Introduction of evidential contribution measures for the input variables used in a CaRBS based analysis: An application in strategic consensus

摘要

CaRBS (Classification and Ranking Belief Simplex) is a novel technique for object classification, which, due to its reliance on the Dempster–Shafer theory of evidence, can operate in the presence of ignorance and ambiguity (uncertain reasoning). In this article, we further the introduction of the CaRBS technique with the development of new measures that quantify the evidential contribution of input variables in terms of the belief that can be assigned to objects being associated with a hypothesis and not-the-hypothesis. The analysis described in the article utilises CaRBS to explore an important management issue: intra-organizational consensus on strategic priorities. Results depicting the contribution of organizational structure, process and environment to the relative degree of strategic consensus within a large public service organization are derived. The efficacy of the introduced measures is then illustrated through comparisons with results from a series of logistic regression and neural network models.