How IQ Tests Work

A professional IQ test is a set of timed and untimed tasks given one-on-one by a trained examiner. Raw points are converted to age-based scores using a large norming sample, then combined into index scores and a full-scale IQ with a mean of 100 and standard deviation of 15. Understanding that process explains why a real test takes an hour or more and why a ten-minute web quiz cannot give the same result.

Plain-language disclaimer: online tools and articles can only estimate where a score sits on a bell curve. A real IQ score comes from a standardized test given and interpreted by a licensed psychologist. Nothing here is a diagnosis or a claim about giftedness.

The basic design of an IQ test

Every standardized IQ test has the same ingredients: a fixed set of tasks, fixed instructions and timing, a scoring manual and a table of norms. Standardization means every test taker gets the same experience, so the scores can be compared. The tasks sample skills such as vocabulary, verbal reasoning, pattern completion, spatial reasoning, working memory and processing speed.

Wechsler tests: subtests and indexes

The Wechsler Adult Intelligence Scale (WAIS) and the Wechsler Intelligence Scale for Children (WISC) are the most widely used individual tests. Recent editions group subtests into index scores, such as verbal comprehension, visual-spatial or perceptual reasoning, fluid reasoning, working memory and processing speed. The index scores combine into a full-scale IQ. Examples of subtests include defining words, copying designs with blocks, completing matrix patterns, repeating digit sequences backwards and matching symbols against the clock. A full administration commonly takes one to a few hours, depending on the version and purpose.

Other widely used tests

The Stanford-Binet Intelligence Scales cover a wide age range. Raven’s Progressive Matrices use only abstract pattern puzzles, which reduces dependence on language. Group-administered ability tests used in schools, such as the Cognitive Abilities Test, are shorter and less precise than individual tests and are used for screening.

From raw score to IQ: norming

Test publishers give the test to a large sample designed to represent the population by age, sex, region, ethnicity and parental education. For each age band they record the distribution of raw scores. A raw score is then converted to a scaled score (mean 10, SD 3 on Wechsler subtests) and the sum of scaled scores to a composite score with mean 100 and SD 15. This means that IQ always shows standing relative to people of the same age. Convert between scales with the IQ score converter.

Step	What happens
1. Raw score	Points earned on each subtest
2. Scaled score	Raw points converted to age-based scaled scores (mean 10, SD 3)
3. Sum of scaled scores	Subtests added within each index
4. Composite score	Sum converted to a standard score (mean 100, SD 15)
5. Percentile and band	Score translated using the normal curve

Reliability and error

Reliability describes how consistent scores are. Full-scale IQ scores on major tests typically show reliability coefficients of 0.95 or higher in the test manuals, which implies a standard error of measurement of about three points. A reported result should include a confidence interval, such as 108 to 116. See the IQ confidence interval calculator.

Validity: do the tests measure what they claim?

Validity is the evidence that interpretations of scores are justified. IQ test scores correlate with each other and with school performance, which supports their use for describing cognitive skills. Validity is not unlimited. Scores do not predict everything, may be affected by language, culture and test familiarity, and should be interpreted by a professional together with other information.

Why online IQ tests are different

Typical web quizzes are short, unsupervised and not normed on a representative sample. Some compute a score from the number correct using an arbitrary formula, and some are designed to flatter. Without proper norms, a result of 130 on a website has no established meaning. They are fine as puzzles, but not as measurements.

Feature	Professional test	Typical online quiz
Administration	Trained examiner, one-on-one	Unsupervised
Norming	Large representative sample	Usually none
Length	About 1-3 hours	5-30 minutes
Reliability data	Published in manual	Rarely published
Use	Clinical and educational decisions	Entertainment

Practice effects and retesting

Taking the same test twice within a short time usually raises scores by a few points because of familiarity. Professionals avoid retesting too soon. Practicing puzzles can improve performance on similar puzzles, but evidence that it raises general cognitive ability is weak.

What high-IQ societies use

Societies such as Mensa admit people who score at or above the 98th percentile on an approved supervised test. The cutoff is defined as a percentile precisely because the number of IQ points differs between tests. See the IQ percentile calculator for what that means on an SD 15 scale (about 131).

Getting tested

If you want a real evaluation, contact a licensed psychologist, school psychologist or neuropsychologist. Tell them why you want the test. A good evaluation explains the score, the margin of error and the strengths and weaknesses behind the total.

Try the IQ tools

Frequently asked questions

How long does a real IQ test take?

Usually between one and three hours, depending on the test and the purpose.

How is IQ calculated?

Raw scores are converted to age-based scaled scores using norm tables, then the sum is converted to a standard score with mean 100 and SD 15.

Can a child take an IQ test?

Yes. Tests such as the WISC and Stanford-Binet are designed for children, and scores compare the child with others of the same age.

Which IQ test is the most accurate?

Individually administered, well-normed tests such as the Wechsler scales and the Stanford-Binet are considered the standard. No test is perfectly accurate.

Why did I get different scores on two tests?

Tests differ in content, norms and error. Differences of a few points are expected, and larger differences should be explained by an examiner.

Do IQ tests measure only school knowledge?

They are designed to reduce dependence on specific schooling, but vocabulary and verbal tasks still reflect learning and language exposure.

Is a free IQ test valid?

Free online tests are generally not validated or normed, so treat results as entertainment.

Sources

Back to the IQ Online Center

Effortless Math Team

11 minutes ago