Yes, there is a zero-inflated logit. Molly covered it in section (see
section 6 notes). You might also consider citing the working paper by Will
Moore and coauthors if you use the model. They "invented" the model (which
they call the Split Population Logit) although its obviously a
straightforward application of previous work. They do provide simulation
results which explore some of its properties though which is nice.
Brandon
On Fri, Apr 15, 2011 at 4:48 PM, Colin Sullivan <cdsulliv(a)gmail.com> wrote:
Hello 2001-ders,
Is "zero-inflated" a general term for mixed models where one distribution
is zero? Or does it only apply to count models?
For instance, is there a "zero-inflated logit" model for binary outcomes?
Thanks,
Colin
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l