What does an IQ test measure?

Short answer: An IQ test measures how you do on a sample of mental tasks, compared with a norm group: formal tests are often "norm-referenced" (APA, 2022), and the WAIS-5 uses age-based norms (Pearson Q-global help). A full test such as the WAIS-5 (Pearson, 2024) samples five abilities and combines them into one Full Scale IQ. Standard tests do not sample creativity, wisdom, practical sense or social sensitivity (APA task force, 1996). Each score is an estimate with a range (Schmukle, Assessment, 2025).

At a glance:

The five WAIS-5 index scores, as Winter and colleagues (Journal of Intelligence, 2025) summarize the manuals:

• Verbal comprehension: solving problems with language and information given in words.
• Visual spatial: solving problems with things you see, such as spatial puzzles.
• Fluid reasoning: solving problems with new information.
• Working memory: holding information in mind and using it on a new problem.
• Processing speed: solving problems quickly with information you already know or can reach easily.
• Full Scale IQ: one score built from seven of the WAIS-5 subtests, meant to represent general intellectual ability (Winter and colleagues, 2025).
• Not measured: creativity, wisdom, practical sense and social sensitivity (APA task force, 1996).

What does an IQ test measure?

An IQ test measures how you do on a sample of mental tasks, compared with a norm group. On the WAIS-5, scores are "derived using age-based norms" (Pearson Q-global help, read 9 October 2026). In its guide to psychological testing in general, the American Psychological Association says formal tests are often described as "norm-referenced", meaning they "have been standardized so that test-takers are evaluated in a similar way, no matter where they live or who administers the test" (APA, 2022).

What sits inside the score depends on the test. A 1996 APA task force report (American Psychologist) said the more familiar measures of general intelligence, "such as the Wechsler tests and the Stanford-Binet, include many different types of items, both verbal and nonverbal." There is also no agreed definition of intelligence: asked to define it, two dozen prominent theorists "gave two dozen somewhat different definitions" (APA task force, 1996).

The five abilities a full IQ test samples

Pearson's WAIS-5, the adult Wechsler test, covers ages 16 to 90 and was published in 2024. It reports five primary index scores, described here as Winter and colleagues (Journal of Intelligence, 2025) summarize the WAIS-5 manuals. Watkins and colleagues (Journal of Intelligence, 2023) give the subtest pairings for the children's version, the WISC-V. Task types follow a 2018 PLoS One table that describes the WISC-IV and WAIS-IV subtests, the previous editions, and no real test item appears here.

• Verbal comprehension. Problem solving "using language and verbally presented information". On the WISC-V it comes from Similarities and Vocabulary. In Similarities, you explain how two words for common objects or ideas are alike.
• Visual spatial. Problem solving with visual information, "such as through spatial puzzles". Block Design asks you to rebuild designs with red-and-white blocks within a time limit.
• Fluid reasoning. Problem solving "with novel information". On the WISC-V it comes from Matrix Reasoning and Figure Weights. More in what fluid intelligence is.
• Working memory. Holding information "presented visually or orally, within the mind" and using it on a new problem. Digit Span forward asks you to repeat numbers in the order you heard them.
• Processing speed. Solving problems rapidly "with known, or easy-to-access, information". Coding asks you to copy symbols paired with numbers, using a key, within a time limit.

Pearson says the WAIS-5 moved "from 4 to 5 factors" (Pearson WAIS-5 FAQ, read 9 October 2026). Researchers debate a similar split on the children's test: factor studies of the WISC-V "have variously suggested either four or five dimensions" (Watkins and colleagues, 2023).

How the parts become one score

Scores on different tasks tend to move together: "people who score high on one such subtest are likely to be above average on others as well" (APA task force, 1996). Researchers call this the positive manifold. Spearman's use of factor analysis in 1904 started a way of modeling intelligence, with layered, unseen factors behind the test scores, that went on to be widely accepted (Schmank and colleagues, Journal of Intelligence, 2019). The shared source is "typically denoted g" (Weber and colleagues, Journal of Intelligence, 2026), and the APA task force (1996) notes that Spearman stressed a general factor, g, "which represents what all the tests have in common".

CHC (Cattell-Horn-Carroll) theory puts g at the top, a set of broad abilities below it, and many narrow abilities below those (Holden and Tanenbaum, Journal of Intelligence, 2023).

The Full Scale IQ is the headline number. On the WAIS-5 it combines seven of the primary subtests and "represents general intellectual ability" (Winter and colleagues, 2025). Like any test score, it is an estimate with measurement error: Schmukle (Assessment, 2025) writes that test administrators should not take an observed score "at face value" and need to compute a confidence interval around it. Pearson's WAIS-5 score report lets the clinician select a 90% or 95% confidence interval (Pearson Q-global help, read 10 October 2026). Your parts will rarely match: the APA task force (1996) wrote that people "rarely perform equally well on all the different kinds of items included in a test of intelligence."

On the WISC-V, index scores sit on a scale with an average of 100 and a standard deviation of 15 (Watkins and colleagues, 2023). For what that scale means, see what the average IQ is.

Do all IQ tests measure the same things?

They overlap, but the mix differs.

• WAIS-5. Five indexes, as above. Its mix changed from the last edition. Winter, Trudel and Kaufman (2024) report that processing speed and working memory made up 40% of the WAIS-IV Full Scale, a value reduced to 28.6% on the WAIS-5.
• Stanford-Binet, Fifth Edition (SB5). Five factors: "knowledge, quantitative reasoning, visual–spatial processing, working memory, and fluid reasoning" (Conte and Harmata, Journal of Intelligence, 2023). The SB5 covers quantitative knowledge, which the WISC-V's primary indexes do not, while the WISC-V covers processing speed and the SB5 does not (Radtke and colleagues, Health Psychology Report, 2025).
• Raven's Progressive Matrices. One kind of task. The APA task force (1996) described it as "a nonverbal, untimed test that requires inductive reasoning about perceptual patterns". More in our guide to Raven's Progressive Matrices.

Results can differ from test to test. Grieder and colleagues (2022) cite three studies that "concluded that any two intelligence tests do not necessarily render comparable FSIQs on the individual level". The APA task force (1996) adds that a person's intellectual performance "will vary on different occasions."

What a matrix-reasoning test measures, and what it leaves out

A matrix puzzle is a grid of shapes with one piece missing. On the Wechsler tests, Matrix Reasoning is listed as "Nonverbal abstract problem solving, inductive reasoning" (PLoS One, 2018), and on the WISC-V it feeds the Fluid Reasoning index (Watkins and colleagues, 2023).

In a study of Raven's Progressive Matrices, Carpenter, Just and Shell (Psychological Review, 1990) found that what separates people is primarily "the ability to induce abstract relations and the ability to dynamically manage a large set of problem-solving goals in working memory."

Researchers disagree on how good a measure that is. A 2020 paper in Brain Sciences reports that Raven's is widely considered "the purest measure of fluid intelligence" (Smirni, 2020). Gignac (Intelligence, 2015) found that Raven's "shared approximately 50% of its variance with g", judged it "not a particularly remarkable test with respect to g", and discouraged researchers from using Raven's alone "if a valid estimate of g is sought."

Pearson notes that the WAIS-5 Nonverbal Index draws on four of the five cognitive domains, with Verbal Comprehension "not included" (Pearson WAIS-5 FAQ). The APA task force (1996) lists Raven's among tests that "use only a single type of item or question", and it follows from that description that a matrix-only test reports no separate verbal comprehension, working memory or processing speed scores.

What IQ tests do not measure

The APA task force (1996) wrote that "standardized tests do not sample all forms of intelligence", naming creativity, wisdom, practical sense and social sensitivity.

• Emotional intelligence. Mayer and Salovey (1997) defined it as the ability to perceive, use, understand and regulate emotions (Haag and colleagues, Frontiers in Psychology, 2025). The five WAIS-5 domains Pearson names are verbal comprehension, visual spatial, fluid reasoning, working memory and processing speed (Pearson WAIS-5 FAQ); emotional intelligence is not among them. It is related, though: a 2021 meta-analysis found that people-centered intelligences, which include emotional intelligence, correlated at r = 0.29 with thing-centered ones such as visuospatial processing, with correlations corrected for unreliability (Bryan and Mayer, Journal of Intelligence, 2021).
• Creativity. Intelligence scores correlate r = .16 with creative achievement across 30 studies (Karwowski and colleagues, Journal of Intelligence, 2021) and r = .25 with divergent thinking across 112 studies (Gerwig and colleagues, Journal of Intelligence, 2021).
• Practical intelligence. Sternberg named analytic, creative and practical aspects of intelligence, "of which only the first is measured to any significant extent by mainstream tests" (APA task force, 1996).

What an IQ score predicts is a separate question, covered in are IQ tests accurate?

Choosing a test for what you want to know

For a school, work or health question, see a psychologist: the APA (2022) says "Licensed clinical psychologists are expertly trained to administer assessments and tests and interpret the results." Pearson's WAIS-5 product page lists the test at Qualification Level C, and Pearson's qualifications policy says "Tests with a C qualification require a high level of expertise in test interpretation" (Pearson, read 9 October 2026).

For a quick reading of one kind of reasoning, WhatIQ's test uses 24 timed picture puzzles, and the site says "It tests reasoning without words or arithmetic." It takes about 15 minutes and gives one estimate on the standard IQ scale with a 90% range, not a clinical assessment (whatiq.com, read 9 October 2026).

Frequently asked questions

How is intelligence measured in psychology?

Psychologists use standardized tests such as the WAIS-5, which is given one to one and scored against age-based norms (Pearson Q-global help, read 9 October 2026). A test can be one part of a wider picture: the APA (2022) says a psychological assessment can include "norm-referenced psychological tests, informal tests and surveys, interview information, school or medical records, medical evaluation, and observational data".

What does an IQ test measure for adults?

Pearson's adult test, the WAIS-5 (2024), covers ages 16 to 90 and samples verbal comprehension, visual spatial, fluid reasoning, working memory and processing speed. In WAIS-5 norm data, verbal skills held up longer than other abilities, while processing speed declined steadily from young adulthood (Winter and colleagues, 2025). That study compared age groups and did not follow the same people over time.

Does an IQ test measure knowledge?

Partly. A full test can sample some acquired knowledge: the Stanford-Binet 5 has a Knowledge factor (Conte and Harmata, 2023). The WAIS-5 Verbal Comprehension index covers problem solving "using language and verbally presented information" (Winter and colleagues, 2025). A matrix test works differently: the APA task force (1996) describes Raven's Progressive Matrices as "a nonverbal, untimed test that requires inductive reasoning about perceptual patterns".

Can an online IQ test measure the same things as the WAIS?

Not a short one built on one task type. The WAIS-5 is given one to one, builds its Full Scale IQ from seven subtests, and reports five index scores (Pearson, 2024). A matrix test such as Raven's uses "only a single type of item or question" (APA task force, 1996). Even two full tests can give one person different scores (Grieder and colleagues, 2022), so treat an online score as an estimate with a range, not a clinical result.

Who made the first IQ test?

Binet and Simon. Their scale of 1905 "was the first formal measure of intelligence" (Holden and Tanenbaum, 2023). It followed a 1904 French commission that decided that no child suspected of an intellectual disability should leave the ordinary school for a special class without a pedagogical and medical examination (Ackerman, Journal of Intelligence, 2022). Stern first presented the intelligence quotient in 1912 (Heinemann, Journal of Intelligence, 2023).

Can your IQ score change between sittings?

Yes. The APA task force (1996) wrote that "a given person's intellectual performance will vary on different occasions". A score is an estimate with a range: Schmukle (Assessment, 2025) advises computing a confidence interval around it rather than taking it "at face value". A 2011 PNAS study by Duckworth and colleagues reported that incentives raised IQ scores; in May 2025 the authors retracted it. Their notice says publication-bias adjustment gives "a more plausible effect size (g = 0.16)", where g is an effect size rather than the general factor, and that its confidence interval "includes zero" (PNAS, 2025).