This paper proposes innovations to parameter estimation in a generalised logistic regression model in the context of detecting differential item functioning in multi-item measurements. The two newly proposed iterative algorithms are compared with existing methods in a simulation study, and their use is demonstrated in a real data example. Additionally the study examines software implementation including specification of initial values for iterative algorithms, and asymptotic properties with estimation of standard errors. Overall, the proposed methods gave comparable results to existing ones and were superior in some scenarios.

翻译：本文针对多项目测量中检测项目功能差异（DIF）的广义逻辑斯蒂回归模型，提出参数估计的创新方法。通过模拟研究比较了两种新提出的迭代算法与现有方法的性能，并在实际数据示例中展示了其应用。此外，研究探讨了软件实现问题，包括迭代算法初始值的设定，以及标准误差估计的渐近性质。总体而言，所提方法在结果上与现有方法具有可比性，并在某些情境下表现更优。