Every candidate who sits a personality test does a bit of quick math first. What are they looking for here? A sales role probably rewards confident and outgoing. A compliance job probably rewards careful and precise. So the honest question is not whether candidates try to shade their answers toward the "right" profile. Most do, at least a little. The real question is whether that shading actually works, and what you can do about it.
The short version: yes, psychometric tests can be gamed, but nowhere near as cleanly as most hiring teams fear. And the most effective fix has almost nothing to do with the test itself. It is about where you put the test in your process. Below is what the evidence says, and the practical moves that follow from it, starting with the one that matters most for fast-moving startup teams who cannot afford a bad read on a new hire.
Across the Equip platform, 2025 to H1 2026
What "gaming" a psychometric test actually means
Gaming a personality test is nothing like cheating on a coding challenge. There is no answer key to steal and no correct response to look up. In fact, psychometric tests on Equip carry no points and cannot be graded. They map behavioural tendencies, not right-or-wrong knowledge.
So gaming here means something narrower: socially desirable responding, or "faking good." A candidate reads a statement like "I stay calm under pressure," senses that agreeing looks better, and nudges the answer up whether or not it is true. Do that across enough items and the profile drifts toward whatever the candidate thinks you want to see. That is the mechanism. The useful part is how much it really distorts the result.
Can they be faked? What the research says
Decades of selection research land on a nuanced answer.
In the lab, faking is easy. In a well-known meta-analysis of 51 studies, Chockalingam Viswesvaran and Deniz Ones found that when people are told to fake good, they can shift their scores by roughly half a standard deviation, around d = 0.50 in between-subjects studies and closer to 0.70 within subjects. That is a real, measurable move. Their other finding is just as important: when people are instructed to fake, every one of the Big Five traits is roughly equally fakable.
In real hiring, faking is much milder. Instructed lab volunteers are not job applicants. When you study actual candidates, the effect shrinks. A meta-analysis by Robert Birkeland and colleagues found that real applicants do inflate their scores, but the lift concentrates on the traits everyone already knows are desirable: conscientiousness and emotional stability. Nobody games their way toward "disorganised and anxious." Traits with no obvious "good" direction barely move, because candidates cannot tell which way to lean.
Pressure is the trigger. This is the finding that should shape how you run your process. In a 2021 study looking only at real applicants, Ping Hu and Brian Connelly showed that the higher the stakes, the more socially desirable the answers became. Faking is not a fixed trait of the test. It is a response to how much the candidate feels is riding on the result.
And the signal survives. Even when some faking happens, the genuine personality read does not vanish. The underlying trait variance still predicts real outcomes. Faking adds noise. It rarely erases the message.
The lever competitors rarely talk about: stakes
Most guidance on faking points at the test. Use smarter question wording. Add proctoring. Randomise items. All fine, but all aimed at the wrong layer. The Hu and Connelly result tells you the real lever is the environment around the test. If pressure drives faking, then the way you frame and place the test decides how much faking you get before a single question is written.
That leads to the tip that does the heavy lifting.
Coding tests, aptitude, role-specific skills. Candidates are meant to try their hardest here, so the pressure is appropriate. You are measuring capability.
Run this as its own invitation, framed as insight rather than a gate. When candidates know it is not graded, they stop performing and start answering honestly.
Bundling personality items inside a scored skills test makes candidates treat the whole thing as an exam. Splitting the round lowers the perceived stakes, which is exactly what the research links to more honest, less inflated answers.
Pro tip: run psychometrics as a separate round
Do not bundle your DISC or OCEAN test inside the same scored assessment as your skills tests. Give it its own round.
The reason is psychological. When a personality test sits between a coding challenge and an aptitude quiz, candidates read the whole experience as one graded exam. They assume every answer is being counted, so they perform, and performing is exactly what corrupts a personality result. Pull the psychometric test out into a separate invitation and the framing changes. Candidates understand it is not scored, the pressure drops, and you get a truer read.
On Equip this lines up with how the product already behaves. Psychometric tests do not contribute to the overall assessment score. The DISC test (24 questions, about 12 minutes) and the OCEAN test (50 questions, about 25 minutes) exist to describe behaviour, not to rank it. So once you have shortlisted candidates on skills, invite that shortlist to a separate assessment for the personality round. The scored decisions are already behind you. The personality read comes with no strings attached, and candidates can simply be themselves.
Five more ways to keep answers honest
Separating the round is the foundation. These build on top of it.
1. Tell candidates there are no right answers
Say it plainly in the invitation. This test is not graded, there is no ideal profile, and honest answers lead to better role fit for them. Candidates who game their way into the wrong job are the ones most likely to leave inside a year. Framing honesty as self-interest works better than hoping nobody tries.
2. Favour formats that are harder to fake
Forced-choice questions, where candidates pick between two equally appealing statements instead of rating each one, hold up better against faking. There is no obviously "better" box to tick, so socially desirable responding has nowhere to go. Where a test offers that format, use it.
3. Never gate a hire on personality alone
Treat psychometric results as one directional signal among several, not a pass-or-fail filter. Pair them with skills evidence, work samples, and structured interviews. A single trait score should open a conversation, never close a candidacy. This is also your strongest defence against the faking that does slip through, because multiple measures are far harder to game than one.
4. Do not lean on "lie detector" scales
Some tests bolt on a social desirability or validity scale that claims to flag fakers. Be careful here. Research on these scales has not found one that reliably catches individual fakers, so a validity score is a weak safety net at best. It is far more effective to remove the reason to fake than to try to police the answers after the fact.
5. Read the DISC natural-versus-adapted gap
DISC results show a natural profile, how someone behaves when relaxed, alongside an adapted profile, how they adjust to their environment. A wide gap between the two is worth a gentle question in the interview. It can point to someone stretching to fit a role rather than settling into it. Used well, a potential faking problem becomes a useful talking point instead.
Match the test to the question you are asking
DISC and OCEAN resist gaming differently because they are built for different jobs. DISC maps communication and working style, which makes it useful for team fit and collaboration. OCEAN, the research-backed Big Five model, gives a deeper and more stable read of underlying traits. Pick the one that answers your real question, and a mildly inflated score becomes far easier to interpret sensibly than if you were using the wrong instrument in the first place.
The honest bottom line
Can psychometric tests be gamed? A little. A motivated candidate can nudge a conscientiousness score upward, and the lab research confirms the shift is real. But the doomsday version, where anyone can fake any profile and your data is worthless, is not what the evidence shows. Real applicants inflate modestly, the genuine signal survives, and the traits with no obvious "good" answer barely move.
More to the point, gaming is a design problem you can solve. Lower the stakes by running personality tests in their own unscored round. Tell candidates the truth about how you use the results. Lean on formats that resist faking, and never let one trait score make the call. Do all that and you are not fighting the candidate's instinct to perform. You are removing the reason to perform in the first place.
That is when a psychometric test does what it is meant to do: show you who the person actually is, not who they think you want them to be.
Read next
AI agents for hiring are everywhere. Most of them aren't agents.
Most recruiting tools added the word "agentic" to their homepage this year. Few changed what the software actually does. This is how to tell the difference before it costs you money or candidates. A field guide for talent leaders.
Resume Parsing Software for Startups: Affordable Solutions Guide
What should a startup look for in an affordable and reliable resume parser?
Conversational AI Interviews vs Traditional Phone Screens: What's Better?
An actionable framework to help you choose.