Game-based assessments (GBAs) do carry a lower faking vulnerability than traditional personality questionnaires, but they are not immune. A peer-reviewed experiment published in the Journal of Business and Psychology (April 2025, doi:10.1007/s10869-025-10019-6) measured faking effects across both formats and found effect sizes of d=0.95 for GBAs versus d=1.62 for traditional questionnaires. Candidates could manipulate both formats; they just found it harder with game-based tasks.

This guide is for recruiters and talent acquisition leads who want to act on that finding. By the end, you will have a practical checklist for managing residual faking and cheating risk across the full assessment lifecycle.

Before starting, it helps to distinguish two separate risks. "Faking" means intentional self-presentation distortion on self-report items. "Cheating" means using external help, AI tools, or secondary devices to influence responses. Both matter, and they require slightly different controls.

What the evidence actually says about format differences

Traditional personality questionnaires are transparent by design. A candidate reading "I work hard even under pressure" can easily infer the desired answer and adjust. GBAs introduce time pressure, interactive decisions, and indirect trait measurement, which constrain that kind of deliberate response editing.

That said, the same 2025 study noted that virtual cues within GBAs produced larger faking effects than elements like economic games or situational judgment tasks (SJTs). Format alone is not a complete safeguard. The specific game mechanics matter.

A practical anti-cheating checklist: before, during, and after

Before the assessment

1. Send a clear candidate communication that states what is permitted, what is not, and what happens if the session is interrupted. Include a refresh/resume policy.
2. Use one-time access links and secure launch flows to prevent link sharing.
3. Where the platform supports it, randomize item order or use procedurally generated variants to limit the value of shared answers.
4. Select assessment formats that resist transparent responding. SJT-style items and economic game tasks generally show stronger faking resistance than virtual-cue or avatar-based items.

During the assessment

5. Apply consistent timing rules and log completion pacing. Unusually fast or unusually slow completion times are both worth flagging.
6. Collect process telemetry, meaning interaction latency, click or touch patterns, and pacing irregularities. These behavioral signals can surface anomalies that raw scores alone would miss.
7. Embed instruction manipulation checks (IMCs) to confirm the candidate is reading and responding genuinely. Design these so they do not penalize candidates facing device or network issues.

After the assessment

8. Set a threshold for escalation. A single anomalous signal is rarely enough. Look for combinations, such as extreme completion time plus atypical interaction patterns plus failed instruction checks.
9. Document every escalation decision, including what triggered it and how it was resolved. This supports fairness reviews and compliance with GDPR automated decision-making requirements.
10. Feed findings back into test design. Repeated exploitation patterns signal a need to update item variants or adjust scoring logic.

Recognizing cheating signals by format

For questionnaires, watch for suspiciously uniform response times (suggesting scripted input) and socially desirable response patterns across all items. For GBAs, signals include unnatural interaction precision, pacing that is inconsistent with genuine gameplay, and repeated failure on embedded instruction checks.

None of these signals are proof on their own. Use risk scoring to prioritize review, then apply human judgment before any adverse decision is made.

When to escalate to proctoring or manual review

Use a tiered approach.

1. Automated flags trigger a review queue, not an immediate disqualification.
2. A trained reviewer checks whether the behavior could be explained by device limitations, accessibility needs, or candidate unfamiliarity with the format.
3. For high-stakes roles, consent-based proctoring (camera, audio, and screen recording) can be activated for borderline cases. Selection Lab's anti-fraud layer, for example, records candidate camera, audio, and screen with explicit consent to verify that answers are the candidate's own. Candidates are also noticeably less likely to attempt fraud once they know recording is active.

All personal data handling should comply with GDPR and, where applicable, align with EU AI Act transparency requirements. Selection Lab stores all personal data in Frankfurt and applies consent-based retention periods per processing purpose.

Candidate communication templates

Pre-assessment message. "Please complete this assessment in a quiet location without assistance from other people or AI tools. If your session is interrupted, use the link in your invitation email to resume. Do not share your access link."

During-assessment reminder. "Follow all on-screen instructions carefully. Some tasks include checks to confirm you are reading the prompts. Complete every step as directed."

If flagged (transparency notice). "Your assessment results have been referred for additional review. This does not mean you have been disqualified. A member of our recruitment team will contact you within [X] business days. If you believe this review is in error, you may contact [contact address] to request a manual review."

Data and privacy statement. "Where proctoring is used, your camera, audio, and screen activity will be recorded with your consent during the assessment session. Recordings are used solely to verify assessment integrity and are retained for [X] days in line with our data retention policy."

What this means for your recruitment process

GBAs reduce applicant faking risk compared to traditional personality questionnaires, but no format eliminates it. The practical advantage shifts when you layer structural controls (access management, randomization, telemetry) with human review and compliant escalation policies. Platforms that automate these controls inside recruiter workflows, with ATS integration for documentation, make the layered approach operationally realistic rather than aspirational. Want to see how Selection Lab's anti-fraud layer fits your process? Book a demo.

Frequently asked questions