Dear experts,
We are conducting an fMRI study where we look at the whole-brain correlation of fMRI activation during a cognitive task with clinical factors by using multiple regression.
Since two regressors of interest are highly intercorrelated, we performed two separate regression analyses, where each of the regressors was used as a covariate of no interest in turn for the other, in line with several previous publications.
A reviewer suggested that this approach is statistically invalid, could anyone comment on that?
If it is in fact incorrect, what would you think could be the appropriate analysis?
Any help on this topic would be most appreciated.
Thank you very much in advance
|