Modeling learner heterogeneity: A mixture learning model with responses and response times

Abstract

The increased popularity of computer-based testing has enabled researchers to collect various types of process data, including test takers' reaction time to assessment items, also known as response times. In recent studies, the relationship between speed and accuracy in a learning setting was explored to understand students' fluency changes over time in applying the mastered skills in addition to skill mastery. This can be achieved by modeling the changes in response accuracy and response times throughout the learning process. We propose a mixture learning model that utilizes the response times and response accuracy. Such a model accounts for the heterogeneities in learning styles among learners and may provide instructors with valuable information, which can be used to design individualized instructions. A Bayesian modeling framework is developed for parameter estimation and the proposed model is evaluated through a simulation study and is fitted to a real data set collected from a computer-based learning system for spatial rotation skills.