[VoxBo] how vbdumpstats works

[Daniel Y Kimberg]

I replied to Ling off-line, but I wanted to take the opportunity to
mention something about vbdumpstats.  It turns out when you do a stat
of series with the IDL-based voxel surfer, it doesn't honor your
detrend flag.  This came up on the mailing list this past summer, and
I probably said I was going to fix it asap, but I still haven't done
anything about it.  So any time vbdumpstats detrends your data, it
will disagree with "stat of series."  The forthcoming updated
vbdumpstats allows you to explicitly turn mean scaling and/or
detrending both off and on.  If you see differences outside of this
known bug, please let me know.

Ling Wong wrote:
> Hi, I just ran vbdumpstats and was wondering about the output.  How are the
> one-tailed and two-tailed p-values calculated, and have the stats been
> corrected in any way?

The t-values and df are calculated as described by Worsley and Friston
in their seminal articles, and the p-values for 1- and 2-tailed tests
are from the regular t tables.  Nothing in vbdumpstats is corrected
for multiple comparisons.

dan