Friday, October 19, 2018
Dartmouth University Program in Quantitative Social Science
We show how to use matching in causal inference to ameliorate model dependence -- where small, indefensible changes in model specification have large impacts on our conclusions. We introduce matching methods that are simpler, more powerful, and easier to understand than existing approaches. We also show that propensity score matching, an enormously popular method, often accomplishes the opposite of its intended goal -- increasing imbalance, inefficiency, model dependence, and bias -- and should not be used in applications. See http://bit.ly/causeI for papers and easy-to-use software to implement all methods discussed.