Date:         Tue, 26 Sep 2006 06:22:24 -0700
Reply-To:     Sarah White <s.white@SGUL.AC.UK>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Sarah White <s.white@SGUL.AC.UK>
Organization: http://groups.google.com
Subject:      Specification of multi-level model in PROC MIXED
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

Dear List

I have a sample of subjects being measured at 3 timepoints.  The
dependent variable is measured at each timepoint and so is
straightforward.  However I have a mix of independent variables, some
which are measured at the same timepoints, others measured only once,
age, gender, etnicity, etc.

My question is: do I specify the static (patient level) and
time-varying (visit level) variables the same way, and SAS just knows
the difference because of how the data is structured?  I am assuming
that for those variables collected at one timepoint only there value is
repeated for the other timepoints, something like
id  visit gender y
1     1      m     53
1      2     m      46
1      3      m      35
2     1      f        24
2     2      f        27
2     3      f       38

What slightly concerns me is that there are some visit level variables
which may or may not change over the time period, ie, whether they take
drugs or not, and then some static variables like gender which wont
because it was only collected once.  Surely it is important to reflect
this in the model specification?

Is this really obvious and I am just getting worried over nothing?  I
would really appreciate some guidance.

Best wishes

Sarah