Confusing sample ROI-to-ROI results

Hi Alfonso and all,

I am a beginner with resting-state analysis and confused with a result I got. I would really appreciate your advice.

I am interested in the relationship between the connectivity of two specific ROIs and a clinical measure. I have extracted the ROI-to-ROI connectivity measures (bivariate correlations) to SPSS to perform the correlational analysis separately for two groups, and got significant (negative) associations. However, as I wanted to interpret these results, I looked at the individual subjects' Fisher-transformed values and saw that some are positive and some are negative (equally distributed...). I am not sure how to proceed regarding the interpretation of this variance, i.e., is it theoretically plausible? 
A previous study that I found showed that these ROIs are negatively correlated during rest, and thus I am afraid that perhaps subjects demonstrating positive ROI-to-ROI correlations were not resting :(
Could there be other explanations for such pattern of results? Am I missing something?

Thank you very much for your help,
Zoi