George L K, Siegler I C, Okun M A
Exp Aging Res. 1981 Fall;7(3):297-314. doi: 10.1080/03610738108259812.
This paper empirically compares the relative advantages of analysis of variance (ANOVA) and multiple regression (MR) approaches to the separation of age, cohort, and time of measurement effects in sequential research designs. The comparison utilizes four synthetic data sets, designed to have specific characteristics. The results support Adam's recent claim that standard ANOVA procedures, as described by Schaie, do not permit the development of accurate decision rules for age-period-cohort analysis. A modified dummy variable regression procedure developed by Mason, et al. is demonstrated to permit accurate attribution of variance among age, cohort, and time of measurement effects in the developmental model.
本文实证比较了方差分析（ANOVA）和多元回归（MR）方法在序列研究设计中分离年龄、队列和测量时间效应方面的相对优势。该比较使用了四个具有特定特征的合成数据集。结果支持了亚当最近的观点，即正如沙伊所描述的标准方差分析程序，不允许为年龄-时期-队列分析制定准确的决策规则。梅森等人开发的一种改进的虚拟变量回归程序被证明能够在发展模型中准确地对年龄、队列和测量时间效应之间的方差进行归因。
