Controlling unobserved heterogeneity and heteroskedasticity in probit models

[Antonio Serrano]

Dear statalisters,

I hope you can help me out with this one. I have a panel data from 1998 to 2007. My model includes one dependent variable, four independent and two control variables. The dependent variable is dichotomous, so I estimated my model using probit model. My regression would be:

xtprobit RADIC DCT EPCT IILR IRI SIZE ABS

I have already solved problems of multicollinearity. My question are:

1. Does it make sense checking unobserved heterogeneity and heteroskedasticity in probit models?

If so,

2. How can I control unobserved heterogeneity? Using Breusch-Pagan test? Which command performs B-P test in probit models?

3. Regarding heteroskedasticity, How can I check it?

4. If there is heteroskedasticity, How can I solve it?

Thanks in advance guys

Kind regards,
Antonio