Thanks to Mark Schaffer and Al Feiveson for shedding light on my -areg-
question.
Phil
--
Phil Ender
UCLA Department of Education
UCLA Academic Technology Services
-----Original Message-----
From: "FEIVESON, ALAN H. (AL) (JSC-SK) (NASA)" <alan.h.feiveson@...>
To: statalist@...
Date: Tue�Apr�26,�2005� 6:35 am
Subject: RE: st: areg question
Using -araeg- attempts to model cluster effects by a constant offset for
each group. There may be further correlation within clusters even
after a
mean level adjustment. If so, using the cluster option would still be
appropriate.
AL Feiveson
-----Original Message-----
From: owner-statalist@...
[mailto:owner-statalist@...]On Behalf Of Mark Schaffer
Sent: Tuesday, April 26, 2005 4:51 AM
To: statalist@...
Subject: Re: st: areg question
It's not a problem, and is in fact a recommendation originally
offered by Arellano (1987). Wooldridge's 2002 book discusses it (I
don't have it handy or I would look up the page ref).
You do need to be sure that you have enough clusters for the
asymptotics to be plausible, but this is a general issue with cluster-
robust and not specific to fixed effects models.
Hope this helps.
--Mark
To: statalist@...
From: Phil Ender <ender@...>
Subject: st: areg question
Date sent: Mon, 25 Apr 2005 17:29:00 -0700
Send reply to: statalist@...
> Dear Statalist,
>
> A student came in with an -areg- model in which the same variable was
> used in both the absorb option and cluster option, sorta like this
>
> areg dv iv1 iv2, absorb(id) cluster(id)
>
> I thought this was a bad idea but could not find a specific reference
> prohibiting it. Can someone point me to a reference about the
> problem or provide an explanation of why it is not a problem.
>
> Phil
> --
> Phil Ender
> UCLA Department of Education
> UCLA Academic Technology Services
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/