What does an IQ test measure?

The tests psychologists actually use

When a psychologist assesses intelligence, they usually reach for one of a small number of published tests. The two best known are the Wechsler scales and the Stanford–Binet.

• WAIS-5 (Wechsler Adult Intelligence Scale, Fifth Edition). Published by Pearson in 2024 for ages 16 to 90. A Full Scale IQ takes seven subtests and about 45 minutes; the ten subtests for all the main index scores take about an hour. Children have their own version, the WISC.
• Stanford–Binet, Fifth Edition (SB5). Published by PAR for ages 2 to 85. It reports a Full Scale IQ, a Verbal IQ and a Nonverbal IQ, built from ten subtests.

Both are given one to one by a trained examiner. Pearson sells the WAIS-5 only to buyers at its highest qualification level (C), which means a relevant doctorate, a licence to practise, or membership of a professional body that requires training in assessment.

What the index scores measure

An IQ test is really a bundle of smaller tests. Scores from related subtests are grouped into index scores, and the overall IQ sits on top. The WAIS-5 describes five areas, and the SB5 has five factors that overlap with them.

Why add these together at all? People who do well on one kind of task tend to do a little better on the others too. The psychologist Charles Spearman described this pattern in 1904 and called the shared part g, for general ability. Full Scale IQ is, roughly, an estimate of it. The index scores matter just as much in practice, because a big gap between them (strong verbal, slow processing speed, say) is often what a psychologist is looking for.

How an IQ score is worked out

Your raw answers mean little on their own. What turns them into an IQ is norming. Before a test is published, it is given to a large sample of people chosen to match the population by age, sex, region, education and other traits. Your result is then compared with people your own age in that sample.

The scores are set so that the average is 100 and the standard deviation (the typical spread) is 15. Because results follow a bell curve, that gives a simple guide:

• About 68% of people score between 85 and 115.
• About 95% score between 70 and 130.
• A score of 130 is about the 98th percentile: higher than roughly 98 in 100 people your age.

Subtests use their own smaller scale (on the SB5 and Wechsler tests, an average of 10 and a spread of 3), which are then combined. A good report also gives a confidence interval, because no test is perfectly precise. "IQ 108, 95% confidence 103 to 113" is more honest than a bare 108.

What IQ predicts, and what it doesn't

The clearest summary is still a report by an American Psychological Association task force (Neisser and colleagues, 1996). It found that IQ scores correlate about .50 with school grades and about .55 with years of education. Those are strong links for psychology, but the report itself points out that a correlation of .50 accounts for only about a quarter of the differences in grades. The rest comes from other things: effort, teaching, health, interest, family and luck.

For job status the same report says IQ accounts for about a quarter of the differences, and for income only about a sixth. So IQ tells you something real about groups, and much less about any one person's future.

Just as important is what IQ tests leave out. The task force named creativity, wisdom, practical sense and social sensitivity as obvious examples, and said there are surely others.

The Flynn effect

If IQ were a fixed, purely inborn quantity, average scores would stay put across generations. They didn't. Over the 20th century, people in many countries scored higher and higher on the same tests, which is why tests have to be re-normed every so often. The pattern is named after the researcher James Flynn, who documented it.

The largest analysis (Pietschnig and Voracek, 2015) combined 271 samples, almost 4 million people in 31 countries, from 1909 to 2013. Full-scale scores rose by about 0.28 points a year, roughly 3 points a decade. Fluid reasoning rose fastest, knowledge-based scores slowest, and the gains have slowed in recent decades.

In some countries the trend has even reversed. Norwegian conscription records for men born from 1962 to 1991 show scores rising until the cohorts born around 1975 and falling after that (Bratsberg and Rogeberg, 2018). Because the same rise and fall shows up between brothers in the same family, the authors conclude the causes are environmental, not genetic, though which parts of the environment remains open.

Why online quizzes and brain games aren't IQ tests

Search for "IQ test" and you will find hundreds of free quizzes, many ending with a precise-looking number. A few questions may even look like the real thing. But the number is missing everything that makes an IQ score meaningful:

• No proper norms. The score isn't compared with a large, representative sample of people your age, so "128" has nothing solid behind it.
• Narrow content. Most quizzes use one kind of puzzle, often pattern matrices. A real test samples five or more areas.
• No standard conditions. You can pause, guess, look things up or take it twice.
• Mixed motives. Some give flattering numbers, or charge to see the result.

Brain games have the same limits, and another one: practice. If you play a game every day you get better at it, so your score rises whether or not anything general has changed. Large reviews agree on this. A study of 11,430 people training online for six weeks found they improved on every task they trained, with no transfer to untrained tasks, even closely related ones (Owen and colleagues, 2010). A major review of the evidence behind brain-training companies found strong evidence that people improve on the trained tasks, less for closely related tasks, and little for distant tasks or everyday life (Simons and colleagues, 2016). More in Does brain training work? and Dual n-back explained.

Why Brainbay gives no IQ score

Brainbay has 37 short games across memory, attention, speed, maths, problem solving and words, plus a five-minute daily workout. It could easily print an "IQ" at the end of a session. It deliberately doesn't, for all the reasons above: its games aren't normed against a representative sample, and practising them makes your scores climb.

What its scores do show is useful in a quieter way: how you do at each game, how far up its levels you have climbed, and how that changes over weeks. The six-area profile shows where you are strongest at these games, not how intelligent you are. If you want to measure yourself against a clear yardstick, timed tasks such as reaction time are easier to compare honestly than an invented IQ.

Sources

• Pearson. Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5), product page.
• Pearson. Qualifications policy.
• PAR. Stanford–Binet Intelligence Scales, Fifth Edition (SB5), product page.
• Neisser, U. et al. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51(2), 77–101.
• Pietschnig, J. & Voracek, M. (2015). One century of global IQ gains: A formal meta-analysis of the Flynn effect (1909–2013). Perspectives on Psychological Science, 10(3), 282–306.
• Bratsberg, B. & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. PNAS, 115(26), 6674–6678.
• Owen, A. M. et al. (2010). Putting brain training to the test. Nature, 465, 775–778.
• Simons, D. J. et al. (2016). Do "brain-training" programs work? Psychological Science in the Public Interest, 17(3), 103–186.