From Wedge Issue to Partisan Divide: the Development of Immigration Policy Opinion After 2016

The Forum 2019; 17(3): 467–486

James G. Gimpel* From Wedge Issue to Partisan Divide: The Development of Immigration Policy Opinion after 2016 https://doi.org/10.1515/for-2019-0024

Abstract: President Trump won an Electoral College majority in 2016 bolstered by voters who supported him, but not the previous nominee, Mitt Romney. Evidence suggests that a campaign promising a more restrictive immigration policy was the key to this improved performance among cross-pressured voters. In the months since inauguration day 2017, however, voters did not remain unaware of the administration’s programmatic steps on immigration and the opposition they encountered. I interpret evidence from a panel survey to suggest that voters gained knowledge about immigration policy after 2016, and began to align their policy views with the positions of their favored political parties. Inasmuch as voters’ policy positions become identical with their party preference, the potential for immigration policy to again act as a wedge issue in 2020 is greatly reduced. President Trump’s 2020 campaign may be able to mobilize more base voters given this increase in policy-party congruence but he may not be as successful as in 2016 in attracting crossover voters.

Introduction

The 2016 election divided the electorate not only by the usual factors that always shape political behavior: party identification, race, religious observance, the influence of neighbors, family and friends, but also by issues the major campaigns identified as important to a governing agenda if elected. Probably most prominent among these was President Trump’s focus on immigration policy (Gimpel 2017). Even after accounting for the conventional explanations for vote choice, whether a voter took a more restrictive or generous stand on immigration helped decide their vote. For weak and leaning partisans and independents whose views on immigration are not already determined by the strength of a party identity, consonance with the president’s thrust on immigration meant a vote for

*Corresponding author: James G. Gimpel, Department of Government, University of Maryland, 3140 Tydings Hall, College Park, Maryland 20742, USA, e-mail: [email protected] 468 James G. Gimpel him in November whereas disagreement pushed you in the opposite direction (Sides, Tesler, and Vavreck 2017). Policy positioning can sometimes be informative particularly if campaigns and candidates are consistent enough to hammer on a few basic themes. Here the question is whether immigration policy continued to inform voter decision making by 2018, well into the first term. It is never obvious that issues matter in elections, particularly given contemporary high levels of political side- taking. After all, if parties take opposing positions on divisive issues, and voters are strong party identifiers with the characteristic high levels of knowledge of campaign and party platforms, issues offer no new information. Issue talk in a campaign is just more hot air. Certainly when asked, highly informed partisans will report that issues matter. That does not mean that they do when voters are reliant on identity-anchored cues. For those voters who are not strong party identifiers, the case for issue voting has to overcome the readily demonstrable ignorance of the electorate. Many studies have pointed to the fact that voters are largely unable to identify the positions taken by candidates, and even their own views, much less make a match between the two (Berelson, Lazarsfeld, and McPhee 1954; Campbell et al. 1960; Sniderman, Brody, and Tetlock 1993; Bartels 1996; Carpini, and Keeter 1996; Kinder 1998; Lenz 2012). Other studies have shown a pervasive tendency for voters to misidentify the issues in a campaign, attributing the wrong positions to the rival candidates (Dalager 1996) and incorrect votes to their members of Congress (Dancey and Sheagley 2016). Clearly not all voting preferences are highly reasoned. Even if voters do not go through rigorously rational comparisons of candidate and party positioning, however, less informed and less polarized voters do gain knowledge from encountering political information from campaign and media sources. Weak party identifiers are less aware than strong partisans of the connection between issue positions and traditional party stands, putting a campaign in a position to offer them additional information that contributes to a decision. Exposure to advertising or attention to other forms of political communication seems to at least have a modest, albeit temporary effect on opinion. These studies clear away some room for messages about issues and policies to be more than just wasted breath. Voters apparently also inject their own personal concerns into a political campaign, mentioning issues they hold deep convictions about even when those have not been actively discussed in a particular contest (Dalager 1996, p. 504; Skitka and Bauman 2008). Issues may become salient to voters because they have heard about them in previous elections, or from any number of non-campaign sources. The issues of a presidential campaign can easily carry-over into thinking about a contemporaneous lower ballot election, and perhaps across short spans From Wedge Issue to Partisan Divide 469 of time into a mid-term election. Some high profile issues are just constantly in the news, or are constant national concerns for an enduring period of time, an indication that national media have influenced what voters consider to be important (Dalager 1996, p. 505–507). A group interest can be vaguely identified as underlying a particular identification but it may not be a contemporary, widely- discussed, interest (Berelson, Lazarsfeld, and McPhee 1954, p. 74). Perhaps policy considerations were important at some point in the past, as partisan identities initially formed, but the original events and circumstances that produced those alignments have largely faded from memory. These are all ways in which issues may play some role in shaping vote choice. Issue-oriented campaign outreach may be informative for voters when positions cut across traditional party lines, or when voters take positions that are incongruent with their generally preferred political party (Hillygus and Shields 2008; Jacobson 2015; Green 2019). This is the familiar notion of cross-pressure, in which a political party alignment exerts influence in one direction, while a preferred policy viewpoint pulls the voter in a different direction. Under these circumstances the ambivalence of voters can open them to persuasion by a party or candidate other than the one they might ordinarily support. Cross-pressured voters have been found to change their minds about candidates up until quite late in the campaign (Chaffee and Rimal 1996). Because of their ambivalence they are politically unstable, and often wind up splitting their tickets (Campbell et al. 1960; Mulligan 2011; Davis and Mason 2016). There is also some potential for cross-pressured voters to exhibit lower interest in elections and not vote at all (Campbell et al. 1960; Margolis 2018). This is an important reason why campaigns that raise cross-cutting issues may not benefit from crossover or split- ticket voting. If the level of mobilization drops then any measureable gains from persuasion may be off-set by a countervalent drop in turnout. In summary, voters do have strong party identities, but they also have inter- ests, collect new information and exercise rationality in sizing up their choices (Berelson, Lazarsfeld, and McPhee 1954, p. 322). The extent to which these influ- ences on decision shape individual political behavior is variable across citizens, and from election-to-election. Contests vary in the extent to which they evoke greater identity-based activism as opposed to cross-group and cross-party movement. Not surprisingly, campaigns also appear to shape the timing of vote decision (Henderson and Hillygus 2016). In midterm elections, we can expect voters to carry over some recall of issues raised in the previous presidential election. Throughout the first 2 years, immigration policy continued to receive a great deal of media attention as the new administration assumed office and made efforts to fulfill campaign promises on items such as building a southern border wall and reforming the nation’s approach to 470 James G. Gimpel illegal and legal immigration. The drop in participation in mid-term elections is of some consequence, however, particularly if the lower rate of involvement is present among those with weaker party identities and among those subject to cross-pressures. An issue that benefits a party in a high-stimulus, high-turnout election may wind up hurting that same party if these same cross-pressured voters do not consistently show up from one election to the next. This may explain why President Trump’s party did not fare well in the mid-term elections even though the immigration issue remained as salient to voting in 2018 as it was in 2016. Immigration may have played a similar role in the decision making of those who voted in both the 2016 and 2018 elections but the 2018 electorate was differently constituted. This would surely be the case if a large share of Trump supporters were low information, peripheral voters, who did not return to vote in 2018 (Fording and Schram 2017). The alternative explanation, to be developed below, is that voters who did participate in both elections have brought their immigration attitudes more closely into line with their party identities. They could have also changed their minds on the immigration issue, but it is more likely that immigration policy has simply lost its influence in their decision making because political party positioning has taken over. Ironically, opinions on an issue can become more polarized, but that issue can also lose political influence if the evolution of opinion is consistent with partisan alignment. Drawing from a panel study across several recent election years, the same voters responding in each wave, I set forth the descrip- tive statistics on the development of immigration opinion on the following pages, then turn to address the possibility that immigration opinion has become more divisive in partisan terms but less influential as a wedge issue.

Data Analysis

The data for this analysis are drawn from the Democracy Fund Voter Study Group surveys carried out by YouGov, combining waves of online panel respondents from 2011–2012, 2016, 2017 and 2018.1 Many of the same questions of interest in this research were repeated across three or more waves of the panel, offer- ing some picture of attitude change among the participants. There was some panel attrition, as not all respondents participated in each wave, but there was also substantial continuity, as more than 4000 panelists surveyed in the initial wave also participated in the subsequent ones. In addition to this opportunity

1 To access the data, see: https://www.voterstudygroup.org/data, accessed 9/18/2019. From Wedge Issue to Partisan Divide 471 to observe opinion change among a representative sample of voters, the 2018 wave also included a large Hispanic oversample, providing a helpful gauge of the opinions of a subgroup thought to have an important stake in immigration policy. The YouGov panel consists largely of respondents who indicated in 2018 that they were registered voters (94 percent), with 78 percent declaring their intention to vote. Since 2016 a number of social science studies have attempted to explain Trump support, with explanations going to symbolic racism and xenophobia, economic hardship, status anxiety, and other issue-related factors aside from party identification (Morgan and Lee 2018; Mutz 2018; Vidal 2018; Wright and Esses 2019; Green and McElwee 2019). Trump’s support was sufficiently dis- tinct from that of the previous Republican nominee, Mitt Romney, to merit these efforts. Not surprisingly, few if any researchers have denied some role for immigration policy attitudes, whether they ultimately attribute those to racial prejudice, to a preoccupation with declining status, or some concep- tion of labor market threat. Given how much emphasis the Trump campaign placed on immigration policy, we should not be surprised to see that it shaped candidate evaluation in 2016, even producing some crossover support (Gimpel 2017). But does it show up to influence voting for other offices in the mid-term election of 2018?

Measure of Immigration Attitudes

To carry out plausible tests of the impact of immigration opinion on voting, a reasonable place to begin is with an additive index out of the responses to three survey items tapping immigration opinion. Because immigration is a multi- faceted area of public policy researchers commonly combine survey measures rather rely upon a single item to gauge opinion. Opinion is also more accurately measured with combinations of similar, scalable, items covering the same theme rather from singular survey questions (Ansolabehere, Rodden, and Snyder 2006, 2008). These three items read as follows, and the crosstabulated results by party identification are reported in the Appendix for each wave: –– Overall, do you think illegal immigrants make a contribution to American society or are a drain? –– Do you favor or oppose providing a legal way for illegal immigrants already in the US to become US citizens? –– Do you think it should be easier or harder for foreigners to immigrate to the US legally than it is currently? 472 James G. Gimpel

As constructed, this measure ranges from 1 to 8, where 8 indicates the most restriction-oriented response and 1 represents opinion-holding favorable to generous immigration.2 There are 2363 respondents for whom the scale can be constructed for each wave of the panel. Histograms of the distribution of responses for each of the four waves are shown in Figure 1. Across the respondents common to all four waves, the distribution of opinion is remarkably similar, suggesting substantial stability and continuity in aggregate opinion. Standard difference in means tests for pairings show statistically significant differences in the distributions, though the substantive differences are argu- able. The mean level of restrictionist sentiment does shift downward slightly from the 2011 (4.93) survey to 2016 (4.88) (p ≤ 0.07), downward again in 2017 (4.44) (p ≤ 0.01), then back up in 2018 (4.72) (p ≤ 0.01), though still not back to the level it started from.3 Focusing only on the means obscures notable aspects of changing variation at the poles of the distribution. The increase in the influence of partisan polarization on the issue from 2011 to 2016 is apparent as it became a high-profile campaign theme, pushing the party faithful in opposite directions. Between the two earliest waves, Republican respondents shift in a decidedly more restrictionist direction while Democrats move opposite. By the 2016 wave, about 57 percent of Strong Republicans scored either a 7 or an 8 on the restriction scale, but by 2018 this was up to 61 percent. Sharper leftward movement is shown among Democrats as indicated by the arrows in Figure 1: in 2016 about 35.8 percent of Strong Democrats took the most generous position on immigration, but by 2018 this jumps to 52.1 percent. Among Democrats (who are not strong) and Democratic leaners, there is also movement toward the liberal pole (see Appendix Tables A1 to A4). Notably, in the aggregate, restrictionist views do not gain any ground from 2016 to 2018 among these panel respondents, remaining at about 31 percent. The most generous immigration position, though, does gain adher- ents, moving from 19 to 25 percent (see Tables A1 to A4). My argument is that this reflects the partisan socialization process at work: as voters become more informed about immigration policy they learn where the political parties and candidates stand. Their views on policy shift into alignment with their partisan commitments.

2 These items scale across each of the four waves at α ≥ 0.66. 3 Statistically significant differences reported here are from two-tailed t-tests on pairings 2011–2016, 2016–2017, 2017–2018. From Wedge Issue to Partisan Divide 473

500 Mean = 4.93 500 Mean = 4.88 Std. Dev. = 2.293 Std. Dev. = 2.328 400 N = 2.363 400 N = 2.363

300 300 requency requency

F 200 200 F

100 100

0 0 0 246810 0246810 Immigration scale from 1–8, for 2011 Immigration scale from 1–8, for 2016

500 Mean = 4.44 500 Mean = 4.72 Std. Dev. = 2.292 Std. Dev. = 2.429 400 N = 2.363 400 N = 2.363

300 300 requency requency F F 200 200

100 100

0 0 0246810 0246810 Immigration scale from 1–8, for 2017 Immigration scale from 1–8, for 2018

Figure 1: Distribution of Opinion on Immigration Restriction across Four Waves of YouGov Democracy Fund Voter Study Group Panel, 2011–2018.

While larger shares of partisans reach polar opposite conclusions about immigration policy by 2018 it remains apparent throughout this period that independents favor the more restrictionist position, giving President Trump an edge with late-deciding voters who might care about this issue. In the 2011 survey, about 38% of independents scored a 7–8, compared to only 12 percent who were at a 1–2. Similar results obtain in 2016, with independents again leaning more toward restriction. In 2017 there is a noticeable drop in the percentage of independents at the restrictive pole, but this rises again in 2018: 38 percent favoring restriction, 18 percent taking the most generous policy position. Across the four waves, the number of independents with moderate immigration views has diminished somewhat, but more independents are now sympathetic to generous immigration policy than were before (see Appendix Tables A1 to A4).

Immigration Policy Declining as a Wedge Issue

The association between party identification and immigration policy views reaches a local peak by 2018. Looking ahead, there is little reason to expect immigration policy will become less politicized in 2020. This could undercut Presi- 474 James G. Gimpel dent Trump’s bid to use immigration as a wedge issue in 2020 to attract crossover support. Plainly there are fewer restrictionist and ambivalent Democrats as party positioning has become more familiar to voters over time. The tendency for Democrats to bring their viewpoints more into alignment with their political party loyalties may help to account for the Republicans’ lackluster performance in 2018 in spite of the president making immigration a major policy focus during his first 2 years. One might counter this reasoning by pointing out that as long as Repub- licans also move in a more polar direction then any losses President Trump may experience among restrictionist Democrats he will make up by attracting more party-congruent Republicans. As a result, the 2020 election will be much more about mobilizing the base than about persuasion. There is some- thing to this argument. But the data examined here seem to suggest a greater movement by Democrats toward the generous end of the immigration con- tinuum than by Republicans toward the restriction end. Republicans were already heavily concentrated at the restriction pole in 2016. It is Democrats who appear to have more closely aligned their immigration views with their party preferences.

Political Preference 2016 and 2018

Examining the data using standard multivariate statistical tools shows some important differences in support from 2016 to 2018. A logistic regression analysis for the Clinton-Trump vote in 2016, and the choice of House candidate in 2018, appears in Appendix Table A5. Explanatory variables include immigration opinion, party identification, economic evaluation, race, age, income, and education. Overall, the effects of immigration on vote choice controlling for these common covariates are very similar across the 2 years. This does speak to the stability of opinions once formed and also when aggregated across the entire voting population. But for subgroups there are also notable changes in the distribution of views on immigration and changes in the extent of their association with expressed political preferences. For instance, Figure 2 illustrates the changes by political party identification between the 2016 and 2018 waves. The Y-axis shows the probability of a Republican vote for president (2016) and House (2018). The X-axis displays positions along the immigration opinion scale from most generous to most restrictive. The solid lines show the probability of Republican support by Republican (red) and Democratic (blue) partisans for 2018, with dashed lines for 2016. From Wedge Issue to Partisan Divide 475

1.00

0.90

0.80

0.70 0.692

0.60 0.532 0.50

0.40 Republican vote (prob.) 0.30 0.317

0.20 0.135 0.10

0.00 Generous 234567Restrictive Immigration restriction

Democrat 2018 Democrat 2016 Republican 2018 Republican 2016

Figure 2: Support for Republicans in 2016 and 2018 by Political Party Identification and Immigration Opinion.

The results are largely consistent with the cross-tabulations reported above. Respondents voting preferences become more consistent with the dominant positions expressed by their political party majorities, and immigration opinion is less influential. Republicans who are for generous immigration policy are more likely to support a Republican House candidate in 2018 than they were to vote for President Trump in 2016. Democrats who favor immigration restriction are less likely to support a Republican in 2018 than they were in 2016. Democrats who are for a generous immigration policy support Democrats both times, and the same for Republicans who favor restriction. All the movement is among those holding party incongruent or cross-pressured positions. Figure 2 strongly suggests that cross-pressures caused by the immigration issue are being resolved in favor of political partisanship. Immigration loses influence as a wedge issue by 2018. But what about other voter subgroups?

Women

Figure 3 shows the same development in the direction of party-consistency for Republican and Democratic women. In 2016 only around 9 percent of Republi- can women who favored generous immigration policy expressed support for Donald Trump. But by 2018, the issue has less pull, and an estimated 22 percent of Republican women are supporting a Republican House candidate. Democratic 476 James G. Gimpel

1.00

0.90

0.80

0.70

0.60 0.599

0.50

0.40 0.410

Republican vote (prob.) 0.30

0.20 0.220

0.10 0.094

0.00 Generous 2 34567Restrictive Immigration restriction

Dem women 2018 Dem women 2016 Rep women 2018 Rep women 2016

Figure 3: Support for Republicans in 2016 and 2018 by Political Party Identification, Gender and Immigration Opinion.

women holding restrictionist views also fall sharply in their Republican support, by nearly 20 points, 60 percent to 41 percent.

Educational Attainment: Some College, No Degree

What about those with moderate levels of educational attainment, specifi- cally those who have had some college, but have not obtained a 2-year or 4-year college degree? These voters are said to constitute a swing constitu- ency, earning lower incomes, and considered vulnerable to labor market adjustments. They also might readily express reservations about immigration on the grounds that it might adversely influence their social and economic status. Figure 4 depicts the relationship and shows this subgroup to be no exception to the general pattern in Figures 2 and 3. Republicans with some college who show sympathy for a more generous immigration policy move back to supporting their party between 2016 and 2018, surging 16 points. Democrats with some college move directly opposite, by a slightly greater magnitude. Even voters thought to be most sensitive to immigration policy move toward conventional party majority positions across the two waves of the YouGov panel. From Wedge Issue to Partisan Divide 477

1.00

0.90

0.80

0.70 0.671 0.60

0.50 0.499

0.40

Republican vote (prob.) 0.30 0.289

0.20 0.124 0.10

0.00 Generous 234567Restrictive Immigration restriction

Dem some college 2018 Dem some college 2016 Rep some college 2018 Rep some college 2016

Figure 4: Support for Republicans in 2016 and 2018 by Political Party Identification, Educational Attainment and Immigration Opinion.

Other Levels of Education

As it turns out, these data serve to document the same pattern of partisan division re-securing a dominant position over a cross-cutting issue for other electoral subgroups. For those with only a high school education, for example, the patterns are much the same as shown in Figure 4. For those with a 4-year college degree there is also a move in the direction of party congruence (not shown).

What about Independents?

Probably the best targets for political persuasion on immigration remain non- aligned or weakly aligned voters who may not be taking much notice of party positions, are not likely to follow political events very closely, and/or may well substitute someone else’s judgment for their own as the election approaches. Figure 5 shows that for independents and leaners with less education but with some inclination to favor a generous immigration policy there is a movement in a Republican direction between 2016 and 2018 (Figure 5). Those with only a high school diploma shift toward Republicans (19 points) more so than those with some college (10 points). There is no movement for those at the restrictionist pole. But these independent voters may be showing an increased willingness to 478 James G. Gimpel

1.00

0.90

0.80

0.70

0.60

0.50

0.40 0.331 0.30 Republican vote (prob.) 0.241 0.20 0.132 0.10

0.00 Generous 2 34567Restrictive Immigration restriction

Ind high school 2018 Ind some coll 2018 Ind some coll 2016 Ind high school 2016

Figure 5: Support for Republicans in 2016 and 2018 by Independent Identification, Educational Attainment and Immigration Opinion.

support Republican candidates even if they hold policy views that run opposite to where a majority of Republicans stand on this issue. These voters are likely to be quite unstable in their political preferences throughout the campaign, however, so no one should read Figure 5 to be anything but a very temporary estimate.

Discussion and Conclusions

Certainly in aggregate the differences reported here between 2016 and 2018 appear to be minimal. The probability of casting a Republican vote in 2018 is very similar to the probability of supporting Donald Trump in 2016 once the usual covariates are included (see Table A6). One might conclude from the results in the appendix tables that nothing appreciably changed. A real test of what’s changed, however, would require an examination of how well the 2016 values of the independent variables explain a 2018 outcome, not how the 2018 versions of those variables do in explaining the outcome. I do not conduct this kind of “true” panel analysis here, but the design of the data collection certainly invites it. Within this subject population there do appear to be changes in relation- ships among explanatory variables between 2016 and 2018 that should not be ignored. Chiefly, the role of party identification, already strong in 2016, grows even stronger. Though the impact of the immigration issue on the vote also seems From Wedge Issue to Partisan Divide 479 to register at about the same magnitude, for some subgroups it also weakens because of greater polarization by party. Resulting shifts may offset each other in aggregate, producing much the same means, but the distributional differences may dictate a very different campaign strategy for 2020, perhaps one that relies on the politics of immigration to activate base voters who have come to care even more about it, but not to play much of a persuasive role for partisans on the other side. If a political preference becomes virtually identical with party identification, then all restrictionists vote one way, and everyone for liberalized immigration votes the other. Immigration opinion and party identification become so mutu- ally reinforcing that there is little possibility for persuasion-oriented movement. Complete partisan polarization is not where opinion stood in 2018 but there is evidence for movement in that direction. There are still partisans who are cross- pressured by holding views opposed by a majority of their co-partisans, but there are fewer of them than in 2016. My best guess is that there will be even fewer still by the Fall of 2020. Wedge issues cannot be cross-cutting and helpful to a campaign’s persuasive efforts if they come to align partisans too perfectly. The augmented partisan side-taking on immigration visible in these opinion surveys comes with a meas- urable reduction in the heterogeneity and inconsistency of opinions necessary to create flows of voters across party lines. If President Trump’s 2016 victory is partly attributable to crossover Democratic voters who will not be as supportive in 2020, these losses will have to be made up somewhere else, perhaps through greater mobilization, or deployment of a new issue with cross-cutting power. Greater partisan intensity on an issue could certainly heighten turnout, resolving ambivalence and contributing to an earlier decision on whom to support. But it is not immediately clear whether such an increase in resolve will benefit one party as opposed to the other if large blocs of voters on both sides wind up with more resolute and party-congruent views. Because most YouGov panelists report being voters, or intending to vote, other research will have to estimate the impact that greater partisan alignment on immigration and other policy issues might have on mobilization in 2020. Although I have made the case that the differences reported here between 2018 and 2016 are the consequence of voters gaining greater knowledge about the positions of the two political parties on immigration policy, these differences could be traceable simply to the important differences between House candidates and the President on this and related policies. In that case, less cross-over support in 2018 than in 2016 might be the result of voters who do not see House candidates campaigning on an issue at all, not because they have obtained a more partisan understanding of it. If these were two independent cross-sectional samples of 480 James G. Gimpel voters, another interpretation would be to attribute increased partisanship from 2016 to 2018 to the voters who have dropped out of the mid-term electorate – less interested and less partisan voters, consistent with the surge-and-decline pattern so commonly observed (Campbell 1960). But given that this is a panel design observing the very same voters from 2016 to 2018 then that usual pattern of mid- term drop-off is not present in these data. Even so, other plausible interpretations of these data are possible, and of course other surveys should be evaluated. The two major parties are always searching for advantages on issues that will give them a reliable electoral advantage in general elections. Sometimes these advantages emerge in one or two elections, but they rarely endure for long as opposing parties react, candidates adapt, voters learn and their viewpoints adjust. Restriction-leaning immigration policies may command the same sizable share of voters in 2020 that gave President Trump the electoral advantage in 2016, but careful study of the changes in opinion since then suggest that his campaign should not take that for granted.

Appendix

Table A1: Immigration Restriction by 7-Point Party Identification for 2011.

7 Point Party ID Total

Strong Democrat Lean Independent Lean Republican Strong Democrat Democrat Republican Republican

Immigration Restriction Scale 3 pt 2016 Generous (1,2) Count 216 55 106 38 7 15 18 455 % 39.6% 20.4% 38.5% 12.2% 2.8% 6.1% 4.2% 19.6% Middle Ground Count 256 137 138 154 124 123 227 1159 % 46.9% 50.7% 50.2% 49.5% 50.0% 50.2% 53.4% 50.0% Restrictive (7,8) Count 74 78 31 119 117 107 180 706 % 13.6% 28.9% 11.3% 38.3% 47.2% 43.7% 42.4% 30.4% Total Count 611 546 270 275 311 248 245 425 % 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

Pearson Chi-Square = 429.10; p ≤ 0.001; Phi = 0.43; p ≤ 0.001. Source: Democracy Fund Voter Study Group Panel, YouGov. From Wedge Issue to Partisan Divide 481

Table A2: Immigration Restriction by 7-Point Party Identification for 2016.

7 Point Party ID Total

Strong Democrat Lean Independent Lean Republican Strong Democrat Democrat Republican Republican

Immigration Restriction Scale 3 pt 2016 Generous (1,2) Count 219 59 94 41 6 15 10 444 % 35.8% 24.3% 38.5% 12.5% 2.7% 5.2% 2.4% 18.8% Middle Ground Count 343 126 133 159 102 139 172 1174 % 56.1% 51.9% 54.5% 48.6% 45.3% 48.4% 40.8% 49.8% Restrictive (7,8) Count 49 58 17 127 117 133 240 741 % 8.0% 23.9% 7.0% 38.8% 52.0% 46.3% 56.9% 31.4% Total Count 611 243 244 327 225 287 422 2359 % 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

Pearson Chi-Square = 590.21; p ≤ 0.001; Phi = 0.50; p ≤ 0.001. Source: Democracy Fund Voter Study Group Panel, YouGov.

Table A3: Immigration Restriction by 7-Point Party Identification for 2017.

7 Point Party ID Total

Strong Democrat Lean Independent Lean Republican Strong Democrat Democrat Republican Republican

Immigration Restriction Scale 3 pt 2017 Generous (1,2) Count 281 69 110 67 12 19 14 572 % 48.5% 29.1% 43.0% 19.5% 5.1% 7.4% 3.1% 24.2% Middle Ground Count 274 116 132 188 130 144 245 1229 % 47.3% 48.9% 51.6% 54.7% 54.9% 55.8% 54.7% 52.1% Restrictive (7,8) Count 24 52 14 89 95 95 189 558 % 4.1% 21.9% 5.5% 25.9% 40.1% 36.8% 42.2% 23.7% Total Count 579 237 256 344 237 258 448 2359 % 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

Pearson Chi-Square = 578.12; p < 0.001; Phi = 0.495; p < 0.001. Source: Democracy Fund Voter Study Group Panel, YouGov. 482 James G. Gimpel

Table A4: Immigration Restriction by 7-Point Party Identification for 2018.

7 Point Party ID Total

Strong Democrat Lean Independent Lean Republican Strong Democrat Democrat Republican Republican

Immigration Restriction Scale 3 pt 2018 Generous (1,2) Count 309 84 120 59 4 11 4 591 % 52.1% 35.1% 46.7% 18.0% 1.7% 4.3% 0.9% 25.1% Middle Ground Count 254 110 127 143 102 121 174 1031 % 42.8% 46.0% 49.4% 43.6% 44.3% 47.6% 38.1% 43.7% Restrictive (7,8) Count 30 45 10 126 124 122 279 736 % 5.1% 18.8% 3.9% 38.4% 53.9% 48.0% 61.1% 31.2% Total Count 593 239 257 328 230 254 457 2358 % 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

Pearson Chi-Square = 843.9; p < 0.001; Phi = 0.60; p < 0.001. Source: Democracy Fund Voter Study Group Panel, YouGov.

Table A5: Odds Ratios Associated with Clinton-Trump Voting in the 2016 November Election and in the 2018 Mid-Term Election for US House.

Independent Variable 2016 Vote 2018 Vote

Immigration Opinion Scale (High is more Restrictive) 2.30** 2.35** Economic Conditions: Same 2.93** 0.07** Economic Conditions: Worse 4.10** 0.03** Black 0.27** 0.25 Hispanic 0.73 1.31 Asian 0.50 1.83 Women 0.49** 0.42** Age 65 and Older 1.08 0.62 Age 18–29 0.48 5.01 Income $10–19,999 1.74 3.46 Income $20–29,999 1.25 2.62 Income $30–39,999 1.11 1.66 Income $40–49,999 2.06 1.73 Income $50–59,999 2.15 1.44 Income $60–69,999 1.76 0.96 Income $70–79,999 3.00 1.91 Income $80–99,999 2.53* 2.40 Income $100–119,999 1.30 1.50 Income $120–149,999 1.75 1.40 From Wedge Issue to Partisan Divide 483

Table A5 (continued)

Independent Variable 2016 Vote 2018 Vote

Income $150,000 and above 1.56 1.18 Democrat 0.08** 0.03** Republican 10.64** 74.24** High School 0.91 1.47 Some College 0.89 0.85 2-Year Degree 0.64 1.78 4-Year Degree 0.97 1.27 Post Graduate Degree 0.66 1.03 N 1811 1717

Dependent variable for 2016: 0 = Clinton vote; 1 = Trump vote; third party voters have been excluded. For 2018: 0 = Democratic; 1 = Republican. Cell entries are odds ratios showing by how much the odds of voting Republican increase/ decrease for each one unit change in x, holding other variables constant. Statistical significance levels: **p ≤ 0.01; *p ≤ 0.05. Source: Democracy Fund Voter Study Group Survey by YouGov.

Table A6: Complete Logistic Regression Model of Clinton-Trump Voting in the 2016 November Election and in the 2018 Mid-Term Election for US House.

Independent Variable 2016 Vote 2018 Vote

Immigration Opinion Scale (High is more Restrictive) 0.8323** 0.8537** (0.0698) (0.0945) Economic Conditions: Same 1.0753** −2.6158** (0.1941) (0.4076) Economic Conditions: Worse 1.4102** −3.4538** (0.3016) (0.3464) Black −1.3080** −1.4061 (0.4142) (0.7428) Hispanic −0.3172 0.2728 (0.4452) (0.4293) Asian −0.6913 0.6052 (0.6049) (0.6271) Women −0.7162** −0.8581** (0.2056) (0.3346) Age 65 and Older 0.0764 −0.4751 (0.1826) (0.2716) Age 18–29 −0.7252 1.6108 (0.8738) (1.3081) Income $10–19,999 0.5534 1.2407 (0.5681) (1.0891) Income $20–29,999 0.2215 0.9644 (0.5712) (0.8419) 484 James G. Gimpel

Table A6 (continued)

Independent Variable 2016 Vote 2018 Vote

Income $30–39,999 0.1077 0.5083 (0.5235) (1.0492) Income $40–49,999 0.7234 0.5453 (0.5572) (0.9998) Income $50–59,999 0.7648 0.3638 (0.5259) (1.1402) Income $60–69,999 0.5632 −0.0397 (0.4897) (0.8864) Income $70–79,999 1.0976 0.6453 (0.5901) (0.9931) Income $80–99,999 0.9265* 0.8750 (0.4608) (1.1242) Income $100–119,999 0.2608 1.4521 (0.4914) (0.9841) Income $120–149,999 0.5615 1.1890 (0.5328) (0.9156) Income $150,000 and above 0.4453 0.5175 (0.5276) (1.0196) Democrat −2.5095** −3.6520** (0.2795) (0.3248) Republican 2.3646** 4.3074** (0.2362) (0.4959) High School −0.0940 0.3883 (0.7788) (0.8634) Some College −0.1216 −0.1678 (0.7598) (0.8188) 2-Year Degree −0.4515 0.5748 (0.8526) (0.9749) 4-Year Degree −0.0307 0.2426 (0.8314) (0.9248) Post Graduate Degree −0.4207 0.0314 (0.7876) (0.8002) Constant −4.1157 −2.0960 (0.9849) (1.4923) Log Pseudolikelihood −395.95 −177.27 N 1811 1717 % Correctly Classified 91.7% 95.5% Null Model 51.0% 55.0%

Dependent variable for 2016: 0 = Clinton vote; 1 = Trump vote; third party voters have been excluded. For 2018: 0 = Democratic; 1 = Republican. Cell entries are logistic regression coefficients (standard errors clustered by state). Statistical significance levels: **p ≤ 0.01; *p ≤ 0.05. Source: Democracy Fund Voter Study Group Panel Survey by YouGov. From Wedge Issue to Partisan Divide 485

References

Ansolabehere, Stephen, Jonathan Rodden, and James M. Snyder Jr. 2006. “Purple America.” Journal of Economic Perspectives 20 (Spring): 97–118. Ansolabehere, S., J. Rodden, and J. M. Snyder. 2008. “The Strength of Issues: Using Multiple Measures to Gauge Preference Stability, Ideological Constraint, and Issue Voting.” American Political Science Review 102 (2): 215–232. Bartels, L. M. 1996. “Uninformed Votes: Information Effects in Presidential Elections.” American Journal of Political Science 40: 194–230. Berelson, B. R., P. F. Lazarsfeld, and W. N. McPhee. 1954. Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago, IL: University of Chicago Press. Campbell, A. 1960. “Surge and Decline: A Study of Electoral Change.” Public Opinion Quarterly 24 (3): 397–418. Campbell, A., P. E. Converse, W. E. Miller, and D. E. Stokes. 1960. The American Voter. Ann Arbor, MI: University of Michigan Press. Carpini, M. X. D., and S. Keeter. 1996. What Americans Know About Politics and Why It Matters. New Haven, CT: Yale University Press. Chaffee, S. H., and R. N. Rimal. 1996. “Time of Vote Decision and Openness to Persuasion.” In Political Persuasion and Attitude Change, edited by Diana C. Mutz, Paul M. Sniderman and Richard A. Brody, 267-291. Ann Arbor, MI: University of Michigan Press. Dalager, J. K. 1996. “Voters, Issues, and Elections: Are the Candidates’ Messages Getting Through?” The Journal of Politics 58 (2): 486–515. Dancey, L., and G. Sheagley. 2016. “Inferences Made Easy: Partisan Voting in Congress, Voter Awareness, and Senator Approval.” American Politics Research 44 (5): 844–874. Davis, N. T., and L. Mason. 2016. “Sorting and the Split-Ticket: Evidence from Presidential and Subpresidential Elections.” Political Behavior 38 (2): 337–354. Fording, R. C., and S. F. Schram. 2017. “The Cognitive and Emotional Sources of Trump Support: The Case of Low-Information Voters.” New Political Science 39 (4): 670–686. Gimpel, J. G. 2017. Immigration Policy Opinion and the 2016 Presidential Vote: Issue Relevance in the Trump-Clinton Election. Washington, DC: Center for Immigration Studies. Green, J. 2019. “Floating Policy Voters in the 2016 US Presidential Election.” Electoral Stud- ies, In press, Posted Online at: https://www.sciencedirect.com/science/article/pii/ S0261379418302907, Accessed September 24, 2019. Green, J., and S. McElwee. 2019. “The Differential Effects of Economic Conditions and Racial Attitudes in the Election of Donald Trump.” Perspectives on Politics 17 (2): 358–379. Henderson, M., and D. S. Hillygus. 2016. “Changing the Clock: The Role of Campaigns in the Timing of Vote Decision.” Public Opinion Quarterly 80 (3): 761–770. Hillygus, D. Sunshine, and Todd G. Shields. 2008. The Persuadable Voter: Wedge Issues in Presidential Campaigns. Princeton, NJ: Princeton University Press. Jacobson, G. C. 2015. “How do Campaigns Matter?” Annual Review of Political Science 18: 31–47. Kinder, Donald R. 1998. “Opinion and Action in the Realm of Politics.” In Handbook of Social Psychology. Oxford: Oxford University Press. Margolis, M. F. 2018. “How Far Does Social Group Influence Reach? Identities, Elites, and Immigration Attitudes.” The Journal of Politics 80 (3): 772–785. Morgan, S. L., and J. Lee. 2018. “Trump Voters and the White Working Class.” Sociological Science 5: 234–245. 486 James G. Gimpel

Mulligan, K. 2011. “Partisan Ambivalence, Split-Ticket Voting, and Divided Government.” Political Psychology 32 (3): 505–530. Mutz, D. C. 2018. “Status Threat, Not Economic Hardship, Explains the 2016 Presidential Vote.” Proceedings of the National Academy of Sciences 115 (19): E4330–E4339. Sides, J., M. Tesler, and L. Vavreck. 2017. “The 2016 US Election: How Trump Lost and Won.” Journal of Democracy 28 (2): 34–44. Skitka, L. J., and C. W. Bauman. 2008. “Moral Conviction and Political Engagement.” Political Psychology 29 (1): 29–54. Sniderman, P. M., R. A. Brody, and P. E. Tetlock. 1993. Reasoning and Choice: Explorations in Political Psychology. Cambridge, MA: Cambridge University Press. Vidal, X. M. 2018. “Immigration Politics in the 2016 Election.” PS: Political Science & Politics 51 (2): 304–308. Wright, J. D., and V. M. Esses. 2019. “It’s Security, Stupid! Voters’ Perceptions of Immigrants as a Security Risk Predicted Support for Donald Trump in the 2016 US Presidential Election.” Journal of Applied Social Psychology 49 (1): 36–49.