Avoiding the “Fallacy of the Wrong Level”
指出在组织数据中，简单回归分析无法判断分析层次，可能导致个体与群体效应混淆。文章介绍了WABA方法，帮助研究者检验数据中的多层次效应，适用于结构方程模型、多层线性模型等分析。
This article illustrates that any bivariate or multiple regression provides inadequate information about the levels of analysis in a data set collected in an organizational setting. As a result, individual-level effects may be incorrectly attributed to the group level, and group-level effects may be incorrectly viewed as being solely individual-level effects. Both of these situations are examples of the “fallacy of the wrong level.” Within and between analysis (WABA) allows levels of analysis to be tested in data. These WABA tests are useful in numerous analytical approaches, including structural equation modeling, hierarchical linear modeling, and various approaches to aggregation. This article provides a decision tree for use in performing tests for multiple alternative levels of analysis in a data set collected in organizations.