Abstract
Summary: In answer to Robinson's trenchant critique of the inadequacies of multivariate statistical decision theory in psychology and the reification of factor structure be replaced with a new way of dealing with data, through structuralndescription and graphics, "visual parables" if you will. In keeping with this structural emphasis, following a brief discussion of notable failings of the traditional approach, a number of examples of descriptive multivariate graphics are given with explanation. Three kinds of data are illustrated: an analysis of paradoxical results in vocal expression of emotion, an experimental study of supercomputer performance data (as a mechanistic analogue to human performance measurement), and a canonical correlation analysis of the correspondence of mutual fund performance to market indices for use in studies of contextually informed market trading behaviour. Future directions and practical applications are discussed.