LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (February 2005)Back to main SPSSX-L pageJoin or leave SPSSX-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:         Fri, 4 Feb 2005 09:51:24 -0500
Reply-To:     Jeffrey Miller <millerjeffm@HOTMAIL.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Jeffrey Miller <millerjeffm@HOTMAIL.COM>
Subject:      SPSS vs. SAS mixed model results
Comments: To: algina@ufl.edu, dmiller@coe.ufl.edu, courtneyzmach@att.net
In-Reply-To:  <s2037885.035@mail.ccsu.nsw.gov.au>
Content-Type: text/plain; format=flowed

Hi all,

I recently ran two simple mixed models in both SAS and SPSS....one with no variables other than the DV, and one with TIME (equidistant) at Level-1. The parameter and variance estimates as well as standard errors were virtually identicial for the first model. However, there was some discrepancy between the results when adding TIME. Both programs are using Full Information Maximum Likelihood. I requested the Unrestricted covariance matrix for both. All in all, the results should be identical. I'm just wondering if the differences are purely due to differences in the ML algorithms between the two programs.

--Let me know if the table is too messy on your screen, and I will gladly send a Word attachment version.

A Comparison of Parameter and Variance Estimates For Two Simple Mixed Models Between Two Programming Languages

SPSS UM SAS UM SPSS UG SAS UG *Intercept Est. 7.3491(.3305) 7.3495(0.3304) 1.844(.4668) 2.1919(0.3944) *Slope Est. 2.339(.2126) 2.3719(0.2037) *Within Variance 51.1085(3.0965) 51.1102(3.0966) 35.2039(2.1048) 32.3382(1.9333) *Intercept Variance 5.9237(2.187) 5.9216(2.1867) .000000.000000 9.16E-17-------- *Slope Variance 2.2906(.82845) 2.7840(0.6893) *Int/Slope Covariance --.7928(1.2023) -0.9457(.8069)

UM is the Unconditional (degenerate) Means model UG is the Unconditional Growth model, which includes 5 equidistant time points. First term is estimate, second is standard error

Thanks in advance, Jeff Miller


Back to: Top of message | Previous page | Main SPSSX-L page