July 2026

Raven's Progressive Matrices

A grid of abstract shapes with one piece missing. No words, no arithmetic, nothing to have studied. It is the purest widely used measure of fluid reasoning, and the benchmark most other tests are validated against.

The format

You are shown a matrix, usually three by three, of abstract geometric figures. One cell is empty. Below it sit several candidate pieces, and your job is to work out which one belongs.

There is nothing to know. No vocabulary, no arithmetic, no general knowledge. The rules governing the pattern are contained entirely within the item, which is what makes the format unusual.

John C. Raven published the first version in 1938. Nearly ninety years later it remains in wide use, which is a rare thing for a psychological instrument.

What it measures, and why it mattered

Matrices target fluid reasoning specifically: identifying rules and relationships in material you have not encountered before.

That isolation is the point. Verbal and knowledge-based items necessarily reward exposure, education and language background. Strip those out and what remains is much closer to reasoning process alone.

This made Raven’s enormously influential. It became the reference measure for fluid ability, which is why newer reasoning tests are routinely validated by correlating against it rather than against something else.

The versions

Version Intended for
Standard Progressive Matrices General adult and adolescent use, 60 items in five sets
Coloured Progressive Matrices Younger children and older adults
Advanced Progressive Matrices Discriminating among high scorers, where the Standard version ceilings

The existence of the Advanced version illustrates something we cover in IQ percentiles: tests run out of discriminating power at the top, and measuring the upper tail requires purpose-built items rather than simply more of the same ones.

The culture-fair claim, honestly

Raven’s is frequently described as culture fair. Culture reduced would be more accurate.

Removing language and learned content strips out a great deal of cultural loading, and that is a genuine achievement. But performance still depends on familiarity with abstract diagrammatic conventions, on formal schooling, and on having encountered the idea of a timed test at all. Cross-cultural score differences on Raven’s are real and have never been fully explained by ability alone.

The honest framing is that it is substantially less culturally loaded than a vocabulary test, which is not the same as being neutral.

Why this test matters to ours

Our own matrix items are generated from the Sandia Matrices, an openly licensed item set released under a BSD-3 licence by Sandia Corporation. Sandia was built explicitly as a free alternative to Raven’s, and the published validation reports a correlation of r = .69 between the two, rising to approximately .93 once corrected for the reliability of both tests.

That is the accuracy figure we quote, and it is why we can quote one at all. Most free online tests cannot name their item source, let alone point to a published correlation against an established instrument. The full sourcing sits on our methodology page.

The figure describes the item set rather than our specific sixteen-question form, which is shorter and therefore carries a wider margin. We say so rather than borrowing the number wholesale.

You can try the format on our free IQ test, where four of the sixteen items are matrix reasoning.

Key takeaways

  • Raven’s Progressive Matrices asks you to complete an abstract pattern with no language or learned content involved.
  • John C. Raven published it in 1938 and it remains in wide use.
  • It targets fluid reasoning specifically and became the benchmark other reasoning tests validate against.
  • Three versions exist: Coloured, Standard and Advanced, the last for discriminating among high scorers.
  • Culture reduced is more accurate than culture fair. Schooling and diagram familiarity still matter.
  • The Sandia Matrices, which our matrix items come from, correlate r = .69 with Raven’s.

Common questions

What is Raven's Progressive Matrices?

A non-verbal reasoning test in which you complete a pattern in a grid of abstract shapes by selecting the missing piece. It requires no language or learned knowledge.

What does it measure?

Fluid reasoning, the ability to identify rules and relationships in novel material. It is widely treated as one of the best single measures of general reasoning ability.

Who created it?

John C. Raven, who published the first version in 1938.

Is it culture fair?

It reduces cultural loading substantially by removing language and learned content, but no test is fully culture free. Familiarity with abstract diagrams and formal schooling still influences performance.

What are the different versions?

Standard Progressive Matrices for general use, Coloured for children and older adults, and Advanced for discriminating at the upper range.

How many questions does it have?

The Standard version has 60 items arranged in five sets of increasing difficulty.

Is it still used today?

Yes, widely in both research and applied settings, and it remains a common benchmark for validating newer reasoning measures.

How does our test relate to it?

Our matrix items are generated from the Sandia Matrices, an openly licensed item set whose scores correlate r = .69 with Raven’s, about .93 corrected for attenuation.

Can you practise for it?

Practice improves performance on matrix items specifically, which is exactly why item security and proper norming matter. It does not indicate a change in underlying ability.

Is a non-verbal test better than a verbal one?

Not better, different. Non-verbal tests isolate fluid reasoning; verbal tests capture accumulated knowledge. Full batteries sample both.