A Survey on Feedback Types in Automated Programming Assessment Systems
Eduard Frankford, Tobias Antensteiner, Michael Vierhauser, Clemens Sauerwein, Vivien Wallner, Iris Groher, Reinhold Plösch, Ruth Breu
TL;DR
The paper addresses how different automated feedback forms in APASs affect student learning in introductory programming. It employs a large-scale, two-university study to compare Compiler Feedback, Unit Test Feedback, and AI-generated Feedback, using TAM surveys and objective performance metrics collected via Artemis. Key findings show Unit Test Feedback is most trusted for perceived usefulness, while AI-generated feedback yields the highest final correctness and encourages deeper, more iterative problem solving, albeit with reliability concerns. The study highlights the value of a hybrid feedback approach and discusses design considerations to improve trust, scalability, and learning outcomes in AI-augmented programming education.
Abstract
With the recent rapid increase in digitization across all major industries, acquiring programming skills has increased the demand for introductory programming courses. This has further resulted in universities integrating programming courses into a wide range of curricula, including not only technical studies but also business and management fields of study. Consequently, additional resources are needed for teaching, grading, and tutoring students with diverse educational backgrounds and skills. As part of this, Automated Programming Assessment Systems (APASs) have emerged, providing scalable and high-quality assessment systems with efficient evaluation and instant feedback. Commonly, APASs heavily rely on predefined unit tests for generating feedback, often limiting the scope and level of detail of feedback that can be provided to students. With the rise of Large Language Models (LLMs) in recent years, new opportunities have emerged as these technologies can enhance feedback quality and personalization. To investigate how different feedback mechanisms in APASs are perceived by students, and how effective they are in supporting problem-solving, we have conducted a large-scale study with over 200 students from two different universities. Specifically, we compare baseline Compiler Feedback, standard Unit Test Feedback, and advanced LLM-based Feedback regarding perceived quality and impact on student performance. Results indicate that while students rate unit test feedback as the most helpful, AI-generated feedback leads to significantly better performances. These findings suggest combining unit tests and AI-driven guidance to optimize automated feedback mechanisms and improve learning outcomes in programming education.