Dear SPM users,
I have a question regarding the logical model underlying a partial correlation. I have two behavioral variables, A and B, and a region C. There is no correlation between A and activity in C and no correlation between B and activity in C, but A and B are correlated. However, when I adjust A for the variance of B and vice versa (i.e. B for the variance of A), both adjusted variables now show a positive correlation with region C. I wonder whether this is logic from a causal analytic point of view? If A and B share common variance, shouldn't there be a correlation between those variables and region C and not only between the adjusted A-/B-values and region C?
Kind regards,
Anne
