afni.nimh.nih.gov

Hi,


Could someone please shed some light on the statistical impact of unequal numbers of events in a multiple regression or deconvolution analysis?

For example if I have 'x' events of condition type A and twice as many events for condition type B, how does this impact the statistical comparison between conditions A and B, say with a linear contrast as [1 -1].

Thanks,
Shruti