델파이 豫測과 베이시안 意思決定 = DELPHI FORECASTS AND BAYESIAN APPROACH

다국어 초록 (Multilingual Abstract)

The Delphi technique which is a method of eliciting and refining group judgments about the future has three features: (1) iteration and controlled feedback: (2) anonymous response: and (3) statistical group response. The primary feature that distinguishes the Delphi from ordinary polling procedures is the feedback of the information gathered from the group and opportunity given participating individuals to modify or refine their judgments based upon their reactions to the collective views of the group. Secondary characteristics, anonymity and collective responses, ovoid undesirable psychological effects for participants. More recently this technique has been used not only to forecast future events but also to generate various types of estimates. Though a number of advantages have teen cited for use of Delphi methodology, the technique itself still is not free of limitations. Many criticisms have been raised about the assumptions and procedures of the Delphi technique or its products.
Major issues associated with the Delphi applications were : (1) difficulties in the selection of the panel and the design of the Delphi questionnaire: (2) the level of measurement and use of point estimates in the response scale : (3) the process for estimating the reliability of a Delphi study: and (4) the validity of the resulting Delphi panel judgment. The validity of the Delphi forecast is typically measured in terms of the degree of consensus among the panelists as a means of quantitatively getting an "objective" opinion.
Regardless of the precision of the forecasting methods and the accuracy of the data produced, the justification of a forecast is its utility in planning and decision making. Recently, the Delphi method has been viewed as a process for group decision making within the Bayesian information processing framework. A number of approaches for attaching personal probability assessments and introducing subjective information to a group decision as a means of enhancing Delphi forecasts have been suggested by Bayesian theorists. These include weighted-average methods, natural-conjugate methods, and calibration methods.
The final fart of the article provides a conceptual model reviewing those theories invlolving Delphi procedures and Bayesian approaches to decision making. In this proposed model, the Delphi method is viewed as a data generating procedure for making a group decision within the Bayesian information processing framework.

목차 (Table of Contents)

• Ⅰ.未來豫測 方法
• 1.探究型
• 2.規範型
• 3.直觀型
• Ⅱ.델파이 方法
• Ⅲ.베이시안 統計方法과 델파이 豫測
• 1.베이시안 意思決定
• 2.델파이에 關한 베이시안 觀點
• 3.集團的 意思決定의 베이시안 方法
• Ⅴ.델파이 豫測의 베이시안 方法