Alex Shackman wrote:
> A further issue to consider when choosing between fMRI and O-15 is the oft
> repeated statement that, unlike PET, fMRI allows for meaningful
> inferential statistics to be performed on individual subject's
> data. In this fashion, one can examine both common activation and
> individual variations -- an advantage which could prove doubly useful in
> the context of experiments employing small sample sizes.
This probably is a big advantage. Then again, it depends on study design and
statistical power. If you are looking directly at drug effects, you may have
essentially one baseline measure and one on-drug measure and you may or may
not have the power you have with repeated sensory or cognitive activations &
baselines.
>
> A perhaps less commonly appreciated point that has been recently discussed
> in print by members of both the London group and the D'Esposito lab, is
> the legitimate statistical generalizability of fixed (between-group)
> versus random (repeated-measures) analyses -- fMRI can be used to perform
> random analyses.
Others know a lot more about this but random effects analysis can be applied
to both.
Thanks,
Kevin Black MD
|