Date: Fri, 23 Sep 2005 21:59:28 0300
ReplyTo: Hector Maletta <hmaletta@fibertel.com.ar>
Sender: "SPSSX(r) Discussion" <SPSSXL@LISTSERV.UGA.EDU>
From: Hector Maletta <hmaletta@fibertel.com.ar>
Subject: Re: GLM Repeated with Categorical Covariates
InReplyTo: <200509240014.j8O0EkqJ007176@listserv.cc.uga.edu>
ContentType: text/plain; charset="USASCII"
Covariates are supposed to be intervallevel variables (they might be
dummies), whereas categorical variables with several categories should be
treated as factors, not as covariates. However, if you convert your
ethnicity variable into a set of three dummies representing the four
categories, you may use the dummies as covariates. Not strictly kosher but
acceptable.
Hector
> Original Message
> From: SPSSX(r) Discussion [mailto:SPSSXL@LISTSERV.UGA.EDU]
> On Behalf Of Stephen Salbod
> Sent: Friday, September 23, 2005 9:15 PM
> To: SPSSXL@LISTSERV.UGA.EDU
> Subject: GLM Repeated with Categorical Covariates
>
> A Friday question: I want to us ethnicity (4 levels) as a
> covariate in an ANCOVA procedure. I think the code is
> correct, but since I've never worked with a set, as
> covariate, I'm a little concerned. The Syntax Guide had
> little to say on this issue (see GLM: Repeated Measures, 726734).
>
>
>
> COMPUTE Ethnic1 = 0.
>
> COMPUTE Ethnic2 = 0.
>
> COMPUTE Ethnic3 = 0.
>
>
>
> DO IF (Ethnic EQ 1).
>
>  COMPUTE Ethnic1 = 1.
>
> ELSE IF (Ethnic EQ 2).
>
>  COMPUTE Ethnic2 = 1.
>
> ELSE IF (Ethnic EQ 3).
>
>  COMPUTE Ethnic3 = 1.
>
> END IF.
>
>
>
> GLM
>
> pretest posttest BY group WITH ethnic1 ethnic2 ethnic3
>
> /WSFACTOR = time 2
>
> /METHOD = SSTYPE(3)
>
> /EMMEANS = TABLES(group*time) WITH(ethnic1=MEAN
> ethnic2=MEAN ethnic3=MEAN)
>
> /CRITERIA = ALPHA(.05)
>
> /WSDESIGN = time
>
> /DESIGN = ethnic1 ethnic2 ethnic3 group .
>
>
>
> Any suggestions will be most welcome.
>
>
>
> TIA.
>
>
>
> Stephen Salbod, Pace University, NYC
>
> __________ Informacisn de NOD32 1.1230 (20050922) __________
>
> Este mensaje ha sido analizado con NOD32 Antivirus System
> http://www.nod32.com
>
>
