I am doing multilevel data analysis using meglm command.
. meglm y x, offset(t) || id:, family(nb)
. meglm y x, offset(t) || id:, family(p)
. meglm y x, offset(t) || id:, family(gamma)
The problem I've faced is the distribution of dependent variable (y). It's orginally from count data, but it was divided by time (t).
I tried to use meglm with a family option: family(nb), family(p), family(gamma) ; however, it doesn't look like negative binomial nor poisson. Actually, menbreg or mepoisson didn't working at all (endless iteration). Oh, I also generated new variable (y_ = y+0.001)
Can anybody help me please?

0 Response to What distribution does it look like?
Post a Comment