“A Unified Adjustment Formula for Generalizing Experimental Results,” Erin Hartman, UCLA

Event time: 
Thursday, April 12, 2018 - 12:00pm to 1:15pm
Institution for Social and Policy Studies (PROS077 ), A002 See map
77 Prospect Street
New Haven, CT 06511
Erin Hartman, Assistant Professor of Political Science and Statistics, UCLA
Event description: 


Abstract: Researchers are often interested in generalizing the average treatment effect (ATE) estimated in a randomized experiment to non-experimental target populations. Previous studies have shown that an unbiased estimate for the population ATE can be obtained if researchers adjust for: (1) all treatment effect moderators (moderator set) or (2) all variables affecting the sampling mechanism (sampling set), both of which are often difficult to perfectly measure. We propose that adjustment be made for a separating set–a set of variables given which decomposes the moderator and sampling variables in to conditionally independent sets–which subsumes previous methods. We show that even when both a moderator set and a sampling set are large and therefore existing approaches might fail, experimental results can be generalized to the target population by adjusting for a potentially much smaller separating set. We develop methods to efficiently find a separating set from data, and apply these methods to simulated and survey experimental data.

Erin Hartman is an Assistant Professor of Political Science and Statistics at UCLA. Her recent research focuses on creating new methods–including both theoretical approaches and new estimation strategies–for identifying and validating causal effects. In particular, she studies the methods under which experimental findings can be extrapolated beyond the experimental sample. She also studies survey design methodologies, including a new survey sampling method that reduces reliance on post hoc weighting methods and alleviate non-response bias, and an transparent weighting methods that automate the selection of the optimal auxiliary vector on which to weight.

In 2012, Erin ran the polling operation for Obama for America’s Analytics department, which very accurately predicted election outcomes in the campaign’s battleground states. She also co-founded a successful analytics and technology start-up, BlueLabs, focused on providing analytics services to clients in politics, issues advocacy, healthcare, and education.

Erin holds a PhD in Political Science and an MA in Statistics from UC Berkeley. She was also a post-doctoral fellow at Princeton University.

This workshop series is being sponsored by the ISPS Center for the Study of American Politics and The Whitney and Betty MacMillan Center for International and Area Studies at Yale with support from the Edward J. and Dorothy Clarke Kempf Fund.