Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Change in explained variance
From
"JVerkuilen (Gmail)" <[email protected]>
To
[email protected]
Subject
Re: st: Change in explained variance
Date
Tue, 18 Sep 2012 00:02:15 -0400
On Mon, Sep 17, 2012 at 10:51 PM, Thomas Didier
<[email protected]> wrote:
>
> I am running a multinomial logit model and would like to determine whether the inclusion of an explanatory variable in the model increases the percentage of explained variance in a statistically significant manner. I am concerned that when running the model with and without the variable of interest, its impact is diluted by a substitution effect.
>
> Would an F-test (Fisher) be appropriate for such measurement and can it be run on Stata?
>
As Rich Williams indicated, -estimates store- and -lrtest- gives you
the likelihood ratio chi square test, which is the analog of the F
test in this case. -estat ic- will give you the AIC and BIC for the
model if you want to go that route.
The multinomial logit model is kind of tricky though, given that if
you have K options, you have K-1 equations for the variable. So
testing for its exclusion is really a K-1 df test and not quite the
same as what most people think of in the context of a regression test.
If you're worried about violations of IIA (not quite sure what you
mean by substitution effect) you should probably check that as if IIA
is violated substantially you probably need a different model. The
manual entry for -suest- gives a very nice walkthrough of how to do
that.
--
JVVerkuilen, PhD
[email protected]
"Out beyond ideas of wrong-doing and right-doing there is a field.
I'll meet you there. When the soul lies down in that grass the world
is too full to talk about." ---Rumi
*
* 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/