Huang Sijia, Luo Jinwen, Jeon Minjeong
School of Education, Indiana University Bloomington, 201 N. Rose Ave, Bloomington, 47405, IN, USA.
University of California, Los Angeles, Los Angeles, CA, USA.
Behav Res Methods. 2025 Jan 9;57(1):54. doi: 10.3758/s13428-024-02555-5.
Educational researchers have a long-lasting interest in the strategies examinees employ when responding to items in an assessment. Mixture item response theory (IRT) modeling is a popular class of approaches to studying examinees' item-response strategies. In the present study, we introduce a response time (RT)-based mixture IRT model for flexible modeling of examinee-and-item-specific item-response strategies. We posit that examinees may alternate between ability-based and non-ability-based strategies across different test items. Our proposed model identifies such within-examinee strategy switches without the need to predefine the non-ability-based strategies. Instead, our proposed approach allows for inferring the nature of these strategies from model parameter estimates. We illustrated the proposed approach using empirical data from PISA 2018 Science test and evaluated it through simulation studies. We concluded the article with discussions of limitations and future research directions.
教育研究者长期以来一直对考生在评估中回答题目时所采用的策略感兴趣。混合项目反应理论（IRT）建模是研究考生项目反应策略的一类流行方法。在本研究中，我们引入了一种基于反应时间（RT）的混合IRT模型，用于灵活地对考生和题目特定的项目反应策略进行建模。我们假定考生可能在不同的测试题目中在基于能力的策略和非基于能力的策略之间交替使用。我们提出的模型无需预先定义非基于能力的策略就能识别考生内部的这种策略转换。相反，我们提出的方法允许从模型参数估计中推断这些策略的性质。我们使用2018年国际学生评估项目（PISA）科学测试的实证数据说明了所提出的方法，并通过模拟研究对其进行了评估。我们在文章结尾讨论了局限性和未来的研究方向。
