Abstract Artificial intelligence (AI) interviewing offers the potential to enhance the depth and quality of survey research, yet it may also introduce respondent fatigue and unintended consequences for the measurement of political attitudes. From the standpoint of respondents, more extensive interviews impose greater cognitive and temporal burdens, which may reduce engagement or increase satisficing. From the perspective of political behavior researchers, deeper probing may shift the subset of beliefs or arguments that respondents access when forming opinions, thereby altering the distribution and apparent stability of measured attitudes. To investigate these dynamics, we conduct a large-scale, fully randomized experiment. We find that AI-mediated interviewing significantly increases both the volume and specificity of respondents’ articulated rationales, but any decline in participant satisfaction is minuscule and we detect no increase in participant rolloff. However, AI interviews appear to shape subsequent responses: After offering justifications through AI interaction, respondents exhibit greater polarization in follow-up items. This result suggests that AI interviewing may alter the cognitive architecture of opinion formation. These findings both highlight the potential for AI interviewing and underscore the need for studies of how AI interviewing may become a source of attitudinal change.
Abstract Standard statistical uncertainty measures, like standard errors, are unable to quantify the uncertainty from an unrepresentative sampling process. Existing work has highlighted the potential risk in large datasets, in which larger surveys resulting in more precise estimates can misleadingly overstate the degree of confidence in biased survey results. In this research note, we introduce a sensitivity framework for researchers to transparently summarize their uncertainty to a biased sampling process. We illustrate the framework on a set of COVID-19 vaccine surveys, and show that had researchers conducted a sensitivity analysis, they would have found that despite substantially smaller standard errors, the larger surveys conducted on less representative samples were more susceptible to potential bias than smaller surveys conducted on more representative samples. We argue that these summary measures should be routinely reported to accurately quantify uncertainty to potential bias, allowing researchers to assess their confidence in results, in large and small datasets.
Research has shown that issue importance is a significant mediator of various political phenomena, such as issue voting, political fragmentation, and substantive representation, among many others. Studies focusing on issue prioritization among voters often relied on the implicit assumption that the issue-carrying capacity of the public-the number of issues the public prioritizes at a time-is relatively fixed over time because of individual-level limitations in the human cognitive architecture. However, prior research testing the limits of this assumption utilized mostly aggregate-level data to examine agenda capacity. We relax this assumption and argue that individuals differ greatly in the number of policy problems they prioritize. Drawing on an original dataset of content-coded verbatim answers to the Most Important Problem (MIP) question in American National Election Studies surveys (1980-2020) and a secondary source of legislative bills, we show that both attention to politics, civic resources, and information environment influence the extent to which respondents prioritize (i) a large number of issues and (ii) issues from different policy areas. These findings advance our understanding of single-issue voters and voters who display more complex issue prioritization behavior, which has important implications for various fields of study, including voting behavior, issue competition, and elite-voter linkages.
In political science, extensive research has explored negativity biases in citizens' attitudes and behaviors. In particular, we know that citizens' political evaluations tend to be more strongly influenced by negative than positive perceptions of traits, events, and policy outcomes. In this paper, I argue that culture is a significant yet understudied correlate of negativity biases. A multilevel analysis using the World Values Survey (WVS) demonstrates that the negativity bias in political trust weakens as a society's level of collectivism increases. Next, I explore the effect of cultural values at the individual level with data from the Asian Barometer Survey (ABS). In line with the results from the WVS, I find that collectivism reduces the negative-positive asymmetry in citizens' political trust based on perceived institutional performance. These results suggest that negativity biases should not be universally assumed as a defining feature of citizens' attitudes toward government across different cultural contexts.
Prior research demonstrates that partisan alignment with the president-presidential copartisanship-has structured Supreme Court job approval and confidence for multiple decades. Presidential copartisans are more approving of the Court than outpartisans, an effect that persists throughout a presidency despite high-salience Court decisions. We examine whether the presidential copartisanship effect persisted into the Biden era despite Dobbs v. Jackson Women's Health Organization (2022) and other conservative rulings. We extend rolling cross-sectional job approval and confidence data through 2024, with surveys before and after Dobbs. We show that the presidential copartisanship effect disappears after Dobbs. Moreover, after Dobbs, partisan divisions reach historic highs, with Democrats dropping to record low approval levels (14 percent). We elaborate on whether this new dynamic will persist and its implications for Supreme Court legitimacy and decision-making.
Across the world, women remain underrepresented in politics. Yet, recent experimental studies of candidate preferences find that citizens favor women over men candidates, with women and left-wing citizens driving aggregate preferences. This raises both theoretical and empirical puzzles. Theoretically, the underlying processes producing heterogeneous preferences across citizens' gender and ideology remain unaddressed, and empirically, conclusions rest primarily on data from the United States and Western Europe, which raises questions about generalizability to democracies in the Global South and other regions. This article reports the results from the most comprehensive and geographically diverse test of citizens' preferences for women candidates to date. We fielded a conjoint candidate choice experiment with a multitude of respondent-level predispositions (e.g., gender, ideology, sexism, and gender typicality) across twenty institutionally, culturally, and economically diverse democracies (N = 14,369). The results advance knowledge about citizens' preferences for women candidates both empirically and theoretically. On average, women candidates are preferred by three percentage points, and this advantage is statistically significant in fourteen countries. Women-especially those identifying as more gender typical-and left-wing citizens-especially individuals low in social dominance orientation, right-wing authoritarianism, and hostile sexism-display particularly strong preferences for women candidates. These results are discussed with respect to their theoretical and practical implications for women's political representation.
Web tracking studies provide unique insights into digital behavior but face challenges related to participant recruitment and attrition. This study examines how different monetary incentive structures-unconditional and conditional-and incentive amounts affect key recruitment stages: consent to participate in web tracking, installation of tracking software, and attrition after installation. To assess the effectiveness of monetary incentives across probabilistic and nonprobabilistic recruitment methods, we implement a preregistered experiment with 1,862 participants in the GESIS Panel.dbd-an academic infrastructure for collecting and linking digital behavioral data with survey data in Germany. Results show that unconditional incentives increase consent rates, whereas conditional incentives more effectively foster installation and, in particular, sustain participation after installation. While probabilistic recruitment poses challenges for initial consent, participation levels hereafter remain comparable across recruitment methods. These findings contribute to the development of cost-effective incentive strategies aimed at enhancing participation in web tracking studies.
The monarchy is a long-standing British institution that enjoys high visibility across the world. However, there is concern that younger generations in the United Kingdom are losing attachment to the monarchy and increasingly support its abolition. However, it could also be that diffuse support for the monarchy increases as people gain experience with it (age effect) or from time-specific factors (period effects). To adjudicate between these possibilities, we estimate age, period, and cohort effects on support for continuing the British monarchy among the UK public. We find that support for continuing the monarchy increases considerably with age. Conversely, we detected few period effects, as public opinion has remained stable over time, and there is little clear evidence that support for the monarchy is lower among younger generations. These findings indicate that support for the monarchy will likely remain stable, and even increase, considering Britain's aging population. The monarchy thus represents a valuable case for understanding how symbolic institutions persist in democratic societies despite lacking procedural fairness, accountability, or instrumental function, a pattern with implications for other non-performance-based institutions like appointed chambers, established churches, and ceremonial offices.
Scholars have suggested that we are experiencing a global crisis in trust in elections, with political polarization and misinformation often cited as reasons for distrust. However, empirical evaluations of perceptions of election integrity over time remain scarce. Using data from the Canadian Election Study from 2008 to 2021 and the Media Ecosystem Observatory from 2021, we show that trust in electoral management bodies, measured in terms of confidence, satisfaction, and fairness, has significantly declined since 2019 among Canadians with a right-wing ideology or who support right-wing parties. We find evidence that rising affective polarization, right-wing and social media consumption, and belief in US-based election fraud narratives are associated with significant trust declines in electoral integrity. The results identify key factors linked to the recent politicization of citizens' perceptions of election fairness and have important implications for the health of our democracy.
Comparative public opinion surveys, including the Comparative Study of Electoral Systems (CSES), often rely on identical labels and question items to measure respondents' symbolic and operational ideology across countries. However, it remains unclear whether common symbolic measures of ideology carry the same meaning across contexts, or whether the issue items used to construct common operational measures are equally relevant in national settings. To address these concerns, we compare symbolic and operational ideology as measured using common cross-national items and country-specific items. We conducted public opinion surveys before and after national elections in seven Asian and Western countries. Our analyses reveal that (i) country-specific measures of ideology are generally more stable over time than common measures, (ii) country-specific symbolic and operational ideology are more strongly associated with each other than their common counterparts, (iii) respondents tend to have a greater understanding of country-specific ideological labels than common ones, (iv) country-specific measures are generally more predictive of vote choice in elections, and (v) ideological extremity, as measured by these two types of indicators, shows different correlations with respondents' attitudes toward democratic values. These findings suggest that researchers should use country-specific measures of ideology in comparative public opinion research.
Why do citizens support negotiated solutions to internal conflicts? We develop a theoretical framework drawing insights from prosocial generalized trust orientations and classic models of public opinion. We expect generalized trust-a baseline belief in the trustworthiness of others as members of a shared moral community-to be positively associated with support for negotiated peace. Original survey and behavioral data collected during the peace talks between the Colombian government and the Fuerzas Armadas Revolucionarias de Colombia (FARC) and survey data following the Guatemalan peace agreement are consistent with this hypothesis. Standard public opinion models suggest that this association should strengthen as negotiated peace grows in salience, a hypothesis compatible with rolling cross-sectional survey analyses and placebo tests surrounding Colombia's 2016 peace referendum. By linking prosocial generalized trusting dispositions to attitudes toward negotiated peace outcomes, our findings advance our knowledge of the microfoundations of international relations and conflict resolution.
Do bilingual ballot designs promote inclusivity? While ethnic politics scholars have argued about the importance of accommodating minorities, there has been little attention paid to one specific institution: the ballot. Likewise, while we know ballot designs are important, the empirical focus has strictly been on monolingual ballots. In this paper, we identify three designs: (1) single: monolingual ballots; (2) separated: one bilingual ballot with two columns, with one language per column; and (3) stacked: one bilingual ballot with one column, with languages collated for each race. We argue that attitudes are most inclusive when ballots are stacked-that is, there are multiple languages sharing the same space. However, this is only the case when language is not a politicized issue. When it is, attitudes in fact become exclusive. To test, we employ three studies that vary on their extent of language politicization: (1) no politicization-the use of an indigenous language and an immigrant language next to Chinese in Taiwan; (2) politicization of the majority language-the use of Spanish alongside English in Texas; and (3) politicization of a minority language-the use of Russian and English in the Republic of Georgia. The results are robust and consistent with our theoretical expectations. Given that the foundation of democracy rests on citizens being able to exercise their voice, it is imperative that we accommodate minority languages effectively on the ballot.
Political scientists often claim that partisanship is explained by nonpolicy factors such as early childhood experiences, parental influence, and group membership. In this paper, we use open-ended survey questions to elicit partisan Americans' own explanations for their affiliations as well as their attitudes toward the opposite party. Policy, values, and ideology are the most common reasons Americans say they identify as a Democrat or Republican and feel the way they do about members of the other party. The rate at which these reasons are cited far exceeds expectations of published scholars of partisanship. Findings do not vary meaningfully across party, region, gender, race, age, or socioeconomic status.
This paper examines whether generational change in Central and Eastern Europe (CEE) is associated with shifts in how political attitudes align along the conventional economic-cultural left-right dimension. Because cohorts in the region were socialized under different political-economic systems, CEE provides a setting to assess whether younger generations display patterns of attitude alignment that differ from those of older cohorts. Using pooled European Social Survey data (2004-2023) and original surveys from Czechia, Hungary, and Lithuania, we combine an age-period-cohort analysis of left-right alignment with a network analysis of associations among thirteen salient issues. We find modest generational differences in the early 2000s, but these diminish over time, with all cohorts exhibiting similarly low levels of left-right alignment. Economic and cultural attitudes remain weakly connected, and EU- and foreign-policy issues often form separate clusters. The findings indicate that generational replacement has not produced a shift toward a more conventional left-right organization of belief systems in the region.
The election of Donald Trump in 2016 inspired sudden interest in the association between voters’ sexism and voting behaviour. Existing research shows that sexism may be a stronger predictor of political attitudes and voting behaviour than a voter’s gender. To what extent do the findings from the US presidential elections apply to other contexts and research questions? To answer this question, we present a systematic review of studies focusing on sexism as a predictor of political attitudes and voting behaviour. In two databases (Scopus and Web of Science), we identify 97 relevant studies, including 42 on the 2016 US presidential elections. The review shows that particularly modern and hostile sexism played a role in Trump's election. However, few other research questions have been answered with similar certainty. Our review discovers gaps in the literature and proposes further directions for this growing research field. To enable future comparative research, we discuss best practices for measuring sexism and show the variety of ways in which sexism is operationalized.
As public opinion research increasingly recognizes the importance of context, this study investigates how the authoritarian past can explain the mobilizing role of political communication in participatory behaviors. Positing that postauthoritarianism can serve as a sociopolitical context shaping citizens' responses to interpersonal discussion and media use, the current research employs cross-national survey data from 14 countries in the Comparative National Elections Project (CNEP). Employing multilevel modeling, this research examines how interpersonal political discussion with strong and weak ties and attitude-congruent media use can be related to two modes of political participation: protesting and voting. The findings suggest that frequent weak-tie political discussion is associated with increased protest participation; this mobilizing role is less pronounced in postauthoritarian countries compared to countries without an authoritarian past. Moreover, exposure to attitude-congruent media exhibits different patterns in facilitating voting regarding previous regime ideologies, left-wing or right-wing. By suggesting postauthoritarianism as a contextual layer, this study advances a more nuanced, cross-national understanding of democratic engagement.
