Standard Error of Measurement


New Member

I am trying to calculate the standard error of measurement for a large data set of test-retest data. I know the usual way to calculate SEM is the sample SD*(sqrt(1-r)). However, because my data is time series, this method does not seem appropriate to me. I 20-50 trials per individual. Since I have direct measurements of test-retest, would the SEM just be the pooled SD of the each individual subject's SD? Or would it be the [ pooled intra-individual SD*(sqrt(10r)) ]?

Thank you!