Hartwig,
>
> Dear SPM experts,
>
> I have a rather basic question on SPM covariance analysis concerning PET
> activation studies. When I like to perform a covariance analysis between
> rCBF and a measure of motor performance during a single PET scan (lets
> say mean movement time of a joystick movement during a single PET scan),
> do I have to establish a normal distribution for the
> covariate-of-interest?
Nein mien frieund, the normal distribution assumption concerns the errors,
and not the predictors.
Sincerely,
Eric
Eric Zarahn
Columbia University
If I need to establish a normal distribution as a
> prerequisite for an analysis of covariance, does this refer to the
> distribution within a single scan or across scans? If the measure of
> motor performance does not follow a normal distribution, is it
> appropriate to transform the performance data into a normal
> distribution?
>
> Thanks
>
> Hartwig Siebner
>
>
>
> Dr. Hartwig R. Siebner
> Department of Neurology
> Motor Control Unit, Technical University
> Möhlstrasse 28,D-81675 Munich,Germany
> e-mail: [log in to unmask]
> FAX: 0049-89-4140-4867
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|