The authors introduce a multidimensional scaling procedure that attempts to derive a spatial representation of stimuli unconfounded by the effect of subjects' degrees of familiarity with these stimuli. The proposed model assumes that stimulus unfamiliarity produces a tendency for a subject to anchor his/her dissimilarity judgements towards a reference value on the response scale. The input data needed to perform such analyses are the degree of stimulus familiarity along with the dissimilarity judgements for all pairs of stimuli. In a Monte Carlo study, the authors investigate the extent to which the procedure recovers known parameters. Furthermore, empirical applications of the model to positioning studies of magazines and banks in The Netherlands are provided. (Author/publisher)
Abstract