Research Summary: Consistently Good vs. Occasionally Great: A Rubric for Open-Ended Feedback Quality from Humans and Machines

A new research paper introduces a five-criterion rubric for evaluating the quality of open-ended feedback, specifically for short answer questions in introductory programming. The rubric aims to guide students towards reattempt success without revealing correct answers, focusing on acknowledging correct portions, identifying flaws, providing actionable guidance, maintaining answer concealment, and using an appropriate tone. This framework is designed to address challenges in delivering high-quality feedback at scale.

Why it matters

This research provides a structured framework for assessing feedback quality, which is critical for educational efficacy and the scalable implementation of learning systems. Understanding and applying such rubrics can enhance learning outcomes across various domains where open-ended responses are common, enabling more effective and consistent educational support.

Key insights

• High-quality feedback is crucial for learning, but its scalable delivery is challenging.
• A five-criterion rubric has been developed to assess open-ended feedback quality.
• The rubric criteria include acknowledging correct parts, identifying flaws, providing actionable guidance, concealing answers, and maintaining appropriate tone.
• The research focuses on feedback for open-ended short answer questions in introductory programming.
• The goal of the feedback is to nudge students towards success on reattempts.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.21850