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: Fixed effects/hausman-test/ different approach?


From   Maarten Buis <maartenlbuis@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Fixed effects/hausman-test/ different approach?
Date   Thu, 19 May 2011 10:50:59 +0200

--- On Thu, May 19, 2011 at 10:22 AM, Ben Ammar wrote:
> Right now I'm analyzing different firms and I want to check whether women or men are doing a better job. I have a panel data set and I use fixed effects for each firm (xtreg,fe). However, different firms have some sort of policy to employ rather women or men and as soon as I'm using fixed effects this performance-difference between men and women is absorbed and no significant difference remains. If I use a pooled regression and OLS or even random effects the difference remains. Statistically, I already could prove that firms have that sort of policy. However, I wondered if there's a way of combining that insight instead of doing first the fixed effects model and then saying: "Oh, by the way, firms rather employ women or men and that's why fixed effects doesn't show any signifance in the explanatory variable anymore." Most disturbing is that Stata or rather the Hausman-test suggests that I should use the fixed-effects which makes my argumentation pretty much two-faced.

The idea of using fixed effects is that you are only comparing
individuals within a firm. That way you are more likely to compare
like with like, but the price you must pay is that when there are only
men or only women within a firm there is nothing to compare with and
all observations from these firms will drop out of your analysis.
Pooled and random effects models also allow the comparison of
individuals across firms. Now you will be able to use all the
observations, but you are less likely to compare like with like. These
are logical consequences of the strategy you used, it is hard to see
how you can have the advantages of both methods simultaneously.

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   |   Site index