There are nearly as many explanations for Donald Trump’s victory in the 2016 presidential election as there are electoral votes that Trump captured. Scholars have pointed to Trump’s ability to capitalize upon the resentment of rural voters (Cramer 2016), his use of the news media to perform key party coordination functions (Azari 2016), his hybrid-media campaign style of engaging in traditional press events and non-traditional “tweet storms” to generate coverage (Wells et al. 2016), his appeal to sexist (Wayne, Valentino, and Ocento 2016) and racist (Schaffner 2016) voters, his support from those who prefer authoritarian, nationalist leaders (Rahn and Oliver 2016) and the traditional “fundamental” explanations of elections (Masket 2016).
At the 2016 Elections Research Center Election Symposium at the University of Wisconsin-Madison in December, 2016, scholars from across the US added explanations of “what happened?” to the mix. Erika Franklin Fowler pointed out that Hillary Clinton followed a non-traditional television advertising strategy of emphasizing the personal characteristics of her opponent rather than her own policy positions, as most candidates have. Samara Klar revealed evidence that Independents who are high self-monitors refrain from visible, public displays of partisanship (i.e. yard signs) while continuing to vote in partisan ways. Katherine J. Cramer presented evidence from conversations, conducted after the election with Trump voters, revealing that many Trump supporters wanted to do something different at the ballot box, even if they do not expect the new president to be able to keep his promises. Young Mie Kim investigated the digital micro-targeting experienced by voters. Barry Burden presented evidence about how gender may have influenced the vote. Byron Shafer and Regina Wagner argued that the 2016 outcome was part and parcel of a long-standing pattern of election results going back decades.
We believe that these explanations are extraordinarily useful in helping scholars, journalists and the public unpack how Donald Trump captured the White House. Our own perspective, however, offers another explanation to consider: Trump held onto traditional elements of the Republican coalition while simultaneously appealing to populist voters in a way that modern Republicans have not been able to do. The evidence we present here, confined to analyses of the 2008 American National Election Study and the 2016 primary election season, build upon our work examining how the ideological heterogeneity of the American electorate helps us understand the conditional mass polarization of the American electorate (Carmines, Ensley, and Wagner 2011, 2012a,b, 2014). In short, we show that ideologically populist Americans – voters who have traditionally made up small portions of the Democratic and (especially) Republican electoral coalitions – have historically held issue preferences that matched the policy positions expressed by Donald Trump in the 2016 primaries. We further present evidence from the 2016 primary season that reveals populists were a far more important part of Donald Trump’s coalition than has been the case for Republican presidential candidates over the past half century.
Ideological Heterogeneity and Conditional Mass Polarization
It is well-established that the conflict space of American party elites is arrayed along a dominant left-right ideological dimension (McCarty, Poole, and Rosenthal 2006). Of course, the meaning of this ideological dimension has changed, as it has incorporated racial, social and religious issues into what was once primarily an economic dimension of conflict. Social and religious issues like abortion and prayer in public schools have not replaced economic issues.
Rather these newer cultural conflicts have joined economic issues in both broadening and re- defining the liberal-conservative dimension of American politics (Highton 2012). Geoffrey Layman and Thomas Carsey have referred to this transformation as “conflict extension” and its existence has been documented among strong partisans in the electorate, party activists and political elites (Layman and Carsey 2002a,b; Carsey and Layman 2006; Layman et al. 2010).
But unlike political elites, mass level preferences on policy issues have not collapsed onto a single liberal-conservative dimension. Instead, among the general public preferences on most salient domestic policy issues vary along not one but two related but separate dimensions, one defined mainly by economic and social welfare issues and the second by social, cultural and religious issues (Shafer and Claggett 1995; and Claggett and Shafer 2010; Carmines, Ensley, and Wagner 2012a). The issues that make up the first dimension are distributional in character; traditionally they focused on the size and scope of government, especially the role of the national governments in intervening in the economy and providing for the general welfare. More recently, these issues have centered on the extent to which government has a responsibility for reducing social and economic inequalities. Economic issues have been at the heart of liberal-conservative conflict in America since the founding but they took on renewed importance during the New Deal era, as the country struggled with what responsibility the national government had in alleviating the effects of the Great Depression. Even more recently, racial issues, which historically had formed a separate, second dimension of political conflict, have fused onto this dimension since many contemporary race-related issues are distributional in character (see Kellstedt 2003; Noel 2014).
The cultural, or social, dimension of conflict – issues that focus on the role of government in enforcing and regulating appropriate moral and social behavior – has also played an important role throughout American political history although only in the contemporary era has it been a defining issue cleavage separating the major parties. More frequently, cultural conflicts have played out within each party’s coalition, setting party factions against one another.
At various times, cultural issues have also been a key motivating factor in the launching of third party efforts. Historically, cultural conflicts primarily involved intra-party rather than inter-party conflicts.
What is distinctive about contemporary American politics, then, is not the existence or even the salience of cultural conflicts – issues such as women’s suffrage, public education and pornography have been contentious topics in American politics at various times – but that in the current era cultural issues have divided rather than cut across party lines. The 1980 presidential election marked a pivotal point in the partisan evolution of cultural conflicts; in the campaign the Republican Party made a major effort to attract the support of cultural and religious conservatives who had been largely apolitical until this point but have since become a core element in the GOP’s electoral coalition. Simultaneously, starting with the 1972 presidential campaign of George McGovern, cultural liberals and religious secularists have moved steadily into the Democratic Party and have become a significant part of the party’s coalition.
Hetherington and Weiler (2009) note that in recent years the rise of these social and cultural issues, which are broadly related to citizens’ support for authoritarianism, has created a deeper, more emotional feeling of polarization among supporters of the two major parties (see also Abramowitz 2013).
Since these two basic dimensions of political conflict have not collapsed onto a single liberal-conservative dimension at the level of the electorate as a whole this must mean that the citizenry is composed not just of liberals and conservatives who have consistently liberal or conservative positions on both economic and social issues – which may be termed the “main diagonal” of ideological conflict since it reflects the structure of political preferences found among party elites and the political class more generally.
Crucially, from our standpoint, the existence of the second dimension means that there are a significant number of citizens whose issue preferences place them in the “off diagonal” policy space, as their combination of policy preferences do not reflect the elite template. Citizens with heterodox policy preferences come in two varieties: those who have liberal positions on economic issues and conservative preferences on social issues or vice versa. We refer to the former as populists and the latter as libertarians (Carmines, Ensley, and Wagner 2012a,b). Populists and libertarians thus have a set of policy preferences that are opposite of one another; they are just as polarized from each other as liberals are from conservatives. But they are alike in one fundamental regard, a regard that sets them apart from liberals and conservatives: neither libertarians nor populists have policy preferences that align with the ideological divide represented by the two major parties. Quite the contrary, libertarians are closer to the Republican position on the economic issue dimension but closer to the Democratic position on social issues while populists are nearer the Democratic Party on economic issues but the Republican Party on social issues. Unlike liberals and conservatives, in other words, libertarians and populists are cross- pressured, having no clear ideological incentive to adopt the polarized beliefs and behaviors of either Republican or Democratic Party elites (see Klar 2014).
Ideological Heterogeneity and Support for Nationalist Public Policies
While most of the Republican candidates for president spent the 2016 primaries running traditional campaigns that had candidates working to raise money, take conservative positions on a variety of issues, advertise and seek elite endorsements, Donald Trump pursued an entirely different strategy. Rather than avoiding controversy, Trump courted it. Instead of working to earn the endorsements of members of the Republican establishment, Trump claimed that party elites were part of the problem. Azari (2016) has argued that Trump used the news media to disseminate his message in a way that helped him to perform vital coordination functions that would normally be conducted by a political party. As Wells et al., (2016) put it, Trump’s “mastery of conventional and digital media – hybrid campaigning – helped drive his coverage to the nomination” (p. 675).
One strategy Trump employed during these media appearances and tweets was to espouse issue preferences consistent with the support of nationalist policies – policies Trump argued were central to the ability to “make American great again.” Rahn and Oliver (2016) wrote in the Washington Post’s political science blog The Monkey Cage that Trump supporters had a strong national identity and supported authoritarian governing styles.
Support for nationalist policies is unevenly distributed across liberals, conservatives, libertarians, populists and moderates. To develop an understanding of the voters who might be most likely to favorably respond to Trump’s message, we examined data from the 2008 American National Election Study (ANES) on respondents’ opinions about four issues: (1) how likely is it that immigration takes away jobs; (2) should the government discourage companies from outsourcing; (3) favor the torture of terrorist suspects; and (4) do Blacks have too much influence on politics. To identify which ideological group a respondent belongs to, we followed the procedures outlined in Carmines, Ensley, and Wagner (2012b). 1 The data in Table 1 are the percentage of White citizens among each ideological type that support/agree with the position. We focus on White respondents given the Trump’s appeal is almost exclusively to those citizens. Table 1 also includes the ratio of Democratic to Republican Party identifiers. 2
White Americans’ Support for Nationalist Policies by Ideological Type, 2008.
|Immigration takes jobs||59%||36%||50%||35%||53%|
|Blacks have too much influence||16%||9%||9%||3%||19%|
|Ratio of D to R identifiers||55/27||43/42||49/38||73/14||16/75|
Data are from 2008 ANES. Cell entries are sample-weighted percentages.
Table 1 illustrates that White populists are the most supportive of the issues that we would associate with Trump’s candidacy; White populists are the most likely to think that immigration takes jobs, to agree that government should discourage outsourcing, and to favor the use of torture. And while Trump’s positions on these issues were attractive to populists, these citizens are the second most pro-Democratic Party group (55 percent identify as Democratic but only 27 percent identify as Republicans). Further, White populists are the second most likely ideological group to think that Blacks have too much influence on politics and thus may not be as bothered by Trump’s racially-oriented campaign (see Schaffner 2016).
We conducted similar analyses using data from the primary election season in 2016. To examine support for nationalist issues, we created a simple summated-rating scale using several issues questions in the 2016 ANES Pilot Study. The “nationalist” issues we examined are (higher values are assumed to be more nationalist):
- Support for allowing Syrian refugees (opposition is higher)
- Whether we should increase federal spending to fight crime (increase is higher)
- Concerns about a local terrorist attack (greater concern is higher)
- Support using troops to fight ISIS in Syria and Iraq (support is higher)
- Support for death penalty in murderer convictions (support is higher)
- Whether legal immigration is generally good or bad for the US (bad is higher)
- Support for an increase in legal immigration (opposition is higher)
Each issue is weighted equally in creating the scale, which is constructed to range between 0 and 1 with higher scores indicating greater support for nationalist policies. It is also worth noting that we used a slightly different approach to assign ideological categories to survey respondents given the paltry number of social/cultural issue questions (only equal pay for women, birth control, and support for businesses denying service to same sex couples were available) in the pilot study. 3 We also used feeling-thermometer questions about feminists, transgendered individuals, and homosexuals as indicators for the social issues scale.
Figure 1 shows that the highest level of support for nationalist policies in 2016 is among populists and conservatives. Liberals exhibit the lowest support for nationalist policies.
Donald Trump’s Appeal to Populists in the 2016 Republican Primary
On the one hand, these results suggest that populists were a group custom-made to respond to Trump’s candidacy. Unanchored in either major political party (Carmines, Ensley, and Wagner 2012b), they are prime candidates for “flanking” strategies aimed at picking up “off- diagonal” voters (Miller and Schofield 2003). On the other hand, our previous work has shown that for the past half century, populists have been a very small part of the Republican Party’s coalition.
Using Axelrod’s (1972) model, which calculates the contribution that different groups make to a political party’s electoral coalition, we estimated the size of each ideological group, as well as their propensity to turn out, their loyalty to a party, and their contribution to a party’s coalition. The group’s contribution is defined as the proportion of a party’s total votes provided by a given group and is based on the three components of the group: its size, turnout, and loyalty. Simply, a group’s contribution to the party’s coalition is greater if the group is large, its turnout is high, and its vote is lopsided in favor of one party. While Axelrod’s model initially was used to calculate the contribution of various demographic groups to the Democratic and Republican electoral coalitions, it can readily be applied to ideological groups. Table 2 shows that in 2012, populists were the smallest ideological group in the electorate and contributed a mere 5 percent to Mitt Romney’s coalition. Only liberals contributed less. Populists turned out less than any other group and were the second least loyal (liberals were the least loyal to the GOP) to the Republican Party. Table 3 shows that for Democrats, populists were the second most loyal but only the fourth largest contribution to Barack Obama’s coalition.
Size, Turnout, Loyalty, and Contribution to Republican Presidential Coalition, 2012.
Size, Turnout, Loyalty, and Contribution to Democratic Presidential Coalition, 2012.
Taken together, these results suggest that populists are a relatively small group that hold greater fealty for the Democratic Party than the Republican Party (see Carmines, Ensley, and Wagner (2014) for similar evidence from 1972 to 1992). It should not be surprising, then, that other GOP candidates did not make a strategic play for populists in the 2016 primaries.
However, in a field of more than a dozen candidates, each vying for the support of what is largely the same slice of the electorate, a play for populists begins to make more strategic sense. We also estimated the support among Republican identifiers, by ideological group, for each Republican candidate in the 2016 primaries. Perhaps not surprisingly, Republican populists were the group most likely to express support for Donald Trump in the ANES pilot. Moderates were the next highest Trump supporters. Notably, 30 percent of conservatives preferred Trump in the primary election season, even in an abnormally large field of conservative candidates.
Though our focus here has been primarily on populists, it is important to keep in mind that Trump did well amongst conservatives and that conservatives are by far the largest and most loyal group of Republican Party voters.
That said, we still wish to note that even after controlling for racial resentment and partisanship, populists were significantly more likely to support Donald Trump in the 2016 primaries. Figure 2 shows that probability that a populist in the Republican primary supported Trump was 42 percent, which is 10 percentage points higher than the next closest ideological group. Full regression results are available from the authors.
What are we as political scientists to make of the mass appeal and electoral success of Donald Trump? To be sure, he did not win the popular vote and takes office with historically low approval ratings. But at the beginning of the primary season few analysts gave Trump much of a chance to win the GOP nomination contest much less gain an Electoral College victory.
Trump’s support among Republican primary voters and probably in the broader electorate, we suggest, only makes sense once we recognize that the political choices offered by a conservative Republican Party and a liberal Democratic Party do not reflect the full extent of the ideological heterogeneity found in the American public. While there are millions of voters holding mainly conservative or liberal issue orientations there is also a sizable segment of the electorate, including self-identified Democrats, Republicans and Independents, whose issue preferences are neither consistently liberal nor conservative. Instead, their heterodox combination of economic and social-welfare issue preferences and their social and cultural preferences reflects a libertarian or populist ideological perspective. In other words, the marketplace of ideas found in the American public is much more varied and heterogeneous than that offered by the two major parties (Carmines, Ensley, and Wagner 2011, 2012a,b, 2014).
By strategic design or dumb luck, the Trump candidacy was able to activate a segment of the electorate that has historically not been part of the GOP electoral coalition. Compared to voters with liberal, conservative, moderate or libertarian views, those citizens holding populist opinions are not only the smallest slice of the electorate and the least likely to turnout, they have been notably disinclined to vote for Republican presidential candidates. But this may have changed in 2016. At least during the primary season Trump with his nationalist policy appeals was able to garner significant support among populist voters.
We do not yet have evidence of voting behavior in the 2016 general election. We do not know if populists increased their turnout, their loyalty to the Republican Party, or their contribution to the Republican coalition. The evidence from the 2016 primary elections suggest that Trump’s message may have appealed to populists in a way that could have fundamentally altered the electorate in states where White populists reside – states like Wisconsin, Pennsylvania and Michigan. We look forward to finding out.
Abramowitz, A. 2013. The Polarized Public: Why our Government is so Dysfunctional. London, UK: Pearson.
Axelrod, Robert. 1972. “Where the Votes come from: An Analysis of Electoral Coalitions, 1952–1968.” American Political Science Review 66: 11–20.
Azari, J. 2016. “How the News Media helped to Nominate Trump.” Political Communication 33: 677–680.
Carmines, E. G., M. J. Ensley, and M. W. Wagner. 2011. “Issue Preferences Civic Engagement, and the Transformation of American Politics.” In Facing the Challenge of Democracy: Explorations in the Analysis of Public Opinion and Political Participation, edited by Paul M. Sniderman and Benjamin Highton. Princeton, NJ: Princeton University Press.
Carmines, E. G., M. J. Ensley, and M. W. Wagner. 2012a. “Political Ideology in American Politics: One, Two, or none?” The Forum 10: 1–18.
Carmines, E. G., M. J. Ensley, and M. W. Wagner. 2012b. “Who Fits the Left-Right Divide? Partisan Polarization in the American Electorate.” American Behavioral Scientist 56: 1631–1653.
Carmines, E. G., M. J. Ensley, and M. W. Wagner. 2014. “Why American Political Parties can’t Get beyond the Left-Right Divide.” In The State of the Parties: The Changing Role of Contemporary American Parties, edited by John C. Green, Daniel J. Coffey, and David B. Cohen. Lanham, MD: Rowman & Littlefield.
Carsey, T. M., and G. C. Layman. 2006. “Changing Sides or Changing Minds? Party Identification and Policy Preferences in the American Electorate.” American Journal of Political Science 50 (2): 464–477.
Claggett, W. J. M., and B. E. Shafer. 2010. The American Public Mind: The Issues Structure of Mass Politics in the Postwar United States. New York: Cambridge University Press.
Cramer, K. J. 2016. The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker. Chicago: University of Chicago Press.
Hetherington, M. J., and J. D. Weiler. 2009. Authoritarianism and Polarization in American Politics. New York: Cambridge University Press.
Highton, B. 2012. “Sorting the American States into Red and Blue: Culture, Economics, and the 2012 U.S. Presidential Election in Historical Context.” Forum 10: 11–19.
Kellstedt, P. M. 2003. The Mass Media and the Dynamics of American Racial Attitudes. UK: Cambridge University Press.
Klar, S. 2014. “A multidimensional Study of Ideological Preferences and Priorities among the American Public.” Public Opinion Quarterly 78 (S1): 344–359.
Layman, G. C., and T. M. Carsey. 2002a. “Party Polarization and Party Structuring of Policy Attitudes: A Comparison of three NES Panel Studies.” Political Behavior 24 (3): 199–236.
Layman, G. C., and T. M. Carsey. 2002b. “Party Polarization and “Conflict Extension” in the American Electorate.” American Journal of Political Science 46 (4): 786–802.
Layman, G. C., T. M. Carsey, J. C. Green, R. Herrera, and R. Cooperman. 2010. “Activists and Conflict Extension in American Party Politics.” American Political Science Review 104 (2): 324–346.
Masket, S. 2016. “The case for ‘Normal’ Elections under a President Donald Trump.” Pacific Standard December 5, 2016. https://psmag.com/the-case-for-normal-elections-under-a-president-donald-trump-9058822b6839#.5b3qgp2fq.
McCarty, N., K. T. Poole, and H. Rosenthal. 2006. Polarized America: The Dance of Ideology and Unequal Riches. Cambridge, MA: MIT Press.
Miller, G., and N. Schofield. 2003. “Activists and Partisan Realignment in the United States.” American Political Science Review 97 (2): 245–260.
Noel, H. 2014. Political Ideologies and Political Parties in America. Cambridge, UK: Cambridge University Press.
Rahn, W., and E. Oliver. 2016. “Trump’s Voters aren’t Authoritarians, New Research Says. So what are they?” Washington Post, The Monkey Cage, March 9, 2016, https://www.washingtonpost.com/news/monkey-cage/wp/2016/03/09/trumps-voters-arent-authoritarians-new-research-says-so-what-are-they/?utm_term=.6c8e20d57723.
Schaffner, B. F. 2016. “White Support for Donald Trump was Driven by Economic Anxiety, but also by Racism and Sexism.” Vox, November 16, 2016. http://www.vox.com/mischiefs-of- faction/2016/11/16/13651184/trump-support-economic-anxiety-racism-sexism.
Shafer, B. E., and W. J. M. Claggett. 1995. The Two Majorities: The Issue Content of Modern American Politics. Baltimore: Johns Hopkins University Press.
Wayne, C., N. Valentino, and M. Ocento. 2016. “How Sexism Drives support for Donald Trump.” Washington Post, The Monkey Cage. October 23, 2016. https://www.washingtonpost.com/news/monkey-cage/wp/2016/10/23/how-sexism-drives-support-for-donald-trump/?utm_term=.91edbe8c4726.
Wells, C., D. V. Shah, J. C. Pevehouse, J. H. Yang, A. Pelled, F. Boehm, J. Lukito, S. Ghosh, and J. L. Schmidt. 2016. “How Trump Drove Coverage to the Nomination: Hybrid Media Campaigning.” Political Communication 33: 669–676.
A summary of the procedure is as follows. We identified survey questions that could be classified as economic/social-welfare issues and social/cultural issues in the 2008 ANES and performed multiple imputation to account for any missing data on the issue survey questions. We then performed a two-factor confirmatory factor analysis using the imputed data sets with the issue questions as indicators for the two dimensions. Based on the factor scores from the confirmatory factor analyses on the imputed data sets, citizens were classified as one of the five ideological types based on their location in the two-dimensional policy space. In all of these analyses, we utilized the sample survey weights.
Respondents that indicated they “leaned” towards one party are classified as partisans.
To locate respondents in the two-dimensional issue space, we handled missing data using confirmatory factor analysis estimated via maximum likelihood with missing data in Stata 12, as opposed to performing multiple imputation before the confirmatory factor analysis. This is the approach used by Layman and Carsey (2002a,b). The choice of how to handle missing data does not affect the results we present here.