Date: Thu, 25 Sep 2003 21:15:19 +0100
Reply-To: Paul Mcgeoghan <Mcgeoghan@Cardiff.ac.uk>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Paul Mcgeoghan <Mcgeoghan@Cardiff.ac.uk>
Subject: Re: repeated measures multiple comparisons
Content-Type: text/plain; charset=US-ASCII
Paul,
I am specifically interested in comparing differences between levels of the
within-subject factor across patient groups.
There are 2 patient groups (study and control) and 2 hospital types (govt and
private).
I want to know if (a) the difference between Visit 1 and Visit 2 is different
when comparing the 2 Study Groups; (b) if there is a difference when comparing
the 2 Control groups and (c) when comparing the Study group and Control group.
Is this possible in the Repeated Measures analysis?
Alternatively, I could work out the difference between Visit 1 and Visit 2
manually and do a 2 way Anova and use the syntax:
/EMMEANS = TABLES(hospital*group) compare(hospital)
/EMMEANS = TABLES(hospital*group) compare(visit)
which would give me this.
Any suggestions/comments on the above?
Paul
==================
Paul McGeoghan,
Application support specialist (Statistics and Databases),
Information Services,
Cardiff University.
Tel. 02920 (875035).
>>> "Paul R Swank" <Paul.R.Swank@uth.tmc.edu> 09/22/03 3:03 PM >>>
If the three way is insignificant then the difference over time by group
does not vary by hospital. By removing the three way interaction but
keeping the two ways (use custom model), you can look at the group by
time affect within each hospital while controlling for the hospital main
effects.
Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Medical School
UT Health Science Center at Houston
-----Original Message-----
From: Paul Mcgeoghan [mailto:Mcgeoghan@Cardiff.ac.uk]
Sent: Friday, September 19, 2003 6:11 PM
To: Paul.R.Swank@uth.tmc.edu
Subject: RE: repeated measures multiple comparisons
No, its not.
==================
Paul McGeoghan,
Application support specialist (Statistics and Databases), Information
Services, Cardiff University. Tel. 02920 (875035).
>>> "Paul R Swank" <Paul.R.Swank@uth.tmc.edu> 09/20/03 12:06 AM >>>
Is the three way interaction even significant?
Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Medical School
UT Health Science Center at Houston
-----Original Message-----
From: Paul Mcgeoghan [mailto:Mcgeoghan@Cardiff.ac.uk]
Sent: Friday, September 19, 2003 4:09 PM
To: Paul.R.Swank@uth.tmc.edu
Subject: RE: repeated measures multiple comparisons
Thanks,
I had originally done that but it still doesn't give me what the
customer asks for.
The closest it would give me is a three way interaction of
VISIT*HOSPITAL*GROUP under within subject repeated contrasts for level 1
to level 2; level 2 to level 3 which is not very clear.
Ideally he wants is a comparison similar to what you get when you do
/EMMEANS =
TABLES(hospital*group*visit) compare(visit)
but also wants to compare the 2 groups in the above syntax.
Hope that makes sense.
Paul
==================
Paul McGeoghan,
Application support specialist (Statistics and Databases), Information
Services, Cardiff University. Tel. 02920 (875035).
>>> "Paul R Swank" <Paul.R.Swank@uth.tmc.edu> 09/19/03 10:20 PM >>>
Include hospital type as an independent variable (fixed effect) along
with group.
Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Medical School
UT Health Science Center at Houston
-----Original Message-----
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Paul Mcgeoghan
Sent: Friday, September 19, 2003 1:33 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: repeated measures multiple comparisons
Hi,
I have a customer who has measured Blood-glucose levels over 3 visits
while Fasting and after eating. Therefore, there are 2 measures named
Fasting and PostPar measured at 3 levels. There are 2 between subject
factors named Hospital (2 groups, private and
government) and Patient Group (control and study).
Within each hospital type, I can work out if there is a difference from
visit 1 to visit 2, visit 2 to visit 3 for the control group and the
study group seperately via the repeated measures syntax: /EMMEANS =
TABLES(hospital*group*visit) compare(visit)
The customer would like for each hospital to compare the study group and
the control group on the mean difference from visit 1 to visit 2 and
visit 2 to visit 3 to see which group has had the biggest increase or
decrease from visit to visit.
The only way in repeated measures I can see of doing this is to Split
the file first by Hospital Type, enter the Patient Group as a Between
Subject factor and choose Repeated Contrasts. Then the Test of
Within-Subject Contrasts gives me the interaction between VISIT*GROUP
comparing Level 1 to Level 2 and Level 2 to Level 3.
Is there another way in repeated measures of achieving this?
Thanks in advance,
Paul
==================
Paul McGeoghan,
Application support specialist (Statistics and Databases), Information
Services, Cardiff University. Tel. 02920 (875035).
