Dissertation

“The Enemy Is Within”: Elite Affective Polarization in American Politics

Recent work has shown that affective polarization is rising and that it has significant consequences for mass political behavior. Still, little work has examined whether and how this construct applies to an increasingly rancorous elite-level politics. In this book dissertation, I explore the causes of elite affective polarization and the consequences of that affect for elite behavior. I draw on original surveys of state legislative candidates, local elected officials, and congressional staff as well as text data and and policy-making behavior. My work contributes to various areas of political science, including work on democratic representation, legislative gridlock and the process of lawmaking, and affective polarization and partisan identity.

"Intra-Party Norms and Affective Polarization" (Job Market Paper)

Samuel Frederick.

Antipathy toward members of the opposing political party, or affective polarization, has risen in recent years in the United States. While previous studies evaluate the individual-level drivers of affective polarization, less work has considered the social determinants of partisan animosity. In this paper, I examine the extent to which intra-party social norms constrain and shape affective polarization at the mass and politician levels. I focus on norms against the expression of partisan incivility. Using original surveys of elected officials and the American public, I show that there is a strong norm against partisan incivility among politicians, but the norm among the mass public favors partisan incivility. Within my surveys, I embed experiments to evaluate the causal effects of priming this social norm on affective polarization. I find tentative support for the idea that social norms around partisan incivility can influence hostility toward the opposing party. In addition, inparty norms increase warmth toward the inparty, suggesting a potential normative basis for inparty identity. My findings shed light on the social roots of affective polarization, offering a deeper understanding of the causes of partisan animosity in American politics.

"Leading the Leaders: Who Drives Affective Polarization Among Politicians?"

Samuel Frederick.

Scholars have recently noted an increase in partisan affective polarization at the mass level. Despite the growth in research about partisan hostility, little work has examined this phenomenon among politicians. In this research note, I present results from an original survey of American local elected officials. I show that politicians are most hostile toward out-party activists, followed by politicians and finally, by voters. My survey highlights an asymmetry in politicians' perceptions of the parties: politicians' feelings toward the out-party are closest, on average, to their feelings toward out-party activists, while politicians seem to distinguish their own party from its activists. This gap in attitudes toward the parties appears to increase affective polarization among politicians.

"Across the Aisle: Affective Polarization and Bipartisanship in American Legislatures."

Samuel Frederick.

Partisan hostility in American legislatures has escalated to the point that some legislators fear violence from their peers across the aisle. Amid this increase in partisan tension, bipartisanship in lawmaking has declined. Existing theoretical frameworks of legislative behavior struggle to explain the extent of inter-partisan animosity, however. In this paper, I argue that partisan identities and affective polarization among legislators can help us understand both inter-partisan animosity and the decline of bipartisanship in lawmaking. Due to the hyper-salience of legislators' political identities, partisan identities may be important in shaping the behavior of American politicians. Using the results of a conjoint experiment in an original survey of state legislative candidates, I show that politicians discriminate against members of the opposing party when selecting partners in the policy process. In particular, more affectively polarized politicians are more likely to choose to work with a copartisan legislator. These results have important implications for the study of legislative politics, affective polarization, and the democratic system.

"All Politics Is Partisan? Ideology, Strategy, and Affective Polarization in Constituency Responsiveness"

Samuel Frederick.

Political scientists and politicians alike have long viewed the provision of constituency service by elected representatives as electorally valuable. Yet, recent research finds that politicians do not provide these services to all constituents equally. Studies show that politicians are less responsive to their constituents from the opposing political party and more responsive to their co-partisan constituents, but previous work has not examined the reasons for this partisan gap. In this paper, I test three reasons why politicians may discriminate against out-party constituents in favor of co-partisans in the provision of constituency service. Specifically, using an original survey of local elected officials, I examine whether partisan discrimination in constituency service can be best explained by electoral strategy, ideology, or partisan identity and affective polarization. My findings suggest that scholars should take seriously the role played by partisan identity in elite decision-making.

Publications

2024

“Does Interaction with Out-Party Elites in a Classroom Setting Diminish Negative Partisanship?”

Samuel Frederick, Michael Miller, and Donald Green.

American Politics Research.

The American electorate is increasingly affectively polarized. Partisans dislike members of the opposing party, even going so far as to discriminate against opposing partisans in nonpolitical domains. Given the potentially pernicious consequences of affective polarization, especially in nonpolitical settings, scholars have suggested facilitating contact between members of opposing parties as a means of reducing affective polarization. We test this approach in the context of a large introductory American government class comprised almost entirely of Democrats. Our pre-registered experiment randomly assigned recitation sections to treatment or control conditions, where the treatment was an hour-long discussion with a county chair from the Republican Party, and the control was an hour-long discussion with an attorney with no party connection. Partisan attitudes and evaluations were measured extensively at baseline, within a few days of the intervention, and two months later. We find no evidence that the treatment changed students’ partisan evaluations. We conclude by considering theoretical avenues for future experiments that change the nature and extent of cross-party social interaction.

2021

“Uncovering the Online Social Structure Surrounding COVID-19.”

Philip Waggoner, Robert Y. Shapiro, Samuel Frederick, and Ming Gong.

The Journal of Social Computing.

How do people talk about COVID-19 online? To address this question, we offer an unsupervised framework that allows us to examine Twitter framings of the pandemic. Our approach employs a network-based exploration of social media data to identify, categorize, and understand communication patterns about the novel coronavirus on Twitter. The simplest structure that emerges from our analysis is the distinction between the internal/personal, external/global, and generic threat framings of the pandemic. This structure replicates in different Twitter samples and is validated using the variation of information measure, reflecting the significance and stability of our findings. Such an exploratory study is useful for understanding the contours of the natural, non-random structure in this online space. We contend that this understanding of structure is necessary to address a host of causal, supervised, and related questions downstream.

Working Papers

"Kill(ed) Bills: How Agenda Control Affects Roll-Call Ideal Point Estimates"

Samuel Frederick and Diana Da In Lee.

Statistical models that estimate spatial locations of individual decision-makers are a critical component of the scientific study of legislative politics. While ideal point estimates based on roll call voting have contributed significantly to the study of Congress, considering only those bills that reach the floor may generate results influenced by the majority party’s agenda choices and over-emphasize intra-party homogeneity and inter-party polarization. We employ natural language processing and predictive machine learning models to impute preferences on all bills introduced between the 103rd and 116th Congresses and estimate the ideal points of members of Congress with these imputed preferences. Using these new ideal point estimates, we uncover various aspects of legislative behavior that were masked by the conventional roll-call-based ideal point estimation, including the multi-dimensionality of the policy space as well as intra-party conflict. We also show that roll call estimates of ideal points display more partisan polarization than our estimates using pre-floor bills. In sum, we find that agenda setting may lead to overestimates of partisan polarization and underestimates of intra-party heterogeneity in floor-based ideal points.

"IMAP: A Tweet-Based Index of Messaging and Affective Polarization Among Partisan Elites"

Samuel Frederick.

Recent research has advanced our understanding of mass-level affective polarization, defined in terms of partisans’ identity-based feelings toward the two major parties in the U.S. Mass-level affective polarization has risen in recent decades and appears to carry real consequences for opinions and behavior. Less is understood, however, about whether political elites are similarly affectively polarized and the consequences this carries for elite behavior and representation. Recent events, including the January 6th Capitol Insurrection, as well as scholarship on the behavioral consequences of partisanship for the public, have laid bare the need to understand partisan animosities among politicians. In this paper, I propose a novel framework for measuring elite partisan messaging and affective polarization which I call the Index of Messaging and Affective Polarization (IMAP). IMAP captures both the prominence and direction of sentiment toward the two major parties in tweets posted by elected officials, combining machine learning methods with Bayesian item-response models. After describing the measure, I demonstrate that the measure comports with the reputations of prominent national political figures as extreme partisans. Next, I illustrate how IMAP relates to a variety of behavioral outcomes, including congressional bill co-sponsorship decisions, public responses to partisan scandals, and mass-level affect. IMAP also appears to capture the behavior of American governors as well. Finally, I conclude by considering potential avenues for future research employing IMAP.

"Reducing Challenges for Challengers? Social Media and the Decline of the Incumbency Advantage"

Samuel Frederick, Gregory Wawro, and Ethan Cooper.

Scholars have recently noted a decline in the incumbency advantage in congressional elections. Conventional wisdom suggests that this decline is due to an increase in partisanship which makes incumbents more susceptible to national partisan tides. However, little work has considered an additional explanation for the decrease in the incumbency advantage: the rise of social media. Social media are a virtually costless tool for candidates to communicate directly with the public, potentially making it easier for challengers to reach constituents and to develop individual brands. As such, social media may lower the barriers to challengers seeking office. In this paper, we draw on a unique database of social media posts by congressional candidates over four election cycles to test the hypothesis that social media reduce the incumbency advantage. Additionally, we present new data from the social media accounts of state legislative candidates in the 2022 elections. We combine these data with a regression discontinuity design to examine the causal effect of social media on electoral performance: do narrow primary election wins by challengers with a social media presence decrease the vote share of incumbent candidates? These analyses allow us to rigorously test whether and how social media contribute to the recent decline in the incumbency advantage.

Works in Progress

“Factional Structure in Primary Elections”

Samuel Frederick and Shigeo Hirano.

“Intra-Party Factional Social Identities”

Samuel Frederick.