Predicting probability of a multivalued endogenous treatment

[Manish Srivastava]

I am trying to figure out how to predict probability of participating in different treatments (or multivalued treatment). Since the entreat ( ) equation does not allow mlogit option, I am not sure how to do it. In the following example , I have created 'program1' a multivalued treatment. But the prediction is not for two different types of programs ( program1==1, program1==2). Any suggestions to graph the probabilities of participation for the two different treatments?
Thank you so much for your kind help.
Sincerely,
Manish

use https://www.stata-press.com/data/r16/class10,clear
clonevar program1= program
replace program1=2 if program==1 & _n <1000
eprobit graduate income i.roommate, endogenous(hsgpa = income i.hscomp) entreat(program1 = i.campus i.scholar income, pocorrelation) vce(robust)
preserve
replace campus=1
replace scholar=1
predict p1 , pr equation(program1)
twoway line p1 income, sort

[Manish Srivastava]

A related post: https://www.statalist.org/forums/for...using-eregress

[Manish Srivastava]

#Solved
Okay, with the help from Stata Technical Support, the problem is solved. Also, note that ERM treats it as an ordinal endogenous treatment not a multinominal endogenous treatment as I had assumed.

use https://www.stata-press.com/data/r16/class10,clear
clonevar program1= program
replace program1=2 if program==1 & _n <1000
eprobit graduate income i.roommate hsgpa, endogenous(hsgpa = income ///
        i.hscomp) entreat(program1 = i.campus i.scholar income, ///
        pocorrelation) vce(robust)
preserve
replace campus=1
replace scholar=1
predict p*, pr equation(program1)
twoway line p1 income, sort name(p1)
twoway line p2 income, sort name(p2)
twoway line p3 income, sort name(p3)