Key takeaways
- An IQ test is a battery of short, separately timed tasks. The Wechsler Adult Intelligence Scale, Fifth Edition, for example, groups its subtests into five domains: verbal comprehension, visual spatial, fluid reasoning, working memory and processing speed.
- The composite score is a rank against a comparison group, not a quantity of anything. Nothing is being counted the way grams or seconds are counted.
- The 'g factor' is one interpretation of a statistical pattern — that scores on unrelated cognitive tasks tend to correlate positively. A 2023 network analysis of CHC measures argues that general intelligence is an interpretation of that pattern rather than its demonstrated cause.
- Motivation, persistence, social judgement, practical skill, creativity and domain expertise are outside the instrument. A battery cannot report on abilities it never sampled.
- Practising arithmetic or memory drills improves arithmetic and memory drills. That is a genuine, useful outcome and it is not an assessment of anything.
An IQ test measures how well one person performed on a specific, fixed set of short cognitive tasks, under standardised conditions, compared with other people of the same age who took the same tasks. That is the whole claim. It is not a reading of a quantity inside anyone's head, and the number it produces is a rank, not an amount.
Almost every argument about IQ becomes clearer once that sentence is taken literally. A rank depends entirely on who is in the comparison group and which tasks were chosen. Change either and the number changes, without anything about the person changing at all.
What is actually inside a test battery
A modern individually administered battery is not one test. It is a collection of short, separately timed subtests, each sampling a narrow skill, whose results are pooled into domain scores and then into a single composite.
Pearson's official description of the Wechsler Adult Intelligence Scale, Fifth Edition — an instrument for ages 16:0 to 90:11 — sets out the structure plainly. Its subtests appraise five cognitive domains: verbal comprehension, visual spatial ability, fluid reasoning, working memory and processing speed. Seven core subtests feed the Full Scale IQ, with further secondary and ancillary subtests available for more specialised questions. Purchasing and administering it requires Qualification Level C, meaning a credentialled professional rather than anyone who can open a web page.
That last detail is not bureaucratic trivia. The qualification requirement exists because the score is only meaningful if the administration conditions were the ones the norms were collected under. A battery administered casually is not a cheaper version of the same measurement; it is a different thing that happens to use similar questions.
The g factor, and what the evidence for it is
If you give a large group of people a dozen unrelated cognitive tasks, the scores tend to correlate positively: people who do well on one tend to do somewhat better than average on the others. This pattern is robust and it is the empirical starting point of the whole field.
What the pattern means is a separate question. The Cattell–Horn–Carroll framework, the model most contemporary batteries are built around, arranges abilities in three strata: individual tests at the bottom, a set of broad abilities above them — fluid reasoning, comprehension-knowledge, visual processing, auditory processing, working memory, retrieval fluency and processing speed — and a general factor at the apex.
A 2023 network analysis of CHC measures by McGrew, Schneider, Decker and Bulut is worth reading carefully on this point. Using network methods rather than conventional factor analysis, the authors recovered seven broad CHC abilities without requiring a dominant general factor to account for the correlations between tests. Their argument is that the positive correlation among tests is the primary fact, and that general intelligence is one interpretation of that fact rather than its demonstrated cause. Comprehension-knowledge, in their model, turned out to be unusually central to the network as a whole.
So "g" is best described as a well-replicated statistical regularity with a contested interpretation — not a substance, and not a thing anyone has located.
The abilities the instrument never touches
A test can only report on what it sampled. That is an obvious point which marketing routinely obscures, so it is worth stating in full.
Not sampled by a standard cognitive battery: motivation, persistence, conscientiousness, emotional regulation, social judgement, practical problem-solving in an unstructured setting, creativity, manual and athletic skill, and any specific expertise that happens not to appear in the subtests. Also not sampled: what a person can do with unlimited time, with tools, with colleagues, or when the task matters to them.
This is why the phrase "measures intelligence" does more work than it should. A battery measures performance on its own contents. Whether those contents add up to the everyday word "intelligence" is a judgement, not a finding.
What the score is useful for, and where that stops
Cognitive measures do relate to real outcomes. A 2026 framework paper in the Journal of Intelligence summarises the position for education: measures of general cognitive ability show moderate-to-strong associations with grades, standardised achievement scores and educational attainment, and adding a cognitive measure to an existing file can produce a statistically meaningful predictive gain, particularly at transition points.
The same paper is the better guide to the limits. Incremental validity is context-dependent, shifting with range restriction, grading differences and unequal opportunity. More importantly, predictive validity is not the same as decision utility — whether a score actually improves a decision someone has to make. Where achievement records already carry most of the signal, adding a test may not change the decision at all; where a target outcome is rare, false positives multiply; and a measure can improve average prediction while worsening outcomes for a subgroup. The authors' conclusion is that testing should be conditional and sequential, used only where it can plausibly change what happens next.
Then there is measurement error. As ETS puts it in its plain-language guide to reliability, test scores are reliable to the extent that they are consistent across occasions, across editions of a test, and across raters. No score is exact, which is why a responsible report gives a range rather than a point.
Where a practice app sits in all this
IQ Test & Brain Training Games is our own app, built by Reign Creative, and this is the honest account of its relationship to everything above: it includes an IQ-style test that samples some of the same kinds of task, without the conditions that make a battery's score interpretable.
The app's IQ-style test covers numerical reasoning, logic, patterns, vocabulary, working memory, spatial reasoning and processing speed, and around it sit memory, logic, focus, vocabulary and mental math games, a memory test and a reaction time test. Some of those areas share names with the domains a standardised battery samples. The procedure does not: the app's test is taken unsupervised, at a moment of your choosing, and its listing documents no comparison group. The listing itself says that IQ test, memory test and other results are estimates based on app activities, not medical, psychological, clinical, educational-placement, employment or professional diagnostic assessments. The app also keeps personal bests, accuracy, streaks, response speed and an assessment history, which describe how you did on its own questions.
Four things are worth keeping apart whenever this subject comes up: practice, entertainment, educational activity and standardised psychological assessment. The app is the first three. It is not, and does not pretend to be, the fourth. The wider question of what practice products can and cannot claim is treated in our guide to whether brain training apps work, and the narrower question of what working memory is, and how contested its trainability is, in our explainer on working memory and training.
How to read a score someone hands you
Three questions do most of the work. Who was the comparison group, and when were they tested? What tasks were sampled, and under what conditions? And what is the margin of error on the reported number?
If the answer to any of those is unavailable, the score is not interpretable — which is the usual situation outside supervised assessment. The companion piece on how an IQ score is calculated works through the arithmetic that turns raw answers into a standard score, and the limitations of IQ testing covers what remains uncertain even when every one of those questions has a good answer. More explainers on testing, reasoning and practice sit in the IQ tests and reasoning hub and across the wider education and brain training section.
Good to know
Frequently asked questions
What do IQ tests measure?
Is the g factor a thing in the brain?
What do IQ tests fail to measure?
Does the IQ Test & Brain Training Games app measure IQ?
Sources
- WAIS-5 — Wechsler Adult Intelligence Scale, Fifth Edition (opens in a new tab)Pearson Assessments (Pearson Clinical Assessment US) — accessed
- A Psychometric Network Analysis of CHC Intelligence Measures: Implications for Research, Theory, and Interpretation of Broad CHC Scores "Beyond g" (opens in a new tab)Journal of Intelligence (McGrew, Schneider, Decker & Bulut, 2023), via PubMed Central — accessed
- From Incremental Validity to Decision Utility: A Framework for Intelligence Testing in Education (opens in a new tab)Journal of Intelligence, via PubMed Central — accessed
- Test Reliability — Basic Concepts (Research Memorandum RM-18-01) (opens in a new tab)Educational Testing Service (Samuel A. Livingston, 2018) — accessed
About the author
Reign Creative LLC — The independent app studio that builds the apps covered on this site and publishes every article here. On Google Play it publishes as Reign Collective Apps.
We use AI tools to help research, draft and edit our articles. Each article lists the sources its factual claims rely on, and details about our apps come from their live Google Play listings. No article on this site has been reviewed by an independent expert unless the reviewer is named on it.
Spotted an error? Tell us — how corrections are handled is set out in our editorial policy.