Statistical treatment of free sorting data: a brief review of methods and a new association model

Abstract

Several statistical procedures, mainly pertaining to multidimensional scaling, multiple correspondence analysis and cluster analysis, have been proposed for the analysis of data from a free sorting procedure. A brief review of these methods is sketched and a statistical model to assess the association between products is discussed. Among other possibilities, these models make it possible to set up a hypothesis testing framework to assess the significance of the effect of external factors on free sorting data. It is also
performed in conjunction with a latent class strategy to identify segments of consumers and better highlight the relationships among the products. An illustration on the basis of a case study is outlined.