Thank you very much for the answer!
I am working on fMRI data.
I was not going to do any intensity normalization beyond the usual default measures in normalization step in SPM2 at the individual pre-processing step (is that right?)
My further question is: is there a way to program the paired-t-test/ 2 sample t-test in the group analysis step (ie arranging all the files in the correct order and summoning them at once)?
To get something like that:
A1 vs A2 B1 vs B2
. 2 sample t-test .
A1[n] vs A2[n] B1[n] vs B2[n]
Because otherwise doing separate statistics first on each individual pair then entering them in the 2 sample t-test matrix seems somewhat a long way of doing it..
----- Original Message -----
From: "Tae Woo-Suk" <[log in to unmask]>
To: [log in to unmask]
Cc: [log in to unmask]
Sent: Tuesday, October 28, 2008 8:49:07 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome / Stockholm / Vienna
Subject: Re: [SPM] Paired t-test analysis and controls
; Are your data PET or SPECT?
I would very much appreciate your insight on the best way to perform a group analysis in the following situation (SPM2):
I have 2 groups of subjects (group A receiving ttt x, group B receiving a placebo) and 1 control group C.
Groups A an B have been scanned at baseline and after ttt/placebo. Control group has been scanned once. All subjects performed the same test.
I would like to know
1)whether there is a difference between groups A and B after ttt
and 2) whether there is a difference between the control group and groups A and B before and after ttt/placebo
For the 1st question I was going to perform a paired t-test analysis (A1vsA2; B1vsB2) since both groups have been scanned twice followed by a 2 sample t-test on the resulting con.images (A.con vs B.con). Would that be correct?
For the second question would it be correct to perform a one-way ANOVA specifying 5 groups (A1 A2 B1 B2 and C)?
Finally, what is the difference between one way ANOVA and one way ANOVA with constant?
; Ans. for Q1) If you want to know the effects of medications, do paired t-test between pre and post medication images within group. But intensity normalization (based on whole brain or white matter)
could be important issue.
Ans. for Q2) Do intensity normalization before statistics.
Masks toolbox could be a good tool for you. Then subtract two paired image (pre - post),
you could get one difference image/subject. If you finished image subtraction,
you can do paired t-test between group A and group B.
Be careful, you should not do any intensity normalization in this step.
Woo-Suk, Tae Ph.D.