Students Who Choose to Challenge Themselves Perform Better

The motivational benefits of goal-setting and utility-value interventions are well established in digital learning platforms at scale. However, autonomy-supportive interventions grounded in self-determination theory remain relatively underexplored. This paper reports on three randomized controlled trials (N = 4,220) examining the impact of autonomy-supportive choices in mastery-based activities. Study 1 examines a pre-mastery decision: whether students' selected mastery threshold (3, 4, or 5 correct in a row) impacts performance on mastery-based activities and subsequent post-test. Studies 2 and 3 examine post-mastery decisions: how performance during mastery-based activities relates to students' choice of post-test difficulty (easy vs. challenging), and how that choice, in turn, predicts post-test performance. Across all three studies, control-condition students were randomly assigned to one of the options. In Study 1, we did not observe significant main effects of access to choice or specific mastery thresholds. However, students who chose higher thresholds of 4 and 5 performed better on the post-test (0.17 SD and 0.19 SD, respectively) than students randomly assigned the same thresholds. Notably, offering choice did not significantly impact student attrition or wheel-spinning on the mastery-based activity. In Studies 2 and 3, higher-performing students tended to choose the more challenging post-test compared to their lower-performing peers. Those who chose the challenging option achieved significantly greater post-test gains (0.20 SD, p < 0.05), even after controlling for prior performance and performance on the mastery-based activity. Overall, our findings suggest that students who voluntarily choose more challenging options outperform peers randomly assigned to the same options, even after accounting for prior performance and mastery efficiency. Importantly, these gains emerged from single-decision-point interventions, highlighting the practical feasibility of embedding learner choice into digital learning platforms at scale.