Constructing a multi-objective measure of research performance

Abstract

This paper focuses on the dichotomy between the multifaceted and multidimensional nature of contemporary R&D activity and unidimensional approaches to the measurement of its performance. While publications in refereed journals and citations are the most preferred indicators of research performance, there are also other indicators such as chapters in edited books, research reports, patents, algorithms, prototypes and designs, etc., which cannot be overlooked. Even when multiple indicators are used, they are used in isolation with the result that one gets only partial views of a multidimensional manifold. Here, a major problem is how to construct a composite measure of research performance, without assigning arbitrary weights to different measures of research output. This problem is particularly important for cross-institutional and cross-national comparisons of research performance. In this paper we have demonstrated the feasibility of constructing a multi-objective measure of research performance using Partial Order Scoring (POSCOR) algorithm developed by Hunya (1976). The algorithm is briefly described and applied to the empirical data on research outputs of 1460 research units in different socio-cultural, institutional and disciplinary settings. The potentialities and limitations of using POSCOR algorithm in scientometric analysis are briefly discussed.