Estimating the joint effect of two interventions using PSM

[Anagaw Derseh]

Dear Colleague

I would like to investigate the effect of two health interventions (X and Y) on health status (Z) using propensity score matching (PSM).
Most of the factors that affect the decision of the individuals to participate in X and Y programs are similar but not exactly the same. So in this case, how can I predict propensity scores showing the probability of participation in the two programs using the same logit model?

As mentioned above, my main interest is to show the join impact of the two interventions (X*Y) on the outcome (Z). Can I also estimate the effect of the two interventions and the joint effect at the same time i.e average treatment effect on tread (ATT) for X, Y and X*Y?

[Fei Wang]

Anagaw, I would treat your case as having three treatment levels (1. X=1 and Y=1; 2. X=1 and Y = 0; 3. X=0 and Y=1) and one control level (X=0 & Y=0). You may generate such a multi-level treatment indicator and use teffects ipwra to estimate the effect of each treatment level.

[Jeff Wooldridge]

I agree with Fei. Matching isn’t well suited to multiple treatments but IPWRA is.