Dear Ayako,
> We are currently working on VOI analysis.
> The mean value extracted from Y.mad file was subjected to ANO0VA.
> However, we found a warning in Kalina Christoff's website syggesting that
> data from Y.mad should not be used for statistical inference.
> I would be grateful if someone comment on this issue.
You are probably referring to the following (bold but cautionary)
statement on my ROI tools webpage:
"Please note: the data extracted from the Y.mad have undergone compression
to save space, which has led to a tiny loss in precision. For instance, if
a value was 4004.0563, it may be extracted as 4004.0561 - something which
should practically be irrelevant if the goal is to plot averages. Data
extracted from the Y.mad file, however, should not be used to compute
statistics such as standard deviation or standard error of the
mean. Anything that involves taking sums of squares may be sensitive to
the compression error."
What I meant was that because of the compression error, there is a
possibility that statistics such as standard deviation may be
overestimated, and therefore, sometimes existing effects may not be
detected. This seems especially a danger if you only compute statistics on
some subsample of your data (e.g., compare the peaks between two
conditions in two average event-related time-courses). The compression
error is really tiny, but then our effects in fMRI are sometimes really
tiny too - so the safest route is to work off the non-compressed data from
the *V*.img files - though of course in practical terms, this is the most
effortful and time-consuming route too.
I myself would be very interested in other people's comments on this.
Kalina
_________________________________________
Kalina Christoff, Ph.D.
MRC Cognition and Brain Sciences Unit
15 Chaucer Road
Cambridge, CB2 2EF, U.K.
Tel: +44 (0) 1223 355294 Ext 723
Fax: +44 (0) 1223 359062
Email: [log in to unmask]
_________________________________________
|