Dear experts, I am performing a correlation analysis and I end up with positive and negative correlation values for the same connected regions in different subjects. For instance: I am calculating the correlation between many seeds. Let's suppose, regions A and B are one pair of connected seeds. For 30 subjects from the control group I get a positive correlation value, and for 10 subjects from the control group I get a negative correlation value. This also happens in the patient group and even in different proportions (i.e., 20 positive and 10 negative). I am concerned about how to treat this values. Should I just average the values regardless of the sign? Should I use absolute values? Or should I exclude patients and controls with negative correlation values and perform comparison after that? Should I just zero the negative and do the comparison? Do you recommend any other approaches? Thank you for any ideas, All the best -- Raphael F. Casseb Medical Physicist, Ph.D. Student Medical Physics Lab - State University of Campinas Contact: +55 19 3521-8246