What Is IQ? How IQ Is Measured and What the Score Means

Where the IQ score came from

The story begins in Paris in 1905. French schools needed a fair way to identify children who would benefit from extra support, and the psychologist Alfred Binet, working with Théodore Simon, put together a series of short tasks of increasing difficulty: following simple instructions, repeating digits, comparing lengths, defining familiar words. In a revision of 1908 they grouped the tasks by age, so that a child's performance could be expressed as a "mental age", the age at which a typical child solves the same set of tasks.

In 1912 the German psychologist William Stern suggested dividing mental age by actual age. In 1916 Lewis Terman at Stanford adapted Binet's tasks for American children, multiplied the ratio by 100, and the term "IQ" entered everyday language. A 10-year-old who performed like a typical 12-year-old scored 120.

The ratio worked reasonably well for children but broke down for adults. Performance on these tasks stops climbing steadily in adulthood, so a ratio would make a perfectly steady 40-year-old look as if they were slipping every year. David Wechsler solved this in 1939 with a different idea. Instead of a ratio, he compared each person with others of the same age and expressed the result as a distance from that group's average. This is the deviation IQ, and almost every modern intelligence test uses it. Corwin Boake's history of intelligence testing follows this path from the Binet-Simon scale to the Wechsler scales in detail.

The deviation IQ: what 100 and 15 actually mean

A raw score on its own means very little. Solving 24 of 32 puzzles could be excellent on a hard test and ordinary on an easy one. To turn raw scores into IQ, test developers give the test to a large reference group, record how the scores spread out in each age band, and then express every new result as a position within that spread.

Two numbers define the scale:

• The center, 100. A score of 100 means performance right in the middle of the reference group for your age.
• The standard deviation, 15. This is the unit of distance. A score of 115 is one standard deviation above the center, 130 is two above, and 85 is one below.

When scores follow the familiar bell-shaped normal curve, those distances translate into predictable shares:

This is why an IQ score is always relative. It does not count how much intelligence you "have" in some absolute unit. It tells you where your performance falls compared with a particular group, at a particular time, on a particular set of tasks.

The time part matters more than you might expect. Average performance on IQ tests rose through the twentieth century in many countries, a pattern known as the Flynn effect. A meta-analysis covering more than a century of data from 31 countries estimated full-scale gains of about 0.28 IQ points per year, close to 3 points per decade, and found that the gains have slowed in recent decades. Because of this drift, established tests are standardized again every so often. Measured against an old reference group, today's test-takers can look better than they would against a fresh one.

What IQ tests actually ask you to do

If you have never seen a professional intelligence test, the tasks are probably less mysterious than you imagine. Most batteries combine several kinds of short tasks:

• Pattern and matrix puzzles. A grid of shapes follows hidden rules, and you choose the piece that completes it. Research on Raven's matrices, one of the best known tests of this kind, found that harder items combine more rules, and that the people who solve them are better at working out abstract rules and keeping several goals in mind at once while they check the options.
• Verbal tasks. Defining words, explaining how two ideas are alike, answering general-knowledge questions.
• Quantitative tasks. Number series, arithmetic problems, comparing amounts.
• Spatial tasks. Rotating a figure in your mind, assembling a shape from parts, imagining how a pattern folds.
• Memory tasks. Repeating sequences of digits or remembering positions.
• Speed tasks. Matching symbols or scanning for a target as quickly as possible.

Each kind of task produces its own score, and the overall IQ, often called the full-scale IQ, combines them. One of the oldest findings in psychology, first described by Charles Spearman in 1904, is that people who do well on one kind of task tend to do at least somewhat well on the others. That shared element is called the general factor, or g, and it is the main reason a single overall score means anything at all. The American Psychological Association's task force report, Intelligence: Knowns and Unknowns, summarizes this and other well-established findings in readable form.

The CHC model in plain words

Treating intelligence as one single thing turned out to be too simple. Treating it as dozens of unrelated skills turned out to be wrong as well. The model behind many of today's major test batteries sits in between. It is called the Cattell-Horn-Carroll model, or CHC, after the three researchers whose work it brings together.

Raymond Cattell, later joined by John Horn, distinguished fluid and crystallized intelligence. Fluid intelligence is reasoning with new material, such as spotting a rule you have never seen before. Crystallized intelligence is the knowledge and skill you have built up through education and experience, such as vocabulary. In 1993 John Carroll published a reanalysis of more than 460 data sets collected over several decades, Human Cognitive Abilities, and proposed a hierarchy of abilities in three layers. Kevin McGrew and colleagues later merged the two lines of work into one framework.

You can picture CHC as a tree:

1. At the top sits a general ability, g, which influences performance on almost every task.
2. In the middle are the broad abilities, more than a dozen in recent versions of the model. Among the best known are fluid reasoning, comprehension-knowledge (the crystallized side), visual processing, quantitative knowledge, working memory, long-term storage and retrieval, and processing speed.
3. At the bottom are many narrow abilities, such as mental rotation, working out a rule from examples, or perceptual speed, which individual tasks tap directly.

The practical value of CHC is that it gives test makers a shared map. A well-designed battery samples several broad abilities, so that the overall score does not rest on one narrow skill, and so that a profile of scores can be described in a common language.

How this relates to the test on this site

The test on iqtest.quest is picture-based. It samples four test areas:

• Pattern reasoning: finding the rule behind a sequence or a grid of pictures and applying it to a new case.
• Spatial thinking: imagining, rotating and comparing shapes in space.
• Number reasoning: recognizing quantities and the relationships between numbers.
• Quick visual matching: finding and matching simple visual details quickly and accurately.

In CHC terms these sit closest to fluid reasoning, visual processing, the quantitative abilities and processing speed. Because the tasks use pictures, shapes and quantities, the test deliberately leaves out vocabulary and general knowledge. That makes it less dependent on language, and it also means it covers a narrower slice of abilities than a full clinical battery.

How a professional IQ test is standardized

Behind a clinical IQ test of the kind psychologists use stands a long development process. The publisher collects data from a large sample chosen to reflect the population by age, sex, region, education and other characteristics, and each age band gets its own norms. The test is checked for reliability (do scores stay consistent?) and validity (does the score relate to what it should?), following professional guidance such as the Standards for Educational and Psychological Testing published jointly by AERA, APA and NCME. A trained examiner then gives the test one to one, in a quiet room, following a fixed script.

No test score is perfectly precise. Every score carries some measurement error, usually summarized as the standard error of measurement, which is why careful tests report a range, often called a confidence interval, alongside the single number.

It is worth being clear about where an online test differs. The test on this site converts your answers to the familiar scale centered at 100 with a standard deviation of 15, but it does so with provisional model parameters for each age group, not with norms collected from a representative population sample. The parameters may change after empirical calibration, and any change comes as a new, versioned release. So the score and percentile you see, and the provisional model range in the detailed report, are estimates for self-discovery, not a clinical assessment. The methodology page explains the scoring step by step.

What IQ does not measure

An IQ score is useful precisely because it is narrow. It describes how well you solved a certain kind of task under certain conditions. Here is some of what it leaves out:

• Knowledge and expertise. A picture-based test says nothing about what you know of medicine, history or carpentry, and even verbal batteries only sample general knowledge.
• Creativity. Producing original ideas is related to, but different from, finding the single correct answer to a puzzle.
• Social and emotional skills. Understanding people, handling conflict and reading a room are not part of standard IQ batteries.
• Motivation, persistence and character. These shape what people do with their abilities, and IQ tests do not measure them.
• Practical judgment. Knowing what matters in a messy real situation is a different skill from solving a well-defined puzzle.
• Your worth or your potential. An IQ score describes performance on a test. It is not a verdict on a person.

IQ scores do predict some things reasonably well. The APA task force reported that IQ correlates with school grades at around 0.5. That is a substantial link, and it still leaves most of the differences in grades to other factors such as effort, teaching, interest and circumstances. It is a fair way to think about IQ in general: a meaningful piece of information, not the whole picture.

Can you trust an online IQ test?

An online test can be genuinely useful, as long as it is honest about what it does. A few questions are worth asking of any test you take:

• Does it explain what it measures and how the score is calculated?
• Does it say whether its scores rest on a representative population sample or on model assumptions?
• Does it tell you how precise the score is, rather than presenting one number as the final word?
• Does it control timing, so that results are comparable from one person to the next?
• Does it avoid diagnostic claims?

An online test cannot reproduce the controlled conditions of an individually administered assessment, and none should be used for clinical, school-placement or employment decisions. If you have a concern about a child's development, or about changes in your own memory or thinking, a qualified professional such as a psychologist or a doctor is the right person to see.

For curiosity and self-discovery, though, a carefully built test can give you a reasonable estimate of how you reason with patterns, shapes and numbers. Just as usefully, understanding what sits behind the number (a scale, a reference group, a set of tasks and a margin of error) lets you read any IQ score, from any test, with the right mix of interest and caution. If you already have a result in front of you, What Does My IQ Score Mean? walks through it piece by piece.