A test of the assumptions of multidimensional scaling by separating sensitivity and response bias

Summary This study investigates one of the main assumptions underlying multidimensional scaling models. It is tested whether the violation of decomposability found in previous studies can be explained by variation in sensitivity or in response bias. Rectangles of different shape served as stimuli in an experiment using a matching procedure. In addition the probability for presenting pairs of identical stimuli was varied. The data were analysed by a model separating response bias and sensitivity. Six versions of the model with different numbers of parameters were applied. The results suggest that the violation of decomposability comes from variation in sensitivity and not in response bias.This research was supported by a grant from the Deutsche Forschungsgemeinschaft to the first author. All computing was done on the Telefunken TR 440 of the Deutsches Rechenzentrum Darmstadt.