�����Ƶ

The 2016 U.S. presidential primaries and general election left the unmistakable impression on many people that Americans were reverting to xenophobic, anti-immigrant sentiments of the kind not seen since the 1920s. In office, President Trump immediately began acting out those sentiments. In his first week in office, the new president issued executive orders directing Border Patrol officers to detain people at the Southern border of the United States and banning entry from seven countries.1 Two weeks later, he initiated the infamous practice of separating children from their parents, sometimes for months.2

Each action met with prompt and often organized opposition. Protests in the streets and motions in the courts halted or delayed the Trump administration’s immigration policies. Polls showed that ordinary people disapproved of the president’s actions. On inauguration day, 41 percent of Americans disapproved of Trump’s performance as president (already high by historical standards); after a month in office, the disapproval rate was 51 percent.3

Which represented American public opinion on immigration: the president’s policies or the demonstrations against them? To answer that, we turn to high-quality surveys of representative samples of Americans. Since 1994, the General Social Survey has asked American adults if they think immigration to the United States should be increased, decreased, or remain the same.4 Figure 1 shows the percentages of all residents, likely voters, and White likely voters saying immigration should be reduced or increased, by year.5 Social scientists have asked about immigration in ways both subtle and blunt, but this simple, direct approach works as well as any and better than most for quantifying positive and negative sentiments.6

Americans’ support for immigration grew from 1994 to 2018 as more residents, voters, and White voters said they thought immigration should be increased, while fewer said the government should reduce immigration. Reducing immigration was, by far, the most popular view in the 1990s; two-thirds of adults thought that was what the government should do. By 2018, roughly one-third (35 percent) held that view; 42 percent thought immigration should remain the same, while 22 percent supported an increase in immigration. The gap between “reduced” and “increased,” once fifty-eight percentage points, narrowed to thirteen percentage points in 2018.

Likely voters–and White likely voters in particular–could conceivably differ substantially from nonvoters.7 The dashed lines in the figure make clear, though, that support for increasing immigration included likely voters and even White likely voters. The changes did not depend on subpopulation; the trend lines for voters and White voters are very close to the trend lines for all residents. To avoid clutter, we do not show the redundant third alternative, “remain the same,” in the graph. The percentage saying “remain the same” rose from 28 percent to 39 percent between 1994 and 2010 and changed only slightly after 2010. Of the three options, the preference for increased immigration has been rising since 2012, while the preference to “remain the same” leveled off and the preference for reduced immigration waned.

American public opinion is notoriously partisan. Republicans and Democrats differ on many issues; liberals and conservatives differ on even more.8 Immigration is no exception. As Figure 2 shows, in 1994, White Republican likely voters were significantly more likely than Democrats to say that immigration should be reduced.9 In the intervening years, what was a gap grew into a chasm. Among Republicans, the percentage saying immigration should be reduced decreased from 76 percent in 1994 to 62 percent in 2000, a significant fourteen-point drop in just six years; it has not changed significantly since then, though, still standing at 62 percent in 2018. White Democrats became less and less likely to endorse reducing immigration, accelerating downward just as Republicans leveled off; by 2018, only 21 percent of White Democrats said immigration should be reduced. From 1994 to 2010, Democrats and independents held very similar views; as Democrats accelerated their shift from opposition to support for immigration in recent years, independents changed more slowly.

We began with a question of whether the public supported President Trump on immigration in the first two years of his administration. The partisan gap in Figure 2 implies “it’s complicated.” Republicans supported reducing immigration as much in 2018 as they did in 2000, so the Trump administration anticipated correctly that their Republican base would approve when they tried to reduce immigration. Protests erupted because the overwhelming majority of Democrats, including almost four out of five White Democrats, either wanted to continue immigration at current levels or see it increased. On many other contentious issues, Republicans moved right over time, while Democrats stood still or moved right more slowly,10 creating partisan sorting or polarization. In the case of immigration, the polarization developed when Republicans stood still and Democrats moved left.

White racial resentment has been another feature of the Trump movement. Three major ethnographies of White Americans’ political perspectives before the 2016 election identified the politics of resentment as the key to politics in the Obama years. And Trump’s success in the election owed a debt to resentment, even though scholars debate the relative contributions of race, economics, and sexuality to the outcome of the 2016 election.11

Sociologist Arlie Hochschild’s Strangers in Their Own Land describes how Louisiana Tea Party activists complained to her that opportunity in America had become a line that stretched on and on.12 People like them were stuck in place, they said, because “line cutters”–Blacks and immigrants–were cutting into the line ahead of them. They resented the special treatment.

Political scientist Katherine J. Cramer’s ethnography of Scott Walker supporters in Wisconsin, The Politics of Resentment, found even stronger resentments.13 There the axis was described to her as rural-urban, but it was patently clear that “urban” was a euphemism for Black. Walker’s small-town base also resented the cosmopolitan elites in Madison and Milwaukee who promoted the cause of the urban poor at the expense of their struggling rural communities. In an appendix, Cramer debunked many of her subjects’ claims about state appropriations. But people spread the false analysis of the state budget because they believed it had to be true.

For their book on The Tea Party and the Remaking of Republican Conservatism, sociologist Theda Skocpol and governance scholar Vanessa Williamson followed Northern and Midwestern Tea Party members, attending meetings and rallies and conducting open-ended interviews with individuals and groups.14 They found the same racially tinged resentment, in their case in the “take our country back” rhetoric they heard in meetings and at rallies. Without saying out loud that they were taking the country back from Blacks and immigrants, Tea Partiers made clear that they were White people taking something back from non-Whites in the strong language they used to describe President Obama and other Democrats.

Racial resentment has also been a staple of quantitative political research.15 Among the many ways researchers measured racial attitudes over the years, items that probed racial resentments have special efficacy.16 Several versions of the scale can be found in the literature;17 we use an adaptation based on items asked in the General Social Survey. The key item in all of these scales is:

Do you agree strongly, agree somewhat, neither agree nor disagree, disagree somewhat, or disagree strongly with the following statement: Irish, Italians, Jewish and many other minorities overcame prejudice and worked their way up. Blacks should do the same without special favors.

Some scholars have questioned whether this item blends race with ideology, psychology, or other confounders. If so, it would complicate interpretation. Recent research by social psychologist Alicia Simmons and social scientist Lawrence Bobo used follow-up questions to better understand Americans’ responses to these questions.18 Their careful parsing of answers to the questions that make up the scale resolve the most important ambiguities in the language of the questions and make clear that conservative and liberal Americans hear the questions similarly and answer in ways that reflect their views on whether Blacks experience racial discrimination today and whether offsetting it should be a priority for public policy.

To create a racial resentment scale, we combined this key item with two others. We gave respondents one point on racial resentment if they agreed somewhat or two points if they agreed strongly with this statement; we subtracted one point if they disagreed somewhat with the statement or two points if they disagreed strongly. We then added another point if they said racial differences in income and poverty come about because “Blacks just don’t have the motivation or will power to pull themselves up out of poverty,” and subtracted a point if they identified racial discrimination as a source of Black-White differences. The resulting scale had a rather nonintuitive range from –3 to +3, so we used a simple linear transformation to put it on a scale that hypothetically runs from zero to one hundred–think of it as analogous to converting degrees Fahrenheit into degrees centigrade.19 Figure 3 shows the results for White likely voters, the subpopulation most at risk of racial resentment.20

Whites’ racial resentment in 1994 was very high: seventy-four among Republicans and sixty-four among Democrats, with independents in between. As with immigration attitudes (see Figure 2), little happened for ten to twelve years, then Democrats began a period of rapid change. By 2018, White Democrats’ score on racial resentment had dropped to forty, while White Republicans’ score held steady (the change of less than a point was not statistically significant).

These quantitative results confirm the political edge to racial resentment found in the qualitative studies,21 but fail to confirm that racial resentment swelled on the right. America’s major parties grew further apart on racial issues because, through the Obama and Trump presidencies, White Democrats gave up their racial resentment.

The racial resentment scale has three elements: the beliefs that Blacks should not get “special favors,” that Blacks lack the will to climb out of poverty, and that discrimination is not important for the Black-White gap. We checked to see if any one of these elements drove the trend among White Democrats while the other two hovered unchanged. The item-by-item trends (not shown) were steepest for the core resentment element (no “special favors”), less so for the individualistic component (“lack will”), and least for discrimination. In short, the results in Figure 3 would look even more dramatic if we used a single item instead of the three-item scale. Further analysis showed that White liberals, a subpopulation that overlaps with White Democrats but not perfectly, reduced their racial resentment even more than White Democrats did.22

Our data were collected over many years, but the samples did not carry over from year to year. In short, we are inferring changes in people’s attitudes from the difference between them and people like them who were interviewed in previous years. We cannot rule out the possibility that some of the patterns we have discussed reflect a party-switching dynamic whereby racially resentful Democrats quit identifying with the Democrats and took their resentments with them to the independent or Republican column. But though we cannot rule out that possibility, we regard it as unlikely. First, there were more Democrats, not fewer, over time. Second, we can show that Democrats who were too young to vote in the 1990s hold significantly less racial resentment than Democrats who were already in the electorate in the 1990s. That new-voter dynamic is consistent with our interpretation of the trends.

To this point, we have shown substantial partisan rifts in White voters’ views of immigration and racial resentment. Both became larger recently and the underlying dynamic of each has been a shift to the left among White Democrats. Race has been a component of every immigration debate in American history, of course.23 The open question is the extent to which White Americans link them in their minds today.

A correlation coefficient is an old and popular statistic that compares covariation between two scores with the variation in the two scores separately. In theory, a correlation coefficient can take any value from 1, the maximum, to –1, the minimum: 1 means that if one of the variables changes, the other changes perfectly proportionately in the same direction; 0 means that one is perfectly independent of the other; and –1 means that if one variable changes, the other changes perfectly proportionately in the opposite direction. We calculated the correlation coefficient between peoples’ immigration preference and their racial resentment for each General Social Survey year that included both measures. The results appear in Figure 4.24

The correlation between immigration preferences and racial resentment was significant in every year. The steady correlation of 0.30 throughout the 1990s and early 2000s was impressively strong by the standards of opinion data of this sort. The rise from 0.30 to 0.50 by 2018 indicates an uncommonly strong relationship. Uncertainty in the data makes it hard to say whether the upward trend started as early as 2006 or as late as 2012. The smoothed trend line supports the idea of small increases followed by an acceleration. The relatively low data point in 2012 hints that maybe change came later and more dramatically. Either way, the increase between 2010 and 2018 is statistically significant by any standard.

Thus, every measure we have indicates that Whites’ views of immigration are closely tied to their views of race. Given the large partisan differences in each opinion (see Figures 2 and 3), the growing correlation might be just another manifestation of partisan sorting. Further analysis (not shown) makes clear that the connection goes beyond partisan sorting. Statistical adjustments for the common dependence on political party identification reduced the net correlation between immigration and racial resentment to 0.40, indicating that partisan sorting accounted for half the increase.25 In recent years, the correlation between immigration preferences and racial resentment has been stronger for Democrats than for Republicans. Thus, the trend is driven by Democrats connecting support for immigration with discrimination and exclusion of Blacks more than by Republicans connecting non-White immigrants with their own racial resentments.

The partisan sorting of immigration and racial attitudes implies that people’s votes will align with their views on these issues. But we cannot say from evidence already in hand how strongly aligned votes were with attitudes or if the alignment is spurious: that is, whether the alignment goes away when we statistically control for other factors that are known to be important predictors of how people vote. So we built a comprehensive multivariate model of vote choice in the 2016 general election and the Republican primaries that year. The variables in the model are listed in the caption of Figure 5; details of our analysis are available in our accompanying SocArXiv paper. As in the preceding analyses, we focus attention on non-Latinx Whites. We switch data sets for this part of the analysis. Up to this point, we have shown results from the General Social Survey. Here we show results from the Cooperative Congressional Election Survey (CCES) because it has far more cases and includes cases from every state. The CCES data also have multiple measures of a conservative approach to race. We use that richness and reflect the switch by referring to conservative racial attitudes when referring to results from the CCES data.

People who favored reducing immigration were significantly more likely to vote for Donald Trump in 2016–in the general election and even more so in the Republican primaries–than were people who favored increasing immigration (see Figure 5). In the general election, 45 percent of Whites at the seventy-fifth percentile of opposition to immigration voted for Trump (remember this marginal percentage removes other factors, including conservative racial attitude, by statistical adjustment); just 36 percent of otherwise identical Whites at the twenty-fifth percentile did, a marginal difference of nine percentage points. The marginal difference in the primaries was twelve percentage points.

Just as strong was the tendency for those who expressed conservative racial views to vote for Trump in the general election; the marginal difference between Whites at the seventy-fifth and twenty-fifth percentiles of racial attitudes was also nine percentage points. The racial-conservatism gradient was less steep in the Republican primaries, mainly because the few White Republicans at the twenty-fifth percentile of the racial attitudes scale had a relatively high probability of voting for Trump for other reasons; the marginal difference was seven percentage points.

The combination of opposing immigration and taking a conservative position on race is a potent expression of the nativist perspective. It is reasonable to ask if the two views compound each other or if having one view or the other is sufficient. We tested that idea statistically by adding an “interaction” between these two factors to our model. In most tests, the interaction was statistically significant but negative. Thus, one or the other sufficed. Either opposing immigration or conservative racial attitudes were sufficient to drive a White voter to Trump. If either view was already strongly conservative, an increase in the other added less to Trump support than if the first view was moderate.

These relationships held when we limited the sample to the swing states.26 This is the third replication of a basic pattern that implies that Trump’s candidacy exerted a strong appeal based on immigration and racial resentment, beyond the usual appeal these factors had for Republican candidates in previous elections or for Republican Party identification. First, in 2016 and 2018, the correlation between opposition to immigration and racial resentment was higher than before and almost as strong with controls for party as without (r = 0.5 without controlling for party and r = 0.4 after controlling for party). Second, marginal differences were as big in the primaries (where all candidates are Republicans) as in the general election (opposition to immigration actually had a somewhat higher marginal difference in the primary results than in the general election, but it was the other way around for racial resentment). Third, the marginal differences in 2016 were as big in swing states as in the nation at large. Together these findings convinced us that opposition to immigration and racial resentment tap into something about Donald Trump’s specific appeal, beyond usual Republican support.

For additional clues about Trump’s candidacy, we separated the items in the CCES racial resentment scale and refit our voting model with one racial resentment indicator at a time.27 Generally speaking, the separate racial items predicted Trump voting at a statistically significant level.28 We note two exceptions. First, people who often felt fearful of other races were almost as likely to vote for Trump versus Clinton in the general election as people who seldom felt fearful, but were more likely to vote for Trump versus all candidates and all options including nonvoters, as well as in the primaries. Second, in the Republican primaries, Trump voters were actually less likely to say that racism is rare, highlighting the more complicated task of distinguishing Trump support from more general Republican orthodoxy in certain respects.

Immigration and race are as entwined now as they have been throughout American history. Their tangle is the organizing principle of this issue of ��æ�岹���ܲ�. Other contributors to this issue have been charged with explicating the history and current manifestations of how immigration and race move together. Our part has been to document how immigration and race shape current politics.

The hallmark of this era is polarization, a combination of strong views, nearly even divide, and an unwillingness to compromise. It is not exactly a pair of opposed silos that agree completely within and disagree completely with the other side. Political scientists have thoroughly debunked that one.29 The mass public is just too hard to constrain.30 American polarization is more subtle. It reflects “partisan sorting” in which views that despite being weakly correlated with one another are nonetheless each correlated with party identification and liberal-conservative ideology.31 Informed by this literature, we approached the politics of immigration and racial resentment expecting to see patterns similar to what data on, say, abortion and taxes reveal. Republicans tend to oppose both abortion and taxes; Democrats tend to support both abortion rights and spending for social programs. But the correlation between attitudes toward abortion and taxes (approximately 0.1) is weak compared with how each variable correlates with political party identification (0.2 for abortion and 0.4 for opposing social spending). This is what scholars mean by weak constraint but partisan sorting among issues.32

In fact, the General Social Survey data reveal more constraint between immigration and racial resentment than between other contentious issues. Among White likely voters, opposing immigration had a correlation with racial resentment of 0.3 in the late 1990s and 0.5 in 2018 (see Figure 4). Meanwhile, the patterns in Figures 2 and 3 translate to correlations of 0.15 in the 1990s and 0.5 in 2018. In short, opposition to immigration and racial resentment showed more constraint than partisan sorting until relatively recently. How can we account for the unusual overlap?

As other essays in this volume make clear, immigration has always been racially inflected. Whites from Spain, England, and France encroaching on, displacing, and colonizing indigenous peoples; Irish, Italians, Jews, Chinese, and others being excluded as “alien races”; Congress imposing quotas in 1924 that set the clock back to 1880. These actions all racialized immigration in different ways.33

More proximately, the survey question that anchors the racial resentment scale ties it to immigration by explicitly (and incorrectly) stipulating that White groups “worked their way up without special favors.” Of course, a naive reading of that question might suggest that in referring positively to immigration, the item encourages a correlation with supporting immigration. But that is not how nativism works.34 The sentiment in the question invokes what the late sociologist Stanley Lieberson called “the great non sequitur”:35

It is a serious mistake to under-estimate how far the new Europeans have come in the nation and how hard it was, but it is equally erroneous to assume that the obstacles were as great as those faced by blacks or that the starting place was the same.36

But the item was not designed to tap into the survey respondent’s ability to reason from evidence. When they penned the question, political scientist Donald Kinder and psychologist David Sears were thinking about “deep-seated feelings of social morality and propriety and in early-learned fears and stereotypes.”37

It is how contemporary race scholars have come to understand the persistence of racism into the twenty-first century.

Although racial affairs cannot be properly understood without a structural perspective on racism, I no longer regard racial domination as just a matter of presumably objective practices and mechanisms driven by the socioeconomic material interests of actors. Racial actors, both dominant and subordinate, simply cannot transact their lives without racialized emotions.38

But deep roots do not imply stasis. In fact, our research uncovered substantial change in a racially progressive direction. We began by noting Americans in general, likely voters, and even White likely voters were becoming more open to immigration in recent years. The anti-immigration agenda of the Trump administration is not responding to the public will; it pushes back against it. The administration was not totally without public support, of course. While most Americans were moving left on immigration, Republicans held their position on the right (not that they moved further right, but held their position on the right). In the presidential primaries of 2016, Republicans who opposed immigration supported Trump. And he has delivered much of what they hoped for. Their sense of urgency and penchant for strong language and false claims stem from both their minority status and their sense of being “strangers.”39

The evidence also shows that racial resentment is waning, with an identical partisan inflection. Republicans expressed significantly more racial resentment than Democrats did in the 1990s. But after Obama’s election–or perhaps Michael Brown’s murder in Ferguson, Missouri–Democrats, and White Democrats, in particular, expressed less and less racial resentment (see Figure 3).

Partisan sorting on immigration and racial resentment is usually interpreted as working to the advantage of Republican candidates. Most social scientists have read the rising predictive power of racial resentment as confirmation that Trump succeeded by giving up on the “dog whistles” of the past and expressing his constituents’ resentments in stark terms. But that conflates the election outcome with the direction of the underlying trends in public opinion. Racial resentment is more predictive because Democrats have staked out a left side of the issue. In the last twenty years, White Democrats dropped from a score of sixty-two on the racial resentment scale (which ranges from zero to one hundred) to forty-one: twenty-one points in twenty years. In the same time period, Republicans dropped just two points. That is the partisan sort that gives racial resentment its predictive power. As White Democrats let go of their racial resentment, racial attitudes come to predict voting outcomes better. For Democrats, racial disparities are no longer about their fellow citizens’ personal failings; they are about ending discrimination, even if it means direct racially targeted acknowledgment and remediation.

authors’ note

Michael Hout presented a preliminary version of this essay at the annual meeting of the National �����Ƶ of Sciences in Washington, D.C., on April 29, 2018. We thank Arlie Hochschild, Paula England, Donald Kinder, and Larry Bobo for useful comments and discussion.

Endnotes

• 1The White House, “Executive Order: Border Security and Immigration Enforcement Improvements,” EO 13767, January 25, 2017; and The White House, “Executive Order Protecting the Nation from Foreign Terrorist Entry into the United States,” EO 13769, March 6, 2017.
• 2The White House, “Presidential Executive Order on Enforcing Federal Law with Respect to Transnational Criminal Organizations and Preventing International Trafficking,” EO 13773, February 9, 2017.
• 3Nate Silver, “” FiveThirtyEight, updated December 23, 2020.
• 4See the General Social Survey, NORC at the University of Chicago, https://www.gss.norc.org. Question wording evolved over time. We combined all three versions for this figure and the others in this essay. For readers interested in the technical details of our analysis, we posted a longer, more technical version of on SocArXiv; it includes the Stata code as a supplement.
• 5The most recent data are from 2018; the GSS skips odd-numbered years, and the 2020 GSS did not occur in time to be included in this essay.
• 6Jens Hainmueller and Daniel J. Hopkins, “,” Annual Review of Political Science 17 (1) (2014): 225–249.
• 7We define “likely voters” as people who told the GSS that they voted. They constitute about 70 percent of all GSS respondents, compared with 50 percent turnout in most U.S. presidential elections between 1992 and 2016. So either some nonvoters erroneously claim to have voted or GSS nonrespondents were mostly nonvoters. We cannot resolve that uncertainty here, so we refer to this group as “likely voters.” When we say “Whites,” we mean non-Hispanic Whites.
• 8Delia Baldassarri and Barum Park, “Was There a Culture War? Partisan Polarization and Secular Trends in U.S. Public Opinion,” Journal of Politics (forthcoming); Claude S. Fischer and Greggor Mattson, “” Annual Review of Sociology 35 (1) (2009): 435–455; and Donald R. Kinder and Nathan P. Kalmoe, Neither Liberal nor Conservative: Ideological Innocence in the American Public (Chicago: University of Chicago Press, 2017).
• 9When we say “significantly” we mean statistically significant at the conventional p < 0.05 level.
• 10Jacob S. Hacker and Paul Pierson, Off Center: The Republican Revolution and the Erosion of American Democracy (New Haven, Conn.: Yale University Press, 2005).
• 11Lawrence D. Bobo, “Racism in Trump’s America: Reflections on Culture, Sociology, and the 2016 U.S. Presidential Election,” The British Journal of Sociology 68 (S1) (2017): S85–S104.
• 12Arlie Russell Hochschild, Strangers in Their Own Land (New York: The New Press, 2016).
• 13Katherine J. Cramer, The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker (Chicago: University of Chicago Press, 2016).
• 14Theda Skocpol and Vanessa Williamson, The Tea Party and the Remaking of Republican Conservatism (Oxford: Oxford University Press, 2015).
• 15Michael Tesler, Post-Racial or Most-Racial? Race and Politics in the Obama Era (Chicago: University of Chicago Press, 2016).
• 16Lawrence D. Bobo, Camille Z. Charles, Maria Krysan, and Alicia D. Simmons, “The Real Record on Racial Attitudes,” Social Trends in American Life: Findings from the General Social Survey since 1972, ed. Peter V. Marsden (Princeton, N.J.: Princeton University Press, 2012), 38–83.
• 17Tesler, Post-Racial or Most-Racial?
• 18Alicia D. Simmons and Lawrence D. Bobo, “Understanding ‘No Special Favors’: A Quantitative and Qualitative Mapping of the Meaning of Responses to the Racial Resentment Scale,” DuBois Review 15 (2) (2018): 323–352.
• 19New score = 90/6Old Score + 50.
• 20The GSS racial resentment scale is an adaptation of more commonly used forms of the scale found in the American National Election Studies (ANES). The ANES data show the same patterns for 1992–2016 as we see in Figure 3 for 1994–2018 (without the very low 2018 data point for Democrats). The Cooperative Congressional Election Study (CCES) has a distinct measure we use in the current study.
• 21Hochschild, Strangers in Their Own Land; Cramer, The Politics of Resentment; and Skocpol and Williamson, The Tea Party and the Remaking of Republican Conservatism.
• 22We classify people by political party instead of by political ideology because Donald R. Kinder, one of the first political scientists to write about racial resentment, does not trust Americans to answer the political ideology question accurately. Donald R. Kinder and David O. Sears. “Prejudice and Politics: Symbolic Racism versus Racial Threats to the Good Life,” Journal of Personality and Social Psychology 40 (1981): 414–431; and Kinder and Kalmoe, Neither Liberal nor Conservative.
• 23Richard Alba, Ethnic America (New Haven, Conn.: Yale University Press, 1990); Alejandro Portes and Rubén G. Rumbaut, Immigrant America: A Portrait,4th ed. (Berkeley: University of California Press, 2014); and Robert N. Bellah, Broken Covenant (Chicago: University of Chicago Press, 1974).
• 24In these computations we included all non-Hispanic White respondents, not just the likely voters, for two reasons. First, if views on either immigration or race affect the likelihood of voting, we could misstate the correlation by excluding nonvoters. Second, the questions were only asked of subsets of GSS respondents; excluding nonvoters left fewer than five hundred cases in some years, leading to imprecise estimates.
• 25Our statistical adjustment consisted of regressing immigration preferences and racial resentment scores (separately) on political party for each year, saving the residuals, and repeating the calculations in Figure 4 for those residuals. The patterns are so similar that the two trend lines would be hard to discern if we put them both on the chart. Thus, we concluded that the trend in Figure 4 is not attributable to partisan sorting.
• 26Nate Silver, “,” FiveThirtyEight, October 31, 2016.
• 27Again, the details of this analysis are not shown here due to space constraints but are available in our SocArXiv paper.
• 28This is a standard expectation of classic measurement theory, which holds that more items imply less measurement error. Measurement errors in predictors bias estimates toward zero.
• 29Morris P. Fiorina and Samuel J. Abrams, Disconnect: The Breakdown of Representation in American Politics (Norman: University of Oklahoma Press, 2011).
• 30Philip E. Converse, “The Nature of Belief Systems in Mass Publics,” Ideology and Its Discontents, ed. David E. Apter (Glencoe, Ill.: Free Press, 1964), 206–261.
• 31Delia Baldassarri and Andrew Gelman, “,” American Journal of Sociology 114 (2) (2008): 408; and Andrew Gelman, Red State, Blue State, Rich State, Poor State (Princeton, N.J.: Princeton University Press, 2009).
• 32Baldassarri and Gelman, “Partisans without Constraint.”
• 33Claude S. Fischer and Michael Hout, Century of Difference (New York: Russell Sage Foundation, 2006).
• 34Bellah, Broken Covenant.
• 35Stanley Lieberson, A Piece of the Pie: Blacks and White Immigrants since 1880 (Berkeley: University of California Press, 1981).
• 36Ibid., 383.
• 37Kinder and Sears, “Prejudice and Politics,” 416.
• 38Eduardo Bonilla-Silva, “Feeling Race: Theorizing the Racial Economy of Emotions,” American Sociological Review 84 (1) (2019): 1–25.
• 39Hochschild, Strangers in Their Own Land.