proximal contact and support for punitive immigration policy

A growing body of work recognizes the impact of having a relational connection with someone threatened by punitive policies on political attitudes and behaviors. Scholars refer to having a loved one who has experience with punitive policies as proximal contact (Walker 2020; Sanchez et al. 2015; Anoll and Israel-Trummel 2019). Researchers examining the political consequences of proximal contact have focused on behavior, finding that proximal contact can both compel people to join civic groups and immigration protests, and degrade the likelihood of voting (Amuedo-Dorantes and Lopez 2017; Street, Jones-Correa, and Zepeda-Millán 2017; Israel-Trummel and Shortle 2019). Proximal contact impacts behavior through two key mechanisms. First, proximal contact can make the material threat posed by punitive immigration policy salient, and it can spur individuals to take defensive action (Walker and García-Castañon 2017; Anoll and Israel-Trummel 2019). Second, proximal contact degrades trust in government, which may in turn manifest as withdrawing from voting, even as individuals engage in system-challenging behavior (Rocha, Knoll, and Wrinkle 2015). This dynamic is compounded by the inclination to avoid activities that risk exposing one’s status or the status of a loved one (Brayne 2014; Pedraza, Nichols, and LeBrón 2017).

Proximal contact impacts collective consciousness and trust in government because, “the design and implementation of public policy constitute important forces shaping citizens’ orientations toward the institutions and policies of government” (Mettler and Soss 2004, 62). Scholars refer to this process as political learning, which suggests that proximal contact with immigration policy may change policy attitudes by educating individuals to the consequences of punitive policies (Rocha, Knoll, and Wrinkle 2015; Lerman and Weaver 2014). However, researchers have yet to establish a link between proximal contact and specific immigration policy attitudes. Moreover, the majority of this work focuses on attitudes held by Latinos.² For insight into whether and how the logic of political learning might impact white Americans we turn to research on the impact of inter-group contact on attitudes towards immigrants.

Inter-group contact and attitudes towards immigration

A deep body of research examines the political consequences of contact with people who are not members of one’s group. In his groundbreaking work on the topic, Allport (1954) theorized that exposure to members of out-groups can reduce perceived social distance, increase affection, and decrease prejudice (Pettigrew 1998; Pettigrew et al. 2011; Page-Gould, Mendoza-Denton, and Tropp 2008; Schlueter and Scheepers 2010).³ Scholars have assessed this theory through a variety of measures of inter-group contact, yielding mixed findings, particularly as it pertains to the relationship between influxes of new immigrants and perceptions towards them (Dixon 2006; Enos 2014; McLaren 2003; Della Posta 2013). The assumption underlying the use of geographic variables to proxy for inter-group contact is that heightened presence of immigrants in a neighborhood increases opportunities for contact, which in turn may override stereotypes and soften negative perceptions. Yet, researchers frequently find the opposite (Enos 2014; Hero and Tolbert 1996; Tolbert and Hero 1996; Campbell, Wong, and Citrin 2006; Hood and Morris 1997; Hopkins 2010; Fox 2004). The relationship some researchers observe between the presence of immigrants and negative out-group perceptions is often attributed to group threat and inter-group conflict (Blalock 1967; Key 1949). Yet, recent work highlights that, “[w]ork on the effects of racial/ethnic context and spacial proximity often assumes that such contextual measures serve as a proxy for personal contact. This assumption often stems from the fact that measures of personal contact … are rare in social surveys” (Gravelle 2016, 2). Gravelle demonstrates that while geographic context can moderate the impact of partisan cues on policy attitudes, measures of personal contact with immigrants do not necessarily impact these same attitudes, concluding that, “these findings challenge the assumption that contextual measures are proxies for interpersonal contact” (2016, 3).

Inter-group contact theory develops the basic idea that interacting with people who are unlike one’s self can enhance empathy through humanizing out-group members (Pettigrew et al. 2011). Empirically, research leveraging relational measures of inter-group contact (as opposed to contextual measures) overwhelmingly demonstrates that having a relationship with an out-group member reduces negative stereotypes and prejudicial attitudes (Dixon 2006; Hewstone and Schmid 2014; Levin, Van Laar, and Sidanius 2003; Neumann and Moy 2018; Page-Gould, Mendoza-Denton, and Tropp 2008; Schlueter and Scheepers 2010; Wright and Tropp 2005). For example, drawing on data from the 2000 General Social Survey (GSS), Dixon (2006) finds that simply knowing someone who is Asian or Latino, reduces prejudicial attitudes towards these groups among whites, as does having a close relationship with someone who is Black. Controlling for context using a matched design at the county level, O’Neil and Tienda (2010) find that sustained contact with immigrants outside the workplace decreases beliefs that immigrants are problematic. The impact of inter-group contact is particularly potent among those who have close, personal relationships with out-group members (Pettigrew 1998).

The capacity for inter-group contact to reduce prejudice holds across a variety of contexts, including with respect to the LGBTQ community, the mentally ill and the disabled (Pettigrew et al. 2011). Hewstone and Schmid (2014) find that more positive contact with Muslims improved out-group trust among non-Muslims in England. Contact with immigrants diminished prejudice among majority group members across several European countries (McLaren 2003; Neumann and Moy 2018; Della Posta 2013). Thus, extensive research confirms the theoretical proposition offered by researchers concerned with proximal contact: relational connections can change individuals’ perceptions of people who are not like themselves. The process of political learning, which researchers have observed among Latinos as a consequence of proximal contact, most likely occurs among whites who have contact with Latino immigrants.

Yet, research linking inter-group contact and policy attitudes is more mixed. A handful of studies demonstrate that the positive effect of inter-group contact on perceptions of immigrants extends to support for more permissive immigration policy (McLaren 2003; Neumann and Moy 2018; Della Posta 2013). However, all of these studies concern countries in Europe. In the U.S., partisanship predominately shapes opinion (Chandler and Tsai 2001; Abrajano and Hajnal 2015). Using Latino ethnicity as a proxy for immigration, Pearson-Merkowitz, Filindra, and Dyck (2016) find that knowing someone who is Latino improves support for permissive policy, but primarily among Democrats. Other studies link changing demographics at the community level to support for restrictive policies, but those that further examine the intervening role of relational measures of contact yield conflicting findings. While Gravelle (2016) finds that living in a county with a large Hispanic population increases support for permissive immigration policies among Democrats, contact has no impact on policy attitudes. At the same time, Stein, Post, and Rinden (2000) find that inter-group contact increases support for permissive immigration policies among those who live in high-immigrant communities.

In sum, research on the political consequences of proximal contact with immigration policy establishes that individuals learn political lessons by watching their loved ones negotiate a punitive immigration environment. These lessons are so poignant that they can shape attitudes towards government and compel political action. Research on inter-group conflict suggests that we should not expect political learning to be limited to target group members, but that it can impact the attitudes of non-target group members. However, neither body of work clearly links contact to policy attitudes, highlighting a need for additional empirical and theoretical work in this area. The assumption underlying research that tries to link inter-group contact to policy attitudes is that individuals extend the lessons learned from inter-group contact to politics more generally. This work instead gets sidetracked by contextual measures of contact and the predominant role of partisanship. The logic of political learning and research on proximal contact clarifies the theoretical leap from attitudes about immigrants to attitudes towards immigration policy. Researchers have yet to explore this underlying, indirect relationship between inter-group contact and policy attitudes. It is to developing, theoretically, and empirically assessing this indirect relationship that we now turn.

Theoretical expectations

Our theory about how proximal contact influences policy attitudes is as follows: Relationships with people impacted by a given policy improves perceptions of people unlike one’s self, on the one hand; and has the potential to educate individuals to the consequences of said policy, on the other hand. When individuals translate positive personal experiences with individuals to the whole out-group, improved perceptions of immigrants as a group can shape attitudes towards immigration policy. In particular, positive perceptions of immigrants may yield improved support for policies that expand access to rights and support the integration of immigrants into society. The relationship between proximal contact and attitudes towards immigration policy is therefore theorized to be indirect, operating via improved attitudes towards immigrants as whole and contingent on positive interactions with individual immigrants.

This argument generates a series of testable hypotheses. The first concerns the impact of proximal contact on attitudes towards immigrants as a whole. We hypothesize the following:

• Proximal contact with a Latino immigrant will be associated with more favorable perceptions of immigrants as a group.

However, for this hypothesis to hold, the proximal contact should be through a meaningful relationship with an immigrant. Indeed, inter-group contact theory holds, “the special importance of cross-group friendship in promoting positive contact effects,” because friendship provides those conditions required for contact to change attitudes (Pettigrew et al. 2011, 275). This generates the following qualifying hypothesis:

• The impact of proximal contact with a Latino immigrant on perceptions of immigrants as a group will be more pronounced among those whose proximal relationship is with a loved one, relative to those whose proximal relationship is with an acquaintance or co-worker.

The above hypotheses are well-trod ground in the inter-group contact literature. Less well understood are the conditions under which proximal contact yields shifts in policy attitudes. We argue that the relationship is indirect, impacting policy positions via improved attitudes towards immigrants as a whole. We therefore hypothesize that:

• Positive attitudes towards immigrants as a group will be associated with support for permissive immigration policies, even if proximal contact itself does not directly influence attitudes towards these same policies.

Again, this hypothesis follows from inter-group contact theory. Previous research has not fully investigated the indirect nature by which proximal contact should influence policy attitudes. The indirect nature of the relationship between proximal contact and policy attitudes theorized here suggests that positive affect towards immigrants as a group is a mediator, generating our final primary hypothesis:

• Positive attitudes towards immigrants as a group mediates the relationship between proximal contact and attitudes towards permissive immigration policy.

Work elsewhere examines the role of living in a community characterized by growth in immigrant and minority populations. When researchers operationalize inter-group contact using geographic measures of proximity to out-group members, they often find that proximity is associated with support for restrictive immigration policy. Geo-graphic measures of proximity fail to capture the most important component of inter-group contact for attitudes, a sustained and meaningful relationship with an out-group member. Thus, the above specified relationship between proximal contact and policy attitudes should persist even after accounting for geographic context.

Finally, a condition embedded in our argument is that the nature of proximal contact is positive (Thomsen and Rafiqi 2018). If people have poor experiences with the Latino immigrants with whom they have contact, we should anticipate the opposite results – when proximal contact is itself negative, it should yield decreased support for permissive policies and increased support for restrictive ones.

Data and methods

Our data are from an original module on the nationally representative 2018 Cooperative Congressional Election Study (CCES) pre-election survey. The survey was conducted online through YouGov and includes 747 self-identified non-Hispanic white respondents.

Proximal contact

Our key independent variable is whether one has a relational connection to an immigrant (proximal contact). A detailed breakdown of these survey questions is available in the appendix, in Tables A1 and A2. We asked all respondents: “Now take a moment to think about all the people in your family, your friends, coworkers, and other people you know. Do you happen to know somebody who is an immigrant?” Since we are interested in the capacity of proximal contact to change the attitudes of white Americans towards racial and ethnic minorities, we then followed up and asked whether the immigrant one knew was Latino/Hispanic. Finally, we inquired about the nature of that relationship, asking: “Thinking about the Latino immigrant with whom you are closest, what is your relationship with the person? Select all that apply: Spouse or romantic partner, other family member, friend, neighbor, co-worker, employee, yourself, other.” Nearly 90% of respondents indicate the person of whom they are thinking is a romantic partner, family member or friend, which we code as having a loved one who is a Latino immigrant.

Our measure of proximal contact is not without limitation. Had we first asked individuals to think of their loved ones and then asked them to identify how many of them were Latino immigrants, we may have arrived at a different distribution on this variable. There is, thus, slippage between the concept/theory. Attitudes are theorized to transform via a close and caring relationship; our measure captures relationships that may not be particularly strong, and may be threatened by social desirability bias where individuals prone to supporting lenient immigration policy are also likely to indicate they have a friend who is a Latino immigrant when they really do not. In this way, our measure is subjective. Yet, attempts to objectively measure relationships with Latino immigrants that exist in the literature fall short at capturing the central relational mechanism posited to affect attitudes. Previous scholars have used context, where living among Latinos is not itself indicative of relationship status. One scholar has introduced exposure experimentally by paying people who are Latino to frequent a train station, and measuring attitudes of patrons pre- and –post-exposure. This again fails to capture the relational mechanism of interest here. Given that we innovate on this measure, we anticipate that any relationship observed here between proximal contact and attitudinal shifts at least provides impetus for future research to continue to improve precision of measure and design when examining the role of relational ties with out-group members in shaping policy attitudes, in this case via the humanization of immigrants.⁴

Our proximal contact variable has four categories: not knowing/not being sure if they know any immigrant (43% of white respondents, N = 318); not knowing a Latino immigrant/but knowing some immigrant (14% of white respondents, N = 102), knowing a Latino immigrant who is not a loved one (19% of white respondents, N = 144), and knowing a Latino immigrant who is a loved one (24% of respondents, N = 181).⁵ In 2015, 17% of all U.S. newlyweds were interracial, and the most common pairing, accounting for 42%, is one Hispanic and one white spouse (Livingston and Brown 2017, May 18). According to the General Social Survey (GSS), in 2000 74% of individuals reported that they knew someone who is Hispanic, and in 2006 33% of individuals indicated that between 1 and 5 people they trusted were Hispanic (Smith, Michael Hout, and Kim 1972–2018). Drawing on a survey from 2018, Walker, Roman, and Barreto (2019) report that 22% of white respondents indicated they knew someone undocumented or who had contact with law enforcement for reasons related to immigration. Thus, that 24% of respondents have a loved one who is a Latino immigrant appears externally valid.

Outcomes

We have two key outcomes of interest. We hypothesize that proximal contact is associated with support for permissive immigration policies via improved perceptions of immigrants as a group. We are therefore interested in the relationship between proximal contact and perceptions of immigrants; and in the relationship between proximal contact, group perceptions, and support for immigration policies.

To measure attitudes towards immigrants as a group, we draw on a series of five questions that measure respondents’ perceptions of traits of Latino immigrants. We assess each item separately and as a scale where the items are added together (Cronbach’s alpha = .88. See appendix Table A3 for item correlations).⁶ Responses are scaled to range from 0 to 1 and coded such that higher values reflect more positive perceptions of Latino immigrants. The question stem was: “Please place your perceptions of Latino immigrants on a seven-point scale, based on how well you think each trait describes the group.” (1) Violent to Non-Violent; (2) Intelligent to Unintelligent; (3) Hardworking to Lazy; (4) Shares my values to Does not share my values; (5) Cares about the future of this country to Does not care about the future of this country.

Our second outcome of interest is attitudes toward immigration policy. These questions are dichotomous, coded such that higher values reflect less punitive and more permissive attitudes toward immigration policy. The first two questions relate to the role of local authorities: “Should YOUR local police take an active role in identifying undocumented or illegal immigrants (0), or should enforcement be left mainly to the federal authorities? (1).” The second question was “Which statement do you agree with most? Sanctuary cities are more likely to increase crime and harbor criminals (0); Sanctuary cities are more likely to reduce crime (1).” Additional questions asked respondents whether they support or oppose each of the following: (1) Increase spending on border security by $25 billion, including building a wall between the U.S. and Mexico (0 = Support; 1 = Oppose); (2) Provide legal status to children of immigrants who are already in the United States and were brought to the United States by their parents. Provide these children the option of citizenship in 10 years if they meet citizenship requirements and commit no crimes (DACA) (Oppose = 0; Support = 1); (3) Reduce legal immigration by eliminating the visa lottery and ending family-based migration (0 = Support; 1 = Oppose); (4) Withhold federal funds from any local police department that does not report to the federal government anyone they identify as an illegal immigrant (0 = Support; 1 = Oppose); and (5) Send to prison any person who has been deported from the United States and reenters the United States (0 = Support; 1 = Oppose). We first evaluate each item independently. We then evaluate an index, created by summing the items and scaling to range from 0 to 1. The Cronbach’s standardized alpha for the scale is 0.88, and correlations between items are displayed in the appendix, in Table A3.

Other factors that impact attitudes towards immigration

We primarily rely on ordinary least squares regression to assess the association between proximal contact, perceptions of immigrants and attitudes towards immigration policy. However, whether a white person has a loved one who is a Latino immigrant is not random, and a variety of other factors may likewise impact attitudes towards immigration policy. We therefore include controls for potential confounding variables.⁷

A key variable previous scholarship has identified as impacting attitudes towards immigration policy and which has been used to measure inter-group contact is the geographic context in which one lives. In order to account for the possibility that context may impact attitudes in ways distinct from proximal contact, we control for the proportion Hispanic and change in the proportion of foreign born in a respondent’s zip code. The other key factor researchers have found overwhelms inter-group contact to shape attitudes towards immigration is partisanship (Pearson-Merkowitz, Filindra, and Dyck 2016; Chandler and Tsai 2001; Abrajano and Hajnal 2015). We control for party affiliation (with Democrat as the reference category) and approval of President Trump, who has contributed to an increasingly polarized climate around issues related to immigration.

Attitudes towards racial out-groups also contribute to positions on immigration policy, prompting us to control for racial resentment (Brader, Valentino, and Jardina 2009; Hutchings and Wong 2014). Finally, we include demographic controls which may likewise impact political attitudes. These include education, income, gender, marital status, whether the respondent owns a home, whether a respondent lives in a city (vs. a suburban or rural area), whether one is registered to vote and political interest. We also control for ACS measures of the median value of housing and the population proportion below the federal poverty line in the respondent’s zip code as additional economic indicators, since previous research has likewise demonstrated that economic factors can impact attitudes towards immigration (Hainmueller and Hiscox 2010; Scheve and Slaughter 2001; Facchini, Mayda, and Puglisi 2009; Citrin et al. 1997).

Results: proximal contact and perceptions of immigrants

We first examine the descriptive relationships between proximal contact with Latino immigrants and immigrant perceptions. Figure 1 displays the raw average score of each outcome variable by level of proximal contact with Latino immigrants. Proximal connections are associated with positive perceptions of Latino immigrants, particularly when that connection is a loved one.

Figure 1.

Average outcome value on trait perceptions of Latino immigrants.

Note: Figure displays the average outcome score by level of contact with Latino immigrants.

Figure 2 displays these relationships after including controls for relevant confounding variables.⁸ The figure displays the coefficient estimate for the average marginal effect of having a loved one who is a Latino immigrant versus not knowing an immigrant, and 95% confidence intervals. Full coefficient results for each outcome and by each category of proximal contact are available in appendix Tables A6 and A7.

Figure 2.

Average marginal effects of having a Latino immigrant loved one on trait perceptions of Latino immigrants.

Note: Figure displays coefficient estimates and 95% confidence intervals from ordinary least squares regressions with covariates.

In Figure 2, each result is positive and statistically significant: Having a loved one who is a Latino immigrant is associated with significantly more positive perceptions of the traits of Latino immigrants as a group.⁹ Respondents with other kinds of proximal contact also hold more positive attitudes about the impact of immigrants on society, but are not significantly more positive about the traits of Latino immigrants as a group (displayed in appendix Table A6). The size of the effects of having a loved one who is a Latino immigrant are generally stronger than other relationships (appendix Figure A4). Thus, the findings from a regression analysis support our first two hypotheses: that proximal contact is associated with improved attitudes towards immigrants as a group, and that this relationship is most pronounced for those whose proximal connection is a loved one.¹⁰

Previous research has leveraged the immigrant population in the community where one lives to proxy for inter-group contact. In contrast to the findings presented here, this work has found that living in such an environment is associated with increasing hostility towards immigrants. To address this, we have included geographic controls in our regression models. However, we subject the robustness of the relational finding to further scrutiny through matching. That is, we match respondents according to the context in which they live and their underlying political orientations, and compare those who have a loved one who is a Latino immigrant to those who do not know any immigrant (N = 479).¹¹

Table 1 displays the results of the matched analysis. The first two rows examine a more limited subset of N = 158 observations for whom we could find exact matches between the treatment and control group on contextual variables. The results of a bivariate analysis post-matching indicate that proximal contact improves attitudes toward Latino immigrants as a whole. We also employ “optimal” matching to avoid the need to find an exact match. This allows the estimate of treatment effects to incorporate a broader set of data (N = 354 matched observations), but the data retain some imbalances. We therefore add the full set of covariates to the regression model on the matched data. The coefficient on proximal contact remains positive and significant.¹²

Table 1.

Matching analysis results.

	Coefficient on proximal contact	Lower CI	Upper CI	N
Bivariate Exact: Trait	0.10	0.04	0.17	158
Bivariate Optimal: Trait	0.10	0.06	0.14	354
Full Covariates Optimal: Trait	0.06	0.02	0.10	354

Policy attitudes

So far, we have found support for the hypothesis that proximal contact can improve perceptions of immigrants as a group. We also have demonstrated that, in keeping with inter-group contact theory, the relationship is most pronounced among people whose proximal connection is a loved one. These relationships persist even accounting for contextual factors. Yet, less well understood in the literature is how proximal contact can shape one’s perspective on policy. We theorize that the relationship between contact and policy is indirect, operating via improved attitudes towards immigrants as a group. Our next testable hypothesis, then, anticipates that positive attitudes towards immigrants as a group will be associated with declining support for punitive immigration policy, even if proximal contact is not itself directly associated with these attitudes.

We first examine the direct relationship between proximal contact and policy attitudes. At the bivariate level having a loved one who is a Latino immigrant is statistically associated with lower levels of support for punitive immigration policy. Appendix Figure A8 displays the average marginal effect of having a loved one who is a Latino immigrant on policy positions, both in a bivariate regression model and in a fully specified regression model. The relationship between proximal contact and support for more permissive, less punitive policy is positive across all items in the bivariate model, and achieves statistical significance with respect to declining support for the use of prison to punish repeat border-crossers, supporting the visa lottery program, supporting DACA, opposing the wall, and the full scale of policy items. However, these relationships are not robust to the inclusion of covariates.

We next test our expectation that the relationship between proximal contact and policy is indirect. While proximal contact is not itself significantly related to support for less punitive immigration policy in a fully specified model, perceptions of Latino immigrants’ traits are positively associated with policy support. Table 2 displays the relationship between proximal contact, perceptions of Latino immigrants, and support for more permissive immigration policy as measured by the full policy index. Coefficients displayed here reflect models including a full battery of controls. These relationships displayed in Table 2 suggest that insofar as proximal contact improves attitudes towards immigrants as a group, it likewise has the capacity to reduce support for punitive immigration policy.¹³

Table 2.

Proximal contact, perceptions of immigrants and support for permissive immigration policy.

	Policy attitudes
Knows non-Latino immigrant	0.003
	(0.024)
Latino immigrant loved one	0.009
	(0.020)
Latino immigrant non-loved one	0.002
	(0.022)
Trait perceptions	0.414∗∗∗
	(0.042)
Observations	692
R2	0.746
Adjusted R2	0.736

Note:

^∗p < 0.1;

^**p < 0.05;

^***p < 0.01.

Models include covariates from fully specified models, located in appendix Table A10.

We next assess the nature of the indirect relationship between proximal contact and policy attitudes by leveraging a mediation analysis. We have theorized that proximal contact is related to policy attitudes insofar as it impacts perceptions of immigrants as a whole. We apply the Imai, Keele, and Yamamoto (2010) causal mediation analysis using the Tingley et al. (2014) R package. We consider “having a Latino immigrant who is a loved one” to be the treatment and “not knowing an immigrant” to be the control. Note, this method requires significant assumptions about the structure of the proposed relationship, particularly, sequential ignorability. We must assume we have included all relevant pre-treatment confounders with the treatment and also assume there are no post-treatment confounders with the mediator and the outcome. That is, if some other mechanism, correlated with our proposed mediator of perceptions of immigrants, also explains the relationship between proximal contact and policy attitudes, then our results are likely biased. This assumption is untestable, and results are suggestive.

We carry out a mediation analyses, using trait perceptions of Latino immigrants as a mediator. The mediation analysis requires two regression models, one wherein the treatment (proximal contact) predicts the mediator (perceptions of immigrants, models displayed in appendix Table A6), and a second wherein the treatment and mediator predict the outcome (policy attitudes, models displayed in appendix Table A10). Figure 3 displays a summary of the results testing specifically for the mediation effect. The figure includes the total effect of proximal contact on policy attitudes, the average direct effect of proximal contact on policy attitudes (ADE), and the average causal mediation effect (ACME). The ACME represents the indirect effect of having a loved one who is a Latino immigrant on policy attitudes through its influence on respondents’ perceptions of immigrants.

Figure 3.

Mediation analysis for policy attitudes.

Note: Figure displays point estimates and 95% confidence intervals.

The ACME is positive and significant (p< 0.01), while the total effect is positive but is not significant. Having a loved one who is a Latino immigrant is weakly associated with less punitive attitudes on immigration policy, but in keeping with our expectations, this effect largely works indirectly through its influence on attitudes about immigrants. Whether proximal contact translates into a shift in policy attitudes appears to depend on how much individuals translate their positive perceptions of the immigrant they know into perceptions about Latino immigrants as a group. This offers support for our final hypothesis – that positive attitudes towards immigrants as a group mediates the relationship between proximal contact and declining support for punitive immigration policy. Figure A10 in the appendix displays sensitivity analyses for the ACME, which assess how robust the effect is to violations of sequential ignorability due to unmeasured confounders. This analysis suggests that the estimates presented above are fairly robust, with the key sensitivity parameter employed by Imai, Keele, and Yamamoto (2010) having to fall above 0.4 to reduce the ACME observed in Figure 3 with trait perceptions as the mediator to zero.

Robustness and placebo analyses

In a final set of analyses, we first test a premise of our theory that proximal contact should influence attitudes in a pro-immigrant direction contingent on the nature of the proximal contact. Specifically, our theoretical expectation is that proximal contact improves support for permissive policies when it manifests as improved perceptions of immigrants as a group. The flip-side of this expectation is that when individuals have negative interactions with whomever they are connected to, such interactions can yield support for punitive policies instead of more permissive preferences. While most respondents in the sample rate the Latino immigrant they know positively overall (see appendix Figure A11 for distribution), this is not uniformly true. We therefore restrict the analysis to individuals with proximal contact, in order to examine variation in trait perceptions of the Latino immigrant the respondent knows, and the consequent relationship to perceptions of Latino immigrants, overall, and policy attitudes.

We expect to find that among those with proximal contact, positive perceptions of the immigrant an individual knows is itself associated with more positive attitudes towards Latino immigrants in general. The left pane of Figure 4 confirms this expectation, displaying the average estimated score on the trait perceptions index, given how one evaluates the immigrant they personally know, controlling on covariates. More positive perceptions of the Latino immigrant a respondent knows translate into more positive perceptions of Latino immigrants as a whole, while more negative proximal contact is associated with more negative perceptions (p< 0.001).¹⁴ The right pane of Figure 4 in turn displays the relationship between how one evaluates the immigrant they personally know and support for permissive policies, where again, negative evaluations are associated with lesser support for permissive policies and positive evaluations are associated with greater support for these same policies (p< 0.001).¹⁵

Figure 4.

Variation within respondents who know a Latino immigrant on attitudes toward Latino immigrants and immigration policy attitudes.

Note: Figure displays point estimates and 95% confidence intervals, as well as points corresponding to perceptions of the immigrant the respondent knows.

These analyses suggest that the impact of proximal contact on policy attitudes is indirect, working via attitudes towards immigrants as a whole, and conditional on the nature of contact. Another potential alternative explanation is that it may be that political ideology, rather than proximal contact, is shaping the way individuals think about immigrants and the challenges they face. To help rule out this alternative pathway, we assess whether proximal contact and trait perceptions are associated with other policy attitudes outside of immigration. If our theory is correct, we should primarily find a significant relationship between proximal contact and attitudes on immigration policy and not in other policy areas.¹⁶ If we do observe significant relationships in other areas, we might worry that we have not adequately accounted for ideology in our models – that something other than proximal contact is driving the relationships observed. We therefore repeat a mediation analysis, but use two different policies as outcome variables: support for Medicare for All, and a measure of gun control, support for universal background checks. These questions were asked during the pre-election wave of the CCES as part of the common content questionnaire. Figure A13 in the appendix displays the results. Proximal contact does not have a direct or indirect effect on attitudes towards either policy. This provides additional support for our theory that it is indeed proximal contact – and not a broader underlying values orientation – driving attitudes towards immigration policy in our study.

Discussion and conclusion

We began by asking: How does exposure to immigrants and their communities (and in particular, Latino immigrants and their communities) impact attitudes towards immigrants and immigration held by white Americans? Drawing on growing work around the impact of proximal contact with punitive institutions on political attitudes, together with the robust literature on inter-group contact theory, we argue that having a loved one who is a Latino immigrant shapes policy attitudes indirectly, via changed attitudes towards immigrants as a group. That is, when individuals have a positive relationship with an immigrant, and they translate that positive experience into improved perceptions of immigrants writ large, that positive relationship can yield support for permissive immigration policy. While this is well-trod theoretical ground in the inter-group contact literature, scholars have struggled to empirically link contact to policy attitudes. Research on the extant effects of proximal contact has not addressed policy attitudes in any significant way.

In order to assess our theory, we draw on an original module of questions fielded on the 2018 CCES. We bring this survey data together with a variety of contextual measures, and leverage questions about personal experiences with Latino immigrants and immigration in one’s geographic community. We work through a progressive set of testable hypotheses, first finding that having a loved one who is a Latino immigrant is associated with improved attitudes towards immigrants as a group. These findings hold across a variety of measures of attitudes towards Latino immigrants and immigrant contributions to society, and are robust to a variety of model specifications.

We then assess the independent effect of both proximal contact and attitudes towards immigrants on policy attitudes, and find that while proximal contact itself is not significantly associated with policy positions after accounting for additional covariates, positive views of immigrants are associated with increasing support for permissive immigration policy. Using a more formal analysis of this relationship, we find evidence that improved perspectives towards immigrants mediates the relationship between proximal contact and policy attitudes. Finally, we demonstrate that the inverse of our theory is also true – for those who view their proximal relationships negatively, attitudes towards immigrants as whole are more negative and support for punitive policies is higher.

This paper contributes to inter-group contact theory empirically, insofar as we demonstrate an indirect link between contact and policy attitudes. Previous work has searched for a direct link, but only consistently found that contact reduces prejudice. We also build on existing work around proximal contact with punitive institutions, which has yet to examine policy attitudes, instead focusing on trust in government and political behavior. We extend this to chart out a path by which policy attitudes may likewise be influenced by contact. Finally, we contribute to work in racial and ethnic politics by developing a theory around how white attitudes towards immigration may be moved from perceiving immigrants negatively as a threat to humanizing and viewing immigrants more positively, and offer preliminary evidence that is supportive of this theory. Immigration is one of the most salient domestic policies driving electoral politics (Abrajano and Hajnal 2015; Matos 2020). A better understanding of ways to mitigate immigration policy preferences among non-Hispanic white Americans, who are the majority of the voting electorate, is crucial. Our paper finds that perceived or even aspiring personal relationships with members of out-groups matter. The changing U.S. electorate is shaped by new immigrants, their families and communities. Our research demonstrates that personal relationships with immigrants prove to be one way to combat “white backlash” (Abrajano and Hajnal 2015) vis-à-vis immigration policy.

Our analyses are not without limitations. While we have leveraged a wide range of identification strategies for observational data, the survey is nevertheless cross-sectional. We cannot demonstrate attitudinal change from our data, despite the theoretical story we posit. Thus, future research is needed to isolate the exact causal chain linking proximal contact with changes in policy attitudes. Panel or repeated cross-sectional data may be particularly well suited to this task, where cross-sectional experimental work faces the challenge of simulating deep and meaningful relationships with an immigrant. Likewise, we draw on a single survey. We do this because this survey included the full suite of variables relevant to our theory, which is in turn well-grounded in the literature. Yet, these findings should be subjected to replication and extension.

Additional areas for future research should focus on identifying the limits and contours of the theory outlined here. Does learning about a policy through a relational connection impact attitudes towards that policy in areas other than immigration? We might expect similar findings when looking at other punitive policies, such as in the areas of welfare and criminal justice. What about when a policy does not obviously single out a group or whose consequences are widespread and esoteric, such as those with respect to the environment? What kinds of relationships create attitudinal change? Are certain demographic groups (e.g., the young, the more/less educated, those with more/less wealth) especially likely to be impacted by proximal contact? As the demographics of the nation continue to change, growing closer to minority-majority status, understanding the factors that shape how white Americans regard their minority neighbors is of heightened importance.