A comparative study of multiple-attribute decision-making techniques using a subjective experiment

The objective of this dissertation is to investigate four classical Multiple Attribute Decision Making (MADM) evaluation methods. These MADM methods are the Analytic Hierarchy Process (AHP), Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), Linear Assignment Method (LAM) and Elimination and Choice Translating Reality (ELECTRE). These four methods represent three different compensatory models, namely, the scoring model represented by AHP, compromising model represented by TOPSIS, and concordance model represented by LAM and ELECTRE. The task is to select the preferred method, the one that best represents the Decision Maker's (DM's) expectations.; In order to find the preferred method, two groups of subjects were formed from students in the Department of Engineering Management at the George Washington University. These groups were used in an experiment to determine which method has the most judgmental agreement with the subjects, i.e., the method that subjectively suites the most subjects is considered the preferred method.; The subjects were given three problems to evaluate. These problems were: (1) resort selection problem, (2) college ranking problem, and (3) apartment selection problem. The subjects' tasks were to develop evaluation attributes for the three problems, elicit attribute weights and values, and rank the problem alternatives based on the attribute weights and values they supplied. As a result, a score was assigned to each method based on the match between the method ranking of alternatives and the subject ranking of alternatives. This was achieved through a measure of effectiveness prepared to rate the four methods.; Major findings of the study demonstrated a consensus between the two groups in which a significant difference was found between AHP and LAM and between ELECTRE and LAM (at the p {dollar}<{dollar} 0.05 level). The two groups were found to prefer AHP over LAM and ELECTRE over LAM. One group also found AHP to be preferable to TOPSIS; the other group was in near agreement with group 1 in this comparison. In addition, the study showed that there is a significant difference between the subjects' prior ranking and post-hoc ranking.; Other findings were that there is no statistically significant difference in preference (at the p {dollar}<{dollar} 0.05 level) with regard to gender, employment, age, experience with MADM decision aides, and knowledge of weighting tools.; Finally, it is suggested that AHP and ELECTRE are the preferred methods, since they were not dominated by each other or by the remaining two methods. In addition, future studies are recommended to include different types of problems, under different circumstances, with subjects displaying a variety of personal and professional characteristics.