Restructuring a correlated multilevel decision hierarchy in multicriteria decision analysis

The main aim of this study is to propose applicable and easily adopted statistical analyses to restructure a sound multilevel decision hierarchy, which reflects decision makers' priorities, achieves a good balance between completeness and conciseness, and takes unconsciously intercorrelated subcriteria into account. A forest planning case was used as a numerical example to demonstrate the role of correlations, and the set-up was based on the analytic hierarchy process (AHP). With the AHP, as well as other hierarchical multicriteria approaches, shape of the hierarchy structure may influence the decision makers and their judgments. In particular, the risk of overrating some main criteria exists, if the same concept is measured within multiple main criteria. We propose an approach, in which the subcriteria are grouped so that correlations between the different subcriteria are taken into account. A new decision hierarchy is constructed based on distance metrics, explanatory factor analysis, and cluster analysis. According the results, all these subsequent analyses supported each other. The tested methods seemed to be advisable tools to reveal the impacts of intercorrelated criteria and to guide the reconstructing a new decision hierarchy without intercorrelated subcriteria underneath main criteria.