American National Election Studies Data Research Articles

Synthetic Replacements for Human Survey Data? The Perils of Large Language Models

Abstract Large language models (LLMs) offer new research possibilities for social scientists, but their potential as “synthetic data” is still largely unknown. In this paper, we investigate how accurately the popular LLM ChatGPT can recover public opinion, prompting the LLM to adopt different “personas” and then provide feeling thermometer scores for 11 sociopolitical groups. The average scores generated by ChatGPT correspond closely to the averages in our baseline survey, the 2016–2020 American National Election Study (ANES). Nevertheless, sampling by ChatGPT is not reliable for statistical inference: there is less variation in responses than in the real surveys, and regression coefficients often differ significantly from equivalent estimates obtained using ANES data. We also document how the distribution of synthetic responses varies with minor changes in prompt wording, and we show how the same prompt yields significantly different results over a 3-month period. Altogether, our findings raise serious concerns about the quality, reliability, and reproducibility of synthetic survey data generated by LLMs.

Trump Voters' social position in U.S. Society: Uniqueness and radical‐right support

Trump Voters' social position in U.S. Society: Uniqueness and radical‐right support

Growing Up in a Polarized Party System: Ideological Divergence and Partisan Sorting Across Generations

AbstractAmerican political elites have increasingly polarized over the past decades, which has inspired much research into mass polarization. We study whether there is a generational component to mass polarization by disentangling period, age, and cohort differences while distinguishing two forms of mass polarization: partisan sorting and ideological divergence. Drawing from General Social Survey and American National Election Studies data, we find that partisan sorting has increased across long-standing and emerging issues, while ideological divergence has not. Contrary to expectations, over-time increases in sorting are clearly driven by changes within generations rather than by generational replacement. On several issues, newer generations turn out to be less sorted than those they replace. This tentatively suggests that, partially as a consequence of demographic changes, generational replacement will gradually lead to less polarization in American public opinion as it converges toward more liberal positions.

Does the Salience of Partisan Competition Increase Affective Polarization in the United States?

We examine if increased salience of partisan competition causes affective polarization in the United States during presidential elections. To do so, we leverage the random and quasi-random timing of survey interviews conducted during election campaigns. We conduct three separate studies. In Study 1, we utilize the 2008 National Annenberg Election Survey (NAES), in which random survey interview timing allows for a credible causal estimate of salience on affective polarization. In Study 2, we employ American National Election Studies (ANES) data from 1980 to 2016, again leveraging survey timing to assess the effect of salience on affective polarization. In Study 3, we examine changes in affective polarization as a result of increasingly salient partisan competition using NAES and ANES panel data from 1980 to 2008. Across the three studies we identify a meaningful increase in affective polarization toward candidates, but not toward parties, as a result of heightened partisan competition.

Are More Educated States More Gay-Friendly? How the Increase in College Attainment Promotes Acceptance of Gay Men and Lesbians

ABSTRACT Attitudes toward gay men and lesbians in the United States have liberalized dramatically over time yet vary substantially between states. Trends in college attainment mirror this uneven progress. This study draws on theories of the university as a cultural institution to examine the relationship between college-educated populations and attitudes toward gay men and lesbians over time and across American states. Employing 1984–2016 American National Election Studies data, this study demonstrates that larger shares of college graduates in a state are associated with more favorable attitudes on average toward gay men and lesbians, including among those who do not have a college degree. These results are robust to competing state-level religious, political, and economic explanations. Results also show that individuals in states with more college graduates are more likely to identify as LGB. This increases the likelihood of knowing someone who is LGB, which in turn leads to more accepting views. Cultural values of tolerance and acceptance espoused by university graduates may create social contexts in which sexual minorities feel safe to come out, thus increasing the visibility of sexual minority communities and enhancing favorable attitudes.

Computationally efficient Bayesian unit-level random neural network modelling of survey data under informative sampling for small area estimation

Abstract The topic of neural networks has seen a surge of interest in recent years. However, one of the main challenges with these approaches is quantification of uncertainty. The use of random weight models offer a potential solution. In addition to uncertainty quantification, these models are extremely computationally efficient as they do not require optimisation through stochastic gradient descent. We show how this approach can be used to account for informative sampling of survey data through the use of a pseudo-likelihood. We illustrate the effectiveness of this methodology through simulation and data application involving American National Election Studies data.

Abortion, religion, and racial resentment: Unpacking the underpinnings of contemporary abortion attitudes

AbstractBackgroundFor many Americans, pro‐life attitudes are directly attached to their religious beliefs, especially white evangelicals. Some have argued that evangelicals came to oppose abortion not simply because of their views on the sanctity of life, but out of a growing racial resentment as government policy and society moved towards greater racial equality.ObjectivesThis study explores the relationship of evangelicalism, racial attitudes, and views on the legality of abortion to explore whether racial resentment is behind evangelical opposition to abortion.MethodsTo carry out this exploration this study employs American National Election Studies data from 2000 to 2020 and the 2020 Public Religion Research Institute (PRRI) Values survey.ResultsWe find no support for the idea that racial attitudes are disproportionately correlated with the abortion views of white evangelicals. Rather, we find that racial attitudes are now correlated with views on abortion for all Americans. Where abortion attitudes are distinctive from attitudes on other policy issues is in having very strong religious determinants, suggesting that genuine religious beliefs do indeed underscore the pro‐life views of white evangelicals.ConclusionThis study provides a good baseline for understanding the relationship between racial attitudes, evangelicalism, and abortion attitudes at the cusp of the Dobbs decision overturning Constitutional protections for abortion, and should be revisited in the post Roe era.

Do Leader Evaluations (De)Mobilize Voter Turnout? Lessons From Presidential Elections in the United States

Do evaluations of presidential candidates in the US affect the level of voter turnout? Voters’ affections towards presidential candidates, we contend, can either stimulate or inhibit voter inclinations to turnout. Voters are more inclined to turn out when they have positive feelings towards the candidate with which they identify because they want “their” candidate to win. But citizens may also be more likely to vote when they dislike the candidate of the party with which they do not identify. In that case, voters are motivated to prevent the candidate from being elected. Utilizing the American National Election Studies data for 1968–2020, the analysis finds that the likelihood of voting is affected by (a) the degree to which voters’ affections towards the candidate differ from one another (having a clear‐cut choice between options) and (b) the nature of the affections (negative or positive) towards both in‐ and out‐party candidates.

Succession effects in presidential elections

What is the impact of ex-presidents on the electoral performance of their party's successor candidates for the office and how long after a president has left office are successor effects electorally significant? Mattei and Weisberg (1994) argue that successor effects are large primarily when the vice president runs for office immediately at the end of the term of the president under whom they served. Timeliness and the association between presidents and their vice presidents account for succession effects. This paper challenges the Mattei-Weisberg successor effects perspective, arguing that significant successor effects can be possible for non-vice presidential candidates and long after a president has left office. Using American National Election Study data, I add numerous successions to those Mattei and Weisberg investigated. Analysis finds that succession effects apply to non-vice presidential candidates. Further, some ex-presidents may influence voting decisions long after they left office. There is also a hint in these data that succession effects have climbed over time, because as polarization has increased, the linkage between presidents and their parties has tightened.

Multidimensional polarization dynamics in US election data in the long term (2012–2020) and in the 2020 election cycle

AbstractWe use a network‐based method to explore bifurcation in the multidimensional opinion‐based political identity structure from 2012 to 2020 in American National Election Studies data. We define polarization as ideological clustering which occurs when attitudes are linked or aligned across group‐relevant dimensions. We identify relevant dimensions with a theory‐driven approach and confirm them with the data‐driven Boruta method, validating the importance of these items for self‐reported political identity in these samples. To account for data sets having different sizes, we bootstrapped to obtain comparable samples. For each, a bipartite projection generates a network where edges represent similarity in responses between dyads. The data provide us with preidentified groups (Republicans and Democrats). We use them as our network communities and to calculate an edge‐based polarization. Results show bifurcation progressively increasing, with a striking increase from 2016 to 2020. We visualize these identity‐related shifts in opinion structure over time and discuss how polarization results from both between‐ and within‐group dynamics. We apply a similar method to a smaller data set (N = 294) to explore short‐term fluctuations before and after the 2020 election. Results suggest that between‐group polarization is more evident after than before the election, because in‐group opinion dynamics result in a more synchronized opinion‐space for Republicans.

Understanding the appeal of libertarianism: Gender and race differences in the endorsement of libertarian principles

AbstractThere is a stereotype of libertarians being young, college educated, white men and that the Libertarian Party lacks appeal among women and individuals of color. There is a great deal of research investigating gender differences in public opinion on a number of issues including the provision of government resources and government spending (Barnes and Cassese; Howell and Day). Nevertheless, there is no work specifically investigating why women and nonwhites do not find libertarianism appealing. We test several hypotheses using 2016 American National Election Study data and 2013 PRRI data. We find a sizeable and significant gender gap and race gap in support for libertarian principles. We investigate several explanations for these gaps finding moderate support for self‐interest, racial attitudes, and egalitarianism as reasons for women and African Americans being less supportive of Libertarian Principles. We believe that the modest success of and media attention garnered by Ron Paul and Rand Paul in recent years along with the success of the Libertarian Party presidential ticket in 2016 highlights the need to understand who is drawn to libertarianism and why.1

Gender and race gaps in voting and over-reporting: An intersectional comparison of CCES with ANES data

ABSTRACT It is widely recognized that women, and perhaps Black women in particular, are more likely to vote than men. However, this is based on self-reports of voting, which can be unreliable. The Cooperative Congressional Election Study (CCES) and American National Election Studies (ANES) help with this issue by validating votes. Here, we use the CCES and ANES data to take an intersectional approach to estimate the probability that men and women of different races over-report voting, and how it affects the gender gap in voting. We find that Anglo women are somewhat less likely to over-report voting, but Black men and women and Latinas may be more inclined to over-report. We find evidence that relying on self-reports may underestimate the degree to which women (Anglo, Black and Latina) are likely to vote. However, the CCES data estimate that Anglo men are more likely to vote than Anglo women. We examine the CCES data to better understand this unique finding and suggest there may be sampling issues with the CCES data.

Polarized Networks? New Evidence on American Voters’ Political Discussion Networks

An important mechanism of mass political polarization involves citizens’ social networks: how politically homogeneous are they, how has this changed over recent years, and which individual and contextual variables predict the degree of homogeneity in social networks? Moreover, what are the consequences of network homophily on political preferences and in and out-group perceptions? In this paper, we address these questions by combining data from the 2000 American National Election Study and original data from the 2016 Cooperative Congressional Election Study. Both surveys ask respondents a battery of questions about the individuals with whom they most frequently discuss politics, including perceived vote choice and level of political knowledge. Using these data, we offer an updated empirical assessment of how polarization is influencing—and is influenced by—social network homophily.

Rights and Tolerance Support in U.S. Presidential Elections, 1992–2016

Do attitudes toward issues such as abortion, LGBTQ (lesbian, gay, bisexual, transgender, and queer) rights, and tolerance of diversity matter to U.S. voters and elections? Scholarship on rights and tolerance support has failed to take into account the potentially confounding influence of racial resentment as a factor behind voter choice. This is a serious limitation that we seek to address with American National Election Studies data for presidential elections from 1992 through 2016. We bring together for the first time rights and tolerance support and racial resentment, alongside further consideration of anti-immigrant sentiments and white identity. Racial resentment and anti-immigrant sentiments shape the behavior of voters, but there is new evidence for the large influence of rights and tolerance attitudes, including with respect to the 2016 presidential election. The results suggest a growing dependence of Democratic candidates on high levels of rights and tolerance support. We discuss implications for theory and research on voter choice.

The Roots of Right-Wing Populism: Donald Trump in 2016

Using survey data from the American National Election Study (ANES) and aggregate data on Congressional districts, this article assesses the roles that economic and social factors played in Donald J. Trump’s 2016 “populist” presidential candidacy. It shows the hollowness of claims that economic issues played little or no role. While agreeing that racial resentment and sexism were important factors, the article shows how various economic considerations helped Trump win the Republican nomination and then led significant blocs of voters to shift from supporting Democrats or abstaining in 2012 to vote for him. It also presents striking evidence of the importance of political money and Senators’ “reverse coattails” in the final result.

Polls and Elections: Did Character Count? Candidate Traits and the 2016 Presidential Vote

AbstractThe 2016 presidential election featured two candidates whose personal qualities Americans held in exceptionally low regard. Voters perceived Hillary Clinton to be knowledgeable, but dishonest. Assessments of Donald Trump were historically poor across the board. Because both candidates suffered from voter perceptions of their weak character, campaign attacks and media coverage combined to elevate character to a leading campaign issue. Yet our analysis of American National Election Studies data indicates that while voters were influenced by their perceptions of candidate character, only a small number of votes were changed by these perceptions. This is because candidate traits are more strongly determined by party identification than ever before, and because increasing party polarization has led to there being fewer genuinely persuadable voters than in past elections.

The Reliability of Same-Race and Cross-Race Skin Tone Judgments

The purpose of this study is to examine the reliability of the skin tone measures in the widely used American National Election Studies data collection (ANES 2016 Time Series). Low reliability in skin tone measurement can lead to false conclusions regarding theoretically important relationships. Consistent with previous reliability analyses based on data from the General Social Survey, we find that different interviewers agree on Black and Latinx respondent skin tone less than 20% of the time and inter-rater reliability coefficients are very low (< .3). We also exploit unique features of the ANES data that allow us to (1) assess intra-rater reliability using Krippendorff’s alpha and (2) compare observer skin tone judgments to respondent self-appraisals. We find that even for cases where the same interviewer judges the same Black or Latinx respondent 2 months later, interviewers agree with their earlier assessment less than 35% of the time—only modestly exceeding expectations based on chance alone. Furthermore, we find weak correlations between how interviewers remember Black and Latinx respondent skin tone and how respondents self-describe. Importantly, our analyses indicate that these data patterns persist regardless of whether or not interviewer race/ethnicity matches that of the respondent. Thus, our results provide little support for the claim that measurement reliability can be significantly improved through a policy of matching respondents to interviewers of the same race and ethnicity. We discuss the implications for future research on skin tone’s relationship with social attitudes and outcomes.

Intersectionality in Action: Gun Ownership and Women's Political Participation

ObjectiveWe hypothesize that gun ownership among women is an important determinant of political engagement.MethodsFirst, using 2013 Pew Research Center data, we examine different types of political participation concerning gun policy. Next, we examine data from a survey experiment embedded in a unique June 2017 national survey of nearly 900 gun owners. Finally, we analyze 2016 American National Election Studies data of behavioral and cognitive forms of political participation.ResultsGun‐owning women exhibit levels of political participation about gun policy and a greater willingness to engage in political discussions about gun control than nonowning women. We also find greater levels of political engagement among gun‐owning women on measures of participation not related to gun policy.ConclusionWe discuss the implications of our findings for research on political participation as well as for gun policy.

Hispanic Millennial Ideology: Surprisingly, No Liberal “Monolith” Among College Students

The goal of this study is to analyze the ideological positions of Hispanic college students in the U.S. Rio Grande Valley (RGV). Building on Feldman and Johnston’s work, where they argue that a unidimensional model of political ideology provides an incomplete basis for study, we employ two dimensions to account for domestic policy preference. The core of the study is a taxonomic analysis of a survey of RGV college students taking government courses, where we find that the political beliefs of Hispanic millennials trend significantly more conservative and, especially, libertarian than expected. Possible implications for the future include a realignment of Hispanics with the Republican Party, a shift in a more libertarian direction within the Democratic Party itself, or an increase in younger Hispanics’ proclivity for independence from the two major parties. Our findings conform to Feldman and Johnston’s argument that the economic and social dimensions of political ideology are separate and distinct, making it essential that researchers analyze both dimensions in order to have an accurate and comprehensive understanding of individuals’ political ideology. In addition, using recent American National Election Studies (ANES) data sets, we compare how our samples are different from the general U.S. population, employing a principal component analysis (PCA).

Rural political participation in the United States: alienation or action?

ABSTRACTGlobally, concern has risen over support for authoritarian populism, in some cases specifically among rural populations. Rural influence, however, may be limited by relatively low political participation. This article examines participation in the United States, where past evidence shows rural Americans have been less involved in politics than their urban counterparts. Analysis of American National Election Studies (ANES) data from 1952 to 2012 shows historically less participation for rural residents relative to suburban and urban residents. Between 1988 and 2008, low political trust, a negative economic outlook, high external efficacy, and being more ideological and partisan are associated with increased political participation. These findings imply rural Americans with polarised views who feel deprived of political and economic power may participate more frequently in future political activities. Concurrently, similarly polarised circumstances in urban areas and low rural political efficacy may lead the rural-urban participation gap to persist.