s-news
[Top] [All Lists]

How to get se's of predictions from lme models

To: <s-news@lists.biostat.wustl.edu>
Subject: How to get se's of predictions from lme models
From: "Hunsicker, Lawrence" <lawrence-hunsicker@uiowa.edu>
Date: Thu, 19 Aug 2004 16:25:51 -0500
Thread-index: AcSGMxpFX8VKcZFdRP+2t//vj9X0CA==
Thread-topic: [S] How to get se's of predictions from lme models
Good day, all.  I'm back to the trough.

I have created a model of blood pressure over time as a function of assigned 
treatment.  With the model I can get estimates for the blood pressure by time 
and group using predict.lme().  But unlike other predict()s, predict.lme() 
doesn't accept the parameter se.fit=T.  Of course, I can extract the vector of 
coefficients and the cov matrix from the model and do the required matrix 
arithmatic.  But it is sort of tedious to build the design matrix for the 
various combinations of treatment, time, and the interactions.  

Am I missing a simpler way to do this, using an extraction function that I have 
missed?  

Alternatively (or additionally), is there an easy way to build a design matrix 
for TRT*VISIT?

As always, many thanks in advance to any of you that can help me.

Larry Hunsicker


<Prev in Thread] Current Thread [Next in Thread>
  • How to get se's of predictions from lme models, Hunsicker, Lawrence <=