08-04-2017 01:30 PM - edited 08-04-2017 01:52 PM
I am reading Chapter 2 page 33 and 34 of Analysis of Observational Health Care Data Using SAS. It gives an example of how to estimate a treatment effect by stratifying on propensity scores with a dichotomous outcome. Could someone tell me how to do this with a continuous outcome. What statistic will give me the pooled estimate (similar to the Mantel-Haenszel statistic used in the example). I understand from the book that I need to estimate a regression for each quintile. I just don't know how to pool the five estimates. Thank you!