A decision maker friendly method of extracting preference information for preference function modeling

Deals with a method of extracting the preference information of a decision maker (DM) for preference function modeling (PFM), as proposed by J. Barzilai (1997, 1998). We suggest a method of asking questions which are easy for a DM to answer. In this method, we propose a method based on a rating scale to decompose each criterion into multiple indexes. Then, for each alternative, questions with selective choices are asked to the DM with respect to each index, as a questionnaire. This is easy for the DM to answer, since questionnaires with selective choices have been widely used in many research areas of social science. Decomposition of each criterion into multiple indexes descends from the abstract level of the criterion to the concrete level of the indexes. In this way, the DM can get a concrete picture of the concerned criterion in terms of the indexes based on his/her past experiences. To obtain multi-criteria measurements, we need to find the weight for each criterion. To evaluate the ratio of two weights, we evaluate the marginal rate of substitution between each pair of criteria by using questionnaires with selective choices. Finally, we construct a preference function by using a one-dimensional scale of each criterion and the weight for each criterion. A concrete example of a preference function model for choosing a DM's place of residence is included in order to show the effectiveness of the method proposed in this paper.