Hello Statalist,

I am investigating the determinants of participation and contributions to pensions using pooled cross-sectional data. I believe the model is best described as a binary model with sample selection model since individuals make a decision to participate, then based on the initial decision choose to contribute or not. I am wondering whether there is any functions within stata that can test for the normality and homoskedasticity of the errors since the theory states that model suffers heavily when deviating from normality.

I understand a RESET test can be used to check for these in a probit case however I am unsure whether the sample selection issue of the model will effect the usefulness of this test.

Thanks in advance

Callum Crosby