Related Experiment Videos
Collusion detection in multiple choice examinations
A Ercole1, K D Whittlestone, D G Melvin
1Clinical and Biomedical Computing Unit, Clinical School, Addenbrooke's Hospital, Cambridge, UK.
Medical Education
|March 1, 2002
Summary
A new statistical method can detect cheating in multiple choice exams by analyzing answer patterns between students. This approach uses answer correlations and timing data for improved collusion detection in online testing environments.
Area of Science:
- Educational Assessment
- Data Science in Education
- Academic Integrity
Background:
- Collusion in multiple choice examinations poses a significant threat to academic integrity.
- Traditional methods for detecting collusion can be resource-intensive and may not always be effective, especially in digital environments.
Purpose of the Study:
- To develop and validate a novel statistical method for detecting collusion in computer-based multiple choice examinations.
- To assess the efficacy of this method using real-world data from medical examinations.
Main Methods:
- Analysis of answer data from two negatively marked medical prize examinations administered electronically.
- Comparison of answer correlations between pairs of students, adjusting for question difficulty, student ability, and risk aversion.
- Integration of web-server data, including answer timing and computer location, as corroborative evidence.
Main Results:
- Significant answer correlations were identified between candidates who admitted to or were linked to evidence of collusion.
- The statistical method successfully highlighted suspicious answer patterns indicative of collusion.
Conclusions:
- A statistical approach analyzing answer patterns between candidates is effective for detecting collusion in multiple choice examinations.
- Leveraging online examination data, such as answer timing and location, provides valuable corroborative evidence for collusion detection.