1.1 Research Background, Motivation, and Significance
Multi-criteria decision making (MCDM) has been widely used in our daily life Stewart, 2008). In the real life, people have to face a plethora of multiple targets or criteria decision makings; moreover, these criteria may conflict with each other. Human beings applied MCDM in many domains, namely, land use analysis (Antoine, Fischer, and Makowski, 1997), Aerodynamics (Stewart et al., 2008), financial planning (Collier and Gregory, 1995), and etc. Take land use analysis for example; when planning land, managers not only have to consider land development and supply planning but also be necessary to trade-off the repellency, negative environmental impacts, and facility investment costs between different land (Wang, 2005).
Due to distinct decision content, MCDM problems can be separated into multi-attribute decision making (MADM) and multi-objective decision making (MODM) (Wang, 2005). MADM is focus on evaluating questions which can be used in known and limited feasible options; these feasible options are dispersed. Though a set of assessing process, we could measure the relative importance of each attribute. In those feasible options, we can utilize required attribute to find out optimal option. However, there are a plethora of MADM tools; each MADM tools are based on distinct theory. As a result, when using different tools to deal with same question, rank reversal might occur to change the utility.
Unlike MADM, MODM is focused on planning. MODM is utilized to handle questions which constraint and decision target are known, but feasible options are unknown. The decision variable of MODM questions are successive and have multiple limited options. As a result, we will have to generate feasible options to acquire more decisions’ information. Constraints and targets can generate area of feasible options and a feasible solution set. In the feasible solution set, decision-makers can follow their preference to make decision. However, they can only select but not design feasible
solution. As a result, they should use efficient optimal algorithm to find out the most optimal solution. Three methods, specifically weighted, sequence, and multi-objective programming, are significant factors to Table optimal solution set. Current MODM decision methods can be classified according to sectors of decision-makers involved.
Firstly, if decision-makers partly understand the questions and express their preference at the beginning, analysts can utilize utility function and target planning to add preference into function directly. Secondly, by interacting with decision-makers, analysts can determine decision-makers preference in every stage to develop new feasible solution set and gradually find out the optimal solution. Lastly, analysts can narrow down efficient solution set to simplify decision questions, such as multiple-objective linear planning.
Currently, MODM decision-making tools can be classified according to sectors of decision-makers involved. If the decision makers on this issue, some understanding, and then to express a preference at the beginning, you can use the utility function and objective programming approach and direct preferences to join this function. While the other is to consider the interaction between analysts and makers, decision-makers through each stage of choice or preference, the development of new and effective solution set and gradually approaching the optimal solution. The final classification of the purpose is to narrow the set of effective solutions, simplify decision-making problems, such as multi-objective linear programming.
However, a plethora of information will make the decision become much more difficult. Because the ability for human being to deal with data is limited, too much information will cause negative impact when making decision. For instance, systematic error (Kahneman and Tversky, 1979) will generate when decision-makers were forced by a plethora of data and have to utilize experimental rules or normal methods to solute problems. According to the past lecture, the rationality of people is bounded (Tversky and Kahneman, 1983), and the cognitive ability of people is also limited. When making significant decision under high pressure, most of decision-makers cannot deal with problems efficiently; moreover, they were used to simplify the problems or automatically filter information. Those activities will let decision-makers get into decision traps, especially easier for high-level managers who have extremely high time cost.
Furthermore, those tools which can inspire preference may also cause errors from
sensitive decision-makers (Anderson and Clemen, 2013). When conducting too many times interactive methods (which can inspire preference) or requiring making decision from immature scheme, high-level managers may have negative emotions and to make wrong preference selection.
According to the effectiveness and validation scheme on multiple criteria decision-making methods (Wu and Tiao, 2015), the research compared several normal methods, namely weighted sum model (WSM), weighted product model (WPM), analytic hierarchy process method (AHP), revised analytic hierarchy process method (RAHP), TOPSIS method, VIKOR method, and pair method. Under the assumption that people are rational, they have evaluated what kinds of methods can be trusted and efficient. However, scientists articulate that cognitive science is having an impact on decision science and cognitive science is able to describe the dynamical nature of the decision-making process.
In the real life, people have easily changed their decision through psychological influence.
Since 1974, Tversky and Kahneman have articulated that people systematically violate the axioms of rational decision theory in fundamental ways.
In conclusion, this research wants to understand what kinds of MCDM methods can help decision-makers do better determination under different context effects. There are several normal context effects, namely attraction, compromise, similarity, attribute-balance effect. While decision theorists have attempted to explain these establishes under single modeling accounts, there has been no empirical evidence suggesting that the four effects can be obtained under the same experimental paradigm. Moreover, whether decision process will be impact through different decision methods will be explored by this research.
1.2 Research Objectives and Scope
When decision-makers' make decisions, most of them didn't prefer to use repetitive interactive methods to find their preference selection. Though interactive methods can lead decision-makers demonstrate completely and exhaustively preference., it might cause decision-makers unable to focus on the main point because of impatient or other
has incorporated motions, and, then influence the result. As a result, it is still a question that whether the interactive approach is really better than other MCDM methods to meet the decision-makers' preference
There are many MCDM methods that can be applied in different fields.
However, there is no enough research to show how close the result of MCDM methods to the selection which decision maker really want under different situation. Therefore, in this study, we want to understand the impact of the decision-makers' preferences to MCDM methods. Moreover, we could also focus on how much the context effects will affect the decision-makers' preference and evaluate the effectiveness of MCDM methods after considering context effect. Taking all mentioned into accounts, in this research, we want to find better multiple criteria decision-making methods to help decision-makers increase their decisions' quality. According to several experiences and methods under diverse situation, we hope to find out which methods could generate smaller error and how to perfectly consider context effect in decision making. After that, we can help decision-makers to demonstrate their preference more effectively.
1.3 Thesis Structure
This study can be divided into five chapters. Our research target is to find out which evaluation methods would be better when decision-makers are facing multiple criteria non-dominated solutions. We also study that whether the result would be influence by different program set. The first chapter is Introduction. This chapter introduces the research background, motivation and objectives. The second chapter is Literature Review.
According to the current classification of MCDM and frequent decision-making tools, we could realize the effect of cognitive bias theory to decision-makers and utilize those tools as the base of research experience. The third chapter is Research Method. Based on the theoretical resources from Chapter2, we could find out the appropriate research methods to design the research framework. We also propose the research method in this chapter.
The forth chapter is Research Result and Analysis. In this chapter, we explain the result after finishing the research and analyze it more specifically. The last chapter is Conclusion.
Based on the result we proposed in Chapter4, we have to figure out whether the design of
this research can help decision-makers or not. Moreover, we explain the research restrict and provide the directions for follow-up research.