s-news
[Top] [All Lists]

power in longitudinal lme

To: "snews" <s-news@lists.biostat.wustl.edu>
Subject: power in longitudinal lme
From: "Bernd Puschner" <bernd.puschner@bkh-guenzburg.de>
Date: Wed, 28 Jan 2004 11:17:10 +0100
Dear S-Plus users,

we are planning a multicenter randomized clinical trial on the effect of a
new discharge planning protocol (primary outcome rehospitalization) in
patients suffering from schizophrenia. We will have two groups (treatment
and control) four centers and plan on five measurement points.
I plan to analyze data via lme in S-PLUS (i.e. mixed effects) and need to
conduct a power calculation for the grant application. I looked at the
S-Plus archives but found nothing directly relating to longitudinal designs.
Also the software designed by Steve Raudenbush (OpDes) or Roel Bosker (PINT)
was of no help. I was told by Steve Raudenbush that there is SAS code for
such a problem, but I do not have SAS. Did anyone ever conduct power
calulation for a mixed effects model with more than two measurement points
in S-Plus? Not really necessary to consider that it will be multi-site.

Any help would be highly appreciated.

Thanks,

Bernd Puschner

____________________________________________
Bernd Puschner
Ulm University, Department of Psychiatry II
Ludwig-Heilmeyer-Str. 2
D-89312 Guenzburg/Germany


<Prev in Thread] Current Thread [Next in Thread>