关键词: 反应风格, 项目反应树, 李克特量表, 项目反应理论

Abstract: Item response tree (IRTree) models offer a refined framework for disentangling and understanding response styles by decomposing Likert-type item responses into multiple decision nodes. The ongoing evolution of the IRTree framework in recent years has greatly advanced response style research. At the node level, the IRTree model has been extended from within-node unidimensional structures to within-node multidimensional structures, and from assuming dominance processes at all nodes to incorporating ideal point processes, thereby providing a richer representation of the theoretical mechanisms underlying item responding. At the response process level, a growing body of research has explored heterogeneity in the response process, including mixture modeling across respondents, modeling of within-response pathway uncertainty, and intraindividual dynamic modeling, which enables increasingly nuanced investigations. Building on a review of these key advances, this article further summarizes a stepwise analytical workflow for IRTree modeling, and provides supporting R functions along with an empirical demonstration. Finally, this article discusses two perspectives on response style modeling—controlling for measurement bias and gaining insight into the response process—and outlines promising directions for future development of IRTree models.

Key words: response styles, item response tree, Likert scale, item response theory