Personality Assessment in Hiring

Theory & origin

Selection testing rests on decades of industrial-organizational psychology, and the finding is nuanced rather than flattering. Structured, validated assessments of job-relevant traits (conscientiousness has the most support) do predict performance, modestly, and add value on top of an interview. But the market is flooded with instruments that fail basic psychometrics. The MBTI, the most popular by far, is not predictive of job performance and sorts the same person into different types on retake, which is why its own publisher warns against using it for hiring. The grey zone runs between two uses of the same category of tool. As one input among several, weighted for a trait the job actually needs, a good instrument reduces the interviewer's gut-feel bias. As a gate, it does three harms at once. It filters for people like the current team, so it quietly narrows diversity. It can discriminate against neurodivergent and disabled candidates, which is a legal exposure in most jurisdictions. And when candidates game it, which they do, it adds noise while feeling like rigor. The test of the test is validation: can the vendor show it predicts performance for this kind of role, and was it checked for adverse impact. No evidence, no gate.

Key components

The parts of the model and what each one means, in plain terms.

Validation evidence

Does the vendor have criterion validity, proof the score predicts performance, for roles like yours, plus an adverse-impact study. This is the gate everything else depends on.

Job relevance

The traits measured are ones the role actually needs, defined before the test, not chosen to match the people already there.

One input, never a gate

The score sits alongside structured interviews, work samples, and references, weighted modestly. A single test as pass or fail is the core abuse of the tool.

Adverse-impact check

Tested for disparate effect on protected groups, and for screening out neurodivergent and disabled candidates. Failing this is a legal exposure, not just an ethical one.

How a consultant runs it

1. 01 Demand the validation evidence before buying. Criterion validity for roles like yours, and an adverse-impact study. No evidence means the tool is decoration at best.
2. 02 Never let a score gate. Personality is one input among structured interviews, work samples, and references, and it should rarely be the heaviest.
3. 03 Assess only traits the job genuinely requires, defined in advance. Fishing for "culture fit" is how you build a monoculture and call it standards.
4. 04 Check for adverse impact and disability discrimination. If the test screens out neurodivergent candidates, it is a legal problem, not a hiring filter.
5. 05 Retire the vanity tools. MBTI and its kin are fine for a team-building workshop and indefensible in a hiring decision. Their own publishers say so.

Worked example

A fast-growing fintech gates every hire on a 16-type personality quiz, auto-rejecting two types, and cannot understand why the team keeps getting more alike and why an autistic finalist filed a complaint. The review finds the tool has zero criterion validity and no adverse-impact study, and that "fit" was doing the discriminating. The rebuild drops the gate entirely and replaces it with a work sample plus a structured interview, keeping a validated conscientiousness scale as 15% of the scorecard, weighted and never decisive. Quality of hire, measured at the 6-month review, goes up. The candidate pool gets more varied. The complaint is settled and the pattern that caused it is gone, because the one test that used to decide now merely informs.

Common pitfalls

1. 01 Gating on a score, which is the single most common and most legally dangerous misuse
2. 02 Buying tools with no validation evidence because the report looks scientific
3. 03 Screening for "fit," which manufactures a monoculture and can hide discrimination
4. 04 Ignoring adverse impact on neurodivergent and disabled candidates, a live legal exposure
5. 05 Treating a workshop tool like MBTI as a selection instrument, against its own publisher's guidance

Sample deliverable

One real engagement, start to finish. Watch the numbers travel from raw input, onto the chart, into the finished artifact.