Date: Thu, 15 Jun 2006 11:04:16 -0300
Reply-To: Pedro Albuquerque <zefiro@BRTURBO.COM.BR>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Pedro Albuquerque <zefiro@BRTURBO.COM.BR>
Subject: RES: All combinations...
Content-type: text/plain; charset=us-ascii
Tks for all answer !!!
De: Sigurd Hermansen [mailto:HERMANS1@WESTAT.com]
Enviada em: Thursday, June 15, 2006 11:01 AM
Para: Pedro Albuquerque; SAS-L@LISTSERV.UGA.EDU
Assunto: RE: All combinations...
You don't hear me say this often.... For this class of problem SAS has a
better procedure than PROC SQL:
v0='0'; v1='1'; v2='2';
proc summary data=test;
class v0 v1 v2;
output out=pattern (drop=_type_ _freq_);
From: email@example.com [mailto:firstname.lastname@example.org]
On Behalf Of Pedro Albuquerque
Sent: Wednesday, June 14, 2006 10:00 PM
Subject: All combinations...
How can I do to make a data file like this:
ALL COMBINATIONS "0" TO "T"
Input x $100.;
Tks very much
De: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] Em nome de Matt
Enviada em: Wednesday, June 14, 2006 8:49 PM
Assunto: Re: Hausman Test for comparing fixed vs. mixed models
Yes, but TSCREG has two drawbacks:
1) requires that the data be balanced
2) dummy variables for fixed parameters are explicity estimated and
reported, requiring excessive cpu usage and output
My data (SIPP) is unbalanced and is extremely large (~2.5 million
Are there any other ways to get the Hausman test?
Ban Cheah wrote:
> Instead of comparing STATA's xtreg with PROC MIXED, the alternative
> comparison may be with PROC TSCSREG. PROC TSCSREG outputs the Hausman
> test by default.
> -----Original Message-----
> From: email@example.com
> On Behalf Of Matt
> Sent: Wednesday, June 14, 2006 3:44 PM
> To: firstname.lastname@example.org
> Subject: Hausman Test for comparing fixed vs. mixed models
> I have searched up, down, left and right for a way to run the Hausman
> test in SAS, for comparing fixed effects (GLM) vs. random effects
> (MIXED). Apparently it can be done, but I can't seem to find how to
> do so. PROC MODEL does not seem to be appropriate for this comparison.
> I know this can be done, the 'long way' (which I've done), but it
> seems like SAS would have developed some sort of option (possibly in
> MIXED) for the Hausman Test, similar to the 'hausman option' in
> xtreg package.
> Thanks for your help in advance.
> Matt Hall
> Penn State Population Research Institute