Dear FSL experts,
I attempt to test FA difference between patients and controls. Controls have the same gender distribution, whereas patients have more male patients than females. By using two groups with dichotomy covariate interaction suggested on FslWiki (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM), I found the slope of gender and FA in control is greater than the slope in patients.
FslWiki Two Groups with continuous covariate interaction states 'Again, if you run this model and the interaction is not significant anywhere in the brain, use the simpler, Two-Group Difference Adjusted for Covariate, the model described above. If the interaction is significant, only the inferences for the interaction are of interest.'. That is, you suggest if there is an interaction, adjusting gender as a single EV in two group comparison can't remove gender effects on group difference? If it's true, is there another way to adjust gender in my case? Thanks.
Best Wishes,
Joyce
|