Dear statisticians
Because of correlated regressors, the diverse regressors of a subject (one session) were estimated each in a separate GLM (multiple GLMs on the same volumes). Now, instead of running groups stats for each GLM, I run group stats using one single model, including con-images of those separately estimated regressors (As a result, I included much more con images, I got higher degrees of freedom and higher stats).
Now, did I violate statistics?
Thanks for any comment
RL
|