Test types

Culture fair IQ tests

Tests built to minimise the role of language and learned knowledge. They solve a real problem, and the name promises more than any test has delivered.

The problem they were built to solve

Early intelligence tests were full of content that assumed a particular upbringing. Items referenced tennis courts, commodores and household objects familiar to some children and not others. Scores from those tests confounded reasoning with exposure.

Raymond Cattell's response, published as the Culture Fair Intelligence Test, was to strip the content down to abstract figures: series, classification, matrices and topological relations, with no words and no cultural referents.

The same principle underlies Raven's Progressive Matrices and the modern non-verbal batteries such as the TONI, NNAT and UNIT.

The 24-point standard deviation

Cattell's test uses a standard deviation of 24 rather than the 15 used by the Wechsler scales, and this single fact is responsible for a remarkable amount of confusion about high IQ scores.

So a reported score of 148 is the 98th percentile on Cattell and roughly the 99.9th on Wechsler — very different claims attached to the same number. Mensa's threshold is the 98th percentile, which is why the organisation's accepted-score tables list different numbers for different instruments.

Whenever you see a high IQ figure quoted without a scale, this is the first thing to check.

What culture-fair does not mean

Removing language and curricular content removes one layer of confound and leaves several others intact.

Abstract classification is learned. Treating a figure as a bearer of rules rather than a picture is a habit acquired through schooling, and it varies enormously across educational systems.

Test-taking conventions are learned. Multiple choice, filling an answer sheet, working under observation, guessing rather than leaving blank — all of this is culturally specific and all of it affects scores.

The Flynn effect settles the argument. Twentieth-century gains were largest on exactly these tests, larger than on tests full of vocabulary and general knowledge. A measure independent of environment could not behave that way.

The honest description is culture-reduced, and some publishers now use that term.

Where they are genuinely the right tool

The limits above do not make these tests less useful. They make the use case narrower and clearer.

• Testing someone whose first language differs from the test language.
• Assessing a person with a speech, language or hearing impairment.
• Cross-national research where a common instrument is needed.
• Screening large groups quickly, since non-verbal tests score mechanically.
• Measuring fluid reasoning specifically, separated from accumulated knowledge.

What they cannot do is produce a score free of environmental influence, and the practical consequence is that a non-verbal result should be interpreted alongside information about schooling and test familiarity rather than instead of it.