I combine computational methods (natural language processing, machine learning, web-scraping, etc.) with causal inference methods (experiments, quasi-experiments, panel data models). I often use the computational approach to collect and measure novel forms of data and the causal inference approach to test substantive theories with this data. The two approaches complement each other well. I have specific expertise in using large language models (LLMs) for measurement.

I examine the ideological landscape of media and how media influences consumers' ideology, perceptions, and political behavior. My focus has been on internet-based text media (e.g., online news articles), but I am also interested in online videos and social media. I have created one of the largest and most diverse datasets of online media available, new LLM methods for measuring media ideology, and tested for two-dimensional ideological gaps in media markets.

I generally use spatial models of voting behavior and a two-dimensional model of ideology (economic/cultural). I am specifically interested in how people perceive party positions and how this influences their political behavior.  I have designed and conducted a survey with eight experiments that tests how media influences spatial voting calculations and voting behavior. I have also conducted quasi-experimental research on spatial voting and public opinion.