[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: Re: Rép. : Re: st: Multicollinearity and logit

From   "Nick Cox" <>
To   <>
Subject   st: RE: Re: Rép. : Re: st: Multicollinearity and logit
Date   Wed, 19 Mar 2008 15:27:43 -0000

-vif- is out-of-date as of Stata 10 (and probably earlier), but is still there. 

-estat vif- is now recommended. 

But -vif-, as said, gives an inappropriate error message after -logit- 
and -estat vif- won't play at all. 

My understanding, like Richard's, is that anything that examines the collective properties 
of the predictors alone should be accessible after a modelling command, 
regardless of the response variable or link function used in that model. 
That may be a counsel of perfection, but it is a clear ideal. 


Richard Williams

At 05:21 AM 3/19/2008, Herve STOLOWY wrote:
>Dear Richard:
>I thank you for your detailed reply.
>For your information, I discovered the -vif, uncentered- because I 
>had typed -vif- after -logit- and got the following error message:
>not appropriate after regress, nocons;
>use option uncentered to get uncentered VIFs
>Best regards

I got the same thing.  It seems like a nonsensical error message to 
get after running logit, which again makes me wonder if there is some 
sort of bug in -vif-.

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index