3dMVM qualitative covariate of no interest

AFNI Message Board
1

Hi Gang,

I am working on a 3dMVM script, with two quantitative covariates of no interest (BTXTime and TotalBTX) - this has worked fine. I would now like to add a qualitative/categorical covariate of no interest with two factors (benefit: yes, no). Would you be able to help me incorporate the qualitative covariate?

I have added it as a wsVars but would appreciate some guidance in adding it to the gltLabel, if needed:

3dMVM -prefix MVM_Group_Results_subtypes_benefit -jobs 1
-bsVars ‘group+BTXTime+TotalBTX’
-qVars ‘BTXTime,TotalBTX’
-wsVars benefit
-SS_type 3
-num_glt 1
-gltLabel 1 Non-Naive_AD_vs_Non-naive_AB -gltCode 1 ‘group : 1BTX_nonnaive_AD -1BTX_nonnaive_AB’
-mask mask_resamp+tlrc.
-dataTable @3dMVM_DataTable_subtypes_benefit.txt

Thanks,
Lena

2

Hi Lena,

I have added it as a wsVars but would appreciate some guidance in adding it to the gltLabel, if needed:

What hypothesis are you trying to assess regarding the factor ‘benefit’?

3

Hi Gang,

I am not assessing any hypothesis with the ‘benefit’ factor. It is simply a covariate of no interest, same as BTXTime and TotalBTX in qVars.

Based on my understanding, qVars only takes quantitative covariates not qualitative/categorical?

Thanks,
Lena

4

HI Lena,

I am not assessing any hypothesis with the ‘benefit’ factor. It is simply a covariate of no interest, same as BTXTime and TotalBTX in qVars.

In that case, you don’t need to change your specification in gltCode.

Based on my understanding, qVars only takes quantitative covariates not qualitative/categorical?

Correct. Any explanatory variables that not defined under -qVars are treated as categorical.

5

Great! Thank you for your help!

Lena

6

Hi again,

I tried adding the categorical covariate “benefit” in -bsVars and did not mention it in -qVars because it’s categorical, but I get an Error saying: Quitting due to model test failure.

Attaching the 3dMVM code and data table. Interestingly, if I simply delete the benefit bsvars and let the model run with the the two qualitative covariates it runs without an issue. Would really appreciate any guidance!! Thanks!

3dMVM -prefix MVM_Group_Results_subtypes_benefit -jobs 1
-bsVars ‘group+BTXTime+TotalBTX+Benefit’
-qVars ‘BTXTime,TotalBTX’
-SS_type 3
-num_glt 1
-gltLabel 1 Non-Naive_AD_vs_Non-naive_AB -gltCode 1 ‘group : 1BTX_nonnaive_AD -1BTX_nonnaive_AB’
-mask mask_resamp+tlrc.
-dataTable @3dMVM_DataTable_subtypes_benefit.txt

Subj group BTXTime TotalBTX Benefit InputFile
s201 BTX_nonnaive_AB 3 3 yes stats.s201+tlrc[7]
s202 BTX_nonnaive_AD 3 36 yes stats.s202+tlrc[7]
s203 BTX_nonnaive_AD 3.5 57 no stats.s203+tlrc[7]
etc.

7

Try without the mask line, and see if it works.

8

Unfortunately, I get the same error without the mask line.

Lena