Print

Print
Hi,

Thank you very much. Now it is clear.

Then, in your later mail (22.10.2010) you wrote about a third option:
“to keep the full paired ttest design, and then add a new EV which is like the [ 1 -1 1 -1....   EV, but with each pair of 1 -1 multiplied by the subject's score” demeaned (i.e. including both the EV of effect ([1 1 1 1 -1 -1 -1 -1]) (EV1) and the EV of demeaned score of subjects multiplied with EV1 (e.g. [2 4 -5 -1 -2 -4 5 1])) (EV2).
I tried this option too. However, I experienced that the contrast for pre-post effect (EV1) showed nothing, though it did with the original setup (wo. the additional covariate).
Right - so this option shows how the pre-post *differences* correlate with the additional covariate.  This could show more or less effects than the test of how the means correlate, as it is a totally different question.
I understand it. I mean, I had no significance for EV1 (the "difference") in this design, but in the "simple" design (wo. EV2 - covariate).

Bests,

Auer, Tibor M.D. Ph.D.
Biomedizinische NMR Forschungs GmbH
am Max-Planck Institut für Biophysikalische Chemie
Am Fassberg 11
37077 Göttingen
Germany
Phone/Work: +49-(0)551-201-1725
Phone/Home: +49-(0)551-387-0076
Mobile: +49-(0)176-8012-7921
Mail: [log in to unmask]