Re: st: too good to be true : lr test in mlogit?

[Maarten Buis <maartenlbuis@gmail.com>]

On Mon, May 16, 2011 at 12:02 PM, John Litfiba <cariboupad@gmx.fr> wrote:
> Well, when I run a xtlogit (I first type xtset id, where id is the
> unique id for each of my individual in my database) with on year of
> data (1million) observations I get the message
>
> Yvar is categorical (Yes=1, No=0) and Xvar is also categorical
> (type1=1, type2=2)

I would recode Xvar to such that type1 = 0 and type2 = 1:
gen byte type2 = Xvar - 1

Remember that with fixed or random effects models you are making an
additional model for the group specific constants. It often helps when
these group specific constant refer to groups that actually occur in
the data.

the logic of calling this variable type2 rather than Xvar is that the
value 1 is often associated with "true" and 0 with "false", so I think
of a variable named type2 as posing a question to each observation
"are you type2?", the answer being either true (1) or false (0).

If recoding your explanatory variable does not work I would contact
Stata's techsupport, as these results suggest that there is something
logically wrong with the model you are trying to estimate, so ad-hoc
solutions like sampling from your data are not advisable until you
understand where the problem comes from. For how to contact Stata's
techsupport, see: http://www.stata.com/support/tech-support/

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany


http://www.maartenbuis.nl
--------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/