Dear SPM users,
I am evaluating the preprocessing steps of my GLM analysis (with SPM) protocol. I found that there are 2 smoothing steps, the first one in the realignment step and the second one in the smoothing step. What is the difference between those two steps?
The aim of the smoothing steps is to decrease the noise. How do you know that the smoothing steps reduce the amount of noise enough? What happens if there is still noise present in the data? If this noise is correlated to the model regressors, the wrong beta's are estimated and the model is not accurate anymore, right? So how do you know that the model gives the right result?
Thank you in advance.
Debby
|