Hello,
I have a question regarding heterogenous treatment effects. First I have estimated the propensity score given a number of covariates. Second, for each group (mothers and non-mothers) i have fitted separate nonparametric regressions of the dependent variable on the estimated propensity score. Third, I have plotted these to fitted regressions against the propensity score. This results in the following 4 graphs (because I have 4 different outcome-variables). Just by looking at the graphs, can I say something about, whether the treatment effects seems to be heterogenous or homogenous? Is it correctly understood, that if the distance between the two lines are different for varying values of the propensity score, then the treatment effects seems to be heterogeneous?

Best regards,
Mette

Array