The Analysis of Proximity Data

Proximity data consists of measures of similarity or dissimilarity between members of a set of stimuli -- individuals or objects of interest -- and occur in many different disciplines, particularly psychology, sociology and market research. In some instances, such data arise from calculations carried out on the usual multivariate data matrix, the elements of which record the values of a number of variables for a number of individuals. In other circumstances, proximity data are collected directly from experiments in which human subjects are asked to make judgements about the similarity or dissimilarity of pairs of stimuli. Uncovering the pattern or structure in this type of data may be important for a number of reasons, in particular for discovering the dimensions on which similarity judgements are made. In this text a variety of methods which are helpful in investigating and exploring proximity data are described, and their use is illustrated on a range of data sets.Our hope is that the material contained in the book will be a helpful introduction to this area for research workers, who are not primarily statisticians but who collect and wish to analyse proximity data, and also to applied statisticians interested in the underlying methodology.