Multi-attribute decision making(MADM) is quite prevalent in daily lives or organizational settings. Researchers have proposed not a few methodologies to assist decision makers in choice making with, usually conflicting, attributes. Many of these approaches require cardinal or numerical data from decision makers. In practice, it is not easy for decision makers to provide such cardinal information and they also feel considerable cognitive burden.
To cope with the problem, a new approach was developed recently so called MADM with incomplete information, shortly denoted MADMII. This method is to evaluate alternatives considered without determining specific weights of attributes. Although the theoretical background of MADMII was verified by some eminent scholars, its advantages and effectiveness in a practical use are not yet examined.
This thesis aims at verifying the validity or applicability of MADMII empirically. For that purpose, we carry out an experimental comparison of MADMII with some other methods: holistic judgment, the ratio method, the swing method, and the pricing out method. An job selection problem is used for that comparison. Subjects are 72 undergraduate students who enrolled in the course of decision analysis. Each method is evaluated in four criteria developed in this thesis: rank-order correlation, convergent validity in determining an optimal alternative, easy of use, and trustworthiness.
In terms of findings, there are four main results: (1) the mean individual Spearman correlations of alternative ranks between MADMII and the other MADM methods are very high, (2) MADMII is highly correlated to the other MADM methods from the comparison of convergent validity in determining an optimal alternative, especially, MADMII makes a mark of the highest convergence among the methods under comparison, (3) MADMII is not good in the criteria of easy of use, and (4) MADMII has the best trustworthiness.
The main recommendation for further application is that (1) MADMII should be implemented as an interactive computerized system for reducing difficulty of use on decision makers, and (2) to incorporate the ratio method into MADMII is suggested in order to be in harmony with the decision maker's behavioral tendency for providing their preference information.