An Evaluation of Some Statistical Techniques Used in the Analysis of Paired Comparison Data

Publisher Website
Google Scholar
Abstract
Various methods of analyzing paired comparison data are discussed and compared including Thurstone-Mosteller, Bradley-Terry, Scheffe, Morrissey-Gulliksen and others. The Morrissey-Gulllksen method is used when not all pairs are compared. Data from a color preference test are analyzed by all of these methods. The response scales obtained from these methods are very similar and the choice of which method to use will depend on other properties such as distribution assumptions, experimental design or computational requirements. Each of the methods discussed is preferable for at least one of these properties.