Statistical methods for analysing discrete categorical data recorded in performance analysis

Abstract

When compiling summaries of notational analyses of sport performance, analysts have traditionally ignored the discrete nature of the indicators collected, and relied upon statistical methods predicated upon continuous, normal distributions. In this paper, we presented more appropriate statistical methods for analysing the categorical differences between indicators, such as the chi-squared test of significance and log-linear/logit models. The editor invited us to write the paper for a special edition on sport performance analysis. Cooper wrote the specific sections on the chi-squared analysis, drawing together the examples and conducting the analyses of the data. Cooper also helped edited the manuscript.