Explaining Ideological Variation in Presidential Campaign Stump Speeches

I use the text of Presidential stump speeches from 1952 to 1996 to quantify within-candidate ideological variation in presidential platforms. I show that the ideological content of these speeches vary substantially from one location to the next and that the policy platforms of Presidential candidates diverge not only across parties, but across the economic and racial divisions of the American electorate. Using the text of these speeches and methods from natural language processing, I find that while both Democratic and Republican candidates campaign on more conservative economic policies in wealthier counties, Republican candidates alone deliver more socially conservative messages to low-income minority counties relative to their low- and high-income white counterparts.

Redlining and Police Shootings

Social scientists are often interested in the long-run effects of political and economic institutions. In this paper, I show how the U.S. Federal Government's historical involvement in private housing markets is associated with contemporary variation in police violence toward American citizens. In particular, I show that redlining scores — neighborhood-level grades assigned by the Federal Housing Administration (FHA) in the 1930s to determine suitability for federally-backed mortgages — predict the number of fatal encounters with police in the 21st century. I employ a Bayesian negative binomial count model and a hierarchical extension to support my argument.

Member-level Effects of Being in the Majority Core

In this paper, I examine whether occupying a privileged ideological position with respect to the social preferences of the legislature comes with financial benefits for members of Congress. I build on work that generalizes the median to spatial models of legislative bargaining in multiple dimensions and provides an algorithm for efficient computation of the majority core. After constructing a dataset that relates core membership, (euclidean) distance to the core, and PAC contributions for all members of Congress from the 107th to 114th Congresses (roughly 2001-2017), I estimate a variety of models to determine whether being in or near the core is associated with a greater dollar value of candidate contributions stemming from organized interests. While most results do not achieve statistical significance, the majority of specifications produce estimates whose sign is consistent with a PAC contribution driven financial "reward" for being in the majority core.