Academic Journal
Detecting Aberrant Test-Taking Behaviors in Computer-Based Testing Using One-Dimensional Convolutional Neural Networks
| Τίτλος: | Detecting Aberrant Test-Taking Behaviors in Computer-Based Testing Using One-Dimensional Convolutional Neural Networks |
|---|---|
| Γλώσσα: | English |
| Συγγραφείς: | Chunyu Piao, Jiwei Zhang (ORCID |
| Πηγή: | Educational Measurement: Issues and Practice. 2025 44(4):5-17. |
| Διαθεσιμότητα: | Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us |
| Peer Reviewed: | Y |
| Page Count: | 13 |
| Ημερομηνία έκδοσης: | 2025 |
| Τύπος εγγράφου: | Journal Articles Reports - Research |
| Descriptors: | Student Behavior, Test Wiseness, Computer Assisted Testing, Item Response Theory, Reaction Time, Prior Learning, Artificial Intelligence, Models, Accuracy, Identification |
| DOI: | 10.1111/emip.70010 |
| ISSN: | 0731-1745 1745-3992 |
| Περίληψη: | In educational assessments, detecting aberrant test-taking behaviors is crucial for ensuring test validity and reliability. We propose a deep learning method based on a one-dimensional convolutional neural network (1D-CNN) combined with a sliding window technique, and detect aberrant behaviors by analyzing responses and response times. To validate our method, we conduct simulations with two typical aberrant scenarios: test speededness and item preknowledge. We compare our method with several established methods, including extreme gradient boosting, logistic regression, decision trees, and multilayer perceptrons, as well as traditional educational methods such as person-fit and cumulative sum (CUSUM) statistics. Results show that the 1D-CNN model outperforms traditional methods in accuracy, sensitivity, specificity, and the area under curve (AUC), demonstrating superior capability in detecting complex aberrant patterns. Furthermore, real data validation confirms the practical effectiveness and applicability of our method. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Αριθμός Καταχώρησης: | EJ1493244 |
| Βάση Δεδομένων: | ERIC |
| ISSN: | 0731-1745 1745-3992 |
|---|---|
| DOI: | 10.1111/emip.70010 |