Dear Stata users,
I am planning to evaluate the impact of a training on agents using the propensity score (then I still have to choose between using the propensity score for matching or weighting but this is a next step).
The thing is that I have panel data, and instead of using two periods only (one pre-treatment and one post-treatment), I'd like to use all my periods, all the more so as the treatment occurred at four different times (4 training sessions at different dates).
Do you know if this is possible with the "teffects" command ?
It seems to me that it is not, and then I plan to estimate the propensity score manually and then use a regression to estimate the effect of the treatment on the treated using DID, most probably using the pscore as inverse probability weight, and bootstrapping the whole procedure to get correct standard errors.
However, if it is possible to do everything with the teffects command, I guess it would be easier.
Thanks for the advice.
I am planning to evaluate the impact of a training on agents using the propensity score (then I still have to choose between using the propensity score for matching or weighting but this is a next step).
The thing is that I have panel data, and instead of using two periods only (one pre-treatment and one post-treatment), I'd like to use all my periods, all the more so as the treatment occurred at four different times (4 training sessions at different dates).
Do you know if this is possible with the "teffects" command ?
It seems to me that it is not, and then I plan to estimate the propensity score manually and then use a regression to estimate the effect of the treatment on the treated using DID, most probably using the pscore as inverse probability weight, and bootstrapping the whole procedure to get correct standard errors.
However, if it is possible to do everything with the teffects command, I guess it would be easier.
Thanks for the advice.
