Determine whether early predictions can support timely instruction

Determine whether predictions generated from the first submission-based segment of a programming exercise can be translated into effective instructional support, such as targeted hints, instructor check-ins, or prioritization of students for assistance.

Background

The study finds that the earliest submission-based segment contains useful but imperfect predictive information about whether students will eventually remain stuck. The conclusion identifies an unresolved deployment question: whether these early predictions can be used during the same early stage to provide effective support. Potential interventions include targeted hints, instructor check-ins, and prioritization of students for assistance.