Comparing Measurement Error Between Two Different Methods of Measurement of Various Magnitudes

Measurement error is a common problem in several fields of research such as medicine, physiology, and exercise science. The standard deviation of repeated measurements on the same person is the measurement error. One way of presenting measurement error is called the repeatability, which is 2.77 multiplied by the within subject standard deviation. In this article, the two different repeatabilities obtained from two different protocols or “tests” are statistically compared. Briefly, the squared differences between protocols are natural log transformed to achieve normality and compared with a paired t-test. The natural log transformation guarantees that the squared differences follow a normal distribution. When more than two repeatabilities are compared from more than two protocols or “tests,” repeated-measures ANOVA comparing the log transformation of the squared differences is used. When comparing repeatabilities between protocols and between different magnitudes of measurement, repeated-measures ANOVA compares the natural log transformation of the squared differences between protocols and between different magnitudes of measurement.