Date:         Wed, 16 Dec 2009 17:22:53 -0800
Reply-To:     leannica <leanne.streja@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         leannica <leanne.streja@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Re: Mixed model for two repeated measures ?
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

On Dec 16, 7:54 am, C <claudiasatiza...@gmail.com> wrote:
> Hello,
>
> I'm trying to choose the best way to analyze data from a longitudinal
> study and I hope you can help me.
>
> My outcome variable is continuous and was measured at two points in
> time. I want to study the association of my depvar with a 3-level
> independent variable adjusting for gender and age.
>
>          depvar (T1 - T2) = indepvar (3 levels) / age sex
>
> I've been reading the advantages of analyzing longitudianal data with
> the mixed procedure, but I don't know if this will still apply for
> just two repeated measures, or if I should just keep it simple using a
> paired t-test / modeling the difference.
>
> (I ran both models and the output is not the same)
>
> Thank you for your comments or suggestions.
>
> Greets,
>
> C

I am not even sure what you are doing with a t-test?
A t-test can only compare the difference between the two time points
for your depvar, but not adjust.

 But Proc mixed makes sense to me.

proc mixed data=blah... method = ml;
class id indepvar;
model depvar =  indepvar age sex(which I hope is coded 0 1 otherwise
put in class statement)/s;
random intercept/subject=id;
run;