Testing agreement between a new method and the gold standard-how do we test?

Abstract

Data analysis can be the most challenging aspect of a research study. Having been taught statistical techniques that tend to be based on finding significant differences or significant relationships, difficulties arise when trying to determine if a newly developed method is equally as good as the established method (the gold standard). Testing for significant differences is rigorous and it would be rare for researchers to report significant differences without using an appropriate statistical test. Testing for agreement is assessed with far less rigour. Analysis of papers in this journal suggests that testing for agreement is an area that could be improved by a better understanding of statistical methods by biomechanics researchers. This perspectives paper focusses on informing the reader about the assessment of agreement between two methods.