Correlation coefficient repeated measures

#1
Hi Everyone,
I would like to measure the correlation between two quantitative variables. These variables measure the same thing but are obtained with two different devices. My problem is that a same subject can have several measures (at different times).
I wonder if there is a way to calculate a correlation coefficient taking account of a possible "subject effect".
I hope that I was clear enough...
Thanks in advance
 

Jake

Cookie Scientist
#2
One approach would be to center all of the measurements around their subject means (thereby removing the subject effects) and then compute the correlation of these transformed measurements.
 

Miner

TS Contributor
#3
Are these different devices two versions of the same measurement device (e.g., two different weighing scales) or are they two different approaches to measuring (e.g., one weighing scale vs. measuring displacement)?
 
#4
Thanks for your responses.
Miner, to answer your question, the two measures are obtained by two different medical machines, that are supposed to give more or less the same results (there are not two different weighing scales). I don't know if that answers your question.
 

Miner

TS Contributor
#5
It does. There are two approaches that I have used to evaluate the equivalence of two measurement devices. The first is a Youden Plot, and the second is the Bland-Altman Plot. These are better approaches than measuring the correlation. Two measurement devices might correlate well, but one could be biased high/low, or the relationship not be 1:1.