2*2 design

Dear teachers,

The design of our study is a two groups (14 patients vs. 14 controls)*two conditions (stimulation A vs. stimulation B) type, and according to Dr. Yan' s comments, we perfomed mixed effect analysis using DPABI (the latest version).  After GRF (voxel P=0.02, cluster P=0.05, two tails), two clusters connecting  ACC remained in condition T  map, while none remained in either group T or interaction F map (single tail). Meanwhile, we performed paired T test within each group based on corresponding masks of one sample T test maps (P=5*10-5, cluter size>10, A+B),  two clusters different from these above remained in patient group while none remained in control (which are consistent with hypothesis). 

My questions are: 1. are statistic procedures above correct? if not, coud you please give me some advices. 2. whether paried T test are meaningful or not when no clusters found in interaction F map?  Thank you!

Forums:

No. In this case, you can only do the mixed effect analysis first to identify the meaningful interaction effect.

No. In this case, you can only do the mixed effect analysis first to identify the meaningful interaction effect.

Thank you so much.

Dear Dr. Yan,

Now I have identified multiple very small clusters (about 10 or less voxels  for each) from interaction F map using GRF (voxel P=0.01, cluster P=0.05, single tailed) correction.  Should I create mask based on these clusters to perform paired T test within each group and independent two sample T test between groups?  Could you please share me some references about mixed effect analysis using DPABI or similar ones?

Thank you again.

Aaron Hsiang

 

You can just extract the values for those clusters (define as ROIs) and plot them or use SPSS to do post-hoc analyses.

Yan, C.G., Rincon-Cortes, M., Raineki, C., Sarro, E., Colcombe, S., Guilfoyle, D.N., Yang, Z., Gerum, S., Biswal, B.B., Milham, M.P., Sullivan, R.M., Castellanos, F.X., 2017. Aberrant development of intrinsic brain activity in a rat model of caregiver maltreatment of offspring. Transl Psychiatry 7, e1005.

Dear Dr. Yan,

Is it right to set voxel P=0.02 and cluster P=0.05 when using GRF correction for T-test maps and voxel P=0.01 and cluster P=0.05 for F-test maps? If not, Could you tell me the right setings? Thank you.

AF Xiang

 

 

 

We put Voxel P at 0.001 and cluster P at 0.05, also check Two-tailed.