From Clicks to Classifications: Examining Effects and Addressing Rapid Guessing in Longitudinal DCMs
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
Date of Graduation
8-1-2025
Semester of Graduation
Summer
Degree Name
Doctor of Philosophy (PhD)
Department
Department of Graduate Psychology
First Advisor
Yu Bao
Abstract
Rapid guessing, a disengaged test-taking behavior commonly observed in low-stakes assessments, poses a significant threat to the validity of inferences. While prior research has examined rapid guessing within IRT frameworks cross-sectionally and evaluated the use of effort-moderated approaches to mitigate its effects, limited attention has been given to its impact on multidimensional, longitudinal, and diagnostic models. This simulation study investigates how longitudinal rapid guessing factors affect sample inferences and individual classifications from the Transition Diagnostic Classification Model (TDCM). It also evaluates the ability of the effort-moderated approach to mitigate these effects. Results indicate that failing to account for rapid guessing leads to underestimation of learning and learning growth, with the magnitude of bias sensitive to true attribute growth—which is unknown in practice. The effort-moderated approach effectively mitigated bias under missing completely at random (MCAR) conditions but tended to overestimate learning and growth under missing at random (MAR) conditions. Its effectiveness was influenced by item quality and the degree of missingness but was not sensitive to true attribute growth.
