Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: mlogit problem


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: mlogit problem
Date   Sun, 17 Feb 2013 23:29:57 -0500

On Sun, Feb 17, 2013 at 7:31 PM, saqlain raza <bhatti_sb@yahoo.com> wrote:
> Thanks JVerkuilen for your response. N=360.

For regression models it's good practice to maintain a minimum of
10-20 observations per parameter, which is ordinarily the number of
variables. With mlogit, you need to multiply the number of variables
by K-1 where K is the number of categories because there is a
parameter for each category by variable combination, less the
reference category. Even so I'd err on the high side. Thus I'd suggest
with your data you can afford something like 6-12 variables to have a
chance at having reasonably well-defined coefficients with sensible
asymptotic standard errors given your sample size.

Even when your model is running, the parameters must be completely
wonky. If the standard errors are huge that's a sign.


Yes I want to do variables selection for my study. If this is not a
good idea, what should I do?
> Thanks again for your cooperation

Others have addressed this point as well as I think is even possible.
You're hoping for a miracle.

Jay
*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index