[Joseph Coveney]

So, if I understand correctly, you can use mixed effects models only on data in which you measure the covariantes AND the outcome repeatedly over time?

The outcome. Just the outcome:

version 15.1

clear *

set seed `=strreverse("1436437")'
quietly set obs 500

generate double pid_u = rnormal()
generate int pid = _n

quietly expand 5
generate byte outcome = rbinomial(1, invlogit(pid_u))

melogit outcome || pid: , nolog

exit

[Joseph Coveney]

That's a bit glib, sorry. Obviously you'd have to assess at least one covariate (the patient's identity) more than once, but the point is: you can measure an attribute, say, a patient's sex, only once and you can still fit a mixed-effects regression model with it as a covariate. It's not required that each covariate be measured each time. It's not required that any covariate change values each time.

[Horea Feier]

OK, now I understand. Thank you so much for pointing me out, gentlemen.