Dear All,
I have some data where a dozen subjects show a sequence of 8 (repeated
measure) responses following treatment. Separate groups of subjects receive
different treatments.
While the intermediate responses meet the necessary assumptions for ANOVA, the
early and late responses do not, being mainly zero with a few positives. In
one experiment the presence of interactions is also shown by the difference in
repeated profiles for different treatments.
The objectives of this work are to assess treatment contrasts, both overall
and at different repeated levels.
Friedman's Analysis, although weakened by the interactions, and Mann-Whitney
do provide useful information.
However, there is substantial variation within subject groups and I am
concerned to perform the analyses in the most powerful manner.
A later extension to the procedure will likely be to add in another factor
(and complicate analysis further) - differing time intervals between a
standard pre-treatment and the treatment under examination. I also may need
to handle the missing data situation.
Any suggestions on the most appropriate way of analysing these data will be
greatly appreciated.
Please reply to [log in to unmask]
Thanks,
Roger M.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|