Tracking Choices: Computational Analysis of Learning Trajectories

This study investigates how variations in students’ trajectories within the tutoring system, Writing Pal, varied as a function of individual differences and ultimately related to changes in the quality and linguistic properties of prompt-based essays. Fortytwo college students interacted freely with the computerized writing tutor for approximately 4 hours. Using a novel statistical technique (random walks), we visualized students’ self-paced trajectories within the Writing Pal system interface. Analyses revealed that students’ self-reported perseverance was predictive of more systematic interaction patterns. Students’ interaction patterns did not directly influence the quality of their writing; however, students’ trajectories within the system was related to changes in the fine-grained linguistic properties of their essays. These findings demonstrate the potential for random walks to provide researchers with a wealth of information about students’ interactions and subsequent learning outcomes within adaptive learning environments.