Grading software engineering group projects based on individual contributions

Abstract

Universities apply multiple grading strategies for group assignments, such as the same grade for all group members, individual grades determined by the teacher, and individ-ual grades determined by the team. The problem with these strategies is that they require extra work to evaluate contributions or do not consider individual contributions. In soft-ware engineering projects, instructors can view individual contributions through the data that is often available in software engineering group projects by analyzing the Git data. To this end, we have developed and evaluated a new strategy that utilizes the available data and takes into account each individual’s contribution by analyzing the Git repository. The proposed method frames the grade division between team members as a fair division prob-lem, which enables drawing solutions from this well-established research topic. More than 30 students following a software development course used a newly developed web-based application that assigned grades based on individual contributions. This application col-lected student preferences and configuration and analyzed the group Git repositories using multiple metrics to calculate the individual contributions. The metrics used are code com-plexity added, lines of code, lines of documentation, and the amount of commits. The individual contributions were then used to calculate a grade by applying fair division prin-ciples. After using the application, the results were shown to the students, and the students were invited to fill in a questionnaire on the tool and metrics. Students valued the metric code complexity the highest and valued the amount of commits the lowest of the available metrics in the tool. The students rated the application slightly higher than acceptable, and the tool was considered easy to use and an acceptable alternative to the same grade for all group members strategy. However, students have also expressed concerns about the tool as it did not reflect entirely what everyone was doing during the project, and pair program-ming was not taken into account.

Date of Award	1 Apr 2022
Original language	English
Supervisor	Efthimia Aivaloglou (Examiner) & Alaaeddin Swidan (Co-assessor)