Hello. I hope someone help me with choosing an
appropriate method of statistical analysis for the
following paradigm.
We have 3 groups of mice (1 control and 2 which are
genetically altered). Each of the 2 eyes of each mouse
has 4 laser spots applied. Laser spots can create a
response known as CNV, which is graded as 0, 1, 2, or
3 (ordinal scale), where 0 means no CNV resulted from
the laser, and 1, 2, 3 refer to increasing intensity
of CNV.
We can assume that for each of the 3 groups of
animals, there is a probability (p1, p2, p3) that each
laser spot response would be 0. I don't know that we
can make this sort of binomial distribution for the
probability of a score of 1, 2, or 3.
Each of the 3 groups contains 5 animals. The animals
are observed at 3 time points. We wish to know if each
of the 2 genetically altered groups of mice are
different from the control group in their CNV response
to laser at each of the 3 time points. One thing we
would like to know if the probability of a 0 score is
less in one of the groups (0 vs. non zero responses).
My problem is that I don't know whether to treat each
laser lesion as a single data point, giving 8 data
points for each animal, or to somehow give a summary
score for each animal and treat the total as 5 data
points, or to use some type of repeated measures
analysis.
Your help would be greatly appreciated.
With regards,
Jayakrishna Ambati, MD
Retina Service
Massachusetts Eye & Ear Infirmary
Harvard Medical School
Boston, MA 02114
[log in to unmask]
__________________________________________________
Do You Yahoo!?
Get Yahoo! Mail – Free email you can access from anywhere!
http://mail.yahoo.com/
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|