So the mri_glmfit results are less significant than the QDEC results? That is a little surprising because the glmfit results will have a higher DOF. The contrast (gamma.mgh) should be the same. It may be that your FWHM increased (compare the fwhm.dat files). You can also run the two group analysis in glmfit to make sure you get identical results to qdec doug

On 06/25/2014 10:15 AM, melinda dora wrote:

Dear list,

I have a question re group comparisons. I have one group factor with three levels. I have done group-wise comparisons with these groups using qdec (ie doing pairwise comparisons, as it only allows group comparisons between two groups as far as I am aware?) as well as using mri glm fit, specifying the contrast as 1 0 -1 to compare two groups without the effect of the other third group being included. I have found that I get different results for these two methods; using mriglmfit I get no vertices surviving multiple comparisons, whilst I get several using qdec. Could this be because of the differences in using FDR/monte carlo simulation or does it point towards a bigger problem, for example me doing something wrong at some stage? Or is there another explanation?

Very grateful for prompt response and help,

Cheers,

Melinda

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer