Assessment Integrity: Research on Cheating and Verification
Last reviewed
Summary
Online, unproctored assessments are convenient but open to cheating and impersonation, and self-report measures are open to response distortion. Verification and monitoring help, but automated checks, especially face-based ones, have documented error patterns that require human review.
What the evidence says
- A review of unproctored internet testing for employment set out its practical advantages and its risks, including cheating and uncertainty about the test-taker's identity, and discussed verification strategies. [1]
- Applicants tend to describe themselves more favorably than non-applicants on personality questionnaires, a form of response distortion that assessment design has to account for. [2]
- NIST's evaluation of face recognition algorithms found demographic differentials in error rates for many algorithms, with large variation between algorithms. [3]
- Research on commercial facial-analysis systems found higher error rates for some demographic groups than for others. [4]
AuraSync's interpretation
Integrity checks are necessary for trustworthy online assessment, but their errors do not fall evenly. We treat every integrity flag as a signal for a person to review, never an automatic finding of misconduct, and design the assessment around observed responses so there is less to game.
What AuraSync claims
- AuraSync verifies the candidate's identity and monitors the session, raising integrity flags for a person to review.
- A flag is a signal, not a finding of misconduct, and the candidate is entitled to explain what happened.
- Facial verification data is not shared with the hiring organization and is deleted within 90 days of capture.
Limitations
The face-recognition research above evaluated many algorithms, not AuraSync's. Ordinary events, such as poor lighting or someone walking behind the candidate, can raise a flag.
Sources
- [1]Tippins, N. T., Beaty, J., Drasgow, F., Gibson, W. M., Pearlman, K., Segall, D. O., & Shepherd, W. (2006). Unproctored internet testing in employment settings. Personnel Psychology, 59(1), 189–225. https://doi.org/10.1111/j.1744-6570.2006.00909.x
- [2]Birkeland, S. A., Manson, T. M., Kisamore, J. L., Brannick, M. T., & Smith, M. A. (2006). A meta-analytic investigation of job applicant faking on personality measures. International Journal of Selection and Assessment, 14(4), 317–335. https://doi.org/10.1111/j.1468-2389.2006.00354.x
- [3]Grother, P., Ngan, M., & Hanaoka, K. (2019). Face Recognition Vendor Test (FRVT) Part 3: Demographic Effects (NISTIR 8280). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.IR.8280
- [4]Buolamwini, J., & Gebru, T. (2018). Gender shades: Intersectional accuracy disparities in commercial gender classification. Proceedings of Machine Learning Research, 81, 77–91. https://proceedings.mlr.press/v81/buolamwini18a.html
This page summarizes third-party research for orientation. It is not an evaluation of AuraSync, and any summary of law is not legal advice.