My research investigates how people think about politics as well as how thoughts and preferences are translated into political activity. My work has been published in Political Research Quarterly and featured in the Washington Post and The Atlantic.
I am especially interested in how context and geography influence behavior. While I use various methodologies in my research, such as standard computation techniques and computational text analysis, I rely mostly upon experiments (lab, survey, and field). See below for additional information regarding my dissertation and other research projects.
Our judgments of others are based largely upon how we perceive the valence of their actions. People tend to place more weight upon negative information when evaluating the world around them.This asymmetry leads to people requiring a lesser amount of negative information (versus positive) about an individual to conclude that said individual has officially changed for the worse (better). While this asymmetric effect appears to be prevalent in the realm of mundane interpersonal evaluation, it remains less clear whether, and to what extent, this phenomenon applies to relatively impersonal political contexts where politicians and political institutions are the object of evaluation. To address this gap, this paper features multiple experimental studies designed to assess the extent to which negativity bias may be operable in politics. Results indicate that less negative information is required to reach a judgement than is positive information. Similarly, our evidence suggests that voters are quicker to punish politicians for negative behavior than to reward them for positive behavior. In most cases, these effects are moderated by partisanship, with negativity bias being more severe against members of the partisan out-group. Overall, we argue that negativity bias is a powerful force in shaping public opinion and one that is relevant to many contemporary issues in American politics.
Despite a large literature on framing effects in political science, we know relatively little about the effcacy of competing frames. This is surprising, as frame selection is critical to the goal of national advocacy groups. We take a step toward bridging this gap by experimentally manipulating two framing strategies employed by the Alzheimer’s Association. We find, contrary to first-hand accounts, that an economic cost frame (i.e., presenting the economic burden of Alzheimer’s disease on all Americans) is no more effective than a human cost frame (i.e., the number of deaths and diagnoses each year). Our results advance the study of political frames in public policy debates and inform the strategic decisions of advocacy organizations in the United States.
Prior research has found that subtle grammatical details (e.g., verbal aspect) can sway public opinion. Though, due to the nascent state of this literature, uncertainty remains regarding the true magnitude, as well as external validity, of these effects. Combined with scientific standards regarding replication and reproducibility, this suggests that more research in this area is necessary to understand the role of grammar in public opinion. Toward this end, this paper presents a series of experimental studies investigating potential grammatical effects—some of which attempt to reproduce the results of previous research in this area. Results indicate that, contrary to previous findings, subtle grammatical differences in the presentation of political information do not significantly influence public opinion. These findings, due to being contrary to other research in this area, cast further uncertainty regarding the role of grammar in political evaluation and suggest that further research in this area is yet needed.
This project, which is funded by the University of Virginia Presidential Fellowship in Data Science, investigates how politicians use emotional rhetoric during U.S. presidential debates and the impact this has on the emotional intensity of viewers. In the first phase of the project, we use a computer-based text analysis program to document and analyze the emotional content of politician’s language for each U.S. presidential debate from 1960 through the present day. In the second and third phases of the study, we utilize data mining and text analysis strategies on social media data to evaluate the types of emotional responses, as well as their intensity, that individuals watching the presidential debates experience.
Prior research has shown that social identities defined by an attachment to place (i.e., “place-based identity) are influential in shaping how citizens understand and think about political topics. Moreover, prior research has also argued that candidates sometimes utilize “place-based appeals” in order to win support among the electorate, and that such appeals are seemingly widespread. While past research has provided a rich understanding of what place-based identity and place-based appeals are, there is a large gap in what we know about the causal effects of such appeals. In this study, we address this gap by testing experimentally the effects of place-based appeals on voters’ evaluation of candidate likeability and ability to understand their constituents, across the broader American patchwork. Using a set of modiﬁed campaign mailer advertisements, we alter whether respondents see an ad that uses rural or urban imagery when introducing a candidate. We then test for the effectiveness of place-based appeals by measuring how respondents from self-reported rural and urban areas evaluate the candidate across the three conditions. Our results indicate that, consistent with existing theory, place-based appeals are impactful in shaping political evaluations among rural voters, but do not appear as relevant for urban voters. Overall, we argue that places–or symbolically charged geographical sites–are a useful, widespread, and potentially powerful political heuristic.
I am currently engaged in a larger project consisting of multiple studies investigating American attitudes toward Muslims—both as a general social group, and as an immigrant population. While I have begun analyzing survey data related to public opinion of Muslims, planned future data collection will augment these survey data with large scale news media content analysis and computational text analysis of social media in order to better understand how conversations regarding Muslims are framed, as well as experimental data that will enable a better understanding of how these frames impact opinion. Additional studies will be conducted to investigate the extent to which (if at all) Muslims are dehumanized and whether dehumanization plays a significant role in determining attitudes toward Muslims in the American context.