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]

st: RE: Interaction terms in fixed effects model


From   "Jacobs, David" <jacobs.184@sociology.osu.edu>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Interaction terms in fixed effects model
Date   Fri, 7 Sep 2012 18:54:04 +0000

You don't need to include separate main effects if they already are in a fixed-effects model.  For example, if one is interacting a variable that is not time invariant by region or time and the region and time "dummies" are already included in the model, then, of course, you can't include them as separate main effects.  A (brief) mention of this can be found in Allison's Sage little paperback book on fixed-effects models if you need a source.

Dave Jacobs

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Laura Brandstetter
Sent: Friday, September 07, 2012 5:49 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: Interaction terms in fixed effects model

Dear Statalist,

I would highly appreciate any comments on the following problem:
I have a panel dataset (6 yrs) and I use a fixed effects estimator including year dummies. I create interaction terms between different dummy variables and the independent variable of major interest. In different specifications, I analyze the effect of the independent variable, and the interaction term, but it's not possible to include the dummy used for the interaction term for several reasons:

Specification 1: dummy1 bisects the observation period; adding the dummy would cause a collinearity problem as year dummies are already included Specification 2: dummy2 is time-invariant and drops out when using fixed effects

Are these specifications still possible, even if I can't include both variables used for the computation of the interaction term?
Unfortunately, I didn't really find a clear answer yet.

Thank you so much for your help!

Laura
*
*   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/

*
*   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