I am encountering the dreaded mri_glmfit error: matrix is ill-conditioned or badly scaled, condno = 1e+08. I have tried to troubleshoot how to solve this problem from other inquiries, however, I still have had no luck. I am trying to run a 4 group 1 variable analysis: with diagnosis being my variable of interest and trying to regress out the effects of sex and age.
1. Command line:
mri_glmfit --y lh.10.73.SZHC.mgh --fsgd g4v1.73.fsgd dods --C SZ-HC.intercept.73.mtx --surf fsaverage lh --cortex --glmdir g4v1.73.lh
2. The FSGD file (if using one) (see attached)
3. And the design matrix:
0.5 0.5 -0.5 -0.5 0 0 0 0 (or see attached)
To the best of my knowledge, all my files are the right format and I believe I have adequate number of participants per group?
I also tried mean centering my age variable. I had success when I ran just a two group analysis (diagnosis) not controlling for sex or age, however, as soon as I try to run four groups, I encounter this error. I have no idea how to proceed.
I am looking forward to your response!
Thanks,
Lindsay
--
Lindsay Hanford, BSc, PhD Candidate
McMaster Integrative Neuroscience Discovery & Study | Department of Psychology, Neuroscience & Behaviour
McMaster University | 1280 Main Street West, PC329 Psychology Building | Hamilton, ON, L8S 4L8