Hi Mauro,
> >To take the behavioral example... stating that they are impaired on
> >spatial working memory is entirely uninteresting.
>
> I totally agree if your assumption that people mean "particularly
> impaired after multiple tests" is correct. They should clearly not
> come to this conclusion if their patients failed lots of tests or if
> they were only tested with 1 target task. Now, if your point is that
> people test lost of tasks then, among those for which they find
> significant differences, they report only what they want, it's no
> longer a statistical issue...
Ah no, my point is that if people report only _significant_ results,
then this can be very misleading, because of the false implication
that significant results are significantly different from results that
are not significant.
> >Obviously I'm drawing a parellel with the thresholded SPM map...
>
> ... which I'm not sure I agree with. In SPM maps, one usually DOES
> test all areas for task A and B, and it is thus reasonnably
> acceptable to says "area X was the only one significantly more active
> during task A than task B".
And that seems to me exactly parallel to the behavioral situation.
I'm just restating, but in imaging, "area X was the only one
significantly more active during task A than task B" implies the added
" and was particularly active compared to the rest of the brain"
otherwise it wouldn't be interesting. The added part is clearly
unsupported without extra evidence, which is almost never presented.
Thanks again,
Matthew
|