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

st: Fixed effect logit


From   Michael Ash <mash@econs.umass.edu>
To   statalist@hsphsun2.harvard.edu
Subject   st: Fixed effect logit
Date   Fri, 12 Mar 2004 03:33:30 -0500 (EST)

- A panel that always has the same outcome does not
contribute to identification in a FE logit because the best
intercept term, or fixed effect, for such a unit would be
-infinity for always 0 and +infinity for always 1.  This
explains the 2642 groups that are dropped: each household is
either always poor or always nonpoor.

- There is a Stata FAQ that addresses the first "note":
http://www.stata.com/support/faqs/stat/clogit.html
Identification in FE logit comes from comparing patterns of
0 and 1 outcomes among households that have the same total
number of 1 outcomes over the course of the panel.  An older
estimation method could run into problems if the number of
1's within a panel was large (>1), hence the warning, but
the FAQ says not to worry.  Also, since your panel is two
years, after you (Stata) drop all of the (0,0)'s and (1,1)'s
you will have only (0,1)'s and (1,0)'s; so there will no
longer be multiple positive outcomes within groups.

Best regards,

Michael Ash, Assistant Professor
  of Economics and Public Policy
Department of Economics and CPPA
University of Massachusetts
Amherst, MA 01003
Tel 413-545-6329 Fax 413-545-2921
Email mash@econs.umass.edu
http://people.umass.edu/maash


> Date: Thu, 11 Mar 2004 22:30:11 -0000
> From: "Ngo,PT  (pgr)" <P.T.Ngo@lse.ac.uk>
> Subject: st: Fixed effect logit
>
>
> I am running xtlogit ...

> fpov is a 1/0 variable indicating whether a household in poor or not. I have a panel of data for two years. So it is possible that in both year a household is either poor or non poor.
>
> I run the following commands.
>
> iis househol
> tis year
> xtlogit fpov riyielA anutot anugood perenn hhsize hhsizesq riceprice irrignew8 roadnew8 disast4 disast8, fe nolog
>
> that is fpov is regressed on a set of variables that changes between the two years.

> note: multiple positive outcomes within groups encountered.
> note: 2642 groups (5284 obs) dropped due to all positive or
>       all negative outcomes.
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index