Fuzzy measure theoretical approach to screening product innovations

Variety of decision models have been proposed in contemporary literature to tackle the problem of screening product innovations. Although linear models have gained considerable attention and recommendation, contemporary literature contains strong evidence in support of nonlinear noncompensatory models. In this paper, the authors first demonstrate how fuzzy measures, which are defined on subsets of decision attributes, and their Choquet-integral formulation, which exhibits both compensatory and noncompensatory properties, have meaningful behavioral interpretations within the context of new-product screening. Then, they show how to address the complex problem of building such measures by applying a learning algorithm that relies on methods of judgment analysis. An accompanying case study demonstrates how organizations may customize a new product decision aid and fine tune their business strategy as actual results accrue. Finally, the authors present the results of analytical studies to compare the Choquet-integral model with other noncompensatory models, such as Martino's extended scoring model and Einhorn's conjunctive model, and heuristic approaches, such as Tversky's EBA and the lexicographic method. For the new-product-decision scenario considered in the study, the Choquet-integral model provided the best fit, measured by Pearson's rank order correlation coefficient, with all of the competing models.