Is personality testing biased? What hiring teams need to know

Bias in hiring is a real and serious problem. Any tool used to screen or evaluate candidates deserves scrutiny, and personality assessments are no exception. The concern is legitimate: if a test consistently produces lower scores for candidates of a particular demographic group, and those scores influence hiring decisions, the test is producing disparate impact regardless of whether that was the intent.

The question is whether personality assessments — and the Big Five framework specifically — exhibit the kinds of bias that should concern hiring teams and legal compliance functions. The answer is more nuanced than either critics or advocates of assessment tend to acknowledge.

What “bias” means in this context

In psychometric testing, bias has a specific technical meaning. A test is biased if it measures different constructs for different groups, or if it systematically under- or over-predicts outcomes for particular groups. Adverse impact — when a test produces lower scores for protected demographic groups — is not by itself evidence of bias. A test can show group-level differences in mean scores while still being an unbiased predictor of job performance, if those differences reflect real variation in the characteristic being measured rather than test construction artefacts.

This distinction matters because conflating adverse impact with bias leads to bad decisions in both directions: discarding valid tests because they show group differences, or retaining biased tests because they appear not to show differences.

The relevant question for hiring is: does this test predict job performance equally well across demographic groups, or does it systematically over-predict for some groups and under-predict for others?

What the Big Five research shows

The Big Five has been more thoroughly tested for bias than virtually any other personality framework used in employment contexts.

On adverse impact: the Big Five does show some mean-level differences across demographic groups, though the pattern is considerably more complex than what’s observed with cognitive ability tests. Foldes, Duehr, and Ones (2008), in a meta-analysis covering multiple studies and demographic comparisons, found that Big Five dimensions show smaller subgroup mean differences than cognitive ability tests, with most effect sizes in the small range (d < 0.3). The dimension showing the most consistent group differences was conscientiousness, where some studies have found small but statistically reliable variation across racial groups.

On predictive bias: the more important finding is that the Big Five shows minimal differential prediction across demographic groups — that is, the relationship between personality scores and job performance is broadly consistent across groups. Ones, Viswesvaran, and Dilchert (2005) reviewed the evidence and concluded that Big Five measures show little evidence of differential validity or differential prediction across gender and racial groups. The test is measuring what it claims to measure, roughly equally, across groups.

This places the Big Five in a substantially better position than cognitive ability tests, which show large and well-documented subgroup differences and generate meaningful adverse impact in hiring contexts.

Where the real risks are

Acknowledging the relatively strong evidence base doesn’t mean personality testing in hiring is without risk. The risks are real — they’re just different from simple demographic bias.

Response coaching. Personality assessments administered with stakes (a job offer on the line) are susceptible to deliberate impression management. Candidates can and do answer how they think the ideal candidate would answer rather than how they would naturally describe themselves. High scores on measures like conscientiousness or agreeableness may in some cases reflect coaching rather than the underlying trait. The Donovan, Dwight, and Schneider (2014) review found that faking has a small but real effect on personality assessments used in hiring, and that it can distort rank-ordering even when it doesn’t substantially change group-level differences.

Using personality to replace skills evaluation. Personality predicts broad tendencies and fits well as one input into a holistic assessment. When organisations use personality scores as a primary or sole filter — rather than as one data point alongside skills assessments, work samples, and structured interviews — they’re using it beyond its appropriate scope. A personality score is not a substitute for evidence of what someone can do.

Gatekeeping versus development. The research base for the Big Five is strongest for predicting broad occupational performance and job satisfaction in aggregate. It is not well-suited to making binary pass/fail decisions about individual candidates, where measurement error and score interpretation become more consequential. Using assessment to identify and develop people — to understand how they work and what environments suit them — is better supported by the evidence than using it as a hard hiring filter.

Legal exposure. In many jurisdictions, using personality assessments as a hiring screen creates legal obligations around adverse impact analysis. Whether or not a test is technically unbiased, if it generates statistically significant group differences in outcomes and those differences affect protected classes, employers may face scrutiny. HR and legal teams should conduct regular adverse impact analyses on any assessment used in selection.

RIASEC interest profiling and bias

Interest-based assessments like RIASEC (Holland Codes) have a different risk profile from personality measures. Because they measure what kind of work someone finds engaging rather than trait dimensions like dominance or agreeableness, they are less susceptible to impression management (it’s harder to fake interest in accounting) and show smaller demographic group differences in mean scores.

Interest assessments are also arguably more appropriate for career guidance than for hiring selection, since they capture alignment between person and work type rather than general performance potential. The Donnay and Borgen (1996) research on interest assessment validity found strong predictive relationships with job satisfaction and career stability, with limited evidence of differential prediction across gender and ethnic groups.

How to use assessment responsibly

None of this means personality testing should be avoided in talent contexts. The evidence supports using it — carefully, with appropriate scope.

Use assessment for development, not just selection. The risk profile of using assessment to help people understand themselves and develop their careers is much lower than using it as a hiring filter. Development use is also where the validity evidence is strongest.

Treat scores as inputs, not verdicts. Personality scores should be one data point in a broader picture that includes skills assessment, structured interviewing, and work sample evaluation. A candidate who scores lower on conscientiousness but has a strong track record of delivering complex projects has provided better evidence of their conscientiousness at work than the assessment alone.

Conduct adverse impact analysis. If you’re using personality assessment in hiring, monitor the outcomes by demographic group. You should know whether pass rates, score distributions, and hiring rates differ across protected characteristics, and be prepared to investigate and adjust if they do.

Choose validated instruments. Not all personality assessments are equal. Tools derived from the Big Five framework and validated for use in work contexts have a substantially stronger evidence base than proprietary models with limited published validity data. The scientific pedigree of an assessment matters.

The bigger picture

The case for using personality assessment in talent contexts is not that it’s perfect — no assessment is. It’s that it adds valid, evidence-based information that other data sources don’t provide, and that it has lower adverse impact than the cognitive ability testing that it typically supplements or replaces.

Used appropriately, the Big Five and RIASEC frameworks contribute to better matching between people and roles: finding environments where someone’s natural working style and interests align with what the work requires. That’s a good outcome for employers and candidates alike.

The use case that carries the most risk and the least robust evidence is using personality as a binary hiring filter. The use case with the strongest evidence and the most ethical standing is using it to help people understand themselves — how they work, what environments suit them, and where they’re most likely to find genuine engagement rather than just competence.

Traitstack is built around the development use case. The Big Five personality assessment and RIASEC interest assessment are designed to give individuals insight into their own working style and career fit — not to produce a score that determines whether someone gets through a hiring screen. The research behind both frameworks, and the way results are presented, reflects that orientation.