An improved mathematical programming formulation for multi-attribute choice behavior

Abstract

Conjoint Analysis and Mathematical Programming approaches have been used extensively in the past for modelling multi-attribute choice behavior. The Mathematical Programming approaches are more versatile in their ability to capture complex behavior but have been limited to dealing with objective attributes. Conjoint Anlysis, though limited by the additive utility assumption, allows for both subjective and objective attributes. In the present article, we modify the existing mathematical models to account for situations where the decision maker may base her decisions on only a subset of the attributes. Identification of non-value added attributes may be helpful in reducing wastage of resources. Further, we enrich the scope of the model by accomodating both subjective and objective attributes. A limitation of the earlier mathematical programming approaches has been the use of interval scale data implying the gap between any two consecutive levels of an attribute are same. In the proposed model we remove this drawback using ordinal scaled data for objective attributes. The resulting MIP problem has been solved using the data provided by Green and Wind (1975) in the context of a Conjoint Analysis study. A comparison of the results of the proposed model and Conjoint analysis is also been provided.