Allstaters,
Please accept my apologies for the second posting. I needed to clarify
something in my previous query. I mentioned Bonferroni adjustment out
of desperation because the standard post hoc tests like Scheffe and
Tukey were not derived for repeated measures. On the other hand, the
standard Bonferroni adjustment appears to be too conservative because,
in the example I quoted, the six possible comparisons are not orthogonal.
I did wonder whether I should adjust for the number of df available
(in this case 3), rather than the number of comparisons .....
Allan
--
Dr. Allan White, Statistical Advisory Service, University of Birmingham
Tel. 0121-414 4750 or 44750 (internal), Email [log in to unmask]
|