Educational Attainment and Social Norms of Voting

Eric R. Hansen, Loyola University Chicago

Andrew Tyner, Center for Open Science

Why are more educated Americans more likely to vote? It seems like a question that political scientists would know the answer to, but definitive answers to this question are surprisingly hard to come by. In this article, we focus on one possible link between education and voting. We argue that education normalizes voting behavior. In other words, we think that people who attend school for longer are more likely to vote because they think that voting is “the right thing to do” or that it’s a person’s “civic duty.”

We offer three supporting pieces of evidence. First, we use data from the 2016 American National Election Study (ANES) and the 2016 Cooperative Congressional Election Study (CCES) to show that a sense of civic duty mediates the effect of education on validated voter turnout. That is, we show that at least a small part (we estimate about 14%) of the reason that educated people vote more often is because they feel socially obligated to.

Next, we study a behavior that public opinion researchers refer to as “overreporting.” It’s common for 90% or more of respondents in nationally representative surveys to tell researchers that they voted in the last presidential election. But some of those respondents must be mistaken, if not outright lying, since actual voter turnout has hovered around 60% in presidential elections for the last century. If more educated people feel like voting is “the right thing to do,” then they should also be more likely to misreport voting to researchers when their behavior doesn’t live up to their own expectations. Using validated voter turnout data from ANES and CCES, we show that college-educated Americans are significantly more likely than others to tell researchers they voted in the last election when they didn’t vote in reality.

Finally, we present the results of a survey experiment that asked respondents to play a game of “Would You Rather?” In a control condition, we asked respondents how likely they were to vote in 2018. In a treatment condition, we asked respondents how likely they were to vote if they had to choose between voting and accepting a $500 cash prize. Unsurprisingly, respondents offered a chance at $500 were much less likely to choose voting. However, college-educated respondents were less likely than others to choose the money over voting, presumably because the respondents with more education thought that choosing the money wouldn’t be “the right thing to do.” (Notably, it wasn’t because the people with less education needed the money more—high-income people, regardless of education, were more likely than low-income people to choose the money over voting.)

One limitation of our study is that we aren’t able to show that education itself normalizes voting. It’s possible that people who think voting is a civic duty also happen to be more likely to decide to spend more years in school. However, we contribute evidence suggesting that the way Americans see the social norms around voting differ by their level of education.

Feeling the Urge to Act: Political Ads, Arousal, and Emotion


Kristyn L. Karl, Stevens Institute of Technology

If you haven’t seen any political ads lately, fear not, lots of them are likely just around the corner. Thanks in part to staggered election cycles, primaries, and the growth of ballot propositions, political advertisements are everywhere. If the thought of more ads makes you shudder, you are not alone.

Most of us understand how a political ad can lead to an emotional response: the attack ad against your preferred candidate might leave you feeling angry, for example. But political ads also lead to physiological responses. In other words, our bodies react to political ads with changes in arousal (as they do with most stimuli). In some cases, you may recognize if a political ad gets your heart racing and palms sweaty, but, for the most part, it is unlikely that you’d notice the changes in your arousal. Despite going unnoticed, however, these small changes inside your body can impact your behavior.

Measures of self-reported emotion rely on asking people how they feel. This is a sensible approach, as it has a high degree of face validity and has produced meaningful findings in political science research. Due in part to the growth of experimental research in political science, we have evidence that advertisements and the emotions they evoke are capable of influencing voter behavior. Specifically, there is much recent work demonstrating the influence of emotion as a short-term motivation for different forms of political participation.

Yet research in psychology has long demonstrated that affect is often experienced quickly and without conscious awareness. With numerous scholars expressing interest, and in some cases concern, over the impact of televised campaign ads on participation, it is vital that our understanding of the effects of political advertising be based on sound assumptions. How does reliance on self-reported measures of emotion impact our understanding of the short-term motivational forces that impact political participation?

This study investigates three key questions. First, does a direct, real-time measure of emotional arousal in response to political ads predict participation? Second, are the two elements of unconscious response (arousal) and conscious response (self-reported emotion) distinct and unique in their explanatory power? And third, to what extent does emotional arousal translate into self-reported emotion? In other words, what is the relationship between the two?

Using a randomized experiment with carefully manipulated campaign advertisements, I find evidence that an alternative measure of emotional response, physiological arousal, is a powerful predictor of citizens’ willingness to participate in politics. Furthermore, arousal is not simply a proxy for self-reported emotion, but rather, a different and complementary measure of the emotional experience. This research makes clear that both physiological arousal and self-reported emotion are motivational and that each captures a different piece of the emotional experience.

I also explore the relationship between arousal and self-reported emotion and find evidence that the translation of arousal into self-reported emotion depends in part on characteristics of the message such as partisan tone and on characteristics of the individual such as political knowledge. Through the use of unconscious physiological indicators of arousal, the results call attention to the need for a closer examination of what self-reported emotions capture, as well as for greater inclusion of additional measures of emotional response in political science research.

In sum, the impact of arousal on participation is short-term yet direct. In the real world, it is likely that a political stimulus such as a single ad, a conversation, or a newspaper article, will have an immediate effect that quickly dissipates. But in many instances, a political stimulus may be encountered at a key moment. For example, an online ad might solicit an immediate donation to a campaign or an arousing political conversation with a canvasser might increase the likelihood a citizen signs their petition on the spot. This research makes it clear that political participation is motivated by the entirety of the emotional process, which includes both unconscious arousal and what people say they feel.





The Activation of Prejudice and Presidential Voting

Dan Hopkins

As a candidate for President, Donald Trump defied recent norms for presidential candidates with several racially charged statements. Did that rhetoric—and the 2016 campaign more generally—affect voters’ support for Trump in his general-election contest with Hillary Clinton? Given prior research on prejudice and priming, a few research questions loom especially large: did the 2016 campaign activate white citizens’ prejudice? If so, did it particularly activate attitudes toward specific minority groups?

To address those questions, my forthcoming Political Behavior article “The Activation of Prejudice and Presidential Voting” uses a population-based panel survey to examine the predictive power of prejudice and other factors in explaining 2016 support for Trump versus Clinton. The panel survey, conducted by Knowledge Networks/GfK, began tracking U.S. citizens’ social and political attitudes in late 2007, allowing me to examine the relationship between various attitudes and support for Trump over Clinton years later. This lag, in turn, helps us address the challenge of what social scientists term “endogeneity”: while people who hold certain attitudes may seek out candidates who share those views, citizens may also adopt the views held by candidates whom they already support. Panel data lets us observe attitudes years before the candidates emerge, and so to minimize this threat.

Empirically, the article focuses on non-Hispanic white citizens. I estimated statistical models of 2016 vote choice which accounted for a range of demographic factors as well as select attitudes measured in or before 2012. Among the control variables are indicators for vote choice in 2012, meaning that these models estimate the shift in vote choice from the 2012 baseline. The key independent variables are anti-Black and anti-Hispanic prejudice, measured via adherence to in-group and out-group stereotypes. The results show that anti-Black prejudice measured in 2012 is a consistent predictor of which voters shift toward or away from Trump in all three 2016 survey waves. In January 2016, for example, a respondent whose 2012 anti-Black prejudice placed her at the 80th percentile supported Trump at a rate 5.7 percentage points higher than an otherwise similar individual at the 20th percentile. These effects are driven both by low-prejudice white citizens shifting away from the GOP and high-prejudice white citizens shifting toward it. This basic pattern proves quite robust, reappearing in a wide variety of different statistical models. But similar models detect no such effect for the 2012 election.

In the January 2016 panel wave, respondents were asked not only about a Trump-Clinton match-up but also about hypothetical general elections pitting Clinton against either Marco Rubio or Ted Cruz. In both cases, the effect of anti-Black prejudice is not distinguishable from zero, suggesting that even when citizens are asked about candidates at the same time and in the same information environment, their responses to different candidates hinge on those candidates’ rhetoric.

Extensive research has examined racial priming, but there is much more limited research on what types of attitudes can be activated. The statistical models uncover no similar effect for anti-Hispanic prejudice, a noteworthy fact given Trump’s rhetoric targeting unauthorized immigrants from Mexico. However, they do find that 2012 anti-Black prejudice was predictive of backing Trump versus other Republican candidates in the 2016 primaries.

Presidential vote choice is the product of various factors, and so explanations for any given election are necessarily multi-faceted. Still, these results do show that anti-Black prejudice was activated by the choices provided by the 2016 election. Even during an election campaign whose rhetoric focused on immigrants, Muslims, and other groups, it is striking that anti-Black prejudice was one robust predictor of vote choice. These results are consistent with the prospect that the way in which voters respond to immigration and related rhetoric may hinge on longstanding political divisions.



The Higher Power of Religiosity Over Personality on Political Ideology


Aleksander Ksiazkiewicz, University of Illinois at Urbana-Champaign

Amanda Friesen, IUPUI

What are the origins of political attitudes and religious orientations, and why do they correlate? For much of the history of political science, the answers to these questions involved individual socialization, elite discourse, and, sometimes, individual dispositions like personality traits. From the socialization perspective, families, historical contexts, institutions and culture drive the interest, motivations, identities, and information people learn, which in turn shape the attitudes they hold, and ideologies to which they ascribe. Alternatively, individual-level traits alter the environments individuals select and guide them to differentially perceive similar environments, thus shaping the attitudes they hold. In addition, even today personality, religion, and politics are typically studied in dyads, leading to an incomplete picture of how these domains intersect.

In this article, we used behavioral genetic methods to revisit some of these longstanding questions in political psychology research and to further understand the nature and sources of covariance between personality, religion, and politics. The foundation for this kind of research was laid in recent decades, demonstrating that each of these traits are influenced by both environmental factors and genetic predispositions. At this point, that’s old news in the study of political behavior; this is not another “hey, look what social traits are heritable” kind of paper. Instead, we moved beyond the dyadic approach in order to integrate the domains of personality, religion, and politics in order to contribute to the discussion of the mechanisms at work in some of the “culture wars” dividing democracies, paying special attention to both genetic and unique environmental effects. Though we did not have data to examine these effects over time, we relied on behavioral genetic analyses of two unique twin studies, with different age cohorts, conducted in the United States and Australia to provide a novel way to address how genetic and environmental factors can help us to understand the covariation between personality, religion, and politics in dyads and, crucially, all at once.

Personality is often considered the first mover or the foundational, stable element to an individual’s personal orientation. The data showed this may not be accurate. When it comes to politics, a desire for religious guidance accounts for far more of the variance in social ideology, even when controlling for personality. We found that religiosity shares unique genetic variance with social ideology when controlling for the personality trait most strongly related to political ideology, Openness to Experience, in the two cultural contexts. The relationship between Openness and social ideology is much weaker than the relationship between religiosity and social ideology in both countries, and the effect of Openness is reduced to non-significant levels controlling for religiosity. This is even true in the Australian context, where religion plays less of a role in politics than in the United States and despite using a relatively more reliable measure of Openness than religiosity (10-item versus four-item) in that sample. Beyond Openness, when we controlled for each of the Big Five personality traits, none of the traits fully (or even mostly) accounted for the relationship between religiosity and ideology in either national context. Taken together, these findings suggest that future research should devote more effort to examining religiosity as a potential mediator between genes and politics. In order to understand “culture wars” rooted in disagreements over social ideology, political psychologists would be well served to focus on religious beliefs and unpacking what it means that there is a genetic correlation between religiosity and social ideology in these two quite different national contexts.


In contrast to the genetic effects, the environmental effects of religiosity and Openness on ideology are entirely independent. That is, even when accounting for Openness, religiosity shares a distinct unique environmental component with ideology, although it is much smaller than the genetic component. For Openness, the pattern is the same in the United States and there is no detectable overlap in the unique environment with social ideology in Australia. This complicates the theory that personality influences ideology through genetics and emphasizes important questions: what are the life experiences that push together Openness and social ideology (at least in the U.S. context)? Why is the effect of life experiences that push together religiosity and social ideology so much larger than this effect on personality and social ideology? Moreover, contrary to past work that has focused on personality as a major pathway for genetic effects on ideology, our data suggest that personality’s unique role in ideology (i.e., net of shared variance with religiosity) could largely be through environmental influences and not shared dispositions, as the genetic effect of Openness on ideology is attenuated but the unique environment effect is not (at least in the U.S.). What is the genetic component that lies at the intersection of personality, religiosity, and ideology, and how does it differ from the unique environmental experiences that seem to play a role in dyads but not all three?

Modeling a genetics and environmental analysis of personality, religion, and politics is a helpful step in understanding how individuals develop predispositions and preferences in these domains. Extending this analysis across two cultures suggests that the strong relationship between religious salience and social ideology in Western nations persists across contexts with dramatically different levels of aggregate religiosity and politicization of religious belief; it need not have been so, it could well have been the case that genetics only explained the correlations with ideology in a national context where religion is more politically salient, but instead genes matter in both contexts. The shared genetic component of this religion-politics relationship also supports the socialization literature that finds political attitudes tied to religious beliefs are more successfully transmitted, and our evidence demonstrates this may have to do with some underlying heritable predisposition. Leveraging the explanatory power of genetic and environmental effects, we suggest that religion scholars and political psychologists are both partially correct in their assessment of the “culture wars” – religion seems to be the driving force, but its influence resides in dispositional mechanisms as much as it does socially.

We hope our tests of genetic and environmental influences of religiosity, personality, and social ideology across two age groups in two cultures will contribute to understanding across political, religious, and personality domains and encourage more scholars to attempt integrated models of these measures. Much remains to be done; unpacking the role of gene-environment interaction and covariation appear important next steps. Identifying how religious and political upbringings alter the environments into which people select, which may modify the expression of genetic proclivities, could answer many questions on how values across religious and political domains become entangled. Looking at personality traits beyond the Big Five, such as cognitive style, may provide different insights into the religion-politics link. And, of course, research in other national contexts and in new samples in the United States and Australia will no doubt add additional clarity to the relationships that we bring into focus in this research.

Views About the Religious Composition of the Parties Affect Polarization

claassen1.pngPresident Donald Trump attends the Liberty University Commencement Ceremony and delivers remarks Saturday, May 13, 2017, Lynchburg, Virginia. (Official White House Photo by Shealah Craighead). Appearances at Liberty University with Jerry Falwell Jr. have become an essential public expression of the importance of evangelicals for Republicans.

By Ryan L. Claassen, Paul A. Djupe, Andrew R. Lewis, and Jacob R. Neiheisel

Is the Republican Party the party of religious Americans? Is the Democratic Party the party of secular Americans? The answer to these questions is a resounding, “NO!” Evidence of religious diversity in American society and within both major political parties abounds. And numerous social scientific studies have done much to unravel persistent stereotypes about the extent of the religious divide in American politics.

But what if many Americans are living their political lives steadfastly believing that the answer to these questions is a resounding, “YES!”? Such misperceptions could certainly be forgiven by anyone familiar with partisan rhetoric regarding the religious divide in American politics. Yet, these misperceptions have the potential to become self-fulfilling. Do beliefs that exaggerate the religious divide in American politics transform a mythical divide into a real one, with the perceived group divide contributing to polarization?

Our forthcoming article in Political Behavior investigates just what Americans believe about the religious make-up of the parties and explores the implications of those beliefs for partisan identity and partisan sentiment.

In three different national surveys, we asked samples of Americans about the percent of Republicans they believe are evangelicals and the percent of Democrats they believe are seculars. Figure 1 demonstrates that perceptions of the religious make-up of the parties span the entire range of possibilities, but there is a weak, statistically significant connection – perceiving that the Republican Party is full of evangelicals is related to perceiving that the Democratic Party is full of seculars.

Figure 1 – How Beliefs about the Evangelical Percent of Republicans and Religious None Percent of Democrats are Distributed and Linked (March 2016 Data)


To assess the connections between these perceptions and partisanship, it is important to analyze them by sub-groups, which we do in Figure 2. While our samples mirror national proportions, Figure 2 demonstrates that even averaging does little to improve the accuracy of Americans’ perceptions. In particular, members of groups involved in the stereotypical divide (evangelicals and seculars) exaggerate the divide in their beliefs, though Evangelicals do so to a greater degree.

Figure 2 – Beliefs about the Percent of Nones in the Democratic Party and Evangelicals in the Republican Party by Subgroups (November 2015 Data)


So, many Americans perceive there to be a religious divide in American politics that does not exist to such a degree, but to what end? Theoretically, we suggest that perceptions about the composition of the parties provide crucial linking information that helps individuals connect their group identities with a partisan identity. The theory builds upon the traditional political science story regarding the centrality of groups in party coalitions that have long been discussed (see e.g., Achen and Bartels 2016; Ahler and Sood 2018; Mason and Wronski 2018).

Using the same three datasets, we estimated models of party identification (standard seven-point scale) and party affect (difference of party thermometers) as a function of religious group identity, perception of the religious composition of the parties, and a series of statistical interactions that allow for the possibility that members of different groups will use those perceptions differently to shape their partisan commitments. Figure 3 shows that group members do in fact use perceptions about the religious composition of the parties in both positive (i.e., by identifying with the party in which they believe there are many coreligionists) and negative ways (i.e., by not identifying with the party in which they believe there are many antagonist group members). For instance, across all three datasets (top row), religious people (solid, black line) who perceived the Democratic Party to have more nones are more Republican. In the bottom row, evangelicals who perceive the Republican Party to have more evangelicals are more Republican, while non-evangelicals are more Democratic.

Figure 3 – The Interactive Effects of Religious Group Membership and Perceptions of Party Composition on Partisanship (All Three Data Sources)


There is a strong tendency for stereotypical beliefs about religious divisions in American parties to reinforce those divisions as group members use those beliefs to find partisan homes. Interestingly, there is an asymmetry in the magnitude of the effect, however, with evangelicals and religious people leaning more heavily on their perceptions than seculars. Countering this rather dismal news, we also note that many Americans hold counter-stereotypical perceptions, which serve to undermine polarization. For example, some evangelicals believe that many Democrats are evangelicals and their perceptions about the religious divide push them toward the Democratic Party. The pattern repeats for seculars who believe many Republicans are secular. Finally, seculars and evangelicals who hold more accurate perceptions are not terribly polarized politically (where the lines cross in Figure 3, the different groups profess similar partisanship).

There is much evidence here to worry about. The hyperbolic quality of political debate in the country is affecting Americans’ perceptions about the religious divide for the worse and these inaccurate perceptions are very polarizing. If beliefs about the state of the world are fueled by rhetoric portraying each side as locked in an existential battle, perceived religious divisions in American politics might prove to be terribly consequential. That is, exaggerated beliefs about the number of Republicans who are evangelicals, for instance, might lead more evangelicals to become Republican over time.

The silver lining here is that we find two possible antidotes to what could be a spiral of polarization. First, counter-stereotypical beliefs push in the other direction. Nurturing these would reduce polarization over time. Second, and even more appealing, Americans with accurate perceptions appear less polarized. A truce in the culture wars is possible, but inaccurate perceptions must be corrected.

About the authors:

Ryan L. Claassen is a professor of Political Science at Kent State University, author of Godless Democrats and Pious Republicans (2015), coeditor of The Evangelical Crackup (2018), and author and coauthor of numerous articles in political science journals. Further information about his work can be found at his website and on Twitter.

Paul A. Djupe, Denison University, is an affiliated scholar with PRRI, the series editor of Religious Engagement in Democratic Politics (Temple), and co-creator of Further information about his work can be found at his website and on Twitter.

Andrew R. Lewis is an associate professor of Political Science at the University of Cincinnati, Public Fellow with PRRI, and author of The Rights Turn in Conservative Christian Politics (2017). Further information about his work can be found at his website and on Twitter.

Jacob R. Neiheisel is an associate professor of Political Science at the University at Buffalo, SUNY. His work has been published in the American Journal of Political Science, Legislative Studies Quarterly, Political Research Quarterly, and American Politics Research. Further information about his work can be found at his website and on Twitter.

Helping the Homeless: The Role of Empathy, Race and Deservingness in Motivating Policy Support and Charitable Giving

Wronski.pngPhoto credits: Rob via Creative Commons 2.0 (left); National Geographic (right)


Kimberly Gross, George Washington University

Julie Wronski, University of Mississippi

While overall homelessness has decreased in recent years, it remains a widespread problem that threatens many low-income Americans. Media depictions of homelessness often convey individual weaknesses, stigmatizing stereotypes, and disgust. To counter this, homeless charities use humanizing appeals that tell the story of specific individuals depicted as deserving of assistance. For instance, the campaign “Rethink Homelessness” asked homeless people on the streets of Orlando, Florida to write down something about themselves that people wouldn’t know just by walking past them, then created a viral video based upon their stories. But does this type of messaging strategy increase sympathy, move attitudes on government assistance, and ultimately motivate personal behavior to help homeless individuals in need? Importantly, is the success of such campaigns contingent upon whether the depicted beneficiaries are from one’s racial in-group (i.e. White) or out-group (i.e. African-American)?

Our Political Behavior article explores how message campaigns that employ humanizing stories work to help the homeless. First, we examine how two features of the message – particularly, the race of the individuals shown, and the degree to which they are portrayed as deserving – matter. Specifically, we examine how these messages influence monetary charitable giving, as well as attitudes toward government assistance and sympathy. We also explore how individuals’ racial resentment and empathetic ability (measured using the “Reading the Mind in the Eyes” test) moderate these message effects.

We conducted two online survey experiments with Whites-only samples: one administered through Time-Sharing Experiments in the Social Sciences (TESS) and the other by Qualtrics Panels. In each, participants were randomly assigned to watch one of four videos depicting homeless individuals in Washington, DC. Each video, modeled after the “Rethink Homelessness” campaign, showed two male and two female individuals (either White or African-American) holding a sign, while each person says “I am homeless.” The homeless individuals in the videos were matched on age and gender by race, and the paired individuals wore the same clothes. The four signs constituted our deservingness manipulation and provided either no information beyond the person being homeless or four distinct reasons why the individual was homeless.

Willingness to donate money to a homeless charity – our behavioral measure – was assessed using a dictator game. Following the video treatments, participants were offered additional funding for the study ($10 in TESS and $4 in Qualtrics), and were given the opportunity to donate none, some, or all of that compensation in one-dollar increments to Miriam’s Kitchen, a private charity that provides food and services to the homeless in Washington, DC. Across both studies more than 70% of participants donated something, totaling nearly $7500 in donations to Miriam’s Kitchen!

We found that the race of the homeless individuals depicted in our videos had no direct effect on government policy attitudes or charitable donations in either study. Videos that featured deservingness information lead to increased expressions of sympathy in both studies, to greater support for government efforts to help the homeless in the Qualtrics study, and to increased donations and propensity to donate in the Qualtrics study.

As the figures below illustrate, there was some variation in donation behavior based upon individual’s racial attitudes, empathetic ability, and which video they viewed. We generally observed a negative relationship between racial resentment and donations (top panel), and a positive relationship between empathetic ability and donations (bottom panel), with a few important caveats. Those at the low end of the racial resentment scale donated about $0.67 more in the Black information condition relative to the White information condition, while those high in racial resentment donated about $0.72 less in the Black conditions than in the White information condition. Racial resentment had no effect on donations when shown Whites providing reasons for their homelessness. For those with the greatest empathetic abilities, donations were approximately $1.85 more in the two White conditions, and $1.73 more in the Black information condition relative to the Black no information condition. Empathetic ability had no effect on donations when viewing African-Americans with no information regarding the circumstances of their homelessness.



On one hand, it is reassuring that our race manipulations did not reveal a systematic advantage for homeless beneficiaries who are members of one’s racial in-group (i.e. Whites). Yet, individuals’ racial biases nonetheless influenced their donation choices. Deservingness appeals worked when the recipients came from a racial group an individual was biased towards, but such appeals were not enough to overcome bias against the group. Further, individuals own capacity to empathize with those in need was associated with increased donations only when there was some relatable factor like a shared racial group or deservingness information. Overall, organizations utilizing humanizing appeals to generate financial support for their cause will do well to feature stories where the individual, regardless of their race, is not seen as responsible for their adverse circumstances.

A Matter of Principle? On the Relationship Between Racial Resentment and Ideology

Adam M. Enders, University of Louisville

Racial resentment is perhaps the most contentious – albeit most frequently used – measure of racial prejudice in American political behavior research. Where proponents see a reasonable measure with desirable statistical properties (e.g., consistently high reliability, unidimensional structure) and predictive power when it comes to things such as attitudes about racial issues, skeptics challenge the validity of the individual questions that compose the scale and suggest that such questions may be conflating several latent constructs. Perhaps the most vocal critics ­have argued that observed responses to the racial resentment questions that suggest racial prejudice are just as likely a product of adherence to conservative ideological principles. I, following others, refer to this perspective as the “principled conservatism thesis.”

While plenty of previous work has engaged this debate, very few scholars have done so with a consideration of the measurement of ideology in mind. Indeed, investigations of the principled conservatism thesis are usually undertaken using a measure of ideological self-identification, instead focusing on the statistical model, control variables, or even employing different measures of racial prejudice. However, we know from previous research that a large proportion of self-identified conservatives are not conservative when it comes to issue attitudes or other orientations toward the government. This puts a low ceiling on any potential support for the principled conservatism thesis. Given high levels of “conflicted conservatism,” and low levels of ideological constraint, I argue that self-identifications are likely an inappropriate operationalization of adherence to ideological principles.

In this manuscript, I construct several measures of adherence to conservative ideological principles using survey questions about government spending and more abstract ideas about the appropriate size and scope of government. Then, I consider the relationships between the two measures of operational ideology, the self-identification measure, and racial resentment. I find that the correlation between operational ideology and racial resentment is much lower than that between symbolic self-identifications and racial resentment. Moreover, this trend holds over time, from 1992—2016.

Finally, I consider the potential impact of principled conservatism on the observed responses to the individual racial resentment items using a methodology developed to investigate differential item functioning (DIF). Simply put, DIF is a situation in which observed responses to a given survey item are the product – either additively, or interactively – of both the assumed latent construct (e.g., racial prejudice) and some other confounding characteristic (e.g., ideology). I find very little evidence for ideology-based DIF over time across the four racial resentment items. Furthermore, what DIF I do observe is largely related to the self-identification operationalization of ideology that is inappropriate for investigating the effect of adherence to ideological principles on racial resentment.

All evidence taken together, I find little support for the principled conservatism thesis. We are, however, left with a question as to what explains the relationship between ideological self-identifications and racial resentment. The answer seemingly does not lie in adherence to ideological principles. One potential explanation is elite partisan cueing. There exists strong evidence for partisan sorting – the increasing congruence between partisan and ideological identities – among the mass public over time. If self-identified conservatives are increasingly identifying as Republicans, they are also likely being exposed to Republican messaging (to which they are more receptive) more frequently. In other words, perhaps people who identify as conservatives “learn” the language of racial resentment from elites? Or, perhaps, it is one of the other ingredients of ideological identity ­– values, group orientations, cultural connotations – that causes the correlation between symbolic self-identifications and racial resentment. Despite another piece of supporting evidence for the validity of the racial resentment scale, there are no shortage of questions regarding the measure of racial prejudice to be investigated.