I have done a heckprobit regression on a cross section data. My main explanatory variable is a district level variable (logtuiexp). Is it possible to additionally use district level dummies in the regression to control for unobserved heterogeneities? I had done so. However, two district dummies are getting dropped. I am getting some results though.

Here is my code
heckprobit govt male rural logtuiexp i.dist logfeediff mpce [pweight=Multiplier_combined], select(curattend = male rural i.dist logtuiexp logfeediff mpce bpl) vce(cluster dist)
note: logtuiexp omitted because of collinearity
note: logfeediff omitted because of collinearity
note: 459.dist omitted because of collinearity
note: logfeediff omitted because of collinearity

Depending on the way I enter i.dist in the regression, my results are changing. If I end it after all other district level variables, stata is unable to identify.

Any guidance would be extremely help me to proceed with my work.

Thanks in advance
Sayoree Gooptu
PhD Research Scholar