Date:         Wed, 1 Jun 2005 08:02:38 -0500
Reply-To:     "Reutter, Alex" <areutter@spss.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Reutter, Alex" <areutter@spss.com>
Subject:      Re: [BULK] Covariance element constrains in mixed models
Content-Type: text/plain; charset="us-ascii"

This functionality does not currently exist in SPSS.  Out of curiosity, is there a strong theoretical basis for the constraints you want to impose, or do you have few observations and want to conserve degrees of freedom (and/or some other reason for needing the constraints)?

Cheers,
Alex


> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
> RJ Wirth
> Sent: Monday, May 30, 2005 3:16 PM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: [BULK] Covariance element constrains in mixed models
> Importance: Low
>
> Hello ~
>   I am hoping someone with more SPSS experience can answer a question for
> me.
>
> I am attempting to estimate a multivariate mixed model with
> heteroscadastic errors over the repeated observations. This part I can do.
> However, I would like to constrain specific elements in the \REPEATED
> covariance matrix (or unconstrain elements depending on the matrix
> structure specified).
>
> e.g.,
> /REPEATED = time | SUBJECT(id) COVTYPE(UN) .
> where UN[3,2] = 0, UN[5,4] = 0, etc.
>
> Is this possible in SPSS? Is there an equivalent option in SPSS
> (unbeknownst to me) to the 'parms' option on Proc Mixed?
>
> I have searched the archive and looked at all of the standard web-pages,
> pdfs, and manuals that have been posted in the past with no luck. Any
> comments or suggestions (or verification that this is not possible in
> SPSS) - would be much appreciated.
> Cheers,
> RJ