Backward stepwise selection for mixed-effect logistic regression

[Abdullah Algarni]

Dear colleagues,

Is there a package that can do backward stepwise selection for mixed-effect logistic regression in stata? I'm aware of fixed-effect model package but did not find similar package for mixed-effect logistic regression


Thank you
Abdullah

[Clyde Schechter]

Stata's -stepwise- command does not support mixed-effects estimation commands. There may be a user-written command that does, but I am not aware of any. -findit- might be helpful for that.

BUT, but, but... please don't do this! See https://www.stata.com/support/faqs/s...sion-problems/ for a list of compelling reasons why stepwise selection is just bogus and should never be used.

[Abdullah Algarni]

Hmmmm, thanks Clyde,

It seems wise doing manual comparison of models

[Carlo Lazzaro]

Abdullah:
the most severe drawback of stepwise approach(es) is to end up with models that have, at best, a tenuous relationship with the original dataset(s).