s-news
[Top] [All Lists]

lme vs proc mixed

To: "S-News" <s-news@lists.biostat.wustl.edu>
Subject: lme vs proc mixed
From: "Julia Reid" <jreid@myriad.com>
Date: Mon, 08 May 2006 10:32:47 -0600
User-agent: Mozilla Thunderbird 0.9 (X11/20041103)
Hello everybody.
I am trying to compare results for a repeated measures analysis using proc mixed in SAS and lme in SPLUS 7 for linux. I believe I have set all the parameters to match the SAS analysis:
options(contrasts = c(factor = "contr.SAS", ordered = "contr.poly"))
corr = corAR1()
method="REML"

However, the coefficients and p-values produced by summary(lme.object) differ slightly from the proc mixed results. Some of the p-values are the same to 4 decimal places but some differ by as much as 0.03.

Does anyone have any suggestions as to how I can reproduce the results more exactly, or is this to be expected from different algorithms in the two softwares?

Thank you for your help.
Julia Reid




--
Julia Reid, M.Stat.
Myriad Genetics, Inc.
320 Wakara Way
Salt Lake City
UT 84108
Ph: 801-883-3348
Fax: 801-584-3640



<Prev in Thread] Current Thread [Next in Thread>
  • lme vs proc mixed, Julia Reid <=