Dear Doug,

I'm performing a linear regression over one sample with a design matrix including columns for one continuous-valued predictor whose effect on neural measures I'm interested in testing, several covariates to adjust for (continuous- or binary-valued such as age, sex, total intracranial volume, and other demographic groupings), and an intercept.

The contrast matrix I'll be testing will have 1 in the column corresponding to the column in the design matrix containing the predictor of interest and zeros in all other columns.

In addition, I was interested in performing the same linear regressions over subsets of the group to determine whether the pattern associated with the predictor-of-interest is persistent. That's why I want to stack surfaces for different sets of subjects (without recreating surfaces registered to fsaverage_sym for each subjects' laterality indices for each instance). This way I could repeat the same contrast, except subsetting rows of the design matrix for subjects in the given sample.

Thank you.

Best,
Chintan

 Fri, 10 Nov 2017 11:28:52 -0800
> That looks right, though I'm not sure what is in the design matrix. Is it a column of +1 and -1?