Is it necessary to have robust standard error when runing probit?

[Kanoknit Tanjan]

An econmomist suggested that I should run the following command when running binary probit or multinomial peobit. So this cam help solve the problem of heteroskedascity ? However, as i understand, heterskedascity is sth that we have to check for linear regression. But for probit (binary or multinomial), does it need robust standard error?

probit y x1 x2, robust


Many many thanks in advance for advice

Kanoknit
Social Reseacher and PhD student
Bangkok, Thailand

[Joro Kolev]

I think in this case the conclusion was that robust variance does not make much sense in probit, because if you have heteroskedasticity this is a misspecification issue, and the standard errors are your last problem. See this blog here:
https://blog.stata.com/2016/08/30/tw...andard-errors/

But on the other hand the computer will not break if you calculate robust variance.

[Kanoknit Tanjan]

Many thanks for your advice.