|Date: ||Thu, 22 Aug 2002 15:59:18 -0600|
|Reply-To: ||Susan Crawford <Susan.Crawford@calgaryhealthregion.ca>|
|Sender: ||"SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>|
|From: ||Susan Crawford <Susan.Crawford@calgaryhealthregion.ca>|
|Organization: ||Calgary Health Region|
|Subject: ||Repeated measures MANCOVA question|
|Content-Type: ||text/plain; charset=us-ascii|
We are going to do a longitudinal follow-up study of three groups of
children: a healthy normal comparison group, and two treatment groups.
We are going to have 75 children per group, and are going to be
assessing them at 18 months of age, 3 years, 5 years, and 7 years of
age. It's going to be a long study!!
We have hypothesized that our two primary outcome variables will remain
fairly constant at the four ages for the normal comparison group, and
also for one of the treatment groups. We expect these two variables
will decline over time for children in the last treatment group. We
also want to see when does this decline occur (if it does in fact
occur). It may be a constant decline over time, or it could be more
pronounced at one of the age ranges (for example between 18 months and 3
years of age).
Initially, we thought that we'd just do a repeated measures MANCOVA, but
now we are wondering if that will get at the question of when the
decline occurs. Can the question of when the decline occurs be examined
through "contrasts" for the time effect in the analysis? Can we even do
this with SPSS?
Susan Crawford, M.Sc.
Behavioural Research Unit
Alberta Children's Hospital