> The reason why I think there is this increase in power is that I am getting rid of the
> variation around the grand mean, and just comparing within-cell variability to
> between-cell variability.

First, how were the two maps color coded? Do they show the spectrum of beta-values and z-scores respectively, or just the t-values for both?

Secondly, are you sure that the two maps have different p-values?

You basically standardized each subject's response magnitudes across all conditions at each voxel by removing the group mean and scaling by the cross-subjects standard deviation.

> I've seen this type of normalization approach used for other types of data
> (psychophysiology, MEG, fMRI timecourse scaling, etc.), but not really for fMRI betas.
> What are your thoughts? Does this seem statistically justifiable?

I fail to see any rationale or benefit for the z-transformation of FMRI data. In addition, there are at least two problems with the z-transformation:

1) You lose the direct interpretation about the effect magnitude (e.g., percent signal change).
2) Some of tests (e.g., pairwise comparisons between conditions) are fine. However, you cannot obtain testings for those individual conditions because the standardization shifted the center in addition to scaling.

Gang



Edited 1 time(s). Last edit at 08/01/2014 10:52AM by Gang.