Sample size requirements to test the equality of raters' precision

Exact and approximate methods are developed to calculate the required number of subjects n in a repeatability study, where repeatability is measured by the precision of measurements made by a rater. The exact method is based on power calculations under the non-null distribution of the multiple coefficient of determination, which requires intensive numerical computation. The approximate method is based on predictions from families of non-linear curves fitted by the method of least squares.