What is the best/fastest way to calculate a voxelwise correlation between task-contrast beta and an external questionnaire?

AFNI Message Board
1

Howdy folks,

I have two subject groups, call them “cat lovers” and “dog lovers.” For each subject’s first level map, I calculated the contrast activation between seeing dogs vs seeing cats, and I have been using 3dMEMA to calculate each group’s second-level map, as well as a group difference map.

I’d like to explore an alternative analysis, where instead of the discrete, two-group comparison, I would pool both groups together and look for an overall voxelwise correlation with subject scores on the external “Canine Affinity Questionnaire.” What is the best/fastest way to obtain a statistical map of voxelwise correlation with a single questionnaire? Could I just use 3dMEMA and only specify ONE group?

Jim

2

To correlate the beta maps for each subject with a questionnaire, you can use 3Tcorr1D[/url]. Some instructions on that method [url=http://blog.cogneurostats.com/?p=76]here.

You can also use 3dMEMA or 3dttest++ with only a single group defined and convert the t-statistic to a correlation. Instructions for that are here.

3

Thank you Peter. I just used the 3dMEMA way to generate what I need, referring only to one -set of data (across both study group subdirectories)

4

Along the same lines, is there a way to include a covariate of non-interest in this correlation? Say I want to correlate task-contrast beta with a performance measure, while “correcting” for age, for example.