How is IQ measured?
Not by counting correct answers. Your raw performance is compared against a reference population, then converted onto a scale fixed at 100. That middle step is what separates a test from a quiz.
The three steps
1. You answer. Items are scored right or wrong. On a clinical test these span several subtests, each contributing to the total: verbal comprehension, perceptual reasoning, working memory, processing speed.
2. Your raw score is compared to a norming sample. This is the step that does the real work, and the one nobody advertises. A large representative group has already taken the test, and their performance is the reference. The comparison is against your own age band, which is how identical raw performance can mean different things at fifteen and at fifty.
3. It is converted onto the standard scale. Your position in that distribution is mapped to a scale with a mean of 100 and a standard deviation of 15. Every score you have ever seen quoted is the output of this step.
Norming is the whole ballgame
A test with bad norms is not a test. It is a quiz with a number attached.
If the reference sample is small, unrepresentative, or decades out of date, every score it produces is misleading, and almost always in the flattering direction. A self-selected internet sample skews towards people who like taking tests, which makes ordinary performance look impressive.
This is the honest dividing line between a serious instrument and something that emails you a certificate. It is not question difficulty, and it is not how clever the puzzles look. It is whether anyone did the unglamorous work of establishing what the answers mean. That is also the single most useful thing to check when you are trying to tell a real test from a scam.
What a proper test looks like
The WAIS for adults and the WISC for children are the standards. Characteristically:
- Administered one to one by a trained professional
- 60 to 90 minutes
- Multiple subtests producing index scores, not a single total
- The examiner observes how you approach problems, not just whether you solved them
- Norms revised periodically as populations shift
That last two points matter more than people expect. A clinician uses the session to watch strategy, persistence and error patterns, which no online format can reproduce. And norms go stale, which is why the Flynn effect forces periodic re-standardisation.
Where short online tests fit
A screener runs the same three steps with fewer items and looser norms. That gets you a reasonable band and a genuinely interesting result. It cannot deliver precision, subtest profiles, or anything usable for a formal decision.
The defensible claim is roughly here on the distribution. Anything more confident than that, from any online test including ours, is overselling. Our own scoring and its limits are documented on the methodology page, down to which item sets we use and under what licence.
Once you have a score, what percentile means is the next thing worth understanding, because that is the part that actually tells you something.
Key takeaways
- IQ is calculated in three steps: answer, compare to a norming sample, convert to the standard scale.
- Norming is what separates a test from a quiz. A bad reference sample makes every score meaningless.
- Comparison is always against your own age group, which is why raw score alone tells you nothing.
- Clinical tests take 60 to 90 minutes, use multiple subtests, and are administered one to one.
- Short online tests can place you in a band but cannot resolve individual points.