One way Anova (4 levels) and Conjunction

[Caroline Landelle]

Dear users,

I made a one-way Anova including repeated mesures: I.Model = 'con_Mod + random(SS|con_Mod)'

The factor Mod has 4 levels (A,B,C,D)

Q1: about 1-way ANOVA

According to the order of the levels :

dat.con_Mod=repmat({'A' 'B' 'C' 'D'},1,length(dat.sbj_name))'   OR  dat.con_Mod=repmat({'D' 'A' 'B 'C'},1,length(dat.sbj_name))'

I found different results for the following contrasts :

'con_Mod$A# con_Mod$D ' 'A=D;
'con_Mod$B# con_Mod$D 'B=D;
'con_Mod$C # con_Mod$D 'C=D';

But the results for the following 3 contrasts seems to be the same regardless of the order of the levels

'con_Mod$A# con_Mod$B   ' 'A=B;

'con_Mod$A# con_Mod$C  ' 'A=C;

'con_Mod$B# con_Mod$C  ' 'B=C;

However this was not the case if I.DoOnlyAll = 1; but values are missing in some voxel

Is there a logical explanation for these worrying results?

Q2: Conjunction analysis

I would like to deal with a conjunction analysis with a logical AND. : (C-A) AND (C-B).

To do this, I thresholded the two contrasts (FWE) and created a mask. I used imcalc to overlay the results and extract the common region between the two. The method is very concervative, is there another less drastic way to do this logical AND between two contrasts? For exemple correct the conjunction map at the end and not the 2 contrasts before the conjunction ?

Thanks  a lot

Caroline L