Multivariate Statistical Methods Applied to Sizing and Fit Testing
The difficulty of developing accurate sizing charts for clothing or equipment is often underestimated. Typically, designers intend for the item to fit a specific range of people. However, accommodation ofthat range is not always achieved. Fit testing is an important part of the design process that allows collection of data where the item is actually tried on and used by people, instead of mannequins. Multivariate statistical procedures are the proper analytic techniques for investigating this fit test data. Multivariate methods are used because univariate tests can cause designers to correct a “problem” fit area, leading to possibly more problems, instead of identifying important variable combinations which may be the true fit problem. Some of these multivariate statistical methods include principal component analysis (PCA), discriminant analysis (DA), Euclidean distance matrix analysis (EDMA), multivariate analysis of variance (MANOVA), and multivariate regression analysis (MRA). This paper discusses why and when to use these techniques and illustrates some of them with case studies.