Summary: I design a new experimental tool for portably identifying motivated reasoning from Bayesian updating, and use this tool to show that politics affects inference about factual questions on a broad range of topics: immigration, income mobility, crime, racial discrimination, gender, climate change, and gun laws. Politically-motivated reasoning is more severe for partisans and leads people to over-trust news more when it reinforces the biases in their beliefs.
Gender Differences in Motivated Reasoning (Revise & Resubmit, Journal of Economic Behavior and Organiztion)
Summary: I show experimentally that there are statistically significant gender gaps in motivated reasoning about outperforming others, but that motivated reasoning about political issues is similar across genders. Motivated reasoning can help explain why there are gender gaps in overconfidence.
Summary: I experimentally test whether people motivatedly reason towards believing the world is a good place for others, and find a well-powered null effect. Via a survey, I show that respondents anticipate that these types of "positive" beliefs make people happier but are not necessarily beliefs that they are motivated to hold.
Polarization and Public Health: Partisan Differences in Social Distancing during the Coronavirus Pandemic (with Hunt Allcott, Levi Boxell, Jacob Conway, Matthew Gentzkow, and David Yang). Journal of Public Economics, 2020.
Summary: Using GPS data, we show that areas with more Republicans have engaged in less social distancing during COVID-19 than areas with more Democrats, and using a new survey, we show that Republican individuals self-report less social distancing and believe that the risk of COVID-19 is lower.
Works in Progress
Inference from Weak Signals and the Demand for Low-Quality Information
Summary: Do people over- or underinfer from information? I provide experimental evidence that people's inference depends on the strength of the information they receive. Experimental subjects act as if they are cognitively imprecise (Khaw et al., 2020) and partly neglect signal strength. This bias leads both to overinference from weak signals and underinference from strong ones. Consistent with the theory, inference becomes less biased as subjects become more experienced, and subjects have excess demand for the low-quality information sources.
n ∈ [6,12] Angry Men: The Importance of Endogenizing Jury Size when Comparing Voting Rules
Summary: I present a model that compares the accuracy and efficiency of jury voting mechanisms when both jury size and voting rule can vary. The main result is that an (n+2)-person jury that allows for one dissenter is both more accurate and less likely to hang than an n-person jury under unanimity. (Slides available upon request.)
The Supply of Motivated Beliefs
And Now for Something Completely Different
Tribone Tilings of Triangular Regions that Cover All but Three Holes. Discrete & Computational Geometry, 2015.
Summary: I show that there are a lot of ways to tile triangular regions with tribones when you leave three spaces uncovered.