search

kellieotto / ModelMatch

5 stars 3 forks source link

implement propensity score matching #1

Open pbstark opened 10 years ago

pbstark commented 10 years ago

To compare with model-based matching, we should implement propensity score matching.

See https://github.com/statsmodels/statsmodels/issues/858 https://github.com/jburroni/statsmodels/tree/psmatch

potential data: Dehejia, Rajeev H, and Sadek Wahba. “Propensity Score-matching Methods for Nonexperimental Causal Studies.” Review of Economics and Statistics 84, no. 1 (2002): 151–161. http://users.nber.org/~rdehejia/nswdata2.html

A Tutorial and Case Study in Propensity Score Analysis: An Application to Estimating the Effect of In-Hospital Smoking Cessation Counseling on Mortality, Peter C. Austin. Multivariate Behav Res. Jan 2011; 46(1): 119–151. Published online Feb 18, 2011. doi: 10.1080/00273171.2011.540480 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3266945/

pbstark commented 10 years ago

implemented; still need to test against public data

kellieotto commented 10 years ago

Propensity scores don't match the Dehejia & Wahba paper; there's a difference of about 0.05 in the mean propensity scores. Other means in the data do match though.

My estimated average treatment effect is within 1 bootstrap SE of the one estimated in the paper.