st: Heckprob postestimation (predict)

[george konteos <gkonteos@panafonet.gr>]

Dear statalisters, 

 

 I am trying to estimate the impact of training programs on employment in Greece and I am using the heckprob command but I am a little bit confused with the postestimation predictions. Which one of all is the average treatment effect on the treated?

 

the Pr(depvar=1, depvar_s=1) or the Pr(depvar=1 | depvar_s=1) ? What is the difference?

 

And what is the average treatment effect? Perhaps the Pr(depvar=1) ?

 

I appreciate any suggestions

 

Many thanks