How IQ score is calculated: the formula and the logic
Your IQ score is not the number of questions you answered correctly. It is a comparison of your raw performance to everyone else in your age group. Test designers collect data from thousands of people, find the middle of the distribution, call it 100, and then calculate how far you sit from that middle. That distance is expressed in units called standard deviations — each one worth 15 IQ points. So a score of 115 means you performed one standard deviation above the average for your age; 85 means one standard deviation below.
From raw score to IQ: what actually happens
Every modern IQ test — Wechsler, Stanford-Binet, Raven's Progressive Matrices — starts the same way: you answer a battery of questions measuring things like pattern recognition, working memory, verbal reasoning, and spatial thinking. The test records how many tasks you completed correctly and how quickly. That number is your raw score.
The raw score on its own means nothing. A raw score of 42 out of 60 tells you little unless you know how other people your age performed on the same test. So the publisher compares your result to a large normative sample — a group of several thousand people, representative of the general population, who took the same test under the same conditions.
That comparison produces a standardised score, and that standardised score is your IQ.
The bell curve and standard deviations
IQ scores follow a roughly normal distribution — the bell curve. The curve is centred at 100, and the spread is controlled by the standard deviation, which is set at 15 points on most widely used tests. This means:
- About 68% of people score between 85 and 115 (within one standard deviation of 100).
- About 95% score between 70 and 130 (within two standard deviations).
- About 99.7% score between 55 and 145 (within three standard deviations).
A score of 130 does not mean you are 30% smarter than average. It means your performance placed you roughly at the 98th percentile — only about 2 people in 100 scored at that level or higher in the normative sample.
Why age matters so much
Raw cognitive performance changes across a lifetime. A 10-year-old who solves a problem that most adults cannot is impressive. A 35-year-old solving the same problem is less notable. IQ tests correct for this by comparing you only to people in your own age band — typically in ranges of a few years. The technical name for the result is a deviation IQ, and it has been the standard method since the 1960s.
Earlier tests used a different formula: mental age divided by chronological age, multiplied by 100. If a 10-year-old performed like a typical 12-year-old, the ratio IQ was (12 ÷ 10) × 100 = 120. This ratio method was abandoned because it broke down badly in adulthood — raw cognitive scores do not keep rising linearly after adolescence, so the formula produced nonsense results for anyone over about 18.
How 100 stays the midpoint over time
Here is a subtlety most people miss. The midpoint of 100 is not a fixed biological constant — it is re-anchored to the current population every time a test is restandardised. Publishers redo the normative samples every 10 to 20 years. This is necessary because raw scores on IQ tests have been rising over generations, a well-documented pattern called the Flynn effect. If the norms were never updated, average scores would drift above 100. By restandardising, test publishers keep 100 as the population midpoint, which is what makes scores from different eras comparable in concept, even if the underlying raw performance has shifted.
Composite scores and subtests
Modern IQ tests do not produce a single number from a single task. They include multiple subtests, each targeting a different cognitive ability. The Wechsler Adult Intelligence Scale (WAIS), for example, groups subtests into four index scores:
- Verbal Comprehension — vocabulary, general knowledge, abstract verbal reasoning.
- Perceptual Reasoning — visual puzzles, pattern completion, spatial tasks.
- Working Memory — holding and manipulating information in short-term memory.
- Processing Speed — how quickly you can complete simple, accurate tasks.
Each index is itself standardised to a mean of 100 and a standard deviation of 15. The Full Scale IQ (FSIQ) is a composite derived from all four, weighted and combined using a formula specified in the test manual. It is not simply an average of the four index scores — the exact algorithm is proprietary and differs between test editions.
What the number does and does not capture
An IQ score is a reliable measure of a specific set of reasoning abilities at a specific moment. Large-scale studies consistently find that it predicts academic performance and certain kinds of job performance better than most other single measures. But it does not capture creativity, emotional regulation, practical judgement, motivation, or social skill — all of which matter enormously for how a person functions in the world.
It is also a snapshot, not a fixed trait. Scores can shift meaningfully with age, health, sleep, anxiety, familiarity with test formats, and education. A score taken when you were ill or under severe stress may underestimate your typical performance. A score taken after intensive practice with similar puzzles may overestimate it.
A rough scale for reference
| IQ range | Classification (Wechsler) | Approximate percentile |
|---|---|---|
| 130 and above | Very Superior | 98th and above |
| 120–129 | Superior | 91st–97th |
| 110–119 | High Average | 75th–90th |
| 90–109 | Average | 25th–74th |
| 80–89 | Low Average | 9th–24th |
| 70–79 | Borderline | 2nd–8th |
| Below 70 | Extremely Low | Below 2nd |
These labels come from the Wechsler test manuals and are used in clinical settings. Other tests use slightly different cut-offs and terminology.
Can an online test give you a real IQ score?
Technically, any test that lacks a large, representative normative sample cannot produce a clinically valid IQ score. Proper standardisation requires thousands of participants, controlled testing conditions, and regular restandardisation. Online tests that have been carefully developed and validated against established instruments can give a reasonable estimate — close enough to be informative and consistent across attempts — but they are not equivalent to a full clinical assessment administered by a psychologist. If you want to get a sense of where you sit, you can take the test at IQMeter and treat the result as an educated estimate, not a clinical diagnosis.
Частые вопросы
Is IQ calculated differently for children and adults?
Yes. Children and adults are compared to normative samples from their own age band — typically a range of a few years. A 9-year-old is compared to other 9-year-olds; a 40-year-old is compared to other adults in a similar age range. The underlying formula is the same (deviation IQ), but the reference group changes with age.
What does a standard deviation of 15 mean in practice?
It means that roughly two out of three people (68%) score between 85 and 115. Each 15-point step up or down moves you approximately one standard deviation from the population midpoint of 100. A score of 130 is two standard deviations above the mean, reached by about 2% of people.
Can my IQ score change over time?
Yes, within limits. Scores measured years apart in the same person tend to correlate strongly — the trait is relatively stable. But meaningful shifts do happen, particularly in childhood and adolescence, with serious illness, after major changes in education or environment, or as a result of ageing. A single test result should not be treated as permanently fixed.
Why is 100 always described as the average if people keep getting smarter?
Because test publishers periodically update the normative samples. Raw scores on IQ tests have risen across generations (the Flynn effect), but each time the norms are updated, 100 is re-anchored to the new average. This keeps the scale interpretable but means that a score of 100 today reflects higher absolute performance than a score of 100 from 30 years ago.
Do different IQ tests give the same score?
Not always exactly. Different tests measure overlapping but not identical abilities, use different normative samples, and weight subtests differently. Scores from two well-validated tests tend to be close — usually within 5 to 10 points — but perfect agreement is not expected. For clinical decisions, a single test from a qualified examiner is the standard.