3dLME identical outputs for different glt

AFNI Message Board
1

Some of my 3dLME outputs are identical, despite coding for non-overlapping groups. I have a suspicion that this has something to do with the way the glt coding is interacting with the specified model. I’m trying to compare, within a disease group, the longitudinal differences between genetic groups.

-model ‘AgeGenetics+AgeDX’ \

these tests among others:
-gltLabel 33 ‘Age_AD_Mut-WT’ -gltCode 33 'Genetics : 1Mut -1WT DX : 1AD Age : ’
-gltLabel 34 ‘Age_CN_Mut-WT’ -gltCode 34 'Genetics : 1Mut -1WT DX : 1CN Age : ’
have the same output as this test:
-gltLabel 25 ‘Age_Mut-WT’ -gltCode 25 'Genetics : 1Mut -1WT Age : ’ \

so obviously, the specification of the DX group is failing. Is this because I do not have DX:Genetics in the model? The AgeDXGenetics model fails, on reading rhe messageboards I believe this is because every subject always has the same for both DX and genetics. Or is there some limitation or have I not properly understood how to use complex tests in LME?

2

Is this because I do not have DX:Genetics in the model?

Yes, you assumed that there is no interaction between DX and Genetics in your model specification.

The AgeDXGenetics model fails, on reading rhe messageboards I believe this is because
every subject always has the same for both DX and genetics.

Do you mean that DX and Genetics are confounded somehow? What does the contingency table look like for DX and Genetics in terms of subjects?

3

The issue has resolved and I don’t know why. AgeDXGenetics failed every attempt, and the messageboards led me to believe it was because DX and genetics had some unacceptable interaction for LME and separating it into AgeDX+AgeGenetics fixed it. AgeDXGenetics work now, and the only thing I’ve changed otherwise are random effects.

Sorry for wasting your time.