Online Proctoring Based on Item Responses and Response Times

Abstract

The surge of online testing in recent years calls for procedures that monitor examinees’ test-taking behaviors in real time. This study presents online proctoring methods that examine response and response-time behaviors during testing. The procedures are developed using sequential quality control techniques and implemented with various sampling designs to improve time sensitivity. Experimentation with simulated data suggests that the proposed methods can provide a powerful online surveillance tool for regulating anomalous behaviors. The procedures showed high sensitivity to aberrances and gave timely signals while maintaining the error rates reasonably small. We demonstrate utility of the procedures using real-assessment data and discuss implications for practical use.

Online Monitoring of Test‐Taking Behavior Based on Item Responses and Response Times

“…For estimating the item parameters, the study applied maximum marginal likelihood estimation (Kang et al., 2020). During monitoring, examinee's trait parameters were estimated via maximum a posteriori (Kang, 2016) to draw joint inference under the hierarchical framework.…”