Statalist


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

Re: st: Why do logit model coefficients produce signs opposite to those obtained from OLS?


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Why do logit model coefficients produce signs opposite to those obtained from OLS?
Date   Thu, 4 Feb 2010 07:26:56 -0800 (PST)

--- On Thu, 4/2/10, Jitian Sheu wrote:
> I am running a very very simple binary model, y=a+bX+e,
> where y is a dummy variable.
> Before performing -logit- command, I estimate the above
> model by a traditional OLS, i.e. linear probability model
> (regress y x1 x2...)
> 
> I knew that OLS is not a good model for fitting this model.
> I just want to get some direction from results obtained
> from traditional OLS
> 
> However, after I perform -regress- and -logit-, I found
> signs of estimated coefficients from these two models are
> not the same.
> 
> I am just wondering whether this is "normal"? or I am doing
> anything wrong?

This is not normal. One possiblity I could imagine is that 
you specified the -or- option (or used -logistic-) and 
forgot to interpret coefficients less than 1 as negative
effects. In both case you will get odds ratios, that is, the
ratio by which the odds of "success" on y changes for a
unit change in x. An odds ratio less than 1 thus means
that a unit increase in x leads to a smaller odds, i.e.
a negative effect, even though the numberical value of the
odds ratio is positive. I wrote a small text about odds, odds 
ratios, and marginal effects (in the context of interaction 
effects), which is available here: <http://www.maartenbuis.nl/
wp/interactions.html>

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/



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