Students’ experiences with automated final answer diagnoses for mathematics tasks

Abstract

Model Backtracking (MBT) is a novel technique for automated detailed diagnoses based on final answers. In this small-scale pilot study, we answer research questions about nine 15-to-17-year-old senior general secondary students’ experiences with MBT diagnoses. The students practiced linear extrapolation in a learning environment that provides error-specific feedback and selects appropriate subtasks using MBT. Data included screen captures of students navigating the environment and interviews on students’ experiences with the environment. Results showed approaches ranging from correcting an error after receiving feedback to trial-and-error behavior while repeatedly consulting the worked-out solution. Furthermore, students preferred error-specific feedback over worked-out solutions. They found that worked-out solutions provide an insightful overview; yet, errors are not pinpointed, and worked-out solutions reduced motivation for further practice.