2024 ASEE Annual Conference & Exposition

Automated systems such as Marmoset, WebCAT, OK, MarkUs, and many others are widely used in assessing programming assignments. Although they enable instructors to assess students' solutions at scale, the core infrastructure of these systems is not much different from a standard build and test environment, which focuses on ensuring correct solutions. However, when it comes to learning, it would be more important to assist students in correcting their misconceptions when their solutions are incorrect, i.e., provide a feedback message accurately showing them what is wrong and what they can do. The latter, which requires high-quality assessment and considerable effort in composing feedback, however, is rarely discussed, not to mention that no tools or support have been developed in these systems to assist in writing them. In this paper, we aim to fill the gap by providing guidance for assessment writers to write effective assessments and feedback for students' solutions. We present an approach to properly organizing the test cases so that automated assessments can identify students' misconceptions accurately, enabling them to provide high-quality formative feedback to rectify students' misconceptions. Following the guidance outlined, we developed assessments for a database course. By comparing student performance with and without the high-quality formative feedback, we observed an overall improvement in RA of $21\%$, with a $73\%$ improvement in query creation and an $11\%$ improvement in ER, with a $32\%$ improvement in composing new relationship sets and/or specializations.