Allstaters,
I have a one-way repeated measures experimental design with
four levels on the independent variable. The appropriate omnibus F-test
is significant. I now wish to look at all possible pairwise comparisons
using matched-pairs t-tests with a Bonferroni correction for the six
comparisons. However, I have a dim and distant memory that this is not
appropriate for a repeated measures design. Can anyone confirm this and
explain the matter to me and, furthermore, suggest an appropriate
procedure?
Thanks in anticipation.
Allan
--
Dr. Allan White, Statistical Advisory Service, University of Birmingham
Tel. 0121-414 4750 or 44750 (internal), Email [log in to unmask]
|