App Stats: Lam on "Estimating Individual Causal Effects"

We hope you can join us this Wednesday, April 10, 2013 for the Applied Statistics Workshop. Patrick Lam, a Ph.D. candidate from the Department of Government at Harvard University, will give a presentation entitled "Estimating Individual Causal Effects". A light lunch will be served at 12 pm and the talk will begin at 12.15.

"Estimating Individual Causal Effects"
Patrick Lam
Government Department, Harvard University
CGIS K354 (1737 Cambridge St.)
Wednesday, April 10th, 2013 12.00 pm


The literature on causal inference has focused primarily on estimating average treatment effects, which aggregate over many individual effects. However, this aggregation often misses treatment effect heterogeneity, which may be of extreme importance. In addition, researchers often estimate average effects but their real quantity of interest is individual effects. In this paper, I develop methods to estimate individual causal effects based on commonly used matching procedures. I show that predictive mean matching performs the best in imputing missing potential outcomes to estimate the individual effects. I then demonstrate the flexibility of estimating individual causal effects and how they can be used to explore questions of interest, recover any other causal quantity, and be adapted to more complicated data structures. I conclude with empirical examples from political science.

Posted by Konstantin Kashin at April 8, 2013 12:41 AM