[R-sig-ME] Measure of influence for lmer models
jsbarret at sfu.ca
Wed Jan 26 02:10:31 CET 2011
That package came close to what I'm trying to do, and was very useful at identifying influential levels of the grouping factor. I'm wondering if there is a way to do the same calculations for individual observations though...any thoughts?
----- Original Message -----
From: "Dennis Murphy" <djmuser at gmail.com>
To: "Jenn Barrett" <jsbarret at sfu.ca>
Cc: "." <r-sig-mixed-models at r-project.org>
Sent: Friday, 21 January, 2011 18:35:17
Subject: Re: [R-sig-ME] Measure of influence for lmer models
Try the influence.ME package at CRAN. Its description file says:
Tools for detecting influential data in mixed effects models
On Fri, Jan 21, 2011 at 1:05 PM, Jenn Barrett < jsbarret at sfu.ca > wrote:
My colleague and I recently fit a linear mixed effects model with lmer(), and after looking at the results, my supervisor is interested in having me examine the influence of individual observations (he suspects that there may be a couple high leverage points); however, I'm at a loss on how to do this with a "mer" object. Is there a function/method for computing Cook's distance, leverage plots or some other measure of influence for "mer" objects? I've searched the forum, but have only come across unanswered questions.
Thanks in advance.
R-sig-mixed-models at r-project.org mailing list
More information about the R-sig-mixed-models