Methodology & Accuracy

This page documents how the IQ Check test is built and scored — including the assumptions, so you can judge the result rather than take our word for it. If a claim on this site ever seems to conflict with this page, this page wins.

The test

The test consists of 35 matrix-reasoning items: 3×3 visual grids governed by one or more rules (progression, rotation, counting, Latin-square membership, positional movement, element arithmetic, XOR composition, and combinations), ordered from easy to hard. Six response options accompany each item; distractors are systematic rule violations — the classic error patterns — rather than random fillers. The format follows the Raven-style figure-matrix tradition used across professional fluid-reasoning scales; the items themselves are original to this site.

We use matrix items because fluid reasoning is the single best-loaded component of general intelligence and requires no language, mathematics or cultural knowledge — which makes a web test meaningful across countries and education levels. The time limit is 25 minutes; unanswered items score as incorrect, and there is no guessing penalty.

Items are generated, not stored. Each of the 35 slots is a rule template: its structure — family, step sizes, operand magnitudes, position in the difficulty ramp — is fixed, while the surface (which shapes, starting angles, mask configurations) is generated fresh for every session from a random seed. Two attempts therefore never present the same items, retakes measure reasoning rather than recall, and circulated "answer keys" are useless by construction. Every template is machine-verified at build time to produce exactly six visually distinct options across sampled seeds.

Scoring model

Your raw score (0–35) is converted to an IQ on the standard scale — mean 100, standard deviation 15 — by a linear normal mapping: IQ = 100 + 15 × (raw − M) / SD, where M and SD are the expected median and spread of raw scores for your age group:

Ages 6–7: median 8, SD 5
Ages 8–9: median 11, SD 5.5
Ages 10–11: median 13.5, SD 6
Ages 12–13: median 15.5, SD 6
Ages 14–16: median 17, SD 6
Ages 17+: median 18, SD 6

These norms are provisional calibration assumptions, set from the published difficulty of comparable matrix item types and pitched deliberately conservatively. They are not yet standardized against a large collected sample — the honest status of essentially every online IQ test, stated here explicitly instead of implied away. As anonymized response data accumulates, we will re-norm the mapping and document each revision on this page.

Why scores cap at 145

A 35-item test contains only a few items hard enough to separate the top fraction of a percent, so estimates beyond roughly +3 standard deviations are statistical noise. Sites that report 160s and 170s from short online tests are flattering their users. When you hit our ceiling, the result reads "145 — ceiling reached", meaning: well above 130, and only a professionally administered instrument with extended norms can resolve how far.

Accuracy and error

Typical uncertainty: about ±7 points against a professionally administered full-scale test — wider near the floor and ceiling, and wider for children.
Condition sensitivity: sleep, stress, device, interruptions and motivation routinely move single-session results by several points in either direction.
Practice effects: retaking matrix tests inflates scores; your first calm attempt is the meaningful one.
One component: we measure fluid reasoning only. A professional battery adds verbal comprehension, working memory and processing speed, which can each diverge from the composite by a full band.

What this test is not

It is not a clinical instrument, does not produce diagnoses, and is not accepted by schools, employers or high-IQ societies (none accept any online test). For decisions — gifted placement, IEP evaluation, disability assessment — the applicable tools are individually administered batteries such as the WISC-V (children) or WAIS-5 (adults), given by a qualified psychologist.

Privacy of results

Scoring runs entirely in your browser: your answers and score are computed client-side and are not transmitted to or stored on our servers. There is no account system. See the privacy policy for the complete picture.