Demographic-stratified evaluation of educational disparities

Determine whether course-specific Retrieval-Augmented Generation systems measurably reduce help-seeking and educational disparities for different learner groups, including students affected by anxiety, low confidence, gender, widening-participation status, mature-student status, disability, or neurodivergence.

Background

The study proposes that Beacon’s private, low-friction, course-aligned support may reduce anxiety- and confidence-related barriers to academic help-seeking. However, the evaluation did not collect demographic data linking system use or perceived benefit to specific learner characteristics. The authors therefore identify whether the proposed mechanism actually narrows educational disparities, and for whom it is most effective, as an unresolved empirical question requiring demographic-stratified evaluation.