Modeling qualitative data in data envelopment analysis for composite indicators

Abstract

Composite indicators (CIs) are useful tools for performance evaluation in policy analysis and public communication. Among various performance evaluation methodologies, data envelopment analysis (DEA) has recently received considerable attention in the construction of CIs. In basic DEA-based CI models, obtainment of measurable and quantitative indicators is commonly the prerequisite of the evaluation. However, it becomes more and more difficult to be guaranteed in today’s complex performance evaluation activities, because the natural uncertainty of reality often leads up to the imprecision and vagueness inherent in the information that can only be represented by means of qualitative data. In this study, we investigate two approaches within the DEA framework for modeling both quantitative and qualitative data in the context of composite indicators construction. They are imprecise DEA (IDEA) and fuzzy DEA (FDEA), respectively. Based on their principle, we propose two new models of IDEA-based CIs and FDEA-based CIs in road safety management evaluation by creating a composite road safety policy performance index for 25 European countries. The results verify the robustness of the index scores computed from both models, and further imply the effectiveness and reliability of the proposed two approaches for modeling qualitative data.