Explore the vast range of titles available.

We propose and explore the possibility that language models can be studied as effective proxies for specific human subpopulations in social science research. Practical and research applications of artificial intelligence tools have sometimes been limited by problematic biases (such as racism or sexism), which are often treated as uniform properties of the models. We show that the “algorithmic bias” within one such tool—the GPT-3 language model—is instead both fine-grained and demographically correlated, meaning that proper conditioning will cause it to accurately emulate response distributions from a wide variety of human subgroups. We term this property algorithmic fidelity and explore its extent in GPT-3. We create “silicon samples” by conditioning the model on thousands of sociodemographic backstories from real human participants in multiple large surveys conducted in the United States. We then compare the silicon and human samples to demonstrate that the information contained in GPT-3 goes far beyond surface similarity. It is nuanced, multifaceted, and reflects the complex interplay between ideas, attitudes, and sociocultural context that characterize human attitudes. We suggest that language models with sufficient algorithmic fidelity thus constitute a novel and powerful tool to advance understanding of humans and society across a variety of disciplines.

The rhetoric of populist politicians is an important part of their appeal; however, little is known about how that rhetoric operates. Drawing on two large experiments conducted with American adults, we show that frames encouraging individuals to consider political problems in dispositional terms prompt populist expressions, while an encouragement to consider these same problems situationally does not. In our second experiment, we connect this framing change to voting intentions and find that subjects exposed to dispositional frames are more likely to express support for Donald Trump and less likely to express support for Hillary Clinton than subjects exposed to situational frames.We find the same pattern when we compare Bernie Sanders with Clinton but not when we compare Trump with Sanders. Importantly, the impact is contingent on preexisting populist attitudes; subjects with lower populist attitudes are more likely to demonstrate an increase in expressed populism and support for populist candidates.

American politicians frequently evoke race in their messages to the public; at the same time, politicians often pay a price for racialized rhetoric. We propose that elites continue to use messages about race because they can mitigate the costs of doing so with justifications for their original statements. Integrating literatures on elite rhetorical tactics and framing, we predict that when justifications and indirect racial messages are combined, elites can mobilize the support for racially resentful Whites without alienating others. In a pair of survey experiments conducted in 2019 and 2020, we examine the effectiveness of justifications in swaying Whites’ attitudes. We find that two different elite justifications bolster support for their messages. Importantly, we also find these tactics do not incur political costs. This provides a compelling reason that political figures continue to use racial messages in politics despite recent social movements and possible shifts in Americans’ attitudes about race.

What comes to mind when people think about rank-and-file party supporters? What stereotypes do people hold regarding ordinary partisans, and are these views politically consequential? We utilize open-ended survey items and structural topic modeling to document stereotypes about rank-and-file Democrats and Republicans. Many subjects report stereotypes consistent with the parties’ actual composition, but individual differences in political knowledge, interest, and partisan affiliation predict their specific content. Respondents varied in their tendency to characterize partisans in terms of group memberships, issue preferences, or individual traits, lending support to both ideological and identity-based conceptions of partisanship. Most importantly, we show that partisan stereotype content is politically significant: individuals who think of partisans in a predominantly trait-based manner—that is, in a way consistent with partisanship as a social identity—display dramatically higher levels of both affective and ideological polarization.

Do events irrelevant to politics affect citizens’ political opinions? A growing literature suggests that such events (e.g., athletic competitions, shark attacks) do in fact shape political preferences. We present an experiment that largely replicates a widely noted irrelevant event effect. Specifically, we find that the outcome of a sporting event (i.e., a football game) affects presidential approval and likely does so by affecting individuals’ moods. We also show that the effect is short-lived.

Stereotypes of the two parties play an important role in political cognition, and a range of recent studies have examined the content and effects of partisan stereotypes. However, little work has studied change in partisan stereotypes over time. We address this question by comparing data on stereotypes of partisans collected before and after the Trump presidency, a time when we might expect individuals' images of the two parties to undergo significant change. Using a structural topic model, we compare responses to open-ended questions asking respondents to list words describing members of the two parties from 2016 and 2021. We find that partisan stereotypes in the 2021 sample are less group- and issue-based and focused more on personal traits. These results suggest that, during the Trump era, members of the mass public came to see the parties in more personalized, character-focused terms, potentially contributing to affective polarization.

Do events irrelevant to politics, such as the weather and sporting events, affect political opinions? A growing experimental literature suggests that such events can matter. However, extant experimental evidence may over-state irrelevant event effects; this could occur if these studies happen to focus on particular scenarios where irrelevant event effects are likely to occur. One way to address this possibility is through replication, which is what we do. Specifically, we replicate an experimental study that showed the outcome of a college football game can influence presidential approval. Our results partially replicate the previous study and suggest the impact is constrained to a limited set of outcome variables. The findings accentuate the need for scholars to identify the conditions under which irrelevant effects occur. While the effects clearly can occur, there relevance to politics remains unclear.

The last sentence of the abstract for “Football and Public Opinion: A Partial Replication and Extension” as published online in the Journal of Experimental Political Science on 8 November, 2017 contained the following error: While the effects clearly can occur, there relevance to politics remains unclear. It should have read as: While the effects clearly can occur, their relevance to politics remains unclear. The author and publisher apologize for this error.

Many democratic theorists suggest that harmonious relationship between groups are critical for democracy; however, far less is known about how everyday experiences promote or impede such intergroup harmony. In this dissertation, I explore a common, but overlooked, form of intergroup contact – casual, brief experiences with outgroup strangers. I propose a theory of how the ease of communication determines how these kinds of social encounters influence political support for outgroups and empirically test that theory with three separate experiments. Each creates outgroup experiences across racial or ethnic lines and experimentally manipulates the difficulty of communicating with the outgroup member. The first design, a lab-based study, indicates that majority-group members respond to easy-to-understand outgroup members with lower support. Using a field experiment, the second study confirms these patterns and adds an additional paradox: more similar outgroup members encourage these kinds of interactions, but those same kinds of outgroup members are most likely to undermine outgroup support as the interactions unfold. The final experiment indicates that standing propensities to avoid outgroups do not moderate the effect of the ease of communicating and that different contexts may be more or less prone to the backlash observed in the prior two chapters. Across all three experiments, I fail to observe any instances where these interactions promote outgroup support; these kinds of social interactions are much more likely to backfire than they are to ameliorate group differences.