年龄-时期-世代模型的发展历程与社会科学应用

作为社会变迁研究的有力分析工具,年龄-时期-世代(APC)模型近年来得到了国内社会科学学者的广泛关注。由于这三个时间维度存在完全共线性关系(年龄=时期-世代),求解该模型始终面临严峻挑战。本文系统回顾了传统APC模型的基本原理和五种最常用的求解方法:受限广义线性回归法(CGLM)、代理变量法、内生因子法(IE)、分层交叉随机效应APC模型(HAPC)以及最新的APC交互模型(APC-I)。此外,本文总结了国内社会科学研究中APC模型的应用情况,并针对国内数据来源的特点介绍了不等间距数据在APC模型中的应用。最后,本文分析了未来APC模型可能的发展前景以及在国内社会科学应用中的机遇和挑战。

Age-period-cohort model(APC)has been widely adopted by domestic social science research in recent years.Although APC model has a long history in western literatures,there is little agreement on how to adequately analyze APC model.This article begins with an overview of classical APC model,discussing the identification problem.We then outline five major techniques to solve the identification problem:(1)restricted generalized linear regression(CGLM);(2)proxy variable method;(3)intrinsic estimator(IE);(4)hierarchical cross-random effect APC model(HAPC);(5)APC interaction model(APC-I).In addition,this article summarizes the Chinese social sciences literature that used APC model,and introduces the application of unequal spacing data in APC model according to the characteristics of Chinese data sources.Finally,we analyzes the possible development prospects of the APC model in the future,as well as the opportunities and challenges in the application of Chinese social sciences.