Test-retest reliability in quantitative physical education research: A commentary

Abstract

This paper highlights an important statistical development for exercise and physical education research. Traditionally, the Pearson and intraclass correlation coefficients have been liberally used by researchers to quantify the test-retest reliability of many performance, behavioural, and physiologically-related measurements. The suitability of these forms of analyses has recently been challenged by British exercise scientists, who argue that they do not really address what they are meant to, that is, the level of agreement between repeated measurements or scores. As a consequence, our existing knowledge of the reliability of such measurements is questionable and deserves to be re-established with a more appropriate statistical technique. Accordingly, the 95% Limits of Agreement method is presented and offered as an essential supplement for future measurement and evaluation research.