www.cmm.bristol.ac.uk/forum

Hi,

I am trying to fit a two-level probit model with a cross-level interaction. One thing I wasn't sure about was in what form I should enter the main effect for the first (individual) level variable used in the interaction. Should it be a fixed or random effect? I have seen it done both ways in literature, but have not come accors of an explanation why. Is there a particular reason to prefer one approach over the other?

Thanks,
Mari

Too bad this question isn't answered, because I've got exactly the same question.