G Model SOCSCI-1612; No. of Pages 10
ARTICLE IN PRESS The Social Science Journal xxx (2019) xxx–xxx
Contents lists available at ScienceDirect
The Social Science Journal journal homepage: www.elsevier.com/locate/soscij
A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty Hovannes Abramyan a,∗ , Gerard Alexander b a b
Reason Foundation, 5737 Mesmer Avenue, Los Angeles, CA 90230, USA University of Virginia, S285 Gibson Hall, 1540 Jefferson Park Avenue, Charlottesville, VA 22904, USA
a r t i c l e
i n f o
Article history: Received 30 October 2018 Received in revised form 1 May 2019 Accepted 2 May 2019 Available online xxx Keywords: Immigration Amnesty Public opinion Prejudice Language Social conservatism
a b s t r a c t In recent years, strong public opposition to amnesty has prevented the passage of any new immigration laws that include a “pathway to citizenship” for the millions of undocumented immigrants who reside in the United States. Much scholarship on the origins of white opposition to immigration has focused on ethnocentrism. However, it has also tended to conflate racial prejudice with potentially non-racial social concerns – even though the latter could, in theory, be ameliorated through policy design in ways that racism could not. Using experiments embedded within a large national survey, we disentangle the effects of both ethnic prejudice and social concerns on public opinion. Our results indicate that significant social concerns exist independent of racial considerations, and attitudes are quite responsive to policy designed to alleviate them. Further, the extent of this responsiveness depends on partisanship. Once these considerations are addressed through policy design, we find measurably wider support for a pathway to citizenship than would be predicted were racial prejudice the underlying motivator of these social concerns. © 2019 Western Social Science Association. Published by Elsevier Inc. All rights reserved.
1. Introduction Americans have long debated how to address the millions of undocumented immigrants residing in the United States. Recent debates have discussed “comprehensive reform” – an approach that combines enhanced border security with the granting of legal status to most undocumented immigrants residing in the country. Despite some bipartisan legislative support and the endorsement of past Republican and Democratic presidents, comprehensive immigration reform has faced difficulty and immigration has remained a polarizing issue. Granting a “pathway to citizenship” is politically unpopular, and public attitudes show clear partisan differences.
∗ Corresponding author. E-mail addresses:
[email protected] (H. Abramyan),
[email protected] (G. Alexander).
In a 2014 poll, 38 percent of Americans said they would be less likely to vote for a candidate who supported a “path to citizenship for illegal immigrants” (“National US Poll,” 2014). This public opposition – which included 52% of Republicans, 41% of Independents, and 22% of Democrats – pressured Republican Congressional leaders to stall a bill under consideration at the time (Walsh & Bohn, 2014). And more recently, political observers have credited public opposition to immigration for Donald Trump’s Republican nomination, and eventually ascension to the White House (Enten & Bacon, 2017). We examine public attitudes on amnesty, specifically by disentangling the effects of ethnic prejudice and social concerns like the common use of the English language. Prior research has generally conflated the two concepts into a single explanatory factor that drives immigration opposition, and compared it with more tangible economic
https://doi.org/10.1016/j.soscij.2019.05.002 0362-3319/© 2019 Western Social Science Association. Published by Elsevier Inc. All rights reserved.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
2
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
concerns. In contrast, our analyses dissect the premise that all social concerns are rooted in ethnic prejudice. The current study relies on nationally-representative cross-sectional survey data of the American adult population, as well as original experiments. We find that ethnic prejudice and social concerns both motivate opposition to amnesty, but they do not interact in the manner that other scholars have presumed. We argue that the distinction in the role of each factor is an important one; unlike ethnic prejudice – which presents a formidable barrier to attitude change – non-racial social concerns can be deactivated through policy design, broadening public support for amnesty. 2. Previous approaches to studying immigration attitudes
ethnic prejudice. Huddy and Sears (1995), who investigate “linguistic threat centered on fears about the spread” of non-native languages (p. 135), provide an exception but only by categorizing social concerns as non-prejudicial if they mirror economic worries. Thus, the scholarship on non-economic drivers of immigration attitudes has generally understood broad social concerns to be rooted in fear that America is becoming a less “Anglo” nation. If this presumption is accurate, highlyintertwined racial and social concerns make it unlikely that the public mood might turn in the direction of support for amnesty, regardless of policy design. However, social concerns that are not anchored by ethnic prejudice – to the extent these exist – could reveal opportunities for attitude change through more palatable policy design. 3. Concern with social cohesion
Research on the drivers of immigration attitudes tends to juxtapose the explanatory power of economic concerns and non-economic factors. Studies analyzing economic concerns generally find that their impact on immigration attitudes lessens when the economy is doing better. These studies look at general economic performance indicators, as well as things like native labor market worries and perceptions of fiscal pressures caused by an increase in immigration (Brucker et al., 2002; Burns & Gimpel, 2000; Citrin, Green, Muste, & Wong, 1997; Dustmann & Preston, 2007; Hainmueller & Hiscox, 2010; Scheve & Slaughter, 2001). Several find that sociotropic concerns are more closely linked to immigration attitudes than the current state of the national economy (Citrin et al., 1997; Espenshade & Hempstead, 1996; Hainmueller & Hiscox, 2010). Among those studies looking at non-economic influences on immigration attitudes, several investigate security-related concerns. They commonly find that anxieties regarding an increase in crime predict negative feelings about immigration, both in the United States and Western Europe, though to varying degrees (Fitzgerald, Curtis, & Corliss, 2012; Mayda, 2006; McLaren & Johnson, 2007; Sniderman, Hagendoorn, & Prior, 2004). Homola and Tavits (2017) detect mixed effects of intergroup contact on perceptions of immigrant threat regarding violence and vandalism. And Lahav and Courtemanche (2012) identify physical security threat as provoking inter-party agreement on the need for restrictions on immigration. Research on non-economic concerns, however, has more often focused on the role of ethnocentrism and racial prejudice – often by combining these concepts with possibly distinct sociocultural attitudes. For example, Chandler and Tsai (2001) consider cultural concerns to be a form of “ethnocentrism” (p. 186). Dustmann and Preston (2007) reduce non-economic concerns into “cultural and racial concerns” (p. 26) associated with “hostility. . . toward immigration groups with largely different cultural and ethnic backgrounds” (p. 2–3) and “racial and cultural prejudice” (p. 3). And Espenshade and Hempstead (1996) investigate perceptions of “negative cultural or personal traits of immigrants” (p. 540). Thus, to the extent that they are studied, these broad social concerns are commonly treated as manifestations of racial or
However pervasive ethnic or racial prejudice is, some Americans may have non-economic social concerns about immigration that originate elsewhere and are responsive to policy design. Arguments from some immigration skeptics are easily interpreted as racially prejudiced, of course (e.g., Buchanan, 2006; Brimelow, 1995). For them, it is unclear that any policy specification could influence their opposition to amnesty for immigrants of certain disfavored races or ethnicities. But other political thinkers and activists have voiced concerns in non-racial terms that would imply susceptibility to positive attitude change on immigration. Notable figures have articulated a consequentialist concern with the importance of a shared core of social norms, and the danger posed by societal fragmentation for civic peace and national cohesion. For example, Schlesinger (1991) argued that even with U.S. social norms subject to “constant revision,” a shared commitment to key political ideals and a shared language are “necessary bond[s] of national cohesion in so heterogeneous a nation as America” (p. 109). Similarly, others have advocated a return to the previous “melting pot” model of assimilation, that presupposes new waves of immigrants gradually alter the corpus of shared social norms (Fonte & Nagai, 2013; Hanson, 2003; Ravitch, 1990; Vigdor, 2009; Wilkinson, 1997). Among those who express such views, there might be support for policies that are perceived to promote effective assimilation. Thus, their resistance to expanded immigration could be explained by fear regarding the loss of essential societal cohesion, and not necessarily by fears regarding the racial or ethnic origins of immigrants or the nation’s resulting racial or ethnic balance. The important point is not the accuracy of these fears regarding social cohesion (see Citrin, Lerman, Murakami, & Pearson, 2007, and Portes & Rumbaut, 2001, for critiques), but whether they are pervasive and genuinely non-racial among members of the American public. We examine this possibility – that such sociocultural concerns provide a distinct and appreciably different motivator of immigration attitudes – and its implications for attitude change. In addition, we analyze the impact and nature of these concerns among members of different partisan groups. Partisanship is likely an important component in understanding how social concerns relate to amnesty opposition
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
specifically and immigration attitudes in general. There is evidence that some segment of the American mass public adopts their positions based on elite partisan cues on the issue of immigration (Druckman, Peterson, & Slothuus, 2013). However, as Slothuus (2010) demonstrates, people also rely on their pre-existing values in the face of partisan cues. And Bullock (2011) finds that partisans “are generally affected at least as much – and sometimes much more – by exposure to substantial amounts of policy information” (p. 512). Thus, not only do we expect differences in values and social concerns displayed by members of different partisan camps to translate into a gap in support for amnesty – particularly between Republicans and Democrats – we expect partisan differences to attenuate when the policy is framed in a manner that addresses those concerns. 4. Methods Our examination of immigration attitudes uses original data collected as part of a broad national survey. It includes interviews from 5,250 participants, and is demographically representative of the adult American population (see Table A1 in the Appendix A). In addition to answering questions related to the current study, respondents also answered a number of questions regarding other policy issues or social attitudes. The survey was administered online by YouGov. Respondents opted in to participate in the YouGov panel, but were selected for participation through a process that mimics a random digit dial (RDD) sample (Vavreck & Rivers, 2008). Rigorous examination demonstrates that this methodology produces highly reliable and representative samples (Kennedy et al., 2016; Rivers, 2016). 4.1. Experimental manipulation Respondents were randomly assigned one of six survey questions measuring support for a “pathway to citizenship.” We utilized a two-by-three design, in which framing about the origin of the immigrants and policy details were manipulated. The baseline conditions read: “In recent years, the number of undocumented immigrants from certain parts of the world—notably [Latin America/Eastern Europe]—has increased. Some lawmakers have proposed providing a ‘pathway to citizenship’ for undocumented immigrants who have been in the country for many years to address this issue. Would you support such a policy to address recent illegal immigration from [Latin America/Eastern Europe]?”1 The second set of questions replace the final sentence with: “Would you support such a policy to address recent illegal immigration from [Latin America/Eastern Europe] if
1
All experimental conditions include both the phrases “undocumented immigrants” and “illegal immigration” as not to sway opinion in a particular direction. See Merolla, Ramakrishnan, and Haynes, (2013) on the effects of frames using different terms.
3
it is coupled with a requirement that all government business has to be conducted in English only and all public school classes taught in English only?” And our third set of questions replace the last sentence in the baseline with: “Would you support such a policy to address recent illegal immigration from [Latin America/Eastern Europe] if it is coupled with a requirement that immigrants take naturalization and American history courses, be pressured to adopt American customs, have all government business conducted in English only, and all public school classes taught in English only?” The intent was to measure public support for policies that varied in the strength of their requirements for amnesty. Manipulation checks confirmed respondents understood the policies as varying in the stringency of their requirements. Our frames also varied the beneficiary of amnesty. Latinos are 51% of the foreign born in the United States (Radford & Budiman, 2018), making them the largest American immigrant ethnic group. It is reasonable to expect that opposition to amnesty, for some people, reflects racial prejudice regarding the largest immigrant group. It is also quite possible that this prejudice anchors some social concerns, as other scholars expect. To test this, our experiment measures attitudes on extending amnesty to migrants from Latin America, as well as a racially white immigrant group (Eastern Europeans). The choice to reference Eastern European immigrants in the alternate frame was driven by our specific research objectives. It allows us to vary the race of the beneficiaries of amnesty across conditions, while keeping constant other important qualities that characterize immigrants from both regions – this includes a non-English native language, as well as a largely non-Protestant religion. We could have used a frame referencing Western European migrants, however, we suspect that doing so would have evoked the image of racially-white migrants who are more culturally similar to the American cultural mainstream than are Latin American migrants. Thus, we decided a frame referencing Eastern Europeans was preferable. Recent U.S. Census data show Eastern Europeans comprise 40% of all immigrants from Europe, or about 4.4% of total foreign born population in the United States (Alperin & Batalova, 2018). Each respondent received only one question about amnesty policy, mentioning either Latin Americans or Eastern Europeans, and were presented only one policy frame. Thus, we provoke no artificial evenhandedness in responses; aggregate differences across conditions reveal the effects of ethnic prejudice, social concerns, and any possible interaction of the two on support for amnesty.2 We intentionally avoided creating a “pure control” condition that might hope to eliminate the influence of either
2 Our design has some similarity to that of Dustmann and Preston (2007). Although their design could, in principle, allow for racial and social features to vary independently, the specific regional categories they chose combine “cultural and ethnic distance,” rather than isolating them from one another.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
4
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
social values or ethnic prejudice. Not only would such a condition prevent us from detecting any target-dependent social concerns, but it would be unrealistic to expect that individual-level public policy preferences could be made irrelevant when evaluating an actual policy. With nearly 900 respondents assigned to each condition, our estimates are precise enough to detect substantively significant differences in support across them, further reducing our need for any “pure control” group. Our design differs from other experiments that use vignettes (e.g., Harell, Soroka, Iyengar, & Valentino, 2012). Though vignettes offer the ability to examine the influence of personal characteristics on acceptance of individual immigrants, deliberations of policies that target large swaths of the population generally do not focus on single individuals. Also, vignettes may elicit person-positivity bias – a greater inclination toward warm feelings and acceptance of the individuals featured that is not transferable to the larger group to which they belong (Iyengar et al., 2013). Therefore, we framed our questions in terms more characteristic of the policy and public debate. 4.2. Measurement of ethnic prejudice and social values We evaluate the impact of ethnic prejudice using evaluations of racial stereotypes. Respondents were presented with two 7-point scales. The first had as its endpoints the terms “intelligent” and “unintelligent”; the second had endpoints labeled “hardworking” and “lazy.” Respondents were separately asked to place whites, blacks, Asians, and Latinos on each scale. Composite scores were created by averaging the two scores given to each group, coded in the direction of negative evaluations. These were then used to create prejudice scores. Anti-Latino prejudice scores were created by subtracting the average of scores assigned to non-Latinos from composite scores given to Latinos. In this way, higher scores represent greater prejudice against Latinos. The survey items and the procedure for measuring overt racial prejudice roughly follow the process of Sears and Henry (2003). Our measures of social values focus on conservative individualism and moral traditionalism. Each are measured with composite scores, averages of responses on two questions that tap into a predisposition. Conservative individualism items asked respondents the degree to which they agree or disagree with the sentiment that we should worry less about equality, and whether government has gotten bigger because it is overly involved or because the problems faced by the country have become bigger. Traditional morality items asked the degree to which respondents agree or disagree that we should adjust morality to the times, and whether we should be tolerant of other people’s moral standards. Our respondents vary in their anti-Latino prejudice and social values scores, but they exhibit patterns when attitudes are aggregated by party. Democrats in our sample score relatively low in anti-Latino prejudice, and very low in conservative individualism and moral traditionalism. Republicans score noticeably higher than Democrats in anti-Latino prejudice, and much higher in conservative individualism and moral traditionalism. And Independents
Fig. 1. Intensity of aggregate policy support, by experimental condition.
average between Democrats and Republicans on all three measures. These primary sample characteristics are provided visually in the Appendix A (Fig. A1). 5. Results We first examine support for amnesty in each of the experimental conditions, an initial determination as to the responsiveness of attitudes to policy design and insight as to whether social concerns depend on the race of the immigrant group that would receive amnesty. Fig. 1, presents average policy support levels across the six experimental conditions. Responses were coded from 1 (“Strongly oppose”) to 4 (“Strongly favor”) and averaged within treatment conditions.3 Additionally, 95-percent confidence intervals are included to provide bounds for our estimates at a standard level of precision. We provide p-values when statistical significance is not obvious and exclude them when they are not needed. Two patterns emerge in Fig. 1. First, differences in attitudes within policy conditions are not generally statistically significant. When the stringency of amnesty requirements is held constant and region of origin is varied, in most cases, our estimates of support are near equal. The one exception is not substantively significant. The second pattern in Fig. 1 is the substantively and statistically significant jump in support for amnesty once a language requirement has been added to the policy frame. Regardless of the origin of the immigrant group, support increases above our scale mid-point (2.5) in this second condition. Similarly, support for the policy with the most stringent requirements is greater than support in our baseline. However, there is no statistically demonstrable difference in support when compared to our second policy condition. Thus, it appears social concerns are limited;
3 “Don’t know; no opinion” responses were omitted. For the Latin American conditions, these constituted 10–13% of responses. For the Eastern European conditions, these constituted 15–16% of responses.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model
ARTICLE IN PRESS
SOCSCI-1612; No. of Pages 10
H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
5
Table 1 Amnesty support in experimental conditions. Condition
Latin American frame
Eastern European frame
p-value (Chi-squared tests)
1. Baseline N 2. English p-value1/2 N 3. English+ p-value1/3 p-value2/3 N
47.8% 784 59.4% 0.00 718 63.5% 0.00 0.13 737
42.7% 759 62.7% 0.00 754 61.5% 0.00 0.66 714
0.05 0.21
0.47
Table 2 Experimental results by party affiliation. Latin American immigrants
Republicans Support Change from ‘Baseline’ Change from ‘English’ N Independents Support Change from ‘Baseline’ Change from ‘English’ N Democrats Support Change from ‘Baseline’ Change from ‘English’ N
Eastern European immigrants
Baseline
English
English+
Baseline
English
English+
27.2%
50.4% +23.2***
24.1%
60.9% +36.8***
183
176
61.9% +34.7*** +11.5* 192
191
183
59.6% +35.5*** −1.3n.s. 143
44.8%
59.2% +14.4**
38%
53.9% +15.9***
243
216
60.4% +15.6** +1.2n.s. 215
244
230
69.7%
65.3% −4.4n.s.
60.3%
71.9% +11.6**
286
269
67.7% −2n.s. +2.4n.s. 265
262
274
56.1% +18.1*** +2.2n.s. 213 69.1% +8.8* −2.8n.s. 255
Note: Statistical significance from a chi-squared test for equality of proportions; n.s. not significant. * p < 0.05. ** p < 0.01. *** p < 0.001.
requirements beyond use of English do not increase general public support for amnesty. Importantly, this responsiveness is not limited to conditions framed as benefiting immigrants from a particular region, which suggests they are not rooted in dislike for a particular group. Table 1 provides a clearer breakdown of support in each experimental condition by collapsing distinctions of attitude strength. It shows that policy support not only increases with the addition of an English language requirement, but it becomes the majority preference. Additional requirements produce only a negligible shift in support, both in terms of magnitude and substantive interpretation. Understanding the recipients of amnesty to be from Latin America rather than Eastern Europe (or vice versa), does not affect this social concern. 5.1. Support by party The Republican Party in particular has seen strong opposition among its ranks to comprehensive immigration reform. Thus, we examine whether social concerns are more or less limited to Republicans. Table 2 breaks down levels of support from each of three partisan groups across policy conditions. Although overall support increases with the addition of “English only” requirements, Table 2 shows that changes in
support are largest among Republicans. For both the Latin American and Eastern European baseline frames, Republicans are least likely to support amnesty, followed by Independents, and then Democrats. However, adding an “English only” requirement increases Republican support by 23.2 percentage points in the Latin American frame (2 = 19.4, p = 0.00), and 36.8 points in the Eastern European frame (2 = 50.5, p = 0.00). Under the most stringent conditions, Republican support increases an additional 11.5 points in the Latin American frame (2 = 4.4, p = 0.04) and was roughly equal to the level of support in the Eastern European frame. Table 2 shows that Independents also increase in support for amnesty with the addition of more stringent conditions. In both baseline conditions, Independents are likely to oppose amnesty for undocumented immigrants – and are less supportive when Eastern Europeans are to be the beneficiaries. Aggregate support increases by 14.4 points (2 = 8.9, p = 0.00) and 15.9 points (2 = 11.5, p = 0.00) for the two frames – making support for amnesty the majority position in the second condition. The addition of requirements beyond English do not make a significant difference in levels of support for either the Latin American (2 = 0.02, p = 0.88) or the Eastern European (2 = 0.13, p = 0.72) conditions.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
6
Of note are the results for Democratic respondents. They are most supportive of a policy that extends amnesty to Latin American immigrants without added conditions; however, support drops 4.4 points (2 = 1.03, p = 0.31) with the addition of English-only requirements. Under the strictest conditions, support drops 2 points (2 = 0.16, p = 0.69) from our baseline estimate. These differences are small and neither is statistically significant. The added conditions have no effect – or possibly even a slightly negative effect – on support for amnesty among Democrats when Latin Americans would benefit. In contrast, Democratic support in the Eastern European conditions increases 11.6 points (2 = 7.49, p = 0.01) in the “English only” condition. Including other conditions beyond English does not substantively affect support (69.1%; 2 = 0.38, p = 0.54). Also interesting is the lower Democratic support for amnesty regarding Eastern Europeans in the baseline condition (9.4 percentage points; 2 = 4.49, p = 0.03). The addition of an “English only” condition for amnesty increases support among Democrats to a level comparable to support in the Latin American baseline condition. One might suspect that this result is driven by the party’s relatively high share of Latinos. However, in the Eastern European condition, when Democratic support is broken down by race, we find that the difference in support between whites (60.3%) and Latinos (59.2%) is negligible. Black Democrats are slightly more likely to support the policy in the baseline condition (63.9%). To contrast, in the Latin American baseline condition, amnesty support from Democrats is 66.9% among whites, 81.3% among Latinos, and 69% among blacks. So, Latinos are surely responsible for pulling up average Democratic support for amnesty in the Latin American condition, but they are not uniquely responsible for the lower support when the benefiting migrants come from Eastern Europe. These findings show that public support for amnesty is very responsive to policy details, regardless of immigrant origin. Social concerns appear to have a large influence, particularly for Republicans. Democrats also show similar social concerns when the immigrants mentioned come from Eastern Europe, but not Latin America. Support for amnesty is greatest across the board (reaching majority levels among all groups) when the policy provides assurances that broad social concerns are addressed. 5.2. Correlates of policy attitudes We further validate our experimental findings by employing a series of logistic regressions on individuallevel data to model responses to the baseline questions. Responses to the question are collapsed into two categories, support and opposition, as done earlier. Polytomous explanatory variables in our models are standardized so that coefficients represent the effect of a 1-standard deviation increase. For dichotomous variables (i.e., each category of partisanship, male gender, white race, and Southerner) coefficients represent the effect of being a member of the group. Table 3 provides results from all of our models, which affirm our expectations. Models 1 and 3 (which exclude conservative social values) mirror our earlier finding of
partisan differences in support for amnesty when the policy in question does not address social concerns. Models 2 and 4, on the other hand, show that the partisan difference virtually disappears with the inclusion of conservative individualism and moral traditionalism. The exception is that Republicans remain less likely to support amnesty for Eastern European undocumented immigrants than nonRepublicans. Converting the coefficient to a proportional odds-ratio tells us that being a Republican corresponds with respondents being almost half as likely to support amnesty for Eastern European undocumented immigrants (e−0.67 = 0.51). The effects of social concerns, on the other hand, are large and robust. Converting the coefficients in column 2 into proportional odds-ratios reveals that a 1-standard deviation increase in either conservative individualism or moral traditionalism cuts the probability of supporting amnesty by slightly more than half (e−0.74 = 0.48) when framed as benefiting undocumented Latin Americans. When amnesty is framed as benefiting undocumented Eastern European immigrants, moral traditionalism has a similar effect on attitudes (e−0.76 = 0.47), and conservative individualism has a slightly lesser effect (e−0.51 = 0.60). Our models also incorporate a variable that allows us to evaluate the effects of anti-Latino prejudice. Earlier results suggest that social concerns are not rooted in the race of the immigrant group who would benefit from amnesty. However, prejudice against Latinos should still be associated with a lower probability of supporting amnesty for undocumented Latin American immigrants. In line with these expectations, we find that anti-Latino prejudice only affects attitudes for the policy framed as benefiting Latin American immigrants. Models 1 and 2 show that a 1-standard deviation increase in anti-Latino prejudice cuts the odds of supporting amnesty for Latin Americans substantially (by 2/5 in model 2, e−0.47 = 0.63). In contrast, anti-Latino prejudice has no statisticallysignificant effect in models 3 or 4, which focus on extending amnesty to Eastern European immigrants. Fig. 2 plots the marginal effects of anti-Latino prejudice on support for amnesty using our full models (2 and 4). The effect is notable, in the expected direction, and statistically significant only for the policy question that mentioned undocumented immigrants from Latin America. In contrast, the effect of anti-Latino prejudice is substantively and statistically insignificant, and also in the opposite direction, when the beneficiaries are Eastern European immigrants. Unlike social concerns, which relate consistently to attitudes regardless of the group mentioned in our amnesty question, anti-Latino prejudice relates differently to policy support depending on the target of amnesty. This provides further evidence that the social concerns we identify as responsive to policy are not themselves rooted in prejudicial racial attitudes. The potential for prejudice to affect support for amnesty for Latin American immigrants should not be understated, however. In fact, it is clearest when comparing predicting probabilities of support at the extreme ends of our antiLatino scale. For example, when we set prejudice at its scale minimum and hold all other model variables at their respective means, there is a 70.9% probability of support-
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
7
Table 3 Predictors of support for amnesty. Baseline condition: Latin American immigrants
Partisanship Democrat Republican Independent Demographics Age Male White Education Southerner Religiosity Predispositions Anti-Latino affect
(2)
(3)
(4)
1.21*** (0.21) −0.52* (0.27) 0.22 (0.25)
0.26 (0.24) −0.16 (0.29) −0.14 (0.27)
0.73*** (0.21) −1.05*** (0.28) −0.26 (0.25)
0.24 (0.23) −0.67* (0.3) −0.3 (0.27)
−0.15 (0.08) 0.18 (0.17) −0.52* (0.2) 0.36*** (0.09) −0.21 (0.18) −0.16 (0.09)
−0.1 (0.09) 0.31 (0.19) −0.16 (0.22) 0.3** (0.1) −0.27 (0.2) 0.2 (0.11)
−0.43*** (0.08) −0.15 (0.17) 0.002 (0.21) 0.35*** (0.08) −0.33 (0.2) 0.33*** (0.1)
−0.29** (0.09) −0.11 (0.19) 0.13 (0.22) 0.26** (0.09) −0.32 (0.21) 0.64*** (0.11)
−0.54*** (0.09)
−0.47*** (0.1) −0.74*** (0.12) −0.74*** (0.12)
−0.02 (0.09)
0.12 (0.1) −0.51*** (0.11) −0.76*** (0.12)
719 −368.45 756.91
719 −315.2 654.39
680 −367.35 754.71
680 −328.59 681.18
Conservative individualism Moral traditionalism Observations Log Likelihood A.I.C.
Eastern European immigrants
(1)
Note: * p < 0.05; ** p < 0.01; *** p < 0.001. Model intercepts are suppressed in order to avoid having to exclude any particular category of partisanship.
Fig. 2. Effect of anti-Latino prejudice on support for amnesty.
ing amnesty for Latin American immigrants. At the highest level of prejudice, there is a 27.3% probability of supporting amnesty. These probabilities are very similar to those found among Democrats and Republicans, respectively, in
our baseline experiment. Put differently, the gap in support between the most pro-Latino and the most anti-Latino respondents mirrors the partisan gap we discussed earlier. At the average level of anti-Latino prejudice (0.512),
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
8
the probability of support is 48.8%, roughly what we determined to be overall support in the baseline experimental condition. 6. Discussion We set out to determine whether social concerns affect public attitudes toward amnesty in a way that can be empirically distinguished from the effects of ethnocentrism or racial prejudice, given the tendency for researchers to conflate the factors. If not, it would signal strong public support is likely unachievable; policy modifications would ultimately fail to pacify underlying ethnic prejudices (specifically against Latinos). In contrast, if social concerns can be distinguished, then support for “comprehensive” immigration reform could be expanded by devising a policy more appealing to a broader public. Our findings show that support for amnesty increases by addressing certain social concerns, specifically the common use of English and especially among Republicans. While our results demonstrate that fears about America becoming less racially white are not the underlying driver of social concerns regarding amnesty, this finding should not be interpreted to mean that prejudicial racial attitudes play no part in policy attitude formation. The opposite is true. We find that people more negative toward Latinos are less likely to support amnesty for undocumented Latin American immigrants, but not racially white immigrants. The important point, though, is that the effect of this prejudice is not manifested in the social concerns we identified as particularly influential. These results are also important in validating the effectiveness of our experimental frames. We were sensitive to the possibility that any responses to questions regarding immigration might be associated with attitudes toward Latinos, even when our frames discussed migrants from Eastern Europe. However, the failure to find an association between anti-Latino prejudice and support for amnesty in the Eastern European frame strongly indicates this was not the case. Finally, our findings show that even Republicans would support an amnesty policy if it addresses certain important social concerns, like the common use of the English language. Even when made explicit that the policy would address “illegal immigration from Latin America,” a majority of Republicans (62%), Independents (60%), and Democrats (68%) were supportive in our study, despite the absence of assurances regarding things like border enforcement. This suggests that lawmakers looking to pass comprehensive immigration reform should focus on alleviating very specific social concerns from members of the public. Of course, it is possible that these concerns are affected to some degree by other conditions unexplored in the current study, such those more economic in nature. However, investigating this idea and other possibilities provides an avenue for future research. Conflict of interest None.
Acknowledgments This work was supported by the Donors Trust [grant number 2013041601]. Appendix A. Table A1 provides a demographic overview of our full sample, as well as partisan subgroups. It illustrates that Democrats comprise the largest partisan subgroup (41%) in our study, followed by Independents (32.5%) and Republicans (26.5%). Average age is 48.6 years, which does not vary significantly by party. Nearly half of our respondents are male (48.2%) – a slight minority in all partisan subgroups, except among Independents (56.2%). Nearly three-quarters of our sample is racially white. Among Republicans, the share of whites is 84.7%, with the next largest group (Latinos) at 9%. A strong majority of Independents are also white (81.1%), with Latinos comprising 7.5% of the partisan subsample. And, in contrast, 57.4% of Democrats in the sample are white. A quarter (25%) of the Democratic subsample are black and about an eighth (12.4%) are Latino. AsianAmericans make up about 1% or 2% of the total in each subgroup. Partisan subgroups are similar, on average, with regard to education and geographic residence, but differ noticeably in terms of religiosity. Table 1A presents the mean educational attainment of respondents on an unstandardized scale ranging from 0 (no high school diploma) to 1 (a post graduate degree). The total sample averages near the center of the scale (0.47), which corresponds to having some college education but falls short of a 4-year degree. Also, a little less than a third of respondents (30.9%) live in the South, defined by the 11 states of the Confederacy that seceded from the Union. The share of Southerners in our sample varies by a few points, depending on partisanship, but is generally in the same range. Religiosity, on the other hand, is much greater among Republicans than it is among Democrats and Independents. This unstandardized scale ranges from 0 to 1 and represents a composite of religious attendance and self-declared importance of religion in a person’s own life. Republicans average above the scale midpoint by 8 points, while Independents and Democrats score below it by 6 and 7 points, respectively. Fig. A1 presents unstandardized average scores on scales measuring anti-Latino prejudice, conservative individualism, and moral traditionalism. In addition to means and 95% confidence intervals for the full sample, the figure provides estimates for each partisan subgroup. Each plot in Fig. A1 also provides a dashed line at 0.5 to indicate each scale’s respective midpoint. Overall, Republicans average higher on all measures than Independents and, especially, Democrats. The partisan gap ranges from about 0.02 on anti-Latino prejudice, to 0.446 and 0.283 on conservative individualism and moral traditionalism, respectively. While all partisan differences are statistically significant at conventional levels of precision, they are most dramatic on the ideological measures. On both of these measures, Democrats average well below the scale midpoint while Republicans average high above it. Independents average closer to Republicans than they
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model
ARTICLE IN PRESS
SOCSCI-1612; No. of Pages 10
H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
9
Table A1 Sample summary statistics (Weighted).
Democrat (%) Republican (%) Independent (%) Age (years, avg.) Male (%) White (%) Latino (%) Black (%) Asian-American (%) Education (0-1, avg.) Southerner (%) Religiosity (0-1, avg.)
All respondents
Democrats
Republicans
Independents
41% 26.5% 32.5% 48.6 48.2% 72.1% 10.1% 12.2% 1.8% 0.47 30.9% 0.47
48.7 42.4% 57.4% 12.4% 25% 1.9% 0.47 30.5% 0.43
51.1 47.7% 84.7% 9% 2.5% 1% 0.47 32.6% 0.58
48.5 56.2% 81.1% 7.5% 5.1% 2% 0.49 31.3% 0.44
Fig. A1. Average anti-Latino prejudice, conservative individualism, and moral traditionalism scores.
do Democrats, but in all cases average between Republican and Democratic scores.
References Alperin, E., & Batalova, J. (2018). European immigrants in the United States. Washington, D.C: Migration Policy Institute. Retrieved from https://www.migrationpolicy.org/article/european-immigrantsunited-states Brimelow, P. (1995). Alien nation: Common sense about America’s immigration disaster. New York: Random House. Brucker, H., Epstein, G. S., McCormick, B., Saint-Paul, G., Venturini, A., & Zimmermann, K. (2002). European attitudes toward immigrants. In T. Boeri, G. Hanson, & B. McCormick (Eds.), Immigration policy and the welfare state (pp. 105–123). Oxford: Oxford University Press. Buchanan, P. J. (2006). State of emergency: The third world invasion and conquest of America. New York: St. Martin’s Griffin. Bullock, J. G. (2011). Elite influence on public opinion in an informed electorate. The American Political Science Review, 105(3), 496–515. Burns, P., & Gimpel, J. G. (2000). Economic insecurity, prejudicial stereotypes, and public opinion on immigration policy. Political Science Quarterly, 115(2), 201–225. Chandler, C. R., & Tsai, Y. (2001). Social factors influencing immigration attitudes: An analysis of data from the General Social Survey. The Social Science Journal, 38(2), 177–188. Citrin, J., Green, D. P., Muste, C., & Wong, C. (1997). Public opinion toward immigration reform: The role of economic motivations. The Journal of Politics, 59(3), 858–881. Citrin, J., Lerman, A., Murakami, M., & Pearson, K. (2007). Testing Huntington: Is Hispanic immigration a threat to American identity? Perspectives on Politics, 5(1), 31–48.
Druckman, J. N., Peterson, E., & Slothuus, R. (2013). How elite partisan polarization affects public opinion formation. The American Political Science Review, 107(1), 57–79. Dustmann, C., & Preston, I. P. (2007). Racial and economic factors in attitudes to immigration. The BE Journal of Economic Analysis & Policy, 7(1), 1–39. Enten, H., & Bacon, P. (2017). Trump’s hardline immigration stance got him to the White House. FiveThirtyEight (September 12). Retrieved from https://fivethirtyeight.com/features/why-polls-showing-daca-aspopular-even-among-republicans-dont-tell-the-whole-story/ Espenshade, T. J., & Hempstead, K. (1996). Contemporary American attitudes toward U.S. Immigration. The International Migration Review, 30(2), 535–570. Fitzgerald, J., Curtis, K. A., & Corliss, C. L. (2012). Anxious publics: Worries about crime and immigration. Comparative Political Studies, 45(4), 477–506. Fonte, J., & Nagai, A. (2013). America’s patriotic assimilation system is broken. Washington, D.C: Hudson Institute. Hainmueller, J., & Hiscox, M. J. (2010). Attitudes toward highly skilled and low skilled immigration: Evidence from a survey experiment. The American Political Science Review, 104(1), 61–84. Hanson, V. D. (2003). Mexifornia: A state of a becoming. San Francisco: Encounter Books. Harell, A., Soroka, S., Iyengar, S., & Valentino, N. (2012). The impact of economic and cultural cues on support for immigration in Canada and the United States. Canadian Journal of Political Science/Revue Canadienne de Science Politique, 45(3), 499–530. Homola, J., & Tavits, M. (2017). Contact reduces immigration-related fears for leftist but not rightist voters. Comparative Political Studies, 51(13), 1789–1820. Huddy, L., & Sears, D. O. (1995). Opposition to bilingual education: Prejudice or the defense of realistic interests? Social Psychology Quarterly, 58(2), 133–143.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002
G Model SOCSCI-1612; No. of Pages 10
10
ARTICLE IN PRESS H. Abramyan, G. Alexander / The Social Science Journal xxx (2019) xxx–xxx
Iyengar, S., Jackman, S., Messing, S., Valentino, N., Aalberg, T., Duch, R., . . . & Kobayashi, T. (2013). Do attitudes about immigration predict willingness to admit individual immigrants? A cross-national test of the Person-Positivity Bias. Public Opinion Quarterly, 77(3), 641–665. Kennedy, C., Mercer, A., Keeter, S., Hatley, N., McGeeny, K., & Gimenez, A. (2016). Evaluating online nonprobability surveys. Pew Research Center (May 2). Retrieved from http://www.pewresearch. org/2016/05/02/evaluating-online-nonprobability-surveys/ Lahav, G., & Courtemanche, M. (2012). The ideological effects of framing threat on immigration and civil liberties. Political Behavior, 34(3), 477–505. Mayda, A. M. (2006). Who is against immigration? A cross-country investigation of individual attitudes toward immigrants. The Review of Economics and Statistics, 88(3), 510–530. McLaren, L., & Johnson, M. (2007). Resources, group conflict and symbols: Explaining anti-immigration hostility in Britain. Political Studies, 55(4), 709–732. Merolla, J., Ramakrishnan, S. K., & Haynes, C. (2013). Illegal,” “undocumented,” or “unauthorized”: Equivalency frames, issue frames, and public opinion on immigration. Perspectives on Politics, 11(3), 789–807. National US Poll: January 8, 2014. (2014). Hamden, CT: Quinnipiac University Polling Institute. Retrieved from https://poll.qu.edu/ national/release-detail?ReleaseID=1993. Portes, A., & Rumbaut, R. G. (2001). Legacies: The story of the immigrant second generation. Berkeley, C.A: University of California Press. Radford, J., & Budiman, A. (2018). Facts on U.S. Immigrants, 2016. Pew Research Center Hispanic Trends (September 14). Retrieved from http://www.pewhispanic.org/2018/09/14/facts-on-u-s-immigrants/ Ravitch, D. (1990). Multiculturalism: E pluribus plures. The American Scholar, 59(3), 337–354.
Rivers, D. (2016). Pew Research: YouGov consistently outperforms competitors on accuracy. YouGov (May 17). Retreived from https://today.yougov.com/news/2016/05/13/pew-research-yougov/ Scheve, K. F., & Slaughter, M. J. (2001). Labor market competition and individual preferences over immigration policy. The Review of Economics and Statistics, 83(1), 133–145. Schlesinger, A. M. (1991). The disuniting of America: Reflections on a multicultural society. New York: W. W. Norton & Company. Sears, D. O., & Henry, P. J. (2003). The origins of symbolic racism. Journal of Personality and Social Psychology, 85(2), 259–275. Slothuus, R. (2010). When can political parties lead public opinion? Evidence from a natural experiment. Political Communication, 27(2), 158–177. Sniderman, P. M., Hagendoorn, L., & Prior, M. (2004). Predisposing factors and situational triggers: Exclusionary reactions to immigrant minorities. The American Political Science Review, 98(1), 35–49. Vavreck, L., & Rivers, D. (2008). The 2006 cooperative congressional election study. Journal of Elections, Public Opinion and Parties, 18(4), 355–366. Vigdor, J. L. (2009). From immigrants to Americans: The rise and fall of fitting in. Lanham, M.D: Rowman & Littlefield Publishers. Walsh, D., & Bohn, K. (2014). Pelosi insists immigration plan include path to citizenship; some House Republicans vow to oppose major reform. CNN Politics (January 29). Retrieved from http://politicalticker. blogs.cnn.com/2014/01/29/pelosi-insists-immigration-plan-includepath-to-citizenship-some-house-republicans-vow-to-opposemajor-reform/ Wilkinson, J. H. (1997). One nation indivisible: How ethnic separatism threatens America. Reading, M.A: Addison-Wesley Publishing.
Please cite this article in press as: Abramyan, H., & Alexander, G. A distinction with a difference: The influence of social concerns and racial prejudice on support for amnesty. The Social Science Journal (2019), https://doi.org/10.1016/j.soscij.2019.05.002