Hi,
> Using a fixed effect analyse of a block design, I have considered
> significantly activated the brain areas surviving to the threshold of
> p < 0.05 corrected for multiple comparison as well as those for which
> I have a strong a priori hypothesis surviving to the threshold of p <
> 0.001 non corrected for multiple comparisons on the whole brain.
> Could you tell me if this approach is still valid using SPM99 because
> a referee has told me that it was not.
I think it the reviewer would be correct to object to the p<0.001
uncorrected threshold, as it has no statistical basis. If you want a
valid statistical threshold for regions you consider to be of
interest, you should use the small volume correction utilities in SPM,
Best,
Matthew
|