To: spm@jiscmail.ac.uk
Subject: bug fix: Bonferroni corrected p-values
Date: Tue, 09 Oct 2001 18:50:06 -0400
From: "Thomas E. Nichols"
Hi all,
In June new functions were released to incorporate Bonferroni
results. These functions (silently) gave SPM corrected thresholds and
p-values based the minimum of Bonferroni and Random Field results.
(Both methods are valid, though conservative to differing degrees;
hence it's OK to take the minimum of each).
Due to a bug/typo, spm_P.m wasn't using Bonferroni-corrected p-values
even if they were smaller. Corrected thresholds are OK, but
corrected p-values did not take advantage of the Bonferroni p-values.
The only impact of this bug was that the corrected p-values in
the tabular output are not as small as they can be: When the Bonferroni
corrected p-value is smaller they still used the random field p-value.
**If** you picked up the new functions in June, please grab this fix at
ftp://ftp.fil.ion.ucl.ac.uk/spm/spm99_updates/spm_P.m
Sorry for any inconvenience.
See this message on the Bonferroni updates:
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0106&L=spm&P=R15124
-Tom
-- Thomas Nichols -------------------- Department of Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
nichols@umich.edu 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029