Hi folks,
I'm hoping someone can help me with this problem. I've scoured
numerous texts and statistics web-sites, but haven't found an
answer...
I measured the rate of oxygen consumption by animals running on a
treadmill at several different speeds. My design was:
* 5 animals
* 3 measurements at each speed
* 8 speeds
I used the same 5 animals at all speeds, so it is a repeated-measures design.
Variation within animals (i.e. between trials at any one speed) was
much less than variation between animals, so I have been using the
mean of the 3 trials for each animal. (I appreciate that this may be
less than optimal).
Similar studies on other species have used (simple linear) regression
analysis to find the relationship between oxygen consumption rates
and speed. However, I understand that linear regression assumes that
the y-values are independent, which they may not be if the
experimental design is repeated-measures.
It has been suggested to me that the best approach is to generate
(simple linear) regression equations for each animal, and then
compare the slopes and intercepts of each regression line. If all
slopes and intercepts are equivalent, then I can generate an overall
regression equation for the species.
I have also toyed with using a repeated-measures ANOVA, but my sample
size is small and so the power of the ANOVA won't be great. Also, in
order to be able to compare my results to previous studies, it would
be nice to have a regression equation.
My questions:
(1) Is there any way to effectively include the repeated-measures
design into a regression analysis?
(2) Is the method of separating the different individuals a suitable one?
(3) Is there any way of using the data from all 3 trials rather than
taking the mean?
Pointers to good stats texts or papers that deal with this issue
would also be most appreciated.
Thanks,
Koa.
--
----------------------------------
Koa Webster
Postgraduate Student.
Dawson Laboratory
School of Biological Science
University of New South Wales
UNSW SYDNEY 2052
phone: 9385 2216 or 9385 2123
mobile: 0412 355 403
fax: 9385 1558
email: [log in to unmask]
|