Naveed,
There are a few issues with your analysis:
1) It's better to use 3dMVM in your case, not 3dLME.
2) To avoid unnecessary complications, simply code varaible 'sex' as a factor, not a quantitative variable.
3) Your model
-model "diag+sex+age+IQ"
presumes that there are no interactions among the variables. Are you sure this is what you want?
4) The specifications for all the post hoc tests are incorrect. For example, in the following test:
-gltLabel 1 'IQ-DIAG' -gltCode 1 'diag : 1*IQ ' \
since 'diag' is a factor, the part after the colon (:) should be used to specific the linear combination of the factor levels. The "1*IQ" part does not make sense to 3dMVM/3dLME.
Gang
Edited 1 time(s). Last edit at 03/31/2015 05:36AM by Gang.