Draxler Clemens, Kurz Andreas
UMIT TIROL - Private University for Health Sciences and Technology, Eduard-Wallnöfer-Zentrum 1, 6060, Hall in Tirol, Austria.
Paris Lodron University Salzburg, Kapitelgasse 4-6, 5020, Salzburg, Austria.
Behav Res Methods. 2025 Jan 8;57(1):50. doi: 10.3758/s13428-024-02551-9.
This article addresses the problem of measurement invariance in psychometrics. In particular, its focus is on the invariance assumption of item parameters in a class of models known as Rasch models. It suggests a mixed-effects or random intercept model for binary data together with a conditional likelihood approach of both estimating and testing the effects of multiple covariates simultaneously. The procedure can also be viewed as a multivariate multiple regression analysis which can be applied in longitudinal designs to investigate effects of covariates over time or different experimental conditions. This work also derives four statistical tests based on asymptotic theory and a parameter-free test suitable in small sample size scenarios. Finally, it outlines generalizations for categorical data in more than two categories. All procedures are illustrated on real-data examples from behavioral research and on a hypothetical data example related to clinical research in a longitudinal design.
本文探讨心理测量学中的测量不变性问题。具体而言，其重点在于一类称为拉施模型的模型中项目参数的不变性假设。它提出了一种针对二元数据的混合效应或随机截距模型，以及一种同时估计和检验多个协变量效应的条件似然方法。该过程也可被视为一种多元多重回归分析，可应用于纵向设计中，以研究协变量随时间或不同实验条件的影响。这项工作还基于渐近理论推导了四种统计检验，以及一种适用于小样本量情况的无参数检验。最后，它概述了对两类以上分类数据的推广。所有程序均通过行为研究的实际数据示例以及纵向设计中与临床研究相关的假设数据示例进行说明。
