Predicting probability of treatment after using eregress

[ghoetker]

I'm still working my way through the power of the new

eregress

commands. Can someone please tell me how I would predict the probability of treatment for each observation after running

eregress alliance_performance i.gov_low i.gov_low#c.($sources) i.gov_low#c.($controls), ///
    entreat(gov_high=$sources $controls $selection) vce(robust)

I suspect I'm missing something obvious, but multiple readings of the documentation have left me unenlightened. Many thanks.

[William Lisowski]

Disclaimer: I am not a user of eregress, but looked at the documentation in the hopes that something you overlooked might leap out at me.

It seems to me that the predict advanced section of the Stata Extended Regression Models Reference Manual documents a way to coerce predict into using the results from the endogenous treatment model for the variable gov_high rather than the results from the main equation.

I stumbled across that by following this hint in the output of help eregress postestimation

[ERM] predict advanced describes less commonly used predictions, such as predictions of outcomes in auxiliary equations.

My apologies if you've been there, read that.

[ghoetker]

Thank you very much for looking through the documentation. I had missed that entirely. It ends up to be rather simple, once I read that part of the documentation.

 
 eregress alliance_performance i.gov_low i.gov_low#c.($sources) i.gov_low#c.($controls), ///     entreat(gov_high=$sources $controls $selection) vce(robust)  predict prob_gov_high, eq(gov_high) pr