Jacob R. Brown

Jacob R. Brown

Postdoctoral Fellow, Princeton University

Jacob R. Brown

Political scientist studying where people live, and how where they live influences their politics.

Selected Working Papers

Partisan Conversion Through Neighborhood Influence: How Voters Adopt the Partisanship of their Neighbors. (Supporting Information). [Covered by the New York Times]. Under Review.
Abstract Recent studies show that American neighborhoods have become politically homogeneous, raising concerns about how geographic polarization divides parties and influences voters. But it remains unclear how voters are influenced by the politics of their neighbors. I argue that voters are influenced by local norms when defining their own partisan affiliations, adopting local partisanship. Panel data on 41 million voters from 2008-2020 and an original survey of 24,623 respondents demonstrate that exposure to partisan neighbors increases party switching. These effects are largest for older voters, voters in single-family communities, and voters with more same-race neighbors. Survey data support mechanisms of social influence: voters accurately perceive local partisanship, interact more with partisans they live near, and are more comfortable when their partisanship matches neighbors' political affiliations. Partisanship is thus shaped by where voters live and who they live close to, demonstrating the behavioral consequences of geographic polarization.

Partisan Segregation and Partisan Activation: How Geographic Polarization Increases Political Engagement.
Abstract Do voters participate more in politics when they live around neighbors who share their party? I argue that voters are more likely to be socialized into politics when they live near more co-partisans. Using administrative data on every registered voter across 30 U.S. states from 2012-2021 and an original survey of 45,139 voters, I measure the effect of partisan neighborhoods on voting and other political engagement. Focusing on voters who do not change residences between elections, I find that increased exposure to co-partisan neighbors makes voters more likely to turnout, volunteer for campaigns, attend protests, participate in local political meetings, and to publicly express their partisanship through lawn signs, clothing, or bumper stickers. Survey data further show that voters respond to in-group exposure by becoming more comfortable in their neighborhood and more likely to politically engage with neighbors, consistent with social influence driving participatory effects.

Measuring and Modeling Neighborhoods. (with Cory McCartan and Kosuke Imai). (Survey tool). Under Review.
Abstract The availability of granular geographic data presents new opportunities to understand how neighborhoods are formed and how they influence attitudes and behavior. To facilitate such studies, we develop an online survey instrument for respondents to draw their neighborhoods on a map. We then propose a statistical model to analyze how the characteristics of respondents and geography, and their interactions, predict subjective neighborhoods. We illustrate the proposed methodology using a survey of 2,572 voters in Miami, New York City, and Phoenix. Holding other factors constant, White respondents tend to include census blocks with more White residents in their neighborhoods. Similarly, Democratic and Republican respondents are more likely to include co-partisan census blocks. Our model also provides more accurate out-of-sample predictions than standard distance-based neighborhood measures. Lastly, we use these methodological tools to test how demographic information shapes neighborhoods, but find limited effects from this experimental manipulation. Open-source software is available for implementing the methodology.

The Increase in Partisan Segregation in the United States and its Causes. (with Enrico Cantoni, Ryan Enos, Vincent Pons, and Emilie Sartre). [Covered by the New York Times].
Abstract This paper provides novel evidence on trends in geographic partisan segregation. Using two individual-level panel datasets covering the near universe of the U.S. population between 2008 and 2020, we leverage information on individuals’ party affiliation to construct two key indicators: i) the fraction of Democrats among voters affiliated with either major party, which reveals that partisan segregation has increased across geographical units, at the tract, county, and congressional district levels; ii) The dissimilarity index, which measures differences in the partisan mix across distinct sub-units and highlights that partisan segregation has also increased within geographical units. Tracking individuals across election years, we decompose changes in partisan segregation into different sources: voter migration, generational change, older voters entering the electorate, and voters changing their partisanship or their registration status. The rise in partisan segregation is mostly driven by generational change, in Democratic-leaning areas, and by the increasing ideological conformity of stayers, in Republican-leaning areas.

Priming Bias versus Post-treament Bias in Experimental Designs. (with Matthew Blackwell, Sophie Hill, Kosuke Imai, and Teppei Yamamoto).
Abstract Conditioning on variables affected by treatment can induce post-treatment bias when estimating causal effects. Although this suggests that researchers should measure potential moderators before administering the treatment in a survey experiment, doing so may also bias causal effect estimation if the covariate measurement primes respondents to react differently to the treatment. This paper formally analyzes this trade-off between post-treatment and priming biases in three experimental designs that vary when moderators are measured: pre-treatment, post-treatment, or a randomized choice between the two. We derive nonparametric bounds for interactions between the treatment and the moderator in each design and show how to use substantive assumptions to narrow these bounds. These bounds allow researchers to assess the sensitivity of their empirical findings to either source of bias. We extend the basic framework in two ways. First, we apply the framework to the case of post-treatment attention checks and bound how much inattentive respondents can attenuate estimated treatment effects. Second, we develop a parametric Bayesian approach to incorporate pre-treatment covariates in the analysis to sharpen our inferences and quantify estimation uncertainty. We apply these methods to a survey experiment on national identity. We conclude with practical recommendations for scholars designing experiments.

Other Writing