量化專家意見與人口推估

Combining Expert Opinions with Stochastic Forecast

鄭宇庭(Yu-Ting Cheng) ； 李政豫(Chen-Yu Lee) ； 余清祥(Jack C. Yue)

《人口學刊》 49期 (2014/12) Pp. 41-68

https://doi.org/10.6191/JPS.2014.49.02

摘要

人口推估通常使用人口變動要素合成法，通常根據現有人口總數及結構，加上對未來生育、死亡、遷移三要素的預期（或情境），推估未來的人口數及其結構；三要素中的生育和遷移變動趨勢較大，較不容易以歷史資料找出趨勢，大多倚賴專家訂定情境假設。但專家意見也有蒐集上的疑慮，例如意見沒有固定的形成格式，推估結果也難以機率角度詮釋發生可能性。為了彌補專家意見的不足，近年不少國家的人口推估使用隨機方法（stochastic methods），這些方法大多與專家意見組合使用，使得推估結果可兼具兩種方法的特色。

關鍵字

人口推估 ； 隨機推估 ； 專家意見 ； 德菲法 ； 電腦模擬

並列摘要

Population projections are important in estimating the population for future dates used by the government for policy planning. The cohort component method is the most popular population projection method, and requires information on future fertility, mortality, and migration. These three factors are usually decided by expert opinions as well as historical data. However, there are concerns in using expert opinions. For example, it is not easy to quantify the expert opinions, the expert opinions do not have the meaning in probability, and the projection results based on expert opinion usually have small variance. In recent years, stochastic methods combined with expert opinions have been widely applied in order to improve the quality of population projections. The data format of quantification also plays a crucial role in quantifying the expert opinion, and in this study we explore the estimation methods for various data formats. We first review some popular methods for quantifying expert opinions, including the focused group interview and Delphi method. Then, we use computer simulation to evaluate the influence of data format on the analysis methods. It seems that, if the expert opinion is quantified in the form of payment card or binary data, the results are not very reliable. Therefore, we propose a sequential design to collect expert opinion and the simulation study shows promising results.

並列關鍵字

population projection ； stochastic projection ； expert opinion ； Delphi method ； simulation

參考文獻

郭孟坤、余清祥(2008)。電腦模擬、隨機方法與人口推估的實證研究。人口學刊。36，67-98。

Betz, F.,Lipps, O.(2004).Stochastic Population Projection for Germany-Based on the QS-Approach to Modeling Age Specific Fertility Rates.,::Mannheim Research Institute for the Economics of Aging.

Couper, M. R.(1984).The Delphi Technique: Characteristics and Sequence Model.Advances in Nursing Science.7(1),72-77.

Dalkey, N. C.(1969).The Delphi Method: An Experimental Study of Group Opinion.Santa Monica, CA:Rand.

De Beer, J.,Alders, M.(1999).Probabilistic Population and Household Forecasts for the Netherlands.Joint ECEEurostat Work Session on Demographic Projections.(Joint ECEEurostat Work Session on Demographic Projections).: