I have a statistical dilemma facing me, and I'm interested in getting some
opinions/ideas about what I can do.
I have been working with a study where peoples physiological responses to a
particular stimuli improve following exercise. The overall study was set up
to look at the differences in this facilitation between control &
experimental groups. However, when you look at the data, there seem to be
two distinct groups coming out in the control subjects, and we are
interested in finding out whether this is a valid difference. The control
subjects come out pretty much clustered in two groups, ones with a low
facilitation (1-3 x their baseline scores) and ones with a high
facilitation (6-9 x their baseline scores), rather than a "normal"
continuum of scores. The analyses done so far indicate that this unusual
outcome of data is not due to instrument error or practice effects, and the
scores are consistent. The design is a bit convoluted - so far it is
repeated measures on 10 subjects, with four "sittings" of every subject.
Each "sitting" consisted of 4 baseline measurements and 4 facilitation
trials.
While I know that throwing around post-hoc hypotheses is philosophically
incorrect, I am very interested to test weather there is a real difference
within these subjects, especially as it could start giving us some leads as
to the underlying cause of the difference (should it be real). Does anybody
out there have any suggestions as to a particular statistical technique I
could apply or anything further that should be done experimentally to
examine my hypothesis?
Thanks for any help in advance. Please e-mail me directly at foc@iname.com
B.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%