Influence of the Different Measurement Scale and Normalization Method on Results in Topsis

Abstract

The present study presents a comparative analysis of different measurement scale (i.e. linear, power, and so on) and normalization method (i.e. vector, linear, and so on.) in TOPSIS, by testing them against conventional TOPSIS assumption (in the absence of any other standards, the priorities provided by the conventional TOPSIS was used as the benchmark). In this study, the following two questions are considered: 1. Does a decision priorities changed when using different measurement scale and normalization method? and 2. Does the same results of the conventional TOPSIS, be obtained from the modified TOPSIS (in other words: TOPSIS with different measurement scale and normalization method)? It is shown that, the different measurement scale and normalization method can be lead to different priorities (rank and preference intensities). Also, the modified TOPSIS priorities (TOPSIS with geometric measurement scale and logarithmic normalization method), the only or best combination that is roughly equivalent priorities as compared with conventional TOPSIS. So, we suggest (after the more experimental research in future) the modified TOPSIS method as the alternative solution.

Keywords: TOPSIS, Modified TOPSIS, Measurement Scale, Normalization Method

View PDF