| Abstract: | A simulation study on the possibilities to reproduce a ‘true’ rank order of preferences and an underlying configurational structure of weighted distances between ideal points and real stimulus objects by PREFMAP-2 (Phase 2) is performed. Eleven hypotheses are evaluated with respect to three measures of reproduction quality. The results are mainly derived from three-way analyses of variance. Input rank orders can be reproduced quite well. This is not true for the underlying configurational designs, from which input rank orders are thought to be derived. Cluster designs, small numbers of stimulus objects and error are main effects to limit the quality of the results. |
