What are Raven's Progressive Matrices?

Short answer: Raven's Progressive Matrices are nonverbal reasoning tests. Each puzzle shows a pattern of abstract figures with one piece missing, and you choose the piece that completes it. They mainly measure fluid reasoning, the ability to work out rules in a problem you have not seen before, which John C. Raven called "eductive ability" (Raven, Mental Health, 1940). Raven's is a strong single test of reasoning, but it is one kind of puzzle, and one analysis found it shares only about half its variance with general intelligence (Gignac, Intelligence, 2015).

At a glance:

• Measures: reasoning with abstract figures, without words. Many researchers call this fluid intelligence (Brain Sciences, 2020).
• Format: a grid with one piece missing, and six or eight answer options on the Standard version (Child Development, 2025).
• Versions: the Standard (60 items; Raven, 1940), the Coloured (36 items; British Journal of Developmental Psychology, 2025), the Advanced (12 practice items plus 36 test items; Brain and Behavior, 2025), and Pearson's current Raven's 2.
• Origin: John C. Raven, first described with Lionel Penrose in 1936 (Penrose and Raven, British Journal of Medical Psychology, 1936) and published as the Standard Progressive Matrices in 1938 (Bilker and colleagues, Assessment, 2012).
• Scoring: correct answers, converted to a percentile or standard score against an age-based norm group.
• Limits: one item type, scores that have risen over the decades, and a range of error around every score.

What do Raven's Progressive Matrices measure?

In 1940 Raven wrote that "visually presented meaningless figures provide the most satisfactory means of estimating a person's innate eductive ability", and that the Progressive Matrices were "designed to measure the accuracy of eduction" (Raven, Mental Health, 1940). Eduction means drawing out a new relation from what is in front of you. The word "innate" was his 1940 view. Later research shows that scores change with schooling and over time.

Many researchers now call the same skill fluid intelligence. A 2020 paper in Brain Sciences says the Raven tests "are widely considered as the most specifically designed test to measure the g factor and as the purest measure of fluid intelligence". For more on the skill itself, see what fluid intelligence is.

Carpenter, Just and Shell (Psychological Review, 1990) studied people's eye movements and errors as they solved the puzzles. They concluded that "The processes that distinguish among individuals are primarily the ability to induce abstract relations and the ability to dynamically manage a large set of problem-solving goals in working memory." You need to spot the rules and hold several of them in mind at once.

How a Raven's puzzle works

Each item is a design with part of it removed (Raven, 1940). On the Advanced version, each item "consists of a 3 × 3 matrix of figural elements with the bottom right entry missing" (Journal of Intelligence, 2025). You look across the rows and down the columns, work out what changes, and choose the piece that fits. The Standard version asks you to "select the correct figure from six or eight options" (Child Development, 2025), and an Advanced problem comes with "8 response alternatives" (Hayes, Petrov and Sederberg, Intelligence, 2015).

Carpenter and colleagues sorted the rules into five types, as summarized in the 2025 Journal of Intelligence paper:

• Constant in a row: the same value runs along a row but changes down a column. Carpenter found this the easiest rule.
• Pairwise progression: something grows or shrinks step by step, such as size, position or number.
• Figure addition or subtraction: one figure is added to or taken away from another to give the third.
• Distribution of three values: three values of a feature are spread across every row.
• Distribution of two values: two values move through a row and the third is empty. Carpenter found this the hardest rule.

Harder items use harder rules, or more of them at once.

Here is a made-up example, written for this article. It is not a Raven's item. Take a 3 by 3 grid. The top row shows a circle with one dot, then two dots, then three dots. The middle row shows squares that do the same. The bottom row shows a triangle with one dot, a triangle with two dots, and a blank. The shape stays the same along each row and the dots rise by one, so the blank is a triangle with three dots. Real Raven's items are published by Pearson, which restricts how its test materials are reproduced (Pearson permissions page, read 8 October 2026), so none appear here.

The "progressive" part is the order. Raven wrote that "The initial tests in each Set are easy so as to be self evident and these are followed by tests of increasing difficulty" (Raven, 1940).

The versions: Standard, Coloured, Advanced and Raven's 2

• Standard Progressive Matrices (SPM). Raven described it in 1940 as "sixty perceptually presented tests", in five sets of twelve, lettered A to E.
• Coloured Progressive Matrices (CPM). A 2025 paper in the British Journal of Developmental Psychology says "The CPM consists of 36 items, which are divided into three 12‐item sets (A, Ab, and B)". It is used with "young children, individuals with intellectual impairments, and the elderly", according to the same paper.
• Advanced Progressive Matrices (APM). A 12-item practice set, then a 36-item test set, citing the 1998 manual (Brain and Behavior, 2025). It was designed "to spread the scores of the top 20% of the population" (Journal of Intelligence, 2025).
• Raven's 2. Pearson's updated edition, with "Updated and colorized items, design, and norms", given on paper or online, according to Pearson's product page (read 8 October 2026).

Sources disagree on the age range of Raven's 2. Pearson's US product page lists ages 4 to 90. Hogrefe, a German-language distributor of the European edition, lists ages 4 to 69 years 11 months (both read 8 October 2026).

How Raven's is scored

The raw score is the number of items you got right. In a 2020 Brain Sciences study of the Coloured version, "One point was given for each correct answer, and the total score was the sum of the correct answers, with a maximum score of 36".

A raw score is then compared with a norm group, a large sample of people the same age. Pearson says Raven's 2 "Provides a standard score as well as a percentile rank." To see how percentiles line up with the standard IQ scale, see what the average IQ is.

The norm group has to be recent. A 2015 meta-analysis of 271 samples, with data from 1909 to 2013, estimated that scores on fluid tests rose by 0.41 IQ points a year, though the gains "have decreased in more recent decades" (Pietschnig and Voracek, Perspectives on Psychological Science, 2015). Against an old norm table, a person today can score higher than current norms would place them.

Every score carries a range of error. Your result can differ between two tests, or between two sittings of the same test.

How long the test takes

Raven designed the test to be untimed. A 2023 review notes that "the test was explicitly designed to be completed untimed", yet "Raven's matrices are often administered with a time constraint" today (Gonthier, Journal of Intelligence, 2023).

• Standard version: one study found a computerized 60-item version took 17 minutes on average (Williams and McCord, 2006, reported by Bilker and colleagues, Assessment, 2012).
• Advanced version: the usual limit is "40 min for the full 36 items" (Gonthier, 2023).
• Raven's 2: Hogrefe lists about 20 minutes for the short digital form and about 45 minutes for the long form in the European edition (read 8 October 2026). Pearson's US page gives no time.

Is Raven's a good IQ test?

It is a well-studied test of one kind of reasoning, not a full IQ test.

For it: Bilker and colleagues note that the 60 items "are used as an indicator of general intelligence throughout the world" (Assessment, 2012). In their data the full Standard version had a reliability of .96, against .80 and .83 for two 9-item short forms.

Against using it alone: Gignac (Intelligence, 2015) found that Raven's "shared approximately 50% of its variance with g" across several large samples, and concluded that "Raven's is not a particularly remarkable test with respect to g". He advised researchers not to rely on Raven's alone for a valid estimate of g. That disagrees with the "purest measure" view above, and both readings are in the literature.

Matrix tests also differ from one another. In a 2025 Journal of Intelligence study of 100 adults, a matrix-style ICAR test taken on smartphones correlated 0.40 with Raven's Progressive Matrices. The authors call that a moderate to strong correlation. For more on reliability, see are IQ tests accurate?.

Is Raven's culture-fair?

Less than its reputation suggests. Raven's was developed as a "culture-free" and nonverbal test (Brain Sciences, 2020). Pearson now claims less: Raven's 2 "minimizes the impacts of language skills and cultural differences."

Research shows that experience still matters:

• Scores rose over time. Fox and Mitchum (Journal of Experimental Psychology: General, 2013) call Raven's "a seemingly culture-free test that registers the largest Flynn effect". In their archival data, people born around 1940 were less able to work with abstract relations than people born around 1990.
• Schooling raises scores. A 2018 meta-analysis of over 600,000 participants found "beneficial effects of education on cognitive abilities of approximately 1 to 5 IQ points for an additional year of education" (Ritchie and Tucker-Drob, Psychological Science, 2018). It did not look at Raven's alone.

The puzzles use no words, but what you have learned still affects how you do.

Can you practice for Raven's?

Scores do go up on a second try. A 2015 study reports that scores on the Advanced version "increase significantly on repeated testing without any targeted training" (Hayes, Petrov and Sederberg, Intelligence, 2015). In their 35 participants, "one-third of the variance of score gains could be attributed to test-taking strategy alone", and once strategy was accounted for, "the residual score gains were no longer significant". A higher retest score does not show that reasoning ability changed.

Pearson builds the digital Raven's 2 forms from an item bank "for greater security and limitation of practice effects."

Raven's vs matrix reasoning on the Wechsler tests

The Wechsler tests include a matrix puzzle, as one part of a larger test. On the WISC-IV and WAIS-IV, Matrix Reasoning asks you to pick the missing piece of a matrix "from five response options" (PLoS One, 2018). On the children's WISC-V, it feeds the Fluid Reasoning Index (Pearson).

A Raven's test is all matrices. The WAIS-5 takes "45 minutes for the 7-subtest FSIQ", with matrix puzzles as one task among several (Pearson, read 8 October 2026).

What a Raven's score can and cannot tell you

A Raven's score shows how you did on abstract pattern puzzles compared with a norm group, on one day, with a range of error around it. It does not cover vocabulary or knowledge, and it does not diagnose anything. For a decision about school, work or health, see a qualified psychologist.

If you want to try matrix-style puzzles, WhatIQ's test has 24 timed picture puzzles of its own, not Raven's items, and takes about 15 minutes (whatiq.com, read 8 October 2026). It gives an estimate on the standard IQ scale with a 90% range (whatiq.com, read 8 October 2026). It is not a clinical assessment, and that score can differ from one you would get on another test.

Frequently asked questions

Is Raven's Progressive Matrices an IQ test?

It is a reasoning test that is widely used as an indicator of general intelligence (Bilker and colleagues, 2012). It uses only one kind of puzzle, so it is not a full IQ test like the WAIS, which combines several tasks. Gignac (2015) advised against using Raven's alone as a measure of g.

Who developed Raven's Progressive Matrices?

John C. Raven. He and Lionel Penrose described the first version in a 1936 paper in the British Journal of Medical Psychology, and the Standard Progressive Matrices was published in 1938 (Bilker and colleagues, Assessment, 2012). Raven wrote in 1940 that the tests were "designed to measure the accuracy of eduction", meaning the ability to draw out new relations. Pearson publishes the current edition, Raven's 2.

What age range is Raven's for?

It depends on the version and edition. Pearson's US page lists ages 4 to 90 for Raven's 2, while Hogrefe lists 4 to 69 years 11 months for the European edition (both read 8 October 2026). The Coloured version is used with young children and older adults (British Journal of Developmental Psychology, 2025).

How long does the Raven's test take?

It depends on the version. The Advanced version's usual limit is 40 minutes for its 36 items (Gonthier, 2023), and one study found a computerized Standard version took 17 minutes on average (Williams and McCord, 2006, reported by Bilker and colleagues, 2012). Raven designed the test to be untimed.

Can you improve your Raven's score with practice?

Scores often rise on a retest. In one study of the Advanced version, one-third of the variance of score gains came from test-taking strategy alone, and the remaining gains were no longer significant once strategy was accounted for (Hayes, Petrov and Sederberg, 2015). A higher second score does not mean your reasoning ability changed, and any score carries a range of error.

Is a Raven's score the same as an IQ score?

Not exactly. Raven's 2 reports a standard score and a percentile rank against its own norm group (Pearson). Raven's also uses only one kind of puzzle, and Gignac (2015) found it shares about half its variance with general intelligence. Scores from different tests, and from different sittings of the same test, can differ, so read any score as a range.