Dear Biomch-L Readers,
I am writing to ask a question about using PCA on repeated measures data.
I have been looking at possibly using PCA in Matlab to analyze
differences in frequency response data from accelerometers. If using
the built in function princomp, you would create a matrix of your
functions, and the analysis would be completed.
However, I was wondering whether this works for repeated measures.
Briefly, in my experiments, I am taking measurements of vibration in
the spine. The accelerometer takes 5 repeated measures of the spine's
response to vibration. Then, it is moved to another position and takes
another 5 measurements. If I want to compare the differences in the
curves between the two positions, should I just put all 10 curves into
the PCA grouped together? Or is there some way that I must group the
repeated measures first?
Thanks for your help in advance. I will post any replies I receive.
Sincerely,
Colleen
Colleen Decker
M.Sc Student, Biomedical Engineering
University of Alberta