Research On Theories And Methods For Behavioral Multi-Attribute Decision Making

Multi-attribute decision making (MADM) refers to the problem of selecting alternatives associated with multiple attributes. It is a problem with extensive theoretical and practical backgrounds, such as selection of investment project, selection of new product development plan and so on. Most of the traditional multi-attribute decision theories and methods are derived from rational decision theory (or Expected Utility Theory), which requires strict assumptions regarding complete rationality of the decision maker (DM). However, some existing empirical studies about behavioral experiments have shown that characteristic of the DM’s psychological behavior is bounded rational, i.e., the DM is bounded rational in decision process. Thus, it is sometimes inappropriate to solve the MADM problem considering behavior (thereafter behavioral MADM) using the rational decision theory. Now, the study of MADM problem considering behavior has attracted the attention of scholars. It has appeared some preliminary research results but not formed systemic research system, and lack of the methodical and technical support. Therefore, it is of great important theoretic and practical significance to research the MADM problem considering behavior systematically and propose the decision analysis methods for the problem.The purpose of this thesis is to make a deep theoretical analysis and methodological research on behavioral MADM. A series of research works are conducted as follows.First, the MADM problem considering behavior is described and the theoretical framework for the research problem is proposed. In order to making the research system clear, the research problem is classified into three categories, i.e., the deterministic MADM problem considering behavior, the risky MADM problem considering behavior and the multi-attribute group decision making (MAGDM) problem considering behavior. In addition, the MADM problem considering behavior is described, and the research framework for solving the problem is proposed. These basic research works mentioned above provide a theoretical guidance framework and analytical frame work to the related problems and lay a solid foundation to extended issues and applications.Second, the study on the theory and method for the deterministic MADM problem considering behavior is conducted. Specifically, it mainly concludes the following three aspects, i.e.,1) a method for MADM considering aspiration-levels based on Prospect Theory is proposed to solve the MADM problem where attribute values and aspiration-levels are represented in two different formats:crisp numbers and interval numbers;2) a method for MADM with multiple formats of attribute aspirations based on Prospect Theory is proposed to solve the MADM problem with three formats of attribute aspirations;3) an extended TODIM method is proposed to solve the hybrid MADM problem where attribute values are in the formats of crisp numbers, interval numbers and fuzzy numbers.Third, the study on the theory and method for the risky MADM problem considering behavior is conducted. Specifically, it mainly concludes the following four aspects, i.e.,1) a method for risky MADM based on prospect stochastic dominance rule is proposed to solve the risky MADM problem where attribute values and aspiration-levels are in the forms of random variables and crisp numbers, respectively;2) a method for risky hybrid MADM based on Prospect Theory is proposed to solve the risky hybrid MADM problem considering the DM’s aspiration information;3) a method for risky MADM based on Regret Theory is proposed to solve the risky MADM problem where attribute values and probabilities of the states are both in the form of interval numbers;4) a method for risky MADM considering regret aversion is proposed to solve the risky MADM problem where attribute values are in the form of interval numbers and probabilities are in the form of crisp numbers.Fourth, the study on the theory and method for the MAGDM problem considering behavior is conducted. Specifically, it mainly concludes the following three aspects, i.e.,1) a method for MAGDM considering decision making participants’aspiration-levels based on Prospect Theory is proposed to solve the MAGDM problem with decision making participants’aspiration-levels;2) a method for group decision making (GDM) with multiple attribute and multiple identifier based on prospect stochastic dominance rule is proposed to solve the multiple attribute and multiple identifier GDM problem with decision making participants’ aspiration information;3) a method for GDM with multiple attribute and multiple identifier based on Regret Theory is proposed to solve the multiple attribute and multiple identifier GDM problem without decision making participants’ aspiration information.The proposed theories and methods for behavioral MADM can be used to solve many practical decision problems, such as risk investment decision, new product development and design, selection of supplier and so on. The research work in this thesis provides a theoretical guidance framework and methodical and technical framework to the study of the MADM problem considering behavior.