About group analysis

Dear AFNI experts,

I tried to do group analysis by using the command 3dANOVA3,but the result didn’t seem good.So i am wondering that is there some problem about the analysis for every single subject? After preprocessing, i analysed the data by 3dDeconvolve and got the results .Cause it has two different kind trials of the data,so i get the β values by using 3dTcat, and then average them separately by using 3dTstat.But i found for some subject i get a very weird results.My questions are as follow:
1, After doing 3dDeconvolve,i get 10 β values for each kind trials, how can i to draw a picture to see which is the biggest one?
2,To do the group analysis,should i use the average value of the 10 β value or choose another method?
3,Before doing group analysis by 3dANOVA3 ,is it right for me to get it separately and average them as i said before, and use the results as the input data for 3dANOVA3?
I really hope any of the experts of afni could help me to solve the problem,best wishes!

-Jack

Jack,

You need to share more information before we could offer more targeted help. Could you offer more details about the experiment? Why do you have “10 β values for each kind trials?” Why do you want “to draw a picture to see which is the biggest one?” How does your 3dAOVA3 script look like?

Mr. Chen,
Sorry i didn’t offer enough information, the experiment is a event-related design, and it has three sessions, including consciousness,sedation and unconsciousness,and for each session has 60 trials,30 trails for calling the own names of the subjects,30 trails for calling unknown names of the subjects.The 10 β values i mentioned before are from the result of 3dDeconvolve.I thought i should get them by 3dTcat and average them to do the group analysis.For now,i am not sure if is it right to do group analysis like so.And the 3dANOVA3 script is as follow:

3dANOVA3 -type 4
-alevels 3 \ #consciousness/sedation/unconsciousness
-blevels 2 \ #own name/ the other name
-clevels 18 \ #subjects
-dset 1 1 1 con_self+tlrc \

-fa state
-fb name
-fab interaction

Really thanks for your attention!

-Jack

The 10 β values i mentioned before are from the result of 3dDeconvolve.I thought i should get them by 3dTcat and average them to do the group analysis.

What are those 10 β values? How did you model the hemodynamic response for each name calling: GAM, BLOCK, or TENTzero?

Mr. Chen
The model is TENTzero.Should i average them all to put the result to 3dANOVA3?

-Jack