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]

Re: st: cluster SE and degrees of freedom


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: cluster SE and degrees of freedom
Date   Tue, 17 May 2011 07:38:41 -0400

Ophelie Desmarais <ophelie.desmarais@googlemail.com>:

If you are testing one coef, you should be fine, assuming the clusters
are of reasonably similar size.

If you are testing multiple coefs, the overrejection rate will rise
dramatically until you run out of df at which point the matrix will
not even have full rank and you will get missing SE.  See e.g.
http://repec.org/usug2007/crse.pdf

For add'l confirmation, note also that if you run -xtreg, fe- and
request robust SE you will be using a cluster-robust SE with M
clusters in a model with M fixed effects and k regressors.  Search
http://www.stata.com/help.cgi?whatsnew10 for "Watson" to see the ref.

On Mon, May 16, 2011 at 11:51 AM, Ophelie Desmarais
<ophelie.desmarais@googlemail.com> wrote:
> dear stata list,
>
> i am trying to estimate a regression with SEs adjusted for 40 clusters.
>
> since i would like to add fixed effects, i'm running out of degrees of
> freedom - if I run the regression, the F-Test hence is reported
> missing (F[40,41]).
>
> the SEs for the regressors are still displayed: i was wondering if
> they are still valid?
>
> thanks a lot!
*
*   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