Hello!

I have a rather "practical" question. If you have a panel data with T=3, but some of the variables are only measured in the first two waves, what would be the best strategy to account for these variables, if the only option is not to discard them?

I estimate a within-between model, namely, certain variables are treated as fixed effects specifications, and some other variables are treated as random effects specifications:

Is it best to add them as they are (FE specifications), in the first two waves, but what would be the interpretation in the three waves setting?

Thanks!