Spearman's student, Penrose's assistant, and a thesis written before its test existed
Raven aimed at the two parts of Spearman's g. Eductive ability, making sense of complexity, would be measured by the matrices. Reproductive ability, storing and recalling information, would be measured by a vocabulary test later called the Mill Hill Vocabulary Scale.
Find the rule, then the missing piece
Raven's items use 2×2 to 6×6 layouts with six to eight options. These three were made for this page and get harder, as items do within a Raven set.
Every item had to keep its rank at every level of ability
Raven wanted the order of item difficulty to be the same for weak and strong takers alike, so the set would work like a ruler. To check, he plotted how the chance of passing each item rose with ability. These are what item response theory later called item characteristic curves. When Georg Rasch developed his model in the late 1950s, he tested it on the matrices.
Slide ability up and down. Good items keep their order. Switch on a badly built item and watch it overtake others: Raven would have changed or dropped it.
Cyclical in the booklet, a single climb for the Army
The Standard booklet runs sets A to E, each starting easy and getting harder, so the taker relearns the format five times. For the wartime services, after validation, a 20-minute version put the items in a single order of difficulty.
Bar heights illustrate the ordering principle only; real item difficulties are not shown.
A research tool that every recruit sat
Because many armies gave the Standard matrices to every conscript for decades, James Flynn could compare generations on an identical test. His 1987 paper on “massive IQ gains in 14 nations” put the rise beyond reasonable doubt. Robert Thorndike, and John Raven Jr. with matrices data, had reported gains before him.
Twin studies found matrices scores substantially heritable, while conscript data showed them rising fast across generations. Raven kept saying that both can be true at once: a trait can be strongly inherited within a generation and still shift a great deal between generations.
Sources cited on this page
- [1]Penrose, L. S., & Raven, J. C. (1936). A new series of perceptual tests: Preliminary communication. British Journal of Medical Psychology, 16(2), 97–104.doiUsed for: Hero header, date.
- [2]Raven, J. C. (1936). Mental tests used in genetic studies: The performance of related individuals on tests mainly educative and mainly reproductive. MSc thesis, University of London.via Wikipedia summaryUsed for: Thesis title and approach.
- [3]Raven, J., & Raven, J. (2018). John Carlyle Raven and his legacy.eyeonsociety.co.ukUsed for: Biography underlying the Wikipedia account.
- [4]Wikipedia. John C. Raven.en.wikipedia.orgUsed for: Fieldwork, criteria, Darwin Trust, item curves, Rasch, 1938 forms, 20-minute version, Crichton Royal, heritability point.
- [5]Wikipedia. Raven's Progressive Matrices.en.wikipedia.orgUsed for: Forms and set structure, 1942 armed forces, WOSB, publishers, 1998 parallel forms, Flynn.
- [6]Raven, J. C. (1939). The R.E.C.I. series of perceptual tests: An experimental survey. British Journal of Medical Psychology, 18(1), 16–34.cited in [4]Used for: 1939 paper.
- [7]Raven, J. C. (1941). Standardisation of Progressive Matrices, 1938. British Journal of Medical Psychology, 19(1), 137–150.cited in [4]Used for: 1938 standardization.
- [8]Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 101(2), 171–191.doiUsed for: Conscript data and gains.
- [9]Carpenter, P. A., Just, M. A., & Shell, P. (1990). What one intelligence test measures. Psychological Review, 97(3), 404–431.doiUsed for: Raven's note on homogeneous overlapping problems (via arXiv review).
What changed in this revision
Neither the 1936 paper nor the MSc thesis was read; their content comes through Raven's biographers and reference entries. No Raven items are reproduced, and the three puzzles are original.
- REMOVED“The world's most-used non-verbal IQ test.” No source given.
- CORRECTEDWhy Raven built it: the old page said the Binet was “too verbal for translators” and performance kits “too heavy.” Raven's own reasons were testing in crowded homes and uninterpretable composite scores.
- FLAGGEDColoured Progressive Matrices date: 1938 first versions per Raven's biographers; 1947 elsewhere, including our CPM page.
- ADDED1936 header, fieldwork story, item-curve simulator, cyclical vs single-order view, form cards, timeline 1902–Pearson, Flynn and heritability, and 9 references.
- KEPTThe three original matrix puzzles.
Related historical tests
Raven built the vocabulary test to measure reproductive ability alongside the matrices' eductive ability. The two were meant to be given and read together.
The CPM adds Set Ab between A and B for young children and older adults. Sources disagree on whether its first version appeared in 1938 or 1947.
Raymond Cattell's culture-fair test chased the same goal of measuring reasoning without words. It added series, classification and conditions items to the matrix format.
Penrose's survey began with the Stanford-Binet, given in families' homes. Raven's frustration with it there is the origin of the matrices.
Kohs's coloured blocks also tested reasoning without words but needed a kit and individual testing. Raven's printed booklet did away with both.
In 1942, as Britain began giving every entrant a 20-minute matrices test, the US Army introduced the AGCT. The two show contrasting national answers to mass screening.