Hello,
thanks a lot for the hint! Actually, the value is meant to be 1.000000e+00, like every other subject. A mistake by copy&paste, I guess. Unfortunately, it doesn't change the problem.
Thanks,
Antonius
Am 18.01.2011 um 11:24 schrieb Matthew Webster:
> Hello,
> The 7.71E02 value in the 1st column of the design looks a little out of place - is this subject meant to be weighted differently?
>
> Many Regards
>
> Matthew
>
>> Dear FSL experts,
>> I'm trying to run a study on the correlation of FA and neuropsychological data based on tbss. I'm a bit puzzled about my results, because all of the results are systematically opposite to my hypotheses (I assumed that better neuropsychological results accompany with a higher FA value, but it is the other way around with all variables). Hence I assume something like a sign error in my data.
>>
>> I demeaned the neuropsychology data in this way: single value minus mean of the values of all subjects.
>>
>> With glm_gui I created the following files:
>>
>> 01.con
>>
>> /ContrastName1 pCorr
>> /ContrastName2 nCorr
>> /NumWaves 2
>> /NumContrasts 2
>> /PPheights 6.000000e+00 6.000000e+00
>> /RequiredEffect 5.564 5.564
>>
>> /Matrix
>> 0.000000e+00 1.000000e+00
>> 0.000000e+00 -1.000000e+00
>>
>>
>> 01.mat
>>
>> /NumWaves 2
>> /NumPoints 16
>> /PPheights 7.710000e+02 2.005000e+01
>>
>> /Matrix
>> 7.710000e+02 1.080000e+01
>> 1.000000e+00 -2.500000e-01
>> 1.000000e+00 -8.250000e+00
>> 1.000000e+00 -9.250000e+00
>> 1.000000e+00 -1.250000e+00
>> 1.000000e+00 3.750000e+00
>> 1.000000e+00 -9.250000e+00
>> 1.000000e+00 2.750000e+00
>> 1.000000e+00 5.750000e+00
>> 1.000000e+00 -8.250000e+00
>> 1.000000e+00 6.750000e+00
>> 1.000000e+00 5.750000e+00
>> 1.000000e+00 4.750000e+00
>> 1.000000e+00 -1.250000e+00
>> 1.000000e+00 -5.250000e+00
>> 1.000000e+00 2.750000e+00
>>
>>
>> I ran randomise with the following command
>>
>> randomise -i all_FA_skeletonised.nii.gz -o 01 -d 01.mat -t 01.con -m mean_FA_skeleton_mask.nii.gz -n 500 --T2
>>
>>
>> Do you have any suggestions what I could have mixed up?
>>
>> Thanks a lot for your kind help!
>>
>> Cheers,
>> Antonius
>>
|