认知诊断计算机化自适应测验的选题策略

随着认知诊断计算机化自适应测验(cognitive diagnostic computerized adaptive testing, CD-CAT)理论与实践的发展, 兼顾知识状态与能力的双目标CD-CAT逐渐受到重视。选题策略是CAT的核心, 通过梳理传统CD-CAT和双目标CD-CAT选题策略的研究, 并对它们的特点、关系及表现进行介绍和评析。最后, 基于认知诊断模型与CAT实践发展指出未来应加强一般化认知模型、复杂测验条件认知诊断模型下选题策略的研究; 应开发双目标诊断测验的项目和测验特征指标; 还应加强非参数选题方法和CD-CAT的实践应用研究。

Item selection methods for cognitive diagnostic computerized adaptive testing

Dual-objective cognitive diagnostic computerized adaptive testing (CD-CAT), which considers knowledge status and ability simultaneously, has become more and more popular with the theoretical and practical development of CD-CAT. Item selection methods play a key role in CD-CAT. This paper systematically reviews existing item selection methods on traditional and dual-objective CD-CAT, and summarizes the types, characteristics, relations, and performance of these methods. Furthermore, several future research directions were illustrated. First, it is necessary to study item selection strategy with general cognitive models and under complex test conditions. Second, it is important to develop indexes representing items and test characteristic of dual-objective diagnostic testing. Finally, it is meaningful to conduct research on non-parametric item selection methods and practical applications of CD-CAT.