Hi,
> I am reading Friston et al. (2005) Conjunction Revisited. I understand
> that if I want to find brain region that responds to the common component
> between two conditions A (component 1 and component 2) and B (component
> 1), put in other words I perform a cognitive conjunction, I can select the
> two contrasts corresponding to the two conditions A and B for conjunction
> in SPM 5 Contrast Manager and then I set "Global" rather than "Conjuncion"
> for "Null Hyp. to asses?".
>
> With this procedure I test Null Hyp.: K=0 effects; K=1 effect and I infer
> K >= 1 effects. The number of contrasts that share some uninteresting
> effect (in my case only the contrast for condition A) is the number of
> effects for H1 (in my case 1).
The global null tests against the null hypothesis that there are no
effects. Thus, if either of your two effects is positive, the test
will tend to reject the null, and give you a significant result. If
you want to test whether both your effects are positive, you'll need
against the conjunction null, which in this case is that there is only
one positive effect (of the two you are testing).
Best,
Matthew
|