Dear Cliff,
> I’ve noticed that, when looking at longitudinal PET scans, some opt to use
> subtraction images and one-sample t-tests to look at change across two time
> points, in lieu of paired-t tests. I am wary of this, and was wondering if
> anyone could comment on the use of subtraction images. Thanks so much.
A one sample t-test on difference scores (aka subtraction images) is
the same as a paired t-test across those images. To convince yourself
of this you can generate some example data in Matlab and try some
t-tests:
% generate some data
time1 = 5 + 2*randn(1,15); % mean = 5, SD = 2
time2 = 8 + 2*randn(1,15); % mean = 8, SD = 2
% calculate a difference score
diff = time2 - time1;
% t-tests
[h,p,ci,stats] = ttest(time1, time2);
[h1,p1,ci1,stats1] = ttest(diff);
% results
fprintf('\n');
fprintf('Paired t-test: t(%i) = %.3f, p = %.3f\n', stats.df, stats.tstat, p);
fprintf('One sample t-test: t(%i) = %.3f, p = %.3f\n', stats1.df,
stats1.tstat, p1);
Hope this helps!
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|