Dear Donald McLaren,

I also have query on this issue.  For a three-group between-subject design, typically a one-way 3-levels anova is firstly used to test whether there are any differences among the 3 groups by modelling all the 3 groups in a design. If significant differences are detected, post-hoc tests (e.g., t-test)   are then further performed to determine which two groups are different. So how to do the post-hoc tests? 

E.g., to compare group1 and group 2, use T-contrast 1 -1 0 in the anova model or use T-contrast 1 -1 by generating a new design to model only group1 and group2?

This is important in the case that some covariates (e.g., age as a covariate, a column in the design matrix) are needed to modeled in the design. If we choose the second method (e.g., use T-contrast 1 -1 by generating a new design to model only group1 and group2), the age effects of group1 will be different between model comparing group1 and group2 and model comparing group1 and group3.

Anyone any comments??

Best regard



2012/6/15 MCLAREN, Donald <[log in to unmask]>
Because you have a between-subject design, the estimate (magnitude) of group 3 does not effect the magnitude of group 1 or 2 or their difference.

However, having group 3 in the model will effect the variance of the contrast and the t-statistic. Selecting assume equal variance or unequal variance will also effect the statistics.

Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Website: http://www.martinos.org/~mclaren
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.




On Fri, Jun 15, 2012 at 4:35 AM, Martin Federbusch <[log in to unmask]> wrote:


---------- Forwarded message ----------
From: Martin Federbusch <[log in to unmask]>
Date: Fri, Jun 15, 2012 at 10:18 AM
Subject: T-contrast between 2 groups in a 3-group-design
To: [log in to unmask]


Dear all,

I am comparing 3 groups in a VBM-analysis (/2nd-level-analysis). When I'm using a T-contrast only between 2 of these groups, does SPM correct for example the degrees of freedom in this case?

e.g.: Group 1, 2 and 3 have 20 subjects each. All of these groups are in one design, so I have a design with 60 subjects.
Is it possible to use a T-contrast 1 -1 0, which would compare group 1 and 2, but leave out group 3 or do I have to do a new design with only 40 subjects?

Thank You in advance,
best,
Martin





--
Jin-hui Wang, Ph.D. Candidate
Major in Cognitive Neuroscience
State Key Laboratory of Cognitive Neuroscience and Learning
Beijing Normal University
E-mail:  [log in to unmask]
            [log in to unmask]

Google Scholar Citations:
http://scholar.google.com/citations?user=wKLnXbMAAAAJ&hl=en