GOVERNMENT 2306(II) ESSAY
Immigration Enforcement and the Redistribution of Political Trust
icy feel alienated. We reevaluate this argument by examining how immigration enforcement policies affect Latino im-
migrants, native-born Latinos, and Anglos. Using data from a 2012 survey of Texas residents and government statistics
on the Secure Communities program, we find that higher removal rates, especially among noncriminal populations, are
associated with negative political orientations among native- and foreign-born Latinos. We also extend social con-
struction theory by arguing that the attitudes of individuals outside the target group are affected by enforcement. Our
findings show that Anglos living in high-enforcement contexts are the most trusting of government and externally ef-
ficacious. We conclude that immigration policy enforcement redistributes trust in government from resource-deprived
immigrants to Anglos. Enforcement practices thus perpetuate existing political inequalities.
In 2008, the Bush administration began the Secure Com-munities program, an immigration enforcement initiative.The Obama administration expanded the program before replacing it with the largely similar Priority Enforcement Program (PEP) in 2015. Now when an offender is booked in a US prison, his or her fingerprints are automatically cross- referenced with an immigration database operated by US Im- migration and Customs Enforcement (ICE). ICE uses this process to identify, apprehend, and sometimes deport unau- thorized immigrants. ICE prioritizes the removal of criminals and those who have committed multiple immigration vi- olations (US Immigration and Customs Enforcement 2012). From its inception through August of 2012, ICE received nearly 20 million sets of fingerprints and removed more than 220,000 individuals from the United States through Secure Communities. More than 75% of deportees were convicted criminals. Approximately a quarter of these criminals were convicted of serious crimes, such as homicide, robbery, and
kidnapping (US Immigration and Customs Enforcement 2012). After a series of Freedom of Information Act (FOIA) requests, ICE released detailed data about Secure Communi- ties enforcement. These data contain information on en- forcement patterns across political jurisdictions, and ICE has never released similar identifiers alongside general enforce- ment data.
The number of deportations is higher today than ever before and a source of controversy.1 Concerns include the mis- taken arrest and removal ofAmerican citizens and recent efforts by some state and local governments to opt out of local- federal partnerships. A 2011 report by a task force on Se- cure Communities offered another worry: “The program [has] eroded public trust by leading to the detention of many immigrants who had not committed serious crimes, after officials said its aim was to remove ‘the worst of the worst’ immigrant criminals from the United States” (Homeland Se- curity Advisory Council 2011). Immigrant interest groups
Rene R. Rocha (rene-rocha@uiowa.edu) is associate professor of political science at the University of Iowa. He studies Latino politics and immigration pol- icy. Benjamin R. Knoll (benjamin.knoll@centre.edu) is an assistant professor in the politics program at Centre College in Danville, Kentucky. His research covers
American public opinion and voting behavior, with an emphasis on race and politics and religion and politics. Robert D. Wrinkle (rdwrinkle@gmail.com) is professor emeritus at the University of Texas–Pan American. His interests include public policy, political behavior, and Latino politics.
Data and supporting materials necessary to reproduce the numerical results in the paper are available in the JOP Dataverse (https://dataverse.harvard .edu/dataverse/jop). An online appendix with supplementary material is available at http://dx.doi.org/10.1086/681810.
1. The number of removals peaked in 2013, and ICE has not released official figures for 2014 (US Immigration and Customs Enforcement 2014).
The Journal of Politics, volume 77, number 4. Published online July 7, 2015. http://dx.doi.org/10.1086/681810 q 2015 by the Southern Political Science Association. All rights reserved. 0022-3816/2015/7704-0002$10.00 901
have echoed this concern. Several public statements by these organizations express profound disappointment with Presi- dent Obama for pursuing such an aggressive and punitive deportation program in the first few years of his presidency.
What has been the effect of this immigration policy en- forcement campaign? In this article, we examine whether Secure Communities has affected trust in government and political efficacy among Latino immigrants, native-born La- tinos, and Anglos (non-Hispanic whites). Below, we discuss what scholars already know about immigration politics and policy. We then sketch out a theory of how immigration policy enforcement affects political behavior and analyze data from a survey of Texas residents. The results show that im- migration enforcement affects political orientations within each subgroup.
IMMIGRATION POLICY ENFORCEMENT AND POLITICAL BEHAVIOR Current literature tells us much about what factors affect immigration policy attitudes. Residential context (Hood and Morris 1997; Hopkins 2010), cultural and economic threat (Burns and Gimpel 2000; Citrin and Sides 2008; Paxton and Mughan 2006), implicit attitudes (Ayers et al. 2009; Brader, Valentino, and Suhay 2008; Lu and Nicholson-Crotty 2010; Perez 2010), media framing (Valentino, Brader, and Jardina 2013), and demographic characteristics (Binder, Polinard, and Wrinkle 1997; Fetzer 2000; McDaniel, Nooruddin, and Shortle 2011; Neiman, Johnson, and Bowler 2006) all shape preferences.
We also know why lawmakers support reforms to cur- rent immigration policy. Constituency demographics, such as Latino and immigrant population size, matter (Casellas 2010; Fetzer 2006; Hero and Tolbert 1995). The character- istics of the legislators themselves, such as their race/ethnic- ity or partisanship, also predict positions on immigration- related legislation (Bratton 2006; Kerr and Miller 1997; Welch and Hibbing 1984). Studies of state and local politics empha- size that policy outcomes reflect citizen ideology (Hero and Preuhs 2007) and the power of industries that rely on immi- grant labor (Nicholson-Crotty and Nicholson-Crotty 2011).
In short, we have a good understanding of public per- ceptions toward immigrants, how elected officials represent those attitudes, and what kinds of policies eventually are adopted. This leaves one important gap: how policy, once implemented, affects politics. Is there a feedback effect? The social construction of policy design theory (Schneider and Ingram 1993) provides strong theoretical support for the belief that political orientations, specifically trust in govern- ment and political efficacy, are associated with immigration policy enforcement.
The social construction of policy design theory (SCPDT) argues that groups are socially constructed to be associated with either positive or negative evaluations. Groups also have different amounts of power and can be classified as “ad- vantaged” (high power 1 positive construction), “contend- ers” (high power 1 negative construction), “dependents” (low power 1 positive construction), or “deviants” (low power1 negative construction).2
SCPDT predicts that groups with positive constructions will receive more government benefits than those with nega- tive constructions. We thus see elites regularly and visibly advocating for policies such as veterans’ benefits, small busi- ness tax cuts, and government-mandated accommodations for the disabled and elderly. Those with positive social con- structions internalize signals, which leads to higher levels of political efficacy, greater trust in government, and gener- ally more positive attitudes toward politics and society. Mem- bers of negatively constructed groups internalize cues too. De- viants, SCPDT argues, are therefore the most likely to express feelings of alienation (Schneider and Ingram 1993, 342).
The possible spillover consequences of social construction– based policy decisions are not well understood. We want to know whether policy outcomes affect individuals outside of the target group. SCPDT predicts that political attitudes among foreign-born Latinos are affected by immigration enforcement because punitive practices send cues about the value of im- migrants. But enforcement practices also send signals to those outside of the target population about whether the govern- ment is willing to allocate burdens onto a negatively con- structed out-group. Seeing negatively constructed out-groups punished can reinforce faith in the political system. Perceiving the government as unable to penalize “deviants” may jeopar- dize trust.
The implications are important. Previous research has demonstrated that Latino immigrants come to the United States feeling more efficacious and trusting of government than native-born residents (Abrajano and Alvarez 2010; Mi- chelson 2003; Wenzel 2006). Trust in government is a key indicator of general support for or alienation from the po- litical system (Miller 1974). Both trust and efficacy predict political participation (Easton and Dennis 1967; Pateman 1976; Pollock 1983). Possessing these psychological resources can compensate for common disadvantages Latino immi- grants face, such as lower levels of income, education, and political mobilization (Barreto 2005; Leal 2002; Michelson 2005).
2. For a detailed explanation of this classification system, see Schneider and Ingram (1993).
902 / Immigration Enforcement and Trust Rene R. Rocha, Benjamin R. Knoll, and Robert D. Wrinkle
Secure Communities has changed the way immigration law is enforced. Perhaps it has changed the way people who care about immigration feel about politics. If it has, one of the few resources immigrants have at their disposal—pos- itive psychological orientations—is likely eroding. As the voice of the Latino immigrant community becomes soft, political inequality in America grows.
HYPOTHESES We leverage the unusually detailed data on Secure Com- munities supplied by ICE to study whether enforcement has affected political attitudes among foreign-born Latinos, native-born Latinos, andAnglos. Specifically, we askwhether deportations initiated as part of the program are related to trust in federal and local government as well as political ef- ficacy, both internal and external.
Foreign-born Latinos Our first prediction is that foreign-born Latinos who live in communities with high levels of Secure Communities en- forcement will have less trust in government and less politi- cal efficacy. This is because intense enforcement reinforces the negative social construction of immigrants. If foreign- born Latinos perceive that American society considers im- migrants to be “deviant,” they will likely feel politically alienated.
H1. Foreign-born Latinos who live in communities with high levels of immigration enforcement are more likely to express negative feelings toward the political system than those who live in communities with low levels of enforcement.
No study has examined how policy outcomes affect im- migrants’ political orientations, but Massey and Sanchez (2010) believe ethnic identity among immigrants is sensi- tive to policy outcomes. Immigration enforcement and so- cial discrimination “promote the formation of a reactive iden- tity that explicitly rejects self-identification as American” (2). This relationship may be an example of a broader trend be- cause blacks express a stronger sense of racial identity in areas where they are most disadvantaged (Gay 2004; Welch et al. 2001).Our argument, therefore,fitswith thebroader literature on residential context and attitudes.
Native-born Latinos Although not directly subject to immigration enforcement, native-born Latinos still may be affected by policy feedback because of linked fate or group consciousness. Linked fate is the perception that one’s fate and self-interest are strongly
connected to the fate, success, and interests of a wider group (Dawson 1994). According to the 2006 Latino Na- tional Survey, the majority of native Latinos report that their “doing well” depends at least “some” or “a lot” on other Latinos doing well (Fraga et al. 2006).
Non-Latinos may treat native- and foreign-born Latinos similarly (Rocha et al. 2011). If the native-born Latinos experience worse interactions with political institutions in anti-immigrant environments, their attitudes may also change. However, we predict that the effect will be smaller than that observed for foreign-born Latinos. This is because the social networks of native-born immigrants are less likely to be directly affected by immigration enforcement (Fix and Zimmerman 2001), making perceptions of en- forcement less closely aligned with actual removal rates.3
This leads us to our second hypothesis:
H2. Native-born Latinos who live in communities with high levels of immigration enforcement are more likely to express negative feelings toward the political system than those who live in communities with low levels of enforcement, but this effect will be smaller inmagnitude than what is observed for foreign-born Latinos.
Anglos Removals, we argue, make Latinos less trusting and effi- cacious, but they should make Anglos feel positive. People support punishing negatively constructed groups with policy-induced burdens. Because Anglos are more likely to negatively construct unauthorized immigrants (Rocha et al. 2011), high enforcement rates should increase Anglo faith in the policy process. Social construction theorists have not fully explored group-based differences in policy feedback, and most empirical studies document effects among target populations (Mettler and SoRelle 2014). This argument helps expand our understanding of how policy feedback forces penetrate many parts of American society.
H3. Anglos who live in communities with high levels of immigration enforcement are more likely to express positive feelings toward the political system than those who live in communities with low levels of enforcement.
Hypothesis 3 is compatible with other theories. We could arrive at the same hypothesis via racial threat theory by as-
3. Unfortunately, our survey data do not allow us to distinguish be- tween second- and third-generation Latinos. We suspect that the negative effect of enforcement on trust and efficacy would be smallest among third- generation Latinos.
Volume 77 Number 4 October 2015 / 903
suming a link between policy enforcement and Anglo per- ceptions of immigrant group size. High levels of Secure Communities enforcement in their areas may lead Anglos to infer that the proportion of foreign-born Latinos is large or on the rise. Latino population size and growth predict at- titudes and policy preferences (Newman 2013). The result may be increased support for a government that is perceived to be actively enforcing immigration law.
In either case, there are significant normative ramifications. Cynicism toward government among native- and foreign- born Latinos is a possible and undesirable consequence of immigration enforcement. Immigrants enjoy few political advantages, and this relationship threatens valuable psycho- logical resources. But what if enforcement also increases these same psychological resources among Anglos, a group that enjoys numerous political advantages relative to Latinos? The effect of enforcement would be to redistribute psychological resources from an already-disadvantaged group (Latino im- migrants) to an already-advantaged group (Anglos).
DATA AND METHOD We test our hypotheses using data from a public opinion survey conducted by the Center for Survey Research at the University of Texas–Pan American. This survey was in the field during the summer of 2012 and used random digit dialing, including cell phone numbers, to sample Latinos and Anglos living in Texas. The response rate, computed using the AAPOR RR3 method, was 8%. Spanish trans- lations of the instrument were available, and native Spanish speakers administered Spanish surveys. Quota sampling was used to ensure an adequate number of Latino respondents. Latinos made up 31% of our final sample and Anglos com- prised 69%. The total number of respondents was 470. A small number of non-Latinos and non-Anglos were sampled (27 blacks, 4 Asian Americans, and 4 American Indians), and we exclude these groups from our analysis. Although the data were collected exclusively from Texas residents, we argue that the sample is well suited for studying the effect of immigration enforcement on political orientations among Latinos and Anglos. Texas has the third-largest immigrant population in the United States, ranking only behind New York and California. Furthermore, ICE officials implement- ing Secure Communities have been more active in Texas than in all but two states: California and Florida (US Im- migration and Customs Enforcement 2012). Still, we repli- cated our analysis with the 2012 American National Election Study (ANES) and unofficial estimates of immigrant remov- als. We describe this check in more detail at the end of our write-up.
Our dependent variables measure political orientations. We rely on four questions that separately capture trust in
different levels of government, external efficacy, and in- ternal efficacy. They are:
1. How often can you trust local government to do what is right? (Never, Some of the Time, Most of the Time, Always)
2. How often can you trust government inWashington to do what is right? (Never, Some of the Time, Most of the Time, Always)
3. Public officials don’t care much what people like me think. (Disagree Strongly, Disagree Somewhat, Nei- ther Agree nor Disagree, Agree Somewhat, Agree Strongly)
4. Sometimes, politics and government seem so com- plicated that a person like me can’t really understand what’s going on. (Disagree Strongly, Disagree Some- what, Neither Agree nor Disagree, Agree Somewhat, Agree Strongly)
These original survey data are supplemented with contex- tual data provided by ICE. We collected county-level data on enforcement efforts conducted as part of Secure Communities. Documentation received from a FOIA request originally filed by the National Day Labor Organization yielded county-level data on the number of individuals arrested and deported through the Secure Communities program. These data also contain information on the number of deportees who had criminal convictions prior to being detained by ICE. General data on deportations have never been reported at any level below the Area of Responsibility (AOR). Most AOR juris- dictions cover multiple states, and the release of county-level data about Secure Communities enforcement offers a unique opportunity, although the data do not include all immigration- related arrests and removals.
The pool of potential deportees varies across Texas; how- ever, there are no county-level estimates of the size of the unauthorized immigrant population. We must therefore ad- just enforcement figures by accounting for the total number of foreign-born individuals within each county. The result is a county-level measure of the deportation rate, which we spec- ify as the number of removals per 1,000 foreign-born res- idents. The number of removals per 1,000 foreign-born residents varies from 0 to 15, with a mean of 4.8. When in- cluding only noncriminal deportees, the rate varies from 0 to 6 with a mean of 1.4. Criminal removals range from 0 to 12, with a mean of 3.4.
No Secure Communities removals originated from Blanco County, which is west of Austin and 15% Latino. By contrast, there have beenmore than 12,000 instances of enforcement, or 17 per 1,000 foreign-born residents, in Harris County (Hous- ton). The figure for Harris County does not offer a complete
904 / Immigration Enforcement and Trust Rene R. Rocha, Benjamin R. Knoll, and Robert D. Wrinkle
portrait of enforcement patterns because 90%of removals have been directed toward individuals charged with criminal vio- lations. The noncriminal removal rate in Harris County is comparatively low: 1.8 per 1,000. The border county of Webb (Laredo) has the highest noncriminal rate at 5.9. However, other counties away from the border also have high non- criminal removal rates. For example, Collin County (north- west of Dallas) has a rate of 4.2 andTravis County (Austin) has a rate of 3.9.
We believe that the effect of deportation rates will vary by ethnicity and nativity. A pair of interaction terms (native-born Latino # removal rate; foreign-born Latino # removal rate) tests the argument. Our models control for individual-level characteristics known to shape trust in government or efficacy, including years of schooling, sex, and partisan affiliation (Binder et al. 1997). Other contextual variables may affect at- titudes toward government such as political empowerment (Bobo and Gilliam 1990; Bowen and Clark 2014; Pantoja and Segura 2003) and in-group size (Leighley 2001; Rocha and Espino 2010; Welch et al. 2001). Accordingly, we account for whether the respondent is represented by a Latino member of Congress and the percentage of Latinos in the respondent’s county.4 Our models also include an indicator for whether the respondent lives in a county along the United States– Mexico border.
The effect of immigration enforcement likely depends on who is removed. Most of the immigrants deported because of Secure Communities have been charged with a crime. But some have not. Noncriminal removals should have a greater effect on political orientations than do criminal removals. When local and federal officers partner to deport unauthorized immigrants who have never committed a criminal offense, Latinos may be less likely to believe that government “can be trusted to do what is right.” The removal of immigrants who have been convicted of a criminal offense in the United States might be perceived as a legitimate government action and be less likely to make Latinos feel cynical.5 Criminals and non- criminals also have distinct social networks (Carrington
2011). Information about the deportation of noncriminals is therefore transmitted to other noncriminal immigrants more readily than news about the deportation of criminals. Anglos may be apt to view all unauthorized immigrants negatively and react favorably to programs targeting non- criminal immigrants rather than solely criminals.
Do individuals accurately perceive removal rates, and are their perceptions especially sensitive to noncriminal re- movals? Our analysis of the 2011 National Survey of Latinos (NSL) conducted by the Pew Hispanic Center addresses these questions. This survey asked respondents, “Do you personally know someone who has been deported or de- tained by the federal government for immigration reasons in the last twelve months?”A quarter of respondents claimed to have known someone who was detained or deported. We conducted a logit analysis using responses to this item as our dependent variable (1p yes, 0p no). Latinos in areas where the Secure Communities removal rate is high are more likely to report knowing someone who has been detained or de- ported (p p .005). The effect holds when the criminal and noncriminal removal rates are analyzed separately (pp .009 for the criminal removal rate, p p .001 for the noncriminal removal rate). However, the substantive effect of noncrimi- nal removals far exceeds that of criminal removals. Figure 1 illustrates this point. The probability that a respondent knows someone who has been deported is slightly more than .2 when the noncriminal removal rate is 1 per 1,000. The probability doubles when the noncriminal removal rate is 5 per 1,000. Because this survey contains only Latino re- spondents, we cannot use it to draw inferences regarding the effect of Secure Communities on perceptions of enforcement among Anglos.6
ANALYSIS Anglos and Latinos, previous work finds, view government differently (Abrajano and Alvarez 2010). Latinos are more externally efficacious and trusting of both the federal and the local government (Wenzel 2006). Among Latinos, immi- grants express the highest amount of trust (Michelson 2007). Does this pattern hold across all policy environments?
4. We also operationalized political empowerment as whether the respondent lived in a municipality with a Latino mayor, which we felt would more accurately predict trust in local government. This measure of empowerment did not produce results that differ from those presented in tables 1 and 2. We substituted our measure of demographic context, % Latino, with % foreign-born. Results were similarly identical.
5. A recent statement by Janet Murguia, president and CEO of La Raza, illustrates this point. Murguia stated: “For, I think, many in our community, there’s a sense that if you don’t have a criminal record and you’ve been here contributing, then you shouldn’t be selectively deported. On the other hand, if there has been a criminal record established by these individuals, I think people understand the laws have to be in place to protect the interests of the U.S.” (Block 2014).
6. The 2011 NSL also asked: “Compared to the George W. Bush ad- ministration, would you say that the Obama administration has deported more, about the same, or fewer immigrants?” We view this question as tapping a general perception about whether deportations have increased since the Secure Communities program began. Of the respondents, 44% said there have been “more” deportations under the Obama administra- tion, 33% responded “same,” and 9% responded “fewer.”We conducted an ordered logit analysis, with this item as the dependent variable (0p fewer, 1 p about the same, 2 p more). The criminal removal rate does not predict this subjective measure of changes in enforcement (p p .12). However, the noncriminal deportation rate was positively associated with this measure (p p .035).
Volume 77 Number 4 October 2015 / 905
Table 1 presents a series of models that predict trust in local government.7 The dependent variables are measured on an ordinal scale and analyzed with ordered logit models. For each model, standard errors are clustered by county.8
Many of the individual-level variables fail to predict feelings toward local government, with sex and Democratic parti- sanship being exceptions. Latino political empowerment and overall group size also are unrelated to trust.
Table 1 shows that immigration enforcement partially determines trust in government. Figure 2 shows the condi- tional marginal effect of foreign-born status on the proba- bility of a respondent declaring that local government can never be trusted to do what is right.9 Anglos are the refer- ence group. We see that Latino immigrants are generally more trusting than Anglos, as foreign-born status reduces the probability of responding “never” to this item. The ef- fect is large in some places. The reduction is more than 20% in areas with little enforcement. However, the attitudes of
Figure 1. Data come from the 2011 National Survey of Latinos conducted by the Pew Hispanic Center. Each plot shows the predicted probability of an interviewee
responding “Yes” to the item “Do you personally know someone who has been deported or detained by the federal government for immigration reasons in the last
twelvemonths?” The probability is calculated for various Secure Communities noncriminal and criminal removal rates. The bands represent 95% confidence intervals.
Both relationships are statistically significant (p ! .05).
7. A joint F-test of the interactive terms allows us to determine whether the interactions are jointly significant. The resulting F-statistic is significant at the .05 level for all models except when predicting internal efficacy.
8. We replicated our models using HLM 7. The variance components in each of our models indicate that multilevel modeling offers little le- verage. HGLM models produce substantively similar results.
9. This and subsequent figures were generated using the observed value approach recommended by Hanmer and Kalkan (2013).
foreign-born Latinos are less distinct from Anglos in areas with high rates of Secure Communities enforcement, and their attitudes are not statistically distinguishable from An- glos in areas where the removal rate is 8 per 1,000 foreign- born residents (the mean is 4.8). Indeed, the parameter es- timates—although not statistically significant—suggest that immigrants are more likely than Anglos to believe that local government can never be trusted in areas where the removal rate is 11 or higher.
Further analysis (available in the online appendix) shows why the effect of foreign-born status varies with enforcement. The probability of an Anglo respondent believing that local government never can be trusted to do what is right is rel- atively high (.28) if she or he lives in an area where no un- authorized immigrants are being deported through Secure Communities (0 removals per 1,000 foreign-born residents). An Anglo residing in a high-enforcement area (15 removals per 1,000 foreign-born residents) is unlikely to hold this belief (predicated probability p .08). The opposite is true for foreign-born Latinos. Foreign-born Latinos residing in areas where enforcement is low are less likely to hold cynical at- titudes toward local government than those who live in high- enforcement contexts.
The conditional marginal effects plotted in figure 3 reveal that native-born Latinos are also less likely than Anglos to believe that local government can never be trusted if the
906 / Immigration Enforcement and Trust Rene R. Rocha, Benjamin R. Knoll, and Robert D. Wrinkle
immigrant removal rate is low. Native-born Latinos and An- glos are similarly distrusting of local government in places where the removal rate is above 5, but native-born Latinos are more distrusting if more than 11 removals occur per 1,000 foreign-born residents. Figures available in the appendix clarify why the marginal effect changes. Native-born Lati- nos living in counties with high enforcement rates likewise
are less trusting than those living where enforcement is rare. The change in probability among native-born Latinos is smaller than what is predicted for foreign-born Latinos, but the difference between native- and foreign-born Latinos is not statistically significant.
These results suggest two important conclusions: (i) im- migration enforcement affects all groups, not just those di-
Table 1. Immigration Enforcement and Political Trust
Local Government Federal Government
Coefficient SE Coefficient SE
Removal rate .133** .044 .170** .047 Native-born Latino 1.494** .383 1.166 .799 Foreign-born Latino 2.990** 1.012 3.464** .780 Native-born Latino # removal rate 2.209** .047 2.156** .057 Foreign-born Latino # removal rate 2.272* .118 2.359** .090 Education 2.043 .049 2.053 .049 Female .796** .211 .858** .251 Democrat .621 .330 1.122** .297 Independent .117 .351 .074 .285 Border county .080 .662 .889 .640 % Latino in county 2.002 .016 2.013 .014 Latino empowerment 2.417 .683 2.172 .577 Cutpoint 1 2.521 2.609 Cutpoint 2 1.644 3.075 Cutpoint 3 3.504 4.869 Pseudo R2 .138 .141
Noncriminal Removals
Removal rate .778** .217 .792** .202 Native-born Latino 1.920** .454 1.247 .886 Foreign-born Latino 3.102** .993 3.136** .772 Native-born Latino # removal rate 21.192** .294 2.763* .335 Foreign-born Latino # removal rate 21.384* .603 21.471** .369
Criminal Removals
Removal rate .126* .057 .174** .063 Native-born Latino 1.094** .373 .816 .711 Foreign-born Latino 2.493** .953 2.917** .745 Native-born Latino # removal rate 2.206** .054 2.153* .061 Foreign-born Latino # removal rate 2.273* .136 2.376** .117
Note. N p 470. Table entries are ordered logit estimates. All standard errors (SEs) are clustered by county. Dependent variable question wording is: “How often can you trust government in Washington to do what is right?” (1–4: Never, Some of the Time, Most of the Time, Always) and “How often can you trust local government to do what is right?” (1–4: Never, Some of the Time,
Most of the Time, Always). * p ! .05. ** p ! .01.
Volume 77 Number 4 October 2015 / 907
rectly connected to immigrant populations, and (ii) these effects are redistributive in that they shift valued psycho- logical resources away from already-disadvantaged groups to already-advantaged groups. The result is a feedback effect that loudens the voice of Anglos and mutes the voice of
Latinos. As Lowi (1964) famously remarked: “Policy deter- mines politics.”
Table 1 also shows that the results are consistent when noncriminal and criminal removal rates are examined sep- arately. Figure 4 illustrates this. Foreign-born Latinos do not
Figure 2. Data come from ICE and a 2012 survey of Anglos and Latinos residing in Texas. The figures show the conditional marginal effects of the variable
“Foreign-born Latino” on the probability of (i) responding “Never” to the item “How often can you trust local government to do what is right?” and (ii) re-
sponding “Agree Strongly” to the item “Public officials don’t care what people like me think.” The bands represent 95% confidence intervals.
Figure 3. Data come from ICE and a 2012 survey of Anglos and Latinos residing in Texas. The figures show the conditional marginal effects of the variable
“Native-born Latino” on the probability of (i) responding “Never” to the item “How often can you trust local government to do what is right?” and
(ii) responding “Agree Strongly” to the item “Public officials don’t care what people like me think.” The bands represent 95% confidence intervals.
908 / Immigration Enforcement and Trust Rene R. Rocha, Benjamin R. Knoll, and Robert D. Wrinkle
express more positive feelings toward local government than Anglos do in counties where the noncriminal removal rate is 2 per 1,000 or higher. Criminal removals have a much smaller moderating effect. Foreign-born Latinos are signifi- cantly more trusting of local government than Anglos are until the criminal removal rate is 6 per 1,000.
Next, we replicate our analysis on our second dependent variable: trust in the federal government. Secure Commu- nities represents a series of local–federal partnerships, and we are not surprised to see that the results are largely con- sistent. Anglo respondents are approximately 25% more likely than Latino immigrants to believe government in Washington never can be trusted to do what is right in areas where the deportation rate is 0. The two groups are statis- tically distinguishable until the removal rate reaches 8 per 1,000 foreign-born residents. Foreign-born Latinos are ac- tually estimated to be more cynical than Anglos are once the removal rate hits 11 per 1,000, although this last difference is not statistically significant. The relationships hold for noncriminal and criminal deportation rates. As was true when examining trust in local government, the noncriminal removal rate has a larger substantive impact. This pattern again occurs because foreign-born Latinos are least trusting in counties where enforcement is high, whereas Anglos are most trusting in those same locations. Native-born Latinos
express the same level of trust in the federal government in high and low enforcement areas.10
Table 2 considers the effect of immigration enforcement on levels of external and internal efficacy. The results for ex- ternal efficacy are consonant with those presented in our analysis of political trust. In some policy environments, Latino immigrants are more externally efficacious than Anglos. In others, they are not. Latino immigrants are 11% less likely than Anglos to “agree strongly” with the assertion that “public officials don’t care what people like me” think when the re- moval rate is 0. The difference between the two groups is no longer significant once the removal rate reaches 7 per 1,000 foreign-born residents. As with our analysis of political trust, the substantive effect of immigration enforcement is much greater when noncriminal removals are isolated. Foreign-born Latinos are nearly 29% less likely than Anglos to offer the least externally efficacious response when they live in a county where there have been no noncriminal removals. They are a statistically insignificant 3% less likely to do the same when the noncriminal removal rate is 5. This 26-point difference com- pares to 3 points–-from 16% to 13%–-when examining crim- inal removals over this same span. The shifting difference in external efficacy between Anglos and foreign-born Latinos
Figure 4. Data come from ICE and a 2012 survey of Anglos and Latinos residing in Texas. The figures show the conditional marginal effects of the variable
“Foreign-born Latino” on the probability of responding “Never” to the item “How often can you trust local government to do what is right?” The marginal
effect of “Foreign-born Latino” is modeled as being conditional on the noncriminal removal rate (left) and the criminal removal rate (right). The bands
represent 95% confidence intervals.
10. These shifts are fully illustrated in the appendix.
Volume 77 Number 4 October 2015 / 909
occurs because Anglos are most efficacious in high enforce- ment areas, whereas foreign-born Latinos are least effica- cious in those same places. When compared to foreign-born Latinos, the effects of criminal and noncriminal removals among native-born Latinos are modest.
Our focus on a single state does limit the inferences we can draw from our analysis. As a check, we examine the effect of enforcement on efficacy among a national sample of respondents gathered by the 2012 ANES. However, we lack removal data for every county in the nation and rely on
Table 2. Immigration Enforcement and Political Efficacy
External Efficacy Internal Efficacy
Coefficient SE Coefficient SE
Removal rate .092* .040 .022 .030 Native-born Latino .483 .406 2.257 .339 Foreign-born Latino 2.173** .768 .203 .496 Native-born Latino # removal rate 2.119** .040 .039 .031 Foreign-born Latino # removal rate 2.210* .091 .056 .079 Education 2.041 .050 .014 .051 Female .687** .166 .467** .161 Democrat .654 .344 .119 .197 Independent .205 .287 .203 .151 Border county .239 .587 .609 .514 % Latino in county 2.581 1.538 21.994 1.029 Latino empowerment 2.145 .564 .108 .388 Cutpoint 1 21.639 21.148 Cutpoint 2 21.004 .423 Cutpoint 3 1.088 1.470 Cutpoint 4 2.635 3.512 Pseudo R2 .155 .017
