多维测验分数的报告与解释:基于双因子模型的视角
Reporting and Interpreting Multidimensional Test Scores:A Bi-factor Perspective

在心理、行为、管理和教育等社科领域，经常使用多维测验。本文评介并比较了各种多维测验的测量模型；总结了基于双因子模型计算得到的统计指标；根据不同研究目的，提出了两个兼顾简洁性和精确性的多维测验分析流程。作为例子，在马基雅维利主义人格量表的研究中，通过双因子模型分析了如何报告、解释多维测验分数以及如何利用多维建模进行后续分析。
Empirical psychological, educational and social scientists are often confronted with multifaceted tests which are comprised of several related dimensions or subtests. This study is organized as follows:Firstly, the measurement models of multidimensional scales were reviewed and evaluated. Secondly, some statistical indices (i.e., ECV, PUC and four types of reliability coefficients) based on bi-factor modeling were calculated and interpreted. Thirdly, based on different research purposes, we summarized and proposed two analysis processes about multidimensional measurement, and both of them took the simplicity and accuracy into account simultaneously. Finally, as an example, the bi-factor model of the Machiavellianism Personality Scale (MPS) was analyzed, and the following questions were clarified for applied researchers:Do total scores generally reflect variation on a single latent variable? Is it justified to report the subscale scores? After partitioning out variability attributed to the total score, could subscales still provide reliable information? Could a complex multidimensional measurement be modeled using a simple unidimensional CFA?