Test types
Raven's Progressive Matrices
The best-known non-verbal reasoning test in the world, built on a simple idea: show a pattern with a piece missing and see whether the rule can be recovered.
What the test is
John C. Raven published the Progressive Matrices in 1938. Each item shows a matrix of figures — usually three by three — with the bottom-right cell missing, and a set of candidate answers. The task is to work out the rule governing the matrix and select the piece that completes it.
There are no words, no arithmetic and no general knowledge. The instructions can be given in any language or demonstrated without language at all, which is why the test has been used more widely across cultures than any other cognitive instrument.
Three main versions exist. Standard Progressive Matrices covers the general adult range. Coloured Progressive Matrices is for children and older adults. Advanced Progressive Matrices is built to discriminate at the top of the range, where the standard version runs out of headroom.
The rule families
Matrix items are not arbitrary puzzles. They are generated from a small number of rule types, and knowing the families is most of what “being good at matrices” consists of.
- Constant in a row. A feature stays the same across a row and changes between rows.
- Quantitative progression. A feature increases or decreases by a fixed step — more dots, larger shapes, additional lines.
- Figure addition or subtraction. The third cell is the first combined with or minus the second.
- Distribution of three. Three values of a feature appear once each in every row and every column, like a small Latin square.
- Distribution of two. Two values appear, with the third cell taking whichever is left over — effectively an exclusive-or.
- Rotation or reflection. A figure turns by a fixed amount, or flips, from cell to cell.
The influential analysis by Carpenter, Just and Shell in 1990 showed that item difficulty on the Advanced matrices is largely predicted by how many rules must be held simultaneously and by whether the distribution-of-two rule is involved. That is a finding about working memory as much as about reasoning, and it is part of why matrix performance correlates so strongly with the general factor.
The culture-fair claim
Raven's matrices are routinely described as culture-fair or even culture-free. The first is defensible and the second is not.
What the format removes is genuine and substantial: no language, no vocabulary, no curricular knowledge, no culturally specific content. That makes the test appropriate where those things would confound a result — testing across languages, assessing someone with a language impairment, comparing groups with different schooling in the test language.
What it does not remove is familiarity with abstract classification, comfort with testing situations, experience of multiple-choice conventions and the habit of treating a figure as a bearer of rules rather than a picture. All of these vary systematically across groups and all affect performance.
The decisive evidence is the Flynn effect. Gains through the twentieth century were largest on Raven's matrices — larger than on tests full of vocabulary and general knowledge. A test measuring something independent of environment could not show that pattern.
Non-verbal is not the same as culture-free. Removing language removes one confound, which is useful and worth doing, and leaves the deeper ones untouched.
Where it is used
Raven's matrices appear in clinical assessment as a non-verbal reasoning measure, in research as a quick proxy for fluid ability, in educational screening, and in occupational testing as part of cognitive ability batteries.
Their enduring appeal is practical. The test is quick, can be group-administered, needs no language, scores objectively, and has decades of normative data behind it across many populations.
Almost every matrix item on every online IQ test, including the two on ours, is a descendant of this format. That is worth knowing when a site describes its matrix items as a proprietary innovation.
Practice effects
Matrices are among the most practice-sensitive item types there are, precisely because they are built from a small closed set of rules.
Once you know that distribution-of-three and figure-subtraction exist, you stop searching for a rule and start checking a list. The gain is real, arrives quickly, and mostly appears on the first repeat.
This is why clinicians use alternate forms and minimum retest intervals, and why a matrix score from someone who has taken four online tests is not comparable to a first attempt. It is also why the Advanced version exists: the standard form's ceiling is too low once familiarity is in play.
Questions
Questions people ask
What do Raven's Progressive Matrices measure?
Non-verbal fluid reasoning — the ability to identify a rule governing a visual pattern and apply it. Performance correlates strongly with the general factor extracted from broader test batteries.
Are Raven's matrices culture-free?
No. They remove language and curricular knowledge, which is a real advantage, and they do not remove familiarity with abstract classification or testing conventions. The Flynn effect was largest on exactly this test.
How many types of rule appear in matrix items?
A small set: constant in a row, quantitative progression, figure addition or subtraction, distribution of three, distribution of two, and rotation or reflection. Difficulty rises with how many must be held at once.
Can you get better at matrix items with practice?
Yes, quickly, because the rule families are a closed set. The gain is on the format rather than the underlying ability, which is why clinical retesting uses alternate forms.
What is the difference between Standard and Advanced matrices?
The Advanced version is built to discriminate at the top of the ability range, where the Standard version has too low a ceiling.
Find out where you land
Ten reasoning questions, an instant band and an explanation for every answer.