Hi all, I'm investigating two different sets of fMRI data (the same data but with different models specified): I have 18 bregressors from physiological signal which were obtained using the RETOIRCOR algorithm. When I compare residual noise (standard deviation of residuals) in the data denoised with physiologcal regressors and (the same) data but without any physiologcal regressors and the results are highly significant - is that a true result or could it be caused by the fact that one model has a lot of regressors and the other has none and this fact alone can cause such differences?
In other words, is the fact that the model with the noise regressors has 18 more regressors than the other model (where there are 0 regressors) a problem (i.e., the reason for diffeences in std. of residuals)? Sample size and data processing are equal in both models and there are no more confounding factos in any model.
Any help is aprpeciated, thanks in advance,
Marie.