1ddot

AFNI Message Board
1

Dear Afni Gurus,

Once we have the correlation matrices for several subjects using 1ddot, how to average them for a group analysis?

Thanks,
Dinesh

2

Dinesh,

I assume that you used option -dem in 1ddot to compute the correlation values.

Once we have the correlation matrices for several subjects using 1ddot, how to average them for a group analysis?

Averaging among what values? Those correlations in the matrix? What hypothesis are you trying to test for the correlation matrix? Could you provide more context as to what you’re trying to achieve?

3

Thanks Gang.

I’m trying to obtain correlation matrix for all possible pairs of 40 ROIs for each subject. The 1ddot command (1ddot -dem -terse roi1.1D roi2.1D roi3.1D roi4.1D … > subj1_corr_matrix.1D) gives the correlation matrix for each subject and now I wonder how to average them for a group analysis? In other words, average correlation coefficient for each ROI pair for the entire group!

Thanks,
Dinesh

4

I wonder how to average them for a group analysis? In other words, average
correlation coefficient for each ROI pair for the entire group!

You can use

1deval -expr ‘log((1+a)/(1-a))/2’ …

to convert the correlation coefficients to Fisher Z-scores, and then run one-sample t-test on those Z-scores.