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 meas- urement models of multidimensional scales were reviewed and evaluated. Secondly, some statistical indices （i. e. , ECV, PUC and four types of reliability coefficients） based on hi-factor modeling were calculated and interpreted. Thirdly, based on different research purposes, we summarized and proposed two analysis processes about multidi- mensional measurement, and both of them took the simplicity and accuracy into account simultaneously. Finally, as an example, the hi-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 varia- ble？ 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？