This course is intended to be a seminar in which we discuss research designs which have succeeded. Few causal inferences in the social sciences are compelling. We carefully examine successful examples to see why they work. The seminar is also a forum for students to discuss the research designs and methods needed in their own work. It should be particularly helpful for students writing their prospectus or designing a major research project. The seminar will be supplemented by lectures to cover the statistical and computational material needed to understand the readings such as matching methods, instrumental variables, regression discontinuity, and Bayesian, maximum likelihood and robust estimation. Applications are drawn from a variety of elds including political science, statistics, economics, sociology, and public health.
Political Science 236A/Statistics 239A (The Statistics of Causal Inference in the Social Sciences) or equivalent. Experience with R is assumed.