Item response theory (IRT) is a widely used approach to measure test takers' latent traits from assessment data. IRT models typically require conditional independence and homogeneity assumptions that may be violated in practice when unobserved interactions between respondents and items exist. To alleviate these assumptions, Jeon et al. (2021) introduced a latent space item response model (LSIRM) for binary item response data by assuming that items and respondents are embedded in an unobserved metric space, called an interaction map, where the probability of a correct response decreasing as a function of the distance between the respondent's and the item's position in the latent space. The R package lsirm12pl implements the original and extended versions of LSIRM, while supplying several flexible modeling options: (1) an extension to the two-parameter model; (2) an extension to continuous item responses; and (3) handling missing responses. The R package lsirm12pl also offers convenient functions for visualizing estimated results, including the interaction map that represents item and person positions and their interactions. In this paper, we provide a brief overview of the methodological basis of LSIRM and describe the extensions that are considered in the package. We then showcase the use of the package lsirm12pl with real data examples that are contained in the package.
翻译:为减轻这些假设,Jeon等人(2021年)为二进制物品反应数据引入了潜伏空间物品反应模型(LSIRM),假设物品和应答者嵌入一个未观测的计量空间,称为互动图,其中反映的项目和应答者正确反应的概率因受访者与该物品在潜在空间位置之间的距离而降低。在本文中,我们简要概述了LSIRM的方法学基础,并介绍了该软件包的扩展情况,同时提供了若干灵活的模型选项:(1) 扩展了两个参数模型;(2) 扩展了连续项目反应;(3) 处理缺失的响应。R包Isirm12pl还提供了可视化估计结果的方便功能,包括反映项目和人的位置及其相互作用的互动图。