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 at the end of May, 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 using pweight


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: mlogit using pweight
Date   Thu, 17 Nov 2011 14:35:45 -0600

On Thu, Nov 17, 2011 at 1:01 PM, Polloni Stefano
<stefano.polloni@umontreal.ca> wrote:
> Greetings Everyone,
> I have a few question regarding the use of pweights in multinomial logistic regressions:
> (1) Why are the computed standard errors necessarily robust?

Because your data are not i.i.d. (some observations are more likely
than others), and the inverse information matrix standard errors only
work with i.i.d. data. See
http://www.stata.com/support/faqs/stat/svyml.html.

> (2) Why is the log-likelyhood as big as if frequency weights were used i.e as if each observations were dulplicated according to their weight

Because in both cases, the objective function (the
pseudo-log-likelihood... or the log-pseudo-likelihood) is = sum over
observations of [ weight * the likelihood contribution of each
observation ].

> (3) How does this affect the comparability off the AIC and BIC tests computed with the fitstat command (SPost.ado by Long & Freese).

Information criteria are not applicable to the survey weighted data.
See http://www.stata.com/support/faqs/stat/lrtest.html.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   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/


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