Tracing students’ practice behavior in an adaptive math learning program Does it mediate the math anxiety–performance link?

Background: It has been suggested that math anxiety negatively affects students’ performance because math anxious students avoid math-related tasks. As most studies that investigated this assumed relationship focused on self-report measures to assess avoidance behavior, approaches that use objective process measures, which are less likely to be affected by biases, are needed. 

Aims: We conceptualized avoidance behavior in terms of trace data within an adaptive arithmetic learning program. We used a three-wave longitudinal study design to approach the question of how math anxiety is associated with learning behavior that contributes to math performance improvement in school. 

Sample: The study comprised 890 fifth-grade students. 

Methods: We traced students’ practice (i.e., total number of tasks completed and average number of tasks completed before quitting a session within the learning program) over 45 weeks. Math anxiety was conceptualized multidimensionally. 

Results: The mediation analyses showed that math anxious students did not complete fewer tasks overall, but completing more tasks was linked to performance improvement. However, math test anxious students, on average, completed fewer tasks before quitting a session, which explained their weaker performance. 

Conclusions: As only math test anxious students displayed higher levels of quitting, we conclude that the perceived context of the program (i.e., as a test) might play a role and that granularity seems to matter regarding the operationalization of process-measured avoidance behavior. Additionally, practitioners need to encourage students to practice persistently with adaptive math learning programs, as this contributes to students’ performance improvement.