st: prvalue after mlogit with a difference-in-difference strategy

[ANNAMARIA MILAZZO <annamaria.milazzo@phd.unibocconi.it>]

Dear Statalisters

I am running a multinomial logit model where the dependent variable is the "highest level of education completed" which has 4 categories.

It's a diff-in-diff model, where I want to see if an educational program had any effect on the probability of completing a given level of education for the Treatment group (compared to the control group) after its implementation.

Therefore, the model looks like: 

mlogit edu_level = beta1 Treat + beta2 Post + beta3 Treat*Post + Controls + e

Then, I want to compute the change in the predicted probabilities (for each of the categories in the outcome variable) for the Treatment group after the implementation of the program COMPARED to the control group.

I think this amounts to using prvalue TWICE, in the following way:

---FOR THE TREATMENT GROUP
prvalue, x(Treat=1 Post=0 Treat*Post=0) save
prvalue, x(Treat=1 Post=1 Treat*Post=1) dif

---FOR THE CONTROL GROUP
prvalue, x(Treat=0 Post=0 Treat*Post=0) save
prvalue, x(Treat=0 Post=1 Treat*Post=0) dif

and then subtracting the difference for the control group to the difference for the treatment group. 

But how can I compute directly the double difference? is there a way to do this with prvalue? how to get confidence intervals for the double difference?

Many thanks!

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/