James G. Gimpel is a professor of government at the University of Maryland, College Park.
- Approaching the 2016 election, immigration policy polarized opinion by partisan identity more than at any other time in contemporary history.
- Election surveys suggest that immigration policy opinion was responsible for moving crossover voters to Donald Trump in the 2016 presidential election, improving his performance over Mitt Romney in 2012.
- Contrary to the post-election conclusions of Romney advisers, Romney would likely have done better, not worse, by campaigning more vigorously for immigration control.
- Women were more likely to be moved to vote for Trump as a result of immigration policy opinion than men. Working class men were more likely to support Trump than Romney, but the difference is not associated with immigration policy per se.
- Voters at higher levels of education voted for Trump over Romney when they were more conservative on immigration.
- Democratic voters supported Trump more than Romney when they were more conservative on immigration.
- Trump did worse than Romney among some groups who held liberal immigration views, including higher income Republicans.
As a policy issue, immigration played an unusually prominent role in the 2016 election, one that issues do not regularly play in presidential contests. Donald Trump made immigration a central theme of his campaign, raising the subject frequently at major rallies and in widely covered speeches. It was one topic of consistent message discipline in an organization that many called untidy and haphazard.
Just as important, the media has also paid a great deal of attention to his pledges on immigration. To be sure, many stories have been written identifying the president's views on infrastructure, taxes, and other issues, but immigration triggers intense reactions among both supporters and opponents. Wall building, border security, and ending DACA draw large and opinionated audiences, whereas road building, the solvency of Social Security, and trade policy do not.
The polarization of immigration along party lines is not something that began with the 2016 election cycle, nor with Trump's announcement speech. Turning to Gallup's periodic surveys, there is abundant evidence back to 2000 of a marked partisan difference in support for increased/decreased immigration. The opinion trend indicates that, after 2001, the gap in party support for immigration widened for the remainder of that decade, though the overall percentage calling for decreased immigration did decline at first.
The blank space in the Gallup survey series in Figure 1 appears because the standard question about immigration levels was not asked in 2009, 2010, or 2011. When they took up the question again in 2012, there was a surge in the percentage calling for decreased immigration, with an evident partisan gulf that widened further in the years leading up to the 2016 election.1 As I indicated in a previous CIS Backgrounder, the Trump campaign seems to have clearly comprehended these trends while rival GOP candidates remained willfully blind, deaf, and mute.2
The partisan gulf in the most recent polling can be seen just as clearly in the growing gap between the party bases calling for increased immigration (see Figure 2). One thing is for certain, the increase in support for more generous immigration over the last four years comes overwhelmingly from within the Democratic Party, not from Republicans.
Immigration now polarizes the mass public to an extent not seen in contemporary times.3 No one contests that the Trump candidacy in the 2016 election was a source of this heightened interest. But there are legitimate questions about what role immigration played in vote choice controlling for traditional influences on voting. For one thing, there is some doubt about whether issues really matter to voters once we control for their deeply rooted partisan identities. If all the people favorable to restrictionist immigration policy would have voted for Donald Trump because they are Republicans, then maybe his policy positioning didn't make the difference some observers claim. Once researchers control for party identification, there may be no remaining role for issues to play in explaining vote choice. On the other hand, given the Republican nominee's widely alleged appeal to non-traditional Republican constituencies and to voters who refused to support Mitt Romney in 2012, there are reasons to think that Trump's policy pronouncements caused real political movement.
As detailed below, data analysis based on the Democracy Fund Voter Study Group survey released in June 2017 suggest that immigration policy attitudes did drive voters to support Donald Trump, controlling for their party identification, sex, age, race and ethnicity, income, and education.4 Perhaps the impact of immigration on this election appears obvious to many observers, but it is not. Messaging commonly thought to be essential to a candidate's victory often turns out not to be because those issues are already embraced enthusiastically by a candidate's supporters. No matter how much a nominee discusses them, they have no persuasive power, and attract no crossover support.
To conduct a plausible test of the impact of immigration opinion on Trump support, a sensible start is to create 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 than rely upon a single item to gauge opinion. These three items read as follows, and the cross-tabulated results by party identification are reported in the appendix tables:
- 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 United States to become U.S. citizens?
- Do you think it should be easier or harder for foreigners to immigrate to the United States legally than it is currently?
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.5 In 2016, about 29 percent of all respondents scored a 7 or 8 on the scale. On the generous end of the continuum, about 18 percent scored a 1 or 2. Similar results were obtained when the same questions were combined in a scale for the earlier wave of the survey conducted in 2011.6
As for presidential preference, higher scorers on this scale are pronouncedly pro-Trump when it comes to their November 2016 vote (see Figure 3). Comparisons of the same immigration opinion scale against the Romney vote show that opinion in the 2016 election was more polarized than in 2012. In truth, the two GOP nominees did not take wildly different positions. The difference was that Romney was advised to remain quiet about the subject on the campaign trail, having been repeatedly warned that his views would alienate Hispanic voters. Throughout the fall, the Romney campaign was observed to be fearful of discussing immigration, avoiding the topic and abruptly cutting-off questions about it.7 Consequently, many otherwise informed voters reached November of 2012 barely aware of Romney's views on a variety of immigration options. His campaign chief, Matt Rhoades, was widely reported to admit regrets about moving the candidate toward a more restriction-oriented immigration policy in the primary.8
The Trump campaign took a very different course, announcing immigration restriction as a major theme on his first day in the race.9 The candidate's aggressive positioning on the issue arguably caused the Clinton team to embrace racial and ethnic identity politics to an even greater extent than they had in response to the primary challenge by rival Bernie Sanders. The end result was far greater polarization between the two major party candidates on immigration policy by November, as shown in Figure 3, where 90 percent of those at the most conservative pole voted for Trump, compared to just 5 percent at the most liberal pole. Figure 3 also shows that those embracing restrictionist views were much more likely to vote for Trump than they were for Romney.
Notably, exit polling indicates that the GOP nominee did not suffer a markedly negative reaction among Latino voters for boldly calling for immigration control.10 Hispanic turnout reportedly surged, and yet the president-to-be still outperformed Romney. In hindsight, Romney's advisers look to have been wrong in pushing him to remain silent on immigration for fear of alienating Hispanics. This bad advice was based on the assumption that Hispanics think and vote alike, and are uniformly favorable toward liberal immigration policies. But this must be wrong-headed if Donald Trump can win 28-30 percent of them without a great deal of outreach.11
Network exit polls are disparaged whenever various groups wind up disliking their estimates, but alternative methods of obtaining estimates also have shortcomings and sources of error.12 Some polls show greater Trump support among Latinos, others less, but certainly there was no severe drop-off of Hispanics reporting Republican loyalty.13 Some Hispanic voters valued other aspects of the Trump candidacy, ignoring his statements on immigration. Others may have taken an intense disliking to his opponent or simply fell back on their party identification like so many other voters do. The various reasons offered for their choice press home the point that Latinos are not single-issue voters, and that the same explanations that are deployed to account for how non-Latino voters behave apply to Latino voters as well.14 Talking about them as if they are a single-minded bloc does not make it so, but neither are they poised to surge into the Republican Party. The emergent picture is one of an enduring political division ranging from about 25 percent to the high 30s in support for Republican candidates nationally, with lower turnout bolstering GOP prospects.
A majority of election post-mortems have settled on lower-middle-income voters as the single most important group that decided the election in Trump's favor. This is a population thought to be animated by associated concerns including a slow economic recovery, a steady influx of illegal and legal immigrants, and long-term economic restructuring with consequent loss of status.15
Researchers might acquire a clearer sense of how much immigration policy mattered to working class men in 2016 by looking at estimates of Trump voting by sex and income, controlling for party identification, age, education, and race. From these figures can be inferred a policy's impact on a specific group's voting controlling for these other characteristics. As described above, the Democracy Fund data provided by YouGov are the basis for this evaluation.16
Figure 4 displays graphs summarizing how men in several middle-income brackets voted, by the liberal or conservative nature of their immigration views. A couple of aspects of the comparisons in these figures stand out. First, as expected, across all the middle-income groups, holding conservative immigration views predicts high vote percentages for both Romney and Trump, while liberal views align voters with Democrats in both elections.
Second, men in the two lower middle-income brackets were uniformly more likely to vote for Trump than for Romney at every position on the immigration scale, even at the liberal end (Figure 4, top). From this evidence a reasonable conclusion is that immigration was not the only thing that attracted middle-income men to Trump, as those who were more liberal on immigration also favored him over Romney. In the two upper middle-income brackets, on the other hand, men were very similar in their voting in both years (Figure 4, bottom). As income rises, differences in voting from one election year to the next grow considerably smaller among men. Immigration policy positioning matters to vote choice among men, but not that much more in 2016 than in 2012 for those in upper middle-income brackets.
Immigration likely made a larger difference in 2016 for women than for men (Figure 5). Women who favor immigration restriction, and about one-third of them do, voted for Trump at an estimated rate of 86 percent, compared to 2012 when they voted for Romney at about 68 percent, a substantial difference of 18 percentage points. Notably, Trump lost the votes of women overall, collecting only 42 percent support, compared with Clinton's estimated 54 percent.17 But Figure 5 indicates that immigration was both more divisive for women in 2016 than in 2012, and drove more women toward Trump than away from him.
African American and Hispanic Voters
By now it is no secret that African Americans express many of the same reservations about immigration that native-born whites do.18 Given this truth, we should not be surprised that a hard-line stance on immigration won Trump some votes in unusual places. Among black voters there is greater evidence of Trump support at higher levels of restrictive opinion in 2016 (75 percent) than in 2012 (32 percent). These are not trivial differences as about 20 percent of African American respondents expressed opposition to immigration at a 7 or 8 on the eight-point scale in 2016. If three-fourths of those respondents voted for Trump, black support among that group would have been at about 15 percent, compared to only about 8 percent for black voters overall.19 Blacks with reservations about immigration can be said to have contributed modestly to the Trump victory, while blacks supportive of open immigration were about as hostile to Trump as they had been to Romney.
As indicated above and in other research, Latinos are not uniformly pro-immigration and are especially divided on undocumented immigration.20 The Democracy Fund survey indicates that about 23 percent of Hispanic respondents can be placed in the three most conservative categories on the immigration scale. Estimates from this analysis also show that this subgroup was supportive of Trump, slightly more than they were of Romney (see Figure 6).
On the other hand, Latinos on the generous end of the opinion scale are decidedly pro-Clinton, similar to African Americans. Trump seems to have alienated more pro-immigration Latinos than Romney did, while winning a slightly larger share of the most conservative Latinos. For all of the Trump emphasis on immigration and wall-building, the net change was surprisingly small. Across elections, Latino support for Republicans rises or falls marginally when support for GOP candidates rises or falls among the broader electorate of which they are a part. To the extent that they can be characterized as a group at all, their behavior runs more consistent with electoral tides than against them.
Strong and Weak Partisans
Despite the partisan divide on immigration identified above, we do see strong Republicans and strong Democrats supporting Trump at higher levels when they are favorable to immigration restriction than when they are not. From Figure 7, for example, an estimated 26 percent of Democratic restrictionists voted for Trump, compared to only around 11 percent who supported Romney four years earlier.
On the Republican side, some fall-off in Trump support is also visible at the left end of the opinion distribution (see Figure 7). Strong Republicans who favored a more generous immigration policy voted for Trump at a rate of about 18 percent, compared to the 40 percent who supported Romney in 2012. At the conservative end of immigration opinion, the strong Republicans were about the same in their support for the Republican nominees both years (see Figure 7).
Among weak partisans, it is pretty clear that weak Democrats were cross-pressured by immigration in 2016 to an extent that they were not in 2012. Figure 7 shows that the probability of weak Democrats voting for Trump steadily rises as they come to hold more restrictionist views. An estimated 46 percent of these Democrats supported the Republican nominee in 2016, compared with only 11 percent four years earlier. This result suggests considerable prescience on the part of political scientists writing in 2014 who found immigration politics to be an ongoing source of white Democratic defection.21 Was this support by Democrats desiring immigration limits counterbalanced by Republicans who held more liberal views on immigration? No. While there was some drop-off in Trump support, a much smaller share of weak Republicans defected to Clinton than weak Democrats supported Trump.
So far, these survey data suggest that immigration was an ideal example of a wedge issue that benefitted the Trump candidacy, moving many ordinary Democratic voters over to the Trump side while alienating fewer Republicans. What about education? Voters in the highest education categories (e.g., bachelor's degree, post-graduate education) are pretty commonly split between the parties, or even lean Democratic. Network exit polling awarded Clinton a nine-point lead (52 percent to 43 percent) among college graduates and the data reported here suggested a Democratic lead about six points greater among those with post-graduate degrees (i.e., M.A., M.S., JD, Ph.D). Even so, the choice for president among the better-educated was clearly related to their immigration views (Table 1). As voters move to the more restriction-oriented end of the opinion scale, we see a notable uptick in support for Trump across all three of these levels of education, with a notable shift toward Trump in the two most conservative opinion categories (see Table 1).
Those with post-graduate degrees were much more likely to support Trump than they were Romney when they were more conservative on immigration (Figure 8). But similar voters who were liberal on immigration were about equally likely to support Clinton in 2016 as they were Obama in 2012. These comparisons suggest that even among the highly educated, President Trump won many voters who had voted for Obama in 2012, and that taking such a prominent stand on immigration probably contributed to that success. Any trade-offs in electoral support resulting from his controversial views worked more in his favor than against him.
Summary and Conclusions
In the above analysis I have taken stock of several ways in which opinion on immigration shaped voting behavior in the 2016 election. This contest was associated with an unprecedented level of partisan division on many topics, including the subject of immigration. The partisan divide both preceded the Trump campaign as a visible trend in available polls, and was heightened by it, as the campaign was an information source driving opinion-holding on the issue. The Pew Research Center's latest report (2017) suggests that this polarization has only been sustained or augmented 10 months into the president's first term.22
Not addressed here is the impact of the 2016 campaign on turnout levels for the candidates. Certainly the Trump campaign might have also improved on Romney's performance based on the higher turnout of certain electoral subgroups, and the diminished enthusiasm of others. Because the Democracy Fund Voter Study Group data constitute a panel of voters, with few non-voters included, it is not a suitable resource for studying differences in participation across the two elections. These data also have all of the weaknesses of other observational studies, including ambiguity about casual direction, and the possibility that important explanatory variables have been omitted. Even so, the results here for immigration policy opinion are too robust to be the consequence of false positivity.
Apparently tapping into widespread discontent with the trajectory of U.S. immigration policy was a winning strategy in 2016, as evidenced by Trump's improvement over Romney among voters with conservative convictions. The Republican nominee's emphasis pushed the Democrats into a difficult corner in which they were forced to choose between white working class voters and racial and ethnic identity politics. That fateful choice, and the election result, indicates that immigration was a wedge issue in the campaign, and one that Democrats will struggle to address going forward.23 Like other wedge issues in past campaigns, however, it is unlikely to be permanent, as electoral circumstances change and parties adapt.24
Immigration politics had a notable impact on the improvement of the Trump vote over the Romney vote for the following subgroups of the population when they were found to adhere to conservative immigration policies: the well-educated, women, African Americans, and weak Democrats. At the same time, when voters were proponents of generous immigration policy, Trump did worse than Romney among the following groups: Hispanics, strong Republicans, and higher income voters. Trump also did better than Romney among lower-middle-income men, but this improvement runs across the entire range of immigration viewpoints, and so cannot be attributed to immigration restriction per se. Related research has shown that the shift of working class men toward Republicans is a long-term phenomenon25 Only further analysis will reveal what other specific issues might have been a source of Trump support among men.
Any campaign strategy emphasizing an incendiary wedge issue will produce trade-offs in support, as a candidate may lose voters that an alternative candidate from their party might have won, while improving on support from other blocs that another candidate would have lost. From the evidence assembled here, the gains from a focus on immigration restriction appear to have been considerably larger than the losses (see Figure 3).
What any campaign manager hopes is that the groups crossing over to benefit their candidate will be larger than those defecting the other way. Alienating a few affluent and liberal-leaning Republicans turned out to be a small price to pay, at least for a campaign that was not required to beg for money. These same affluent elites are now scrambling to regain control of their party from a populist hijacking. Success will require that they take a less libertarian view of the nation's border, favoring modest but reasonable immigration control measures. Immigration is not the only issue on which Republican elites will have to make adjustments in Donald Trump's direction, but doing so will be a step back toward regaining that most important of political resources: electoral trust.
1 See Bradford S. Jones and Danielle J. Martin, "Path-to-Citizenship or Deportation? How Elite Cues Shaped Opinion on Immigration in the 2010 US House Elections", Political Behavior, 39: 1: 177-204, 2017; and Christopher P. Muste, "The Dynamics of Immigration Opinion in the United States, 1992-2012", Public Opinion Quarterly 77: 1: 398-416, 2013.
2 James G. Gimpel, "Immigration Opinion and the Rise of Donald Trump", Center for Immigration Studies Backgrounder, July 2016.
4 "2016 VOTER Survey Data Set, Top Lines, and Crosstabs", Democracy Fund Voter Study Group, June 2017. Accessed October 17, 2017. The survey was conducted by YouGov, of Palo Alto, Calif.
5 These items scale at reliability value Cronbach's alpha=.66.
6 This scale is also highly correlated with other indicators of immigration policy opinion, such as the desire to see the Republican and Democratic parties adopt more restrictive positions on immigration.
7 Brett LoGiurato, "Mitt Romney Is Terrified Of Talking About Immigration", Business Insider, June 20, 2012; and Evan McMorris-Santoro, "Following Obama's Move, How Will Mitt Romney Break His Silent Treatment Of Immigration?" , Talking Points Memo, June 15, 2012. Accessed October 16, 2017.
8 For example, see: Jon Ward, "Mitt Romney Campaign Manager Matt Rhoades Laments Hard-Right Shift On Immigration", Huffington Post, December 3, 2012, accessed October 16, 2017; and Jeff Zeleny, "Romney Campaign Manager Says He Regrets Immigration Stance", The New York Times, December 3, 2012, accessed October 17, 2017.
9 James G. Gimpel, "Immigration Opinion and the Rise of Donald Trump", Center for Immigration Studies Backgrounder, July 2016.
10 See Harry Enten, "Trump Probably Did Better With Latino Voters Than Romney Did", FiveThirtyEight, November 18, 2016, accessed October 18, 2017.
11 For example post-election stories about Trump and Latino voting, see: Kevin Drum, "Trump Really Did Win 28% of the Latino Vote", Mother Jones, March 6, 2017, accessed October 18, 2017; Marcela Valdes, "'We're Looking at a New Divide Within the Hispanic Community'", The New York Times, November 16, 2015, accessed October 18, 2017; Thomas B. Edsall, "The Democrats' Immigration Problem", The New York Times, February 16, 2017, accessed October 18, 2017; and Alejandra Matos, "On the El Paso border, Trump's appeal with Latinos defies expectations", The Washington Post, December 9, 2016, accessed October 18, 2017.
12 Grace Sparks, "Pollsters Can't Agree On How Latinos Voted In 2016", Huffington Post, December 12, 2016, accessed October 18, 2017.
13 See, for example: "Latinos Will Never Vote For A Republican, And Other Myths About Hispanics From 2016", NPR's "Morning Edition", December 22, 2016, accessed October 18, 2017; Francisco Navas, "What Hispanic people think of Trump: US Latinos give a wide range of answers", The Guardian, January 23, 2017, accessed October 18, 2017; and, Brittny Mejia. "Despite scorn and criticism, these Latino voters don't regret voting for Trump", Los Angeles Times, January 17, 2017, accessed October 18, 2017.
14 Angel Saavedra Cisneros, ed., Latino Identity and Political Attitudes, London, UK: Palgrave Macmillan, 2017.
15 Justin Gest, The New Minority: White Working Class Politics in an Age of Immigration and Inequality, New York: Oxford University Press, 2016. Jeff Manza and Ned Crowley, "Working Class Hero? Interrogating the Social Bases of the Rise of Donald Trump", The Forum, 15(1), pp. 3-28, 2017. Shannon M. Monnat and David L. Brown, "More than a Rural Revolt: Landscapes of Despair and the 2016 Presidential Elections", Journal of Rural Studies 55: 227-236, 2017. L. Collingwood, T. Reny, and A.A. Valenzuela, "Flipping for Trump: Immigration, Not Economics, Explains Shifts in White Working Class Votes", July 7, 2017. Shad Turney, Frank Levy, Jack Citrin, and Neil O'Brian, "Waiting for Trump: The Move to the Right of White Working-Class Men, 1968-2016", The California Journal of Politics and Policy, UC Berkeley, Institute of Governmental Studies, 2017. Accessed October 17, 2017.
18 Vincent L. Hutchings and Cara Wong, "Racism, Group Position and Attitudes about Immigration among Blacks and Whites", Du Bois Review: Social Science Research on Race 11: 2 419-442, 2014.
19 Alec Tyson and Shiva Maniam, "Behind Trump's Victory: Divisions by Race, Gender, Education", Pew Research Center, November 9, 2016. Accessed October 16, 2017.
20 Angel Saavedra Cisneros, ed., Latino Identity and Political Attitudes, London, UK: Palgrave Macmillan, 2017. Mark Hugo Lopez, Rich Morin, and Paul Taylor, Illegal Immigration Backlash Worries, Divides Latinos, Washington, D.C.: Pew Hispanic Center, October 28, 2010. S.M. Rouse, B.C. Wilkinson, and J.C. Garand, "Divided Loyalties? Understanding Variation in Latino Attitudes Toward Immigration", Social Science Quarterly, 91:3: 856-882, 2010.
21 Zoltan Hajnal and Michael U. Rivera, "Immigration, Latinos, and White Partisan Politics: The New Democratic Defection", American Journal of Political Science, 58: 4 773-789, 2014.
23 Peter Beinart, "How the Democrats Lost their Way on Immigration", The Atlantic, July/August 2017. Accessed October 18, 2017. Thomas B. Edsall, "The Democrats' Immigration Problem", The New York Times, February 16, 2017. Accessed October 18, 2017. William A. Galston, "The Populist Moment", The Journal of Democracy, 28: 2: 21-33, 2017.
24 D. Sunshine Hillygus and Todd G. Shields, The Persuadable Voter: Wedge Issues in Presidential Campaigns, Princeton, N.J.: Princeton University Press, 2009.
25 Shad Turney, Frank Levy, Jack Citrin, and Neil O'Brian, "Waiting for Trump: The Move to the Right of White Working-Class Men, 1968-2016", The California Journal of Politics and Policy, Berkeley, Calif.: UC Berkeley, Institute of Governmental Studies, 2017. Accessed October 17, 2017.