Raven's Progressive Matrices is one of the most widely used IQ tests in the world and arguably the cleanest measure of fluid intelligence ever designed. It's a non-verbal, culture-fair pattern-completion test — you see a 3×3 grid with one cell missing, and you pick the answer that completes the pattern. This guide explains how the test works, why psychologists trust it more than most other IQ measures, what it actually measures (and what it doesn't), the three main versions used today, and how to interpret your score.
What Raven's Progressive Matrices Is
The test was developed by John C. Raven in 1936 as part of his doctoral work. Raven wanted a measure of cognitive ability that didn't depend on language, culture, education, or prior knowledge. The format he settled on — incomplete visual patterns with multiple-choice completions — became one of the most enduring contributions to psychometrics.
Each item shows a 3×3 (sometimes 2×2) matrix of geometric shapes following some underlying logical pattern, with one cell empty. You pick the option from a set of choices below that completes the pattern. The items get progressively harder — early items are obvious; late items require holding multiple abstract rules in mind simultaneously.
The test's design is deliberately minimal: no words, no real-world objects, no cultural references. Theoretically, an isolated tribe with no written language could take it and produce a meaningful score relative to anyone else. In practice, exposure to the format and Western abstract-puzzle conventions still matters somewhat — but far less than for any verbal IQ test.
What Raven's Measures: Fluid Intelligence
Psychologists distinguish two broad components of cognitive ability:
- Crystallized intelligence (Gc): accumulated knowledge, vocabulary, learned reasoning patterns. Grows through life until quite late in old age.
- Fluid intelligence (Gf): raw reasoning ability on novel problems with no prior preparation. Peaks in the late teens to mid-20s and slowly declines from there.
Raven's Matrices is one of the purest measures of fluid intelligence ever designed. There's literally nothing to memorise; the only way to score well is to extract abstract rules in real time from material you've never seen before. This is why psychologists trust it more than most other tests when they want to measure raw cognitive capability stripped of education, culture, and accumulated knowledge.
Raven's also correlates extremely well with the g factor — Spearman's general intelligence factor that emerges from any battery of cognitive tests. In factor-analytic studies, Raven's typically loads on g at 0.7-0.8 — among the highest of any single test. That's why a Raven's score gives a remarkably accurate read on overall cognitive ability for a test that takes 30-60 minutes.
The Three Main Versions
Raven's has been published in three main forms, each targeting a different ability range:
Standard Progressive Matrices (SPM)
The original, designed for ages 6 to adulthood across the full ability range. Contains 60 items in five sets of 12, each set getting progressively harder. Most commonly administered as a 40-60 minute timed test. The standard version most people mean when they say "the Raven's test."
Coloured Progressive Matrices (CPM)
An easier version designed for children (ages 5-11), older adults, or anyone with cognitive impairment. Contains 36 items, many with coloured backgrounds to maintain visual interest for younger test-takers. Used heavily in pediatric and gerontological assessment.
Advanced Progressive Matrices (APM)
The hardest version, designed to discriminate among high-ability adults where the standard version would be too easy. Two sets of 12 (Set I) and 36 (Set II) items, with the harder Set II commonly used alone for gifted assessment. APM is the version typically used in Mensa-style cognitive screening and research on high cognitive ability. If you've heard of a "hard IQ test that's just patterns," this is probably what was meant.
Why Psychologists Trust Raven's
A few reasons Raven's keeps showing up in serious research and clinical assessment:
It correlates strongly with g. A single Raven's score predicts performance on a full IQ battery at r ~ 0.7 — meaning you can replicate most of what a multi-hour Wechsler tells you in under an hour.
It's relatively culture-fair. The format eliminates obvious cultural-knowledge confounds. Cross-cultural differences on Raven's are smaller than on verbal IQ tests, though not zero.
It's hard to coach. Because the format is non-verbal and the rules are different on every item, practice effects are smaller than on tests like the SAT. You can improve a bit by getting used to the format, but you can't memorise your way to a high score.
It works across the lifespan. The three versions cover ages 5 to 90+, making longitudinal studies possible.
It works under time pressure. The processing-speed component of fluid intelligence shows up clearly in timed Raven's administration, which is one reason it's used in cognitive-decline research — slowing on Raven's is one of the early markers of certain dementias.
What Raven's Doesn't Measure
The narrowness that makes Raven's a clean fluid-intelligence measure also makes it incomplete:
- Verbal ability. Two people with the same Raven's score can have wildly different vocabulary, reading comprehension, and verbal reasoning. This is why a full IQ battery (like the WAIS) includes multiple separate subtests.
- Crystallized knowledge. Doctors, lawyers, and historians draw heavily on accumulated knowledge that Raven's deliberately strips out. Past-knowledge expertise can compensate for lower fluid intelligence in many real-world tasks.
- Practical intelligence. Reading people, navigating ambiguous situations, making good real-world decisions — none of this shows up on Raven's. Robert Sternberg's "successful intelligence" framework explicitly criticises Raven's for measuring only one slice of cognitive ability.
- Creative thinking. Raven's tests convergent reasoning (one correct answer per item). It says nothing about divergent thinking (generating multiple novel solutions).
- Emotional and social processing. EQ and Raven's are essentially uncorrelated.
The practical implication: a high Raven's score is strong evidence of high fluid intelligence, but it doesn't tell you whether the person is wise, articulate, socially skilled, or knowledgeable. Those have to be measured separately.
How to Interpret a Score
Raven's scores are usually reported either as raw scores (number correct out of 60 for SPM) or as percentile ranks within an age-matched norm sample. The percentile is what matters for comparison:
- 50th percentile = average for your age group, roughly corresponding to IQ 100
- 84th percentile = one standard deviation above average, ~IQ 115
- 98th percentile = two standard deviations above, ~IQ 130 (the conventional "gifted" threshold)
- 99.5th percentile = top 0.5%, ~IQ 140+
Two important caveats:
Age matters a lot. Fluid intelligence peaks around 20-25 and declines steadily after that. A raw Raven's score of 50/60 might place you at the 80th percentile at age 25 and the 95th percentile at age 65 — same score, different relative standing because your peers have lost more ground.
Confidence intervals are real. Even professionally administered Raven's has a measurement error of about ±5 IQ points. Two-test administrations on the same person can differ by that much purely from noise. Don't over-interpret a single number.
The Flynn Effect on Raven's
One of James Flynn's most striking findings was that scores on Raven's Matrices have risen faster than scores on almost any other IQ test over the 20th century. Average performance rose roughly 5-7 points per decade on Raven's, compared to ~3 points per decade for general IQ. The leading explanation: modern environments train abstract pattern recognition much more heavily than premodern ones, and Raven's is precisely the test that rewards that habit.
The implication: if you take a modern Raven's test and score at the 80th percentile, you're scoring better in absolute terms than someone at the 80th percentile of the 1950 norming would have. The "average" Raven's performance has shifted dramatically.
Where to Take Raven's Today
Official Raven's is administered by qualified psychologists and requires a fee. Several free unofficial versions exist online with varying quality. The honest considerations:
- Free online "Raven's-style" tests can give a directional read, but their norming is often weak and their scores should not be trusted as clinically valid
- The genuine Raven's APM is paywalled and proctored — if you need a defensible high-ability score (Mensa, research enrollment), you need the real version
- For a free, scientifically structured measure of your reasoning across multiple subscales (not just visual patterns), our free IQ test takes 20 questions and gives an instant breakdown across numerical, verbal, logical, and pattern-recognition reasoning
Frequently Asked Questions
What is Raven's Progressive Matrices?
A non-verbal IQ test that asks you to complete visual patterns by selecting the missing piece from a set of options. Designed by John Raven in 1936 to measure fluid intelligence with minimal cultural and educational confound.
How long does Raven's take?
The Standard version (SPM) typically takes 40-60 minutes for 60 items. The Advanced version (APM) is similar. The Coloured version (CPM) for children is shorter, about 15-30 minutes.
What's a good Raven's score?
It depends on which version and what comparison group. Roughly: anything above the 84th percentile (about 50/60 on SPM for adults) is one standard deviation above average — strong. Above the 98th percentile (~57/60) is gifted-range.
Can you practise for Raven's?
Slightly. Practice effects on Raven's are smaller than on verbal IQ tests, but familiarity with the format does help. Coaching doesn't transfer well — there's no content to memorise. The largest gains come from a few hours of exposure, then plateau quickly.
Is Raven's the best IQ test?
It's the best single measure of fluid intelligence and one of the strongest correlates of overall g. But it doesn't measure verbal ability, crystallised knowledge, working memory in language tasks, or practical intelligence. A full IQ battery (like WAIS) gives a richer picture; Raven's gives a remarkably accurate quick read.