Gang Wrote:
-------------------------------------------------------
> > Gang: can I use a dataset as covariate in
> 3dttest++?
>
> Yes. Read the 3dttest++ help for the usage and
> details.
>
> I don't know the context of your analysis.
> However, it's generally not considered a good idea
> to perform analysis on statistics including
> t-stats. Instead, it makes more sense to make
> inference on effect estimates such as beta values.
He is only asking for a diagnostic to evaluate similarities between regression coeffcients from two different studies on the same space, not a formal statistical test. I don't think that there is anything wrong with it. The reason I was suggesting calculating the correlations between the t-statistics as opposed to the beta's is because the latter have far less chance of being identically distributed (since they do not even have the same variance) than the former.
Actually, if he can account for the spatial correlation between the observations, then nothing would be wrong with developing a formal test also
Best wishes,
Ranjan
