Probit with Panel Data

[Jay Moorcroft]

Hi Everyone,

I am using two waves of panel data to assess the impact of socio-demographic characteristics of households on a categorical dependent variable.
I am using XTProbit to do so.

I have aggregated individual data to household level, and used the household reference person as the observation for individual characteristics. - Should i cluster the standard errors in the regression?

Thanks.

[Phil Bromiley]

You didn't get a quick answer. Whether you should cluster depends on your substantive question. It is quite likely that you have different error variances in different households and/or correlated errors within households so clustering is a good idea.