Dear John,
> When I do a second-level one-sample t-test I get the area X significant
> let's say at T value of 3.5.
> When, using the same images, I do a second level correlation with a
> psychometric / behavioral measurement (i.e. one value per subject) I get the
> same (or similar area) with a T value lets say 4.8.
> I would like to show that the psychometric measurement indeed offers
> additional information, or in other words that the response at the area X
> depends on the psychometric characteristic of each subject. The measurement
> is continuous.
> How could somebody implement this analysis within SPM?
There are two independent things which you can test: one is whether
the mean across subjects differs from 0 (the basic one-sample t-test).
The other is whether neural responses can be predicted by behavioral
measurements. These don't necessarily have any relation to each
other: you could have a case in which neural activity, on average, did
not differ from zero, but where behavioral scores nonetheless were
related to individual differences in activity. Or conversely, a case
in which mean activity was highly significant, but unrelated to
behavioral scores. Based on what you say, it would seem that you have
a case where mean activity is greater than zero, and also related to
behavioral scores.
The best way to test this would be to use a single model of a
one-sample t-test, adding in a mean-centered covariate for the
behavioral scores. A contrast of [1 0] will test whether the mean
activation is greater than zero; a contrast of [0 1] (or [0 -1]) will
test whether there is a relationship between activity and behavioral
scores.
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|