Hi all,

I am getting an error with my matrix when running mri_glmfit (ERROR: matrix is ill-conditioned or badly scaled, condno = 37799.3). I have tried both DODS and DOSS to counteract this, and I have also checked the correlation between variables (the highest was about .65).

I have 8 classes (2 gender and four data collection sites) and 5 continuous variables. I want to examine the relationship between thickness and the last continuous variable (performance on a cognitive task) regressing out class and the other 4 continuous variables.

Attached are my fsgd file, contrast files for DODS and DOSS and matrix files for DODS and DOSS. My command line is: mri_glmfit --y cognitive_correlations/lh.thickness10.mgh --fsgd sdmttest.fsgd --C sdmt_DODS.mtx --surf hdaverage_final lh --cortex --glmdir cognitive_correlations/lh

Any help would be greatly appreciated!!

Thanks, Eli Johnson