Hi everybody,
I've run an fMRI study in which I made two models: model A the hrf alone
and model B the hrf + a parametric modulation. In addition to that I've
also ROIs from a localizer -Is that alright to compare the beta values
for each condition between models ?? (I ran an ANOVA with all factors
including the model factor on beta values from the hrf alone or the
hrf+modulated response)
As a result I've got from the model with the hrf alone a main effect,
e.g. A>B (indeed A1>B1 and A2>B2) and from the model with the hrf + a
parametric modulation an interaction A1>B1 but A2<B2. Note there is no
main model effect, i.e. the fit is not better .. right? so should I
forget the modulation effect or investigate further the modulation vs
hrf in my model B?
thanks a lot for your help
Cyril
|