Open-ended questions have long been used in survey research, but their use increased significantly with the shift from interviewer-driven telephone interviews to respondent-driven online surveys. Shift in the survey environment, coupled with advancements in natural language processing and text analysis methods, makes it easier for researchers to ask and analyze open-ended questions, but the quality concerns regarding open-ended responses remain, which limits the applications of this type of question. In three experiments, I explore whether providing incentives can improve the quality of open-ended responses, thereby increasing the usefulness of open-ended questions. I find that incentives induced respondents to spend more time answering the questions, provide longer responses, and use a richer vocabulary. Moreover, incentives improved the structural topic models fitted to the responses regarding key quantitative measures. Taken together, the results support the utility of incentives in survey research with open-ended questions.

Partisan bias enters political news coverage through selective coverage and biased presentation, and understanding the contribution through each channel necessitates estimating the difference in ideological leanings conditional on covering the same events. Political events, however, are an underlying characteristic that is unavailable to researchers absent large-scale hand-coding. In this paper, I propose a novel method to estimate quantities of substantive interest conditional on a hidden characteristic of the text data. The method overcomes difficulties of existing methods that are sensitive to critical tuning parameters and produce biased estimates. Applying the proposed method to the study of media bias, I find the ideological difference between media organizations is not always driven by how events are covered. Instead, the contribution of presentation bias to the total difference in ideological leanings between news articles published by media organizations varies in ways that are consistent with the structural difference between different media types. My results shed light on the exposure of different types of voters to the partisan bias of political news coverage.

Do Americans support small individual donations over other sources of political fundraising? Small online contributions are becoming more prevalent, and political elites and the media often idealize them as leveling the playing field in the American political ecosystem. However, we have little understanding of whether and, if any, how much the public supports small donations as a campaign funding source over others and whether such preferences translate into tangible changes in political behavior. Using a conjoint experiment via a nationally representative survey of U.S. citizens, we test whether candidates with higher dependence on small individual donors are more likely to be chosen. Surprisingly, candidates relying more on small donors attract a higher likelihood of vote choice and candidate ratings, not just within primaries or for Democrats, but across primaries, general elections, and all partisan affiliations. Moreover, the public believes that there should be more small donations in American elections and that, compared to the current baseline, the ideal composition of campaign funding should rely less on PACs and large individual donations and more on small donations and other sources such as candidate self-financing. Such beliefs are unshaken when presented with information about lawmakers with the highest reliance on small donors, who are generally perceived as outsiders or ideologically extreme.

Belief in conspiracy theories can be highly detrimental not only for the affected individual, but also for the wellbeing of society as a whole. Despite the importance of understanding the factors that contribute to their prevalence, we currently lack an integrated characterization of which underlying cognitive factors render an individual vulnerable to such beliefs. To address this gap, we utilized a battery of six cognitive tasks targeting distinct but overlapping cognitive attributes spanning different aspects of information processing from information perception, information seeking, and the evaluation and integration of information to make decisions. We then administered this task battery to a large online cohort. When integrating across tasks using unsupervised machine-learning and multivariate statistical methods, we find evidence to suggest that conspiracy believers exhibit a relatively specific but domain general deficit in the ability to effectively integrate and utilize empirical evidence, as well as an increased tendency to misattribute the consequences of their own actions to the interference of others. These effects were found to be associated with conspiracy beliefs even after controlling for the effects of other psychiatrically relevant symptom dimensions such as schizotypy, anxiety, worry and depression. Our findings thus suggest the existence of a specific cognitive phenotype underlying susceptibility to conspiracy beliefs, which in turn could point the way toward more targeted interventions to mitigate their influence.