Hi Afni folks

I am comparing two groups (patients vs. normal controls). Each individual completes four tasks set up as a blocked design. I want to compare differences between groups on these four tasks.

Through reading the message board, I understand 3dANOVA does not work with unequal sample sizes, in my case (n=14 and 16). I would think the best design, in my study, would be a mixed design with tasks and groups as fixed factors, subjects as a random factor and with subjects being nested in groups. So I would use 3dANOVA3 -type 5 and then compare differences between groups on the separate tasks using t-tests. Am I correct to assume that I cannot use this program because of my unequal sample sizes???

My other approach would be to use a mixed design (tasks fixed and subjects random), so a 3dANOVA2 -type3, separately for the patients and the NC and then do t-tests to examine group differences on the different tasks. Is that a statistically correct solution? Will I mess up my error terms???
What in your opinion is the most appropriate analysis???

One last question not related to the ANOVA, but related to clustering. A reviewer is giving me grief regarding how I clustered my data and chose my thresholds. I performed Alpha Sim using my original space voxel size but, clustered the data in Talairach space. As suggested, by Doug Ward (posting on April 29, 2002) I use a smaller rmm in Talairach space when clustering to correct for this. The reviewer is confused by the selection of cluster size in original space and applying it to Talairach space. Do you have any suggestions on how to make this clearer?? I understand that when the data is transformed into Talairach space, you are actually just reslicing your orignal voxels into 45 1x1x1 cubic voxels, so your resolution has not changed, correct??? Last question, is it fair to say that clustering your data, results in a corrected p-value, if you use the Alpha Sim approach???

Thanks alot for indulging my questions...

Jennifer