Political Science PhD candidate focusing on American institutions


A selection of ongoing research projects


My dissertation looks at how Congress uses public policy information goods produced by organizations outside of Congress. I present a theory of ideological and interest based think tank coalitions, and explore how Congress relies on legislative subsidies produced by these outside actors. Using a dataset of over 200,000 white-papers and 14,000 congressional committee reports, this project employees text reuse detection and latent network inference methods to infer a legislative subsidy transmission network from the outside organizations that produce these white papers to the committees that produce reports on bills as they are sent to the floor. I also use survey data and survey experiments to study the staffer decision-making involved in the subsidy uptake process, focusing on the role of motivated reasoning and confirmation bias in subsidy source selection among staffers.

Paper 1 — Biasing Their Bosses: Staff Ideology and the Distortion of Information in Congress.

Congressional Capacity Project

In the fall of 2017 I fielded the 2017 Congressional Capacity Survey as part of a larger joint New America and R Street project on Congressional Capacity. This survey sought to find out more about the backgrounds, career paths, policy views, and job experiences of congressional staffers. I have ongoing projects using data from this survey investigating partisan and ideological selection in information usage and trust, ideological diversity among members' staffs and how staffer issue knowledge effects information use patterns. 

(with Tim Lapira, Lee Drutman, Alex Hertel-Fernandez and Kevin Kosar)

Semantic Smith Waterman

This project introduces a new method for detecting text reuse and paraphrasing in political texts, a critical measurement task in leveraging new text data to observe patterns of coordination, knowledge uptake, and policy transmission and diffusion. The method proposed is an extension of the Smith-Waterman local alignment algorithm with semantically aware mismatch penalties. This modification enables detection of instances of text reuse in which words are changed to semantically similar alternatives to fit new contexts or disguise the source of the text. Our implementation allows for arbitrary user defined semantic spaces, as well as arbitrary weighting of semantic dimensions. This enables users to tailor the semantic weighting to meet their domain specific needs. We apply this technique to find instances of text reuse from Congressional Research Service (CRS) Reports in congressional Committee Reports. The discovery of these instances of reuse enable us to explore when and under what conditions Committees rely on analysis produced by CRS.

(with Benjamin Edwards)

For decades, critics of pluralism have argued that the American interest group system exhibits a significantly biased distribution of policy preferences. We evaluate this argument by measuring groups’ revealed preferences directly, developing a set of ideal point estimates, IGscores, for over 2,600 interest groups and 950 members of Congress on a common scale. We generate the scores by jointly scaling a large dataset of interest groups’ positions on congressional bills with roll-call votes on those same bills. Analyses of the scores uncover significant heterogeneity in the interest group system, with little conservative skew and notable inter-party differences in preference correspondence between legislators and ideologically similar groups. Conservative bias and homogeneity reappear, however, when weighting IGscores by groups’ campaign contributions and lobbying expenditures. These findings suggest that bias among interest groups depends on the extent to which activities like contributions and lobbying influence policymakers’ perceptions about the preferences of organized interests.

The current working paper is available here.

(with Jesse Crosson and Geoff Lorenz)

MLscores: interest group ideal points