年龄—时期—世代模型是研究社会变迁的重要工具，但年龄、时期和世代之间存在完全共线性，使得求解该模型面临严峻的挑战。本文回顾了现有关于该模型的各种求解方法，通过蒙特卡洛模拟研究发现，传统的理论驱动求解方法实施条件过于严格，缺乏现实可操作性；各种方法驱动的求解方法都暗含一些不太合理的参数假定，因此很容易得出误导性结论；以往学者在实践中常常使用的经验法则不可靠。基于对这些方法进行的批判性反思，本文介绍了一种被称作边界分析的新的求解思路，并为这一求解思路提供了新的拓展方向。

Age-period-cohort model is an important tool for studying social changes, but the existence of complete covariance among age, period and cohort makes solving the model a serious challenge. This article reviews various existing solution methods for this model through Monte Carlo simulation studies and finds that: the traditional theory-driven solution methods are implemented under too strict conditions and lack practical operability; various method-driven solution methods imply some less reasonable parameter assumptions and thus are prone to misleading conclusions; and the rules of thumb often used by scholars in practice in the past are also unreliable. Based on a critical reflection on these methods, this article introduces a new solution idea for APC analysis called bounding analysis and provides a new direction for its expansion.

欢迎阅读《社会学研究》！您是该文第2335位读者！