Re: Re: st: specifying the nested design for xtmelogit

[Clyde Schechter <clyde.schechter@gmail.com>]

Sorry to come late to the party.

Dana Shills inquired a couple of days ago whether

xtmelogit Y X ||industry: ||firm:

is a correct specification for his model given that some firms may be
associated with multiple industries.

Alfonso Sanchez-Penalver has responded, advising Dana to use, instead,

xtmelogit Y X industry#firm || industry: || firm:

But I don't think either of these is right.  The problem is that there
need to be crossed random effects, and both of the above
specifications still imply nesting of firm within industry.  I think
what Dana needs to do is

xtmelogit Y X || _all: R.industry || firm:

(I'm assuming that there are fewer industries than firms.  If the
reverse is true, then use -xtmelogit Y X || _all: R.firm: ||
industry:-   Actually, either version is theoretically correct, but
computational efficiency will differ.)

If Dana has version 13, looking at pages 32 and forward in the ME
section of the manual will explain this, and other helpful information
about crossed random effects.  (Since -xtmelogit- is now -melogit- in
version 13, I'm assuming Dana has an older version.  I don't have any
older versions installed on my current machine.)

Clyde Schechter
Dept. of Family & Social Medicine
Albert Einstein College of Medicine
Bronx, New York, USA
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/