How IQ tests work
A good IQ test is more than a set of puzzles. Here's how tests are designed, scored and checked, and how IQ Atlas does it.
1. Measuring several abilities
Intelligence isn't one skill. Modern tests sample several broad abilities from the Cattell–Horn–Carroll model and combine them into an overall score. IQ Atlas uses six sections, all language-free.
2. Questions of rising difficulty
Questions range from very easy to very hard, so the test can tell apart people across the whole range. Each IQ Atlas section starts easy and ends very hard.
3. Scoring with item response theory
Simple tests count correct answers. Better tests use item response theory (IRT), which weighs each answer by the question's difficulty, how well it separates people and how easy it is to guess. A lucky guess on a hard question counts for less than a consistent run of correct answers. IQ Atlas scores every answer this way.
4. Norms and age
Scores are compared with a reference group and placed on a scale with average 100 and standard deviation 15. Because abilities change with age, people are compared with their own age group.
5. Reliability and fairness
A reliable test gives similar results on different occasions, and a fair test doesn't depend on language or culture. IQ Atlas uses only shapes, digits and patterns, checks every question for a single defensible answer, and makes sure wrong options always look clearly different.
Online tests: an honest note
An online test is taken without a supervisor, under unknown conditions, and its norms are less established than those of a clinical test. Treat any online result, including ours, as an informative estimate, not a diagnosis.
Frequently asked questions
How long does a real IQ test take?
Clinical tests take about 60–90 minutes with a psychologist. The IQ Atlas test takes about 45 minutes and can be split across several sittings.