Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: FE and groups w/out variation

From   "Guillaume Frechette" <>
Subject   Re: st: FE and groups w/out variation
Date   Sun, 20 Oct 2002 18:56:05 +0000

Thanks to Scott and Mark, Scott's answer was about regressors w/out variation which is not what I had in mind (my fault, I should have been clearer). Mark is closer to what I was asking about, except that in my case its not simply singletons (although there are some), but some groups, which have more than one observations, have always the same value for their dependant variable. BTW my intuition is the same as what Mark expressed. I'll now try to track down David's post. Thanks again.


From: Mark Schaffer <>

My recollection is that this was posted by David Drukker (apologies, David, if I'm in correct in
attributing this to you), and the advice was indeed not to drop singleton groups.

It's not clear to me why you would ever want to drop singleton groups. A group with one member is
an observation that still has some information in it. If you throw away that observation, you're
throwing away information unnecessarily.

Another way to put it is to consider what happens when you estimate the fixed effects model after
transforming the data into mean-deviation form, so that all the FEs disappear. The FE estimation
is obtained by simple OLS on the transformed data. After the MD transformation, you can't tell
just by looking at the data which datapoints are several observations from one group, and which
datapoints are observations from groups with just one member; in a sense, after the transformation,
they're all independent observations. You wouldn't want to throw away observations here, since the
attribute that distinguishes these obs (they are the sole representatives of their groups) has been
transformed out of the data.


Quoting Guillaume Frechette <>:

> Dear Statalisters: I thought I read on the list not so long ago someone who
> said that when estimating FE one shouldn't drop groups that do not have
> variation (I looked for it in the archive but couldn't find it). Did I
> invent this? At any rate, could someone tell me what is the 'best'
> procedure, keeping all groups or only the ones that have variation, and what
> is the logic for the answer. I would also appreciate it if someone could
> point me to a book or article that spells this out in writing.
> Thanks in advance,
> g
> _________________________________________________________________
> Broadband? Dial-up? Get reliable MSN Internet Access.
> *
> * For searches and help try:
> *
> *
> *

Prof. Mark Schaffer
Director, CERT
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3008


This e-mail and any files transmitted with it are confidential
and intended solely for the use of the individual or entity to
whom it is addressed. If you are not the intended recipient
you are prohibited from using any of the information contained
in this e-mail. In such a case, please destroy all copies in
your possession and notify the sender by reply e-mail. Heriot
Watt University does not accept liability or responsibility
for changes made to this e-mail after it was sent, or for
viruses transmitted through this e-mail. Opinions, comments,
conclusions and other information in this e-mail that do not
relate to the official business of Heriot Watt University are
not endorsed by it.
* For searches and help try:

Get faster connections -- switch to MSN Internet Access!

* For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index