Hi SPMers
I'd like to ask your opinions on a conjunction analyses
I have a task performed in 3 modalities, in 2 different versions
Let's call them 1A, 1B, 1C and 2A, 2B, 2C
(I have 15 subjects)
What I'm currently interested in is regions where the conjunction null of
tasks 1A,B+C differs from the conjunction null of tasks 2A,B+C
I can think of two ways of doing this (bearing in mind conjunctions don't
seem to produce tangible image files)
I wanted to check with the experts whether these two options are
statistically equaivalent or whether there are serious differences (and thus
which is 'correct')?
Here are my suggestions.....
option one: Run the preliminary first level analyses for each subject. Then
run an extra fixed effects analysis for each subject in which I calculate a
difference image for each modality, i.e. 1A-2A, repeat for 1B-2B and 1C-2C
Then take these 3 series of 'difference' con images through to a random
effects analysis and perform my conjunction
option two: Run the preliminary first level analyses for each subject. Then
model all six versions of the tasks in a large random effects analysis.
Perform the t-contrasts 1 1 1 0 0 0 and 0 0 0 1 1 1
I look forward to your opinions!
Many thanks
Rachel Mitchell
-----------------------------------------------------------------------
Dr Rachel L. C. Mitchell
Lecturer in Cognitive Psychology, University of Reading
Honorary Research Fellow, Institute of Psychiatry
Research Psychologist, Berkshire Healthcare NHS Trust
Correspondence Address:
School of Psychology
Whiteknights Road
University of Reading
Reading
Berkshire
RG6 6AL
Tel: +44 (0)118 378 8523
Direct Dial: +44 (0)118 378 7530
Fax: +44 (0)118 378 6715
-----------------------------------------------------------------------
|