"Should I just enter the vectors for proficiency for all participants in any order?"
This is a little confusing; if it's a group-level analysis with one image per subject (i.e., it's _not_ repeated measures), then there should only be one vector, with one element (number) per subject.
Yes, the order matters; the order of the elements in the vector should match the order of the subject images.
"...or should I split my variable (proficiency) into 2 different covariates, (one covariate of proficiency for the 'young' group and one covariate of proficiency for the 'old'group)?"
If you don't split, then you have a kind of "main effect" of the covariate. If you do split, then you're modeling a group-by-covariate interaction. So they're different statistical models. It's your choice. Note that you have to be careful interpretting the results of ANCOVA models.
|