Cognitive Ability Tests in Hiring: Validity, Fairness and the Law

Pre-employment cognitive ability tests predict job performance, but more modestly than was long claimed. The classic estimate from Schmidt and Hunter (1998) was a validity of about 0.51, which made general mental ability look like the single best hiring tool. A 2022 reanalysis by Sackett, Zhang, Berry and Lievens lowered it to about 0.31 and ranked structured interviews higher. In the US, employers using such tests must also be able to show they are job-related if the tests screen out protected groups at higher rates.

One point first: the test on this site is a self-insight tool. It is not designed, normed or validated for employment decisions, and it must not be used for hiring, promotion or any other decision about a person.

What a pre-employment cognitive test looks like

Commercial tests vary, but most fall into a few formats:

• Short general-ability tests. A mix of verbal, numerical and logical items under tight time limits, often 10 to 20 minutes. The Wonderlic Personnel Test, which dates to the 1930s, is a long-standing example of this format.
• Separate reasoning tests. Numerical reasoning (interpreting tables and charts), verbal reasoning (judging whether a statement follows from a passage) and abstract or inductive reasoning (finding the rule in a sequence of shapes). Graduate and professional hiring often uses these.
• Adaptive online tests. Items get harder or easier depending on your answers, which shortens the test.
• Game-based assessments. Short tasks presented as games that record speed, accuracy and choices. Their measurement properties vary more and are less well documented in independent research.

Many of these resemble the item types used in IQ tests, which we describe in types of IQ questions. The main differences are tight timing, unproctored online delivery and norms based on job applicants rather than the general population.

How well do cognitive tests predict job performance?

The classic view: Schmidt and Hunter (1998)

Frank Schmidt and John Hunter summarized 85 years of selection research and put the validity of general mental ability at about 0.51 for predicting job performance, higher still for complex jobs. Structured interviews and work-sample tests came out similar, and combining a cognitive test with an integrity test or a structured interview added useful prediction. For two decades this paper anchored the claim that cognitive tests were the best single predictor available.

The revision: Sackett et al. (2022)

Paul Sackett and colleagues re-examined how those older estimates were corrected for range restriction, the statistical adjustment for the fact that studies only see the performance of people who were hired. They argued the corrections had been applied where they did not fit, inflating many validity estimates. Their revised figures reorder the field:

A validity of 0.31 still means cognitive tests carry real information about future performance, and they remain cheap and quick to administer. But they are no longer clearly ahead of alternatives that also tend to produce smaller differences between groups. The revision has its critics, and the debate over the right corrections continues, but it has shifted mainstream advice toward combining methods rather than leaning on one test.

For context on what ability does and does not predict over a career, see how IQ relates to income and success and average IQ by profession.

Adverse impact and US law

Cognitive ability tests often produce average score differences between racial and ethnic groups, and selecting from the top of the score range can therefore exclude members of some groups at much higher rates. These differences are associated with documented factors such as unequal schooling, economic disadvantage and test familiarity; their causes are debated and not fully understood. Under US law the cause of the difference does not change the employer's obligations.

Griggs v. Duke Power Co. (1971)

Duke Power required a high school diploma and passing scores on two general aptitude tests, the Wonderlic and the Bennett Mechanical Comprehension Test, for jobs outside its lowest-paid department. Neither requirement had been shown to relate to performance in those jobs, and both excluded Black employees and applicants at far higher rates. The Supreme Court held that Title VII of the Civil Rights Act of 1964 prohibits practices that are neutral on their face but discriminatory in effect, unless the employer shows they are related to the job and consistent with business necessity.

What the rules require today

• Disparate impact. The Civil Rights Act of 1991 wrote the disparate-impact framework into Title VII. If a test has adverse impact, the employer must show it is job-related and consistent with business necessity; even then, a plaintiff can prevail by showing a less discriminatory alternative practice — in practice, one that is substantially equally valid — that the employer refused to adopt.
• The four-fifths rule. The federal Uniform Guidelines on Employee Selection Procedures (1978) treat a selection rate for any group that is less than four-fifths (80%) of the rate for the group with the highest selection rate as general evidence of adverse impact. It is a rule of thumb used by enforcement agencies, not a legal safe harbor. Federal enforcement priorities shifted in 2025, when Executive Order 14281 directed agencies to deprioritize disparate-impact enforcement, but the statutory standard in Title VII and private lawsuits are unaffected.
• Validation. The Uniform Guidelines describe how employers can document that a test is job-related, typically through a job analysis and a validation study.
• No group-based score adjustments. The 1991 Act prohibits adjusting scores or using different cutoffs by race, color, religion, sex or national origin.
• Disability accommodations. Under the Americans with Disabilities Act, applicants with disabilities may be entitled to reasonable accommodations, such as extra time or an alternative format, unless the test is meant to measure the very skill affected.

This is a general description, not legal advice. Employers should get advice from an employment lawyer or an industrial-organizational psychologist before adopting or changing a selection test, and state and local rules can add further requirements.

If you are taking a pre-employment test

A few practical points can help you perform closer to your real level:

1. Learn the format in advance. Familiarity with item types and timing reduces errors from surprise. Practice tests published by the employer or the test provider are the most relevant.
2. Manage the clock. Many hiring tests have more items than most people can finish. Skipping a hard item and coming back is often better than stalling.
3. Ask about accommodations early if you have a disability that affects test-taking.
4. Keep the score in perspective. A test is one input. Employers typically combine it with interviews, work samples and references.

For general preparation advice, see how to prepare for an IQ test.

IQ and leadership

Intelligence is related to leadership, but less strongly than many expect. A meta-analysis by Timothy Judge, Amy Colbert and Remus Ilies (2004) found a corrected correlation of about 0.27 between intelligence and leadership, lower than many earlier reviews had implied. The authors noted that personality traits such as extraversion and conscientiousness have shown links to leadership of similar or larger size.

There may also be a ceiling effect. In a study of business leaders rated by their subordinates, John Antonakis, Robert House and Dean Simonton (2017) found that effective leadership behavior, as perceived by teams, rose with intelligence up to a point and then declined. They suggested that very high scorers may pitch ideas or solutions in ways their teams find harder to relate to. The finding comes from one sample and should be treated as suggestive.

Team intelligence: the collective intelligence debate

In 2010, Anita Woolley and colleagues reported in Science that groups working on a varied set of tasks showed a general "collective intelligence" factor, analogous to the general factor in individuals. It was predicted less by the average or maximum IQ of group members than by members' social sensitivity, how evenly they took turns in conversation and the proportion of women in the group, a link largely explained by social sensitivity.

The claim has been contested. Bates and Gupta (2017), in a set of replications, found that the average ability of group members was the main predictor of group performance and found little support for the social-sensitivity effect. A 2021 meta-analysis by Riedl, Woolley and colleagues, pooling many more groups, again supported a collective intelligence factor. The fair summary is that groups do differ in how well they perform across tasks, and that both member ability and how members interact probably matter, but the relative weight is unresolved.

Problem-solving and remote work

Employers increasingly list "problem-solving" as a core skill, especially for remote roles where people work with less supervision. Cognitive tests capture part of this: reasoning with unfamiliar information. Remote work also leans on skills that tests do not measure, such as written communication, self-organization and knowing when to ask for help. There is little research isolating cognitive ability's role in remote-work performance specifically, so claims that remote roles demand a higher IQ are not supported by direct evidence. Structured interviews and realistic work samples remain the most informative ways to see how someone tackles the problems a particular job presents.