Here we are going to look at several diagnostic plots are helpful when attempting to answer a causal question. They can be used to visualize the target population, balance, and treatment effect heterogeneity. Setup I’ve simulated data to demonstrate the utility of the various plots. In each simulation, we have four pre-treatment variables: var1, var2, var3, and var4, a treatment, t, and an outcome y. I have also fit a propensity score model for each and calculated ATE, ATT, and overlap weig...
