Comparing generalized procrustes analysis and statis

We consider a model for sensory profiling data including translation, rotat ion and scaling. We compare two methods to calculate an overall consensus f rom several data matrices: Generalized Procrustes Analysis (GPA) and STATIS (Structuration des Tableaux a Trois Indices de la Statistique). These meth ods are briefly illustrated and explained under our model. A series of simu lations to compare their performance has been carried out. We found signifi cant differences in performance depending on the variance of random errors and on the dimensionality of the true underlying consensus. Therefore, we i nvestigated on the dimensionality of the calculated group averages. We foun d both methods gave too many dimensions compared to the true consensus. Thi s finding is supported by some theoretical considerations. Finally we propo se a combined approach which takes advantage of both methods and which gave better results in the simulations. (C) 1999 Elsevier Science Ltd. All righ ts reserved.