Abstract: Multidimensional constructs are frequently applied to the studies of psychology, education, management and organizational behavior. The conventional measurement methods, including the total score approach, the individual score approach and the higher-order factor model, can’t solve the bandwidth-fidelity dilemma in multidimensional testing. A better solution is to build a bi-factor model for the multidimensional construct. The bi-factor model is potentially applicable when (a) there is a general factor that is hypothesized to account for the commonality of all items; (b) there are multiple domain specific factors, each of which is hypothesized to account for the unique influence of the specific domain. The relationship between the bi-factor model and the higher-order factor model is discussed from the perspective of the concepts, mathematical models, parameters, and practical applications. The applications of the bi-factor model are demonstrated or summarized, including in the reliability study, the method effects of balance scales, exploratory factor analysis and item response theory. An example of bi-factor model is illustrated to explore the structure of Rosenberg’s self-esteem scale and show the existence of effects associated with positive and negative wording besides of the self-esteem trait.

Key words: multidimensional construct, bi-factor model, general factor, specific factor, higher-order factor model, factor analysis

摘要： 双因子模型是一种既有全局因子又有局部因子的模型，近年来有了许多应用。本文讨论了双因子模型和高阶因子模型在数学模型、参数之间的关系，概念上和应用上的差异；概述了双因子模型在信度研究、平衡量表、探索性因子分析和项目反应理论中的应用。作为例子，在Rosenberg自尊量表结构的研究中，通过双因子模型分析了自尊特质效应与项目表述方法效应。

关键词: 多维构念, 双因子模型, 全局因子, 局部因子, 高阶因子模型, 因子分析