Overview

Types of IQ tests

“IQ test” covers at least six unrelated things with different purposes, formats and levels of evidence. Knowing which family you are dealing with answers most questions about it.

Individual clinical tests

Administered one-to-one by a licensed psychologist over one to two hours. The WAIS for adults and the WISC for children are the reference instruments in English-speaking countries, with the Stanford–Binet also in wide use.

These are the only tests whose results carry weight for clinical, educational or legal purposes. The reason is not that the questions are better. It is that the administration is controlled, the norms are stratified to match census data, the examiner observes how you approach the tasks, and the result comes with a confidence interval and an interpretation.

They cost money, require an appointment, and are not available online in any form. Any website offering a WAIS is not offering a WAIS.

Group-administered tests

Designed to be given to many people at once, usually with a strict time limit and a machine-scorable answer sheet. The CogAT in schools and the Wonderlic in hiring are the best-known examples.

They trade depth for throughput. A group test cannot observe how you worked, cannot adapt, and cannot distinguish a wrong answer from a skipped one in the same way an examiner can. What it can do is screen a thousand people in an afternoon, which is what it is for.

Results from group tests are routinely used as a first filter, with individual testing reserved for cases near a decision boundary. Where they are used as the sole basis for a placement decision, the measurement error is doing more of the sorting than anyone intends.

Non-verbal and culture-reduced tests

Built to minimise the role of language and acquired knowledge. Raven's Progressive Matrices is the archetype; Cattell's Culture Fair test was designed on the same principle.

They are genuinely useful where language is a confound: testing across languages, assessing someone with a speech or hearing impairment, or working with recent migrants.

“Culture-fair” overstates what they achieve. Familiarity with abstract classification, with testing situations and with the conventions of multiple-choice reasoning all vary systematically across groups and all affect performance. The strongest evidence against the culture-free claim is that these tests showed the largest Flynn effect gains of all — which would be impossible if they measured something environment-independent.

Achievement and aptitude tests, which are not IQ tests

Achievement tests measure what you have learned in a domain. Aptitude tests predict performance in a specific area. Both correlate with IQ, often strongly, and neither is an intelligence test.

The SAT is the interesting case. It was explicitly derived from the Army Alpha, correlates with IQ measures at something like 0.7 to 0.8, and can be converted to an IQ-like score with reasonable accuracy for cohorts before its 1995 recentering. The College Board has spent decades insisting it is an achievement test, which is true of its content and not the whole story about what it measures.

Online tests

Everything from carefully constructed instruments with published methodology to lead-generation quizzes that return a flattering number regardless of what you answered.

The category is not inherently worthless. A well-built online reasoning test with thirty-plus items can place you in a band reliably. What no online test can do is control administration, norm against a representative sample, or prevent retakes — and those limits are structural, not fixable by better questions.

Our side-by-side reviews go through the main ones and what each actually gives you.

Which one do you need?

The honest decision tree is short.

• You need a result that carries weight somewhere — school placement, an accommodation request, a legal matter, a diagnosis. Only an individual clinical test will do, and you need a licensed psychologist.
• Your child is being assessed by a school. It will probably be a group test first, possibly followed by an individual one. Ask which instrument and when it was last normed.
• You are applying for a job that tests. Expect a timed group test such as the Wonderlic or a similar cognitive ability battery. Familiarisation with the format is legitimate and worth doing.
• You want to join a high-IQ society. Supervised testing through the society, or a prior result from their approved list. No online test qualifies.
• You are curious. A free online test is a perfectly reasonable first read, as long as nobody pretends it is more than that.