POL 684: Causal Inference

This course addresses causal inference in quantitative analysis of public policy and program evaluation. We begin with a review of key questions in regression analysis as well as the problem of selection bias in evaluating treatment effects in the real world versus an ideal experiment. We then turn to randomization as a potential solution but address difficulties in theory and practice. Thus, we spend the bulk of the semester focusing on quasi-experimental techniques - such as regression discontinuity, difference-in-difference, matching, and IV - that allow researchers to exploit instances of natural experiments to (arguably) identify a causal relationship, paying special attention to key testable and untestable assumptions that may underlie each technique. Students will learn the theory behind the techniques, apply techniques to real-world policy data, and critically read cutting-edge policy studies that utilize the methods across a wide range of real-world topics including health insurance, education, environment, housing, and crime.