University of Wisconsin, Madison, WI, USA.
Psychometrika. 2018 Jun;83(2):453-475. doi: 10.1007/s11336-017-9586-5. Epub 2017 Sep 25.
While item complexity is often considered as an item feature in test development, it is much less frequently attended to in the psychometric modeling of test items. Prior work suggests that item complexity may manifest through asymmetry in item characteristics curves (ICCs; Samejima in Psychometrika 65:319-335, 2000). In the current paper, we study the potential for asymmetric IRT models to inform empirically about underlying item complexity, and thus the potential value of asymmetric models as tools for item validation. Both simulation and real data studies are presented. Some psychometric consequences of ignoring asymmetry, as well as potential strategies for more effective estimation of asymmetry, are considered in discussion.
虽然项目复杂性通常被认为是测试开发中的一个项目特征，但在测试项目的心理测量建模中，它很少受到关注。先前的工作表明，项目复杂性可能通过项目特征曲线（ICC；Samejima 在 Psychometrika 65:319-335，2000 年）的不对称表现出来。在本文中，我们研究了不对称的IRT 模型在实证上提供潜在项目复杂性信息的可能性，因此，不对称模型作为项目验证工具的潜在价值。同时提出了模拟和真实数据研究。在讨论中考虑了忽略不对称的一些心理测量后果，以及更有效地估计不对称的潜在策略。
