A Decision Support System to Aid Detection and Substantiation of Contract Cheating in Higher Education

Contract cheating involves outsourcing an assignment for partial or entire completion to an unacknowledged third-party author and is a pernicious form of academic misconduct. It not only devalues the formal qualifications provided by higher education institutions, but has profound consequences for public health and safety when unqualified students enter the labour market. Thus, accurate detection is imperative to mitigate instances of contract cheating. However, detecting outsourced assignments is onerous and demanding, leading to the development of software tools and methodologies to aid the detection of contract cheating behaviour (e.g., authorship investigation tools, keystroke dynamics, and stylometry). Yet, it is unfeasible to rely solely on these tools for the detection of contract cheating, as they are contingent on human judgement and academic expertise to function accurately. Hence, this research proposes an intelligent decision support system that corroborates evidence from different sources to improve the efficacy of detecting, reporting, and substantiating contract cheating. A pilot study was conducted at a UK university, where the process of detecting and substantiating contract cheating involves several stages. Initially, a marker observes cues of contract cheating and invites the student for a viva-voce. If suspicion persists, the marker is required to complete a standard referral form and share this with the relevant administration team. The administration team later shares this form with an academic integrity officer who is tasked with reviewing the submission and determining the likelihood of academic malpractice. Finally, the case is referred to a panel for further investigation and a final decision. This process is laborious and impractical, particularly following the COVID-19 pandemic’s onslaught of high-volume cases and limited resources which may deter staff from pursuing cases. The current process is also contingent on the accuracy of the recorded evidence and decisions by the appropriate staff. Thus, the proposed rule-based expert system supports the decision-making at all levels of the referral process. The system generates an academic integrity score for each case and indicates whether further action is necessary, and shortlists suspected students for viva-voces. The integrity score is calculated by the combination of evidence (e.g., irregularities in the assignment or inappropriate references or methodologies) and learning analytics (e.g., their engagement levels, grade history, and the Turnitin plagiarism software) and is illustrated on a dashboard. The system also assists the autocompletion of the referral forms based on the entered evidence. The algorithm was tested on a sample of graded assignments from previous years, and a preliminary evaluation of the design was conducted by a lecturer and academic integrity officer. Feedback from the academic staff encourages the implementation of the system at the university, noting its success. A large-scale study involving a range of disciplines and assessment methods is recommended to establish a wider scope of efficacy for the decision support system proposed in this paper.