18 Dec
2004
18 Dec
'04
10:13 a.m.
If Rob wrote a procedure in Gauss to do it, you should be able
to translate that into R. Or you can run the procedure in Gauss
to replicate.
----- Original Message -----
From: "Andrew Eggers" <aeggers(a)gmail.com>
To: "Gary King" <king(a)harvard.edu>
Cc: <gov2001-l(a)lists.fas.harvard.edu>
Sent: Saturday, December 18, 2004 2:44 PM
Subject: Re: [gov2001-l] pcse in R with missing data
Certainly, but the work I'm replicating
didn't do this so I'm
still
looking for the non-imputing approach.
Andy
On Sat, 18 Dec 2004 14:36:05 -0500 (EST), Gary King
<king(a)harvard.edu> wrote:
for any stat'l procedure for which you need fully observed
data, you can
multiply impute and treat each imputed data set as if it were
fully
observed.
Gary
On Sat, 18 Dec 2004, Andrew Eggers wrote:
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
I am trying to implement panel corrected standard
errors
(pcse) in R
where there is missing data. Does anyone know of an
existing program
that does this so I don't have to write it myself?
Franzese has done this in Gauss
http://www-personal.umich.edu/~franzese/pcse.documentation.pdf.
I am basically trying to replicate Stata's xtpcse.
Thanks,
Andy
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l