How this IQ test works
We think an online IQ test should be honest about what it is and how it works. This page explains where our questions come from, how your score is calculated, and what the test can and cannot tell you.
Where the questions come from
Our items come from published, openly licensed cognitive-ability research. In other words, these are not puzzles we invented on a weekend — the visual items in particular have been studied and validated against established measures. The sixteen questions sample four types of reasoning, four questions each.
Matrix reasoning
Complete a visual pattern in a 3 × 3 grid.
Number and letter series
Identify the rule behind a sequence.
Verbal reasoning
Reason about relationships between words and concepts.
3D rotation
Decide whether two solid figures are the same object, turned.
Using four different item types matters. Intelligence is not one narrow skill, so a good screener samples across reasoning styles rather than testing the same thing sixteen times.
How your score is calculated
Raw correct answers are not your IQ. Here is the path from answers to score.
- 1
You answer
Each of the sixteen items is scored right or wrong. There is no partial credit and no weighting by difficulty — a short test does not have enough items to estimate difficulty reliably.
- 2
We compare to a reference distribution
Your raw score is placed against how a broad sample performs on items of this kind, using the published norming data for the source item sets.
- 3
We convert to the standard scale
The result is mapped onto a scale centred on 100 with a standard deviation of 15 — the same convention mainstream tests use. So 100 is exactly average, and every 15 points is one standard deviation from the middle.
Your percentile then simply says what share of people you would score above.
What this test is, and what it is not
We would rather set expectations than oversell.
| This test is | This test is not |
|---|---|
| A research-backed screener | A clinical diagnosis |
| A fast, solid estimate of reasoning ability | A proctored, professionally administered IQ test |
| Free, private, and instant | A measure of creativity, character, or worth |
Table 1 · Scope of a sixteen-item online screener.
A short online test cannot replace a full assessment given by a qualified psychologist. What it can do, because it is built on validated items, is give you a genuinely useful estimate in about ten minutes. If you need an official score — for a gifted programme or a formal evaluation — see a licensed professional.
How accurate is it?
The most useful number we can give you is about the items, not about us. The Sandia matrix items our visual questions are generated from correlater = .69 with Raven's Progressive Matrices — roughly.93 once corrected for the reliability of the two tests. Raven's is one of the most established measures of abstract reasoning there is.
That figure describes the item set, not this specific sixteen-question form. Ours is shorter, so it carries a wider margin of error. We quote the number because it is published and you can check it — not because we are claiming to inherit it whole.
Why the scale stops at 145
Sixteen items cannot reliably tell 150 from 160. Beyond about three standard deviations there simply are not enough questions left to discriminate, so every additional point would be noise dressed as precision. Our score table therefore runs from 70 to 145 and stops.
You can cheat, and it costs you
The test runs entirely in your browser so that your result never touches a server. The unavoidable consequence is that the answers are in the page, and anyone determined to look can. We think the privacy is worth more than the security — and cheating on a test you took for your own curiosity only deprives you of a real number.
Sources and licences
Both external item sets are used under open licences that require attribution. This is that attribution, and it doubles as the reading list.
- Matrix reasoning
Matzen, L. E., Benz, Z. O., Dixon, K. R., Posey, J., Kroger, J. K., & Speed, A. E. (2010). Recreating Raven’s: Software for systematically generating large numbers of Raven-like matrix problems with normed properties. Behavior Research Methods, 42(2), 525–541.
Sandia Generated Matrix Tool — BSD-3-Clause, © 2010 Sandia CorporationItems rendered as SVG from the documented relation set. - 3D rotation
Ganis, G., & Kievit, R. (2015). A New Set of Three-Dimensional Shapes for Investigating Mental Rotation Processes: Validation Data and Stimulus Set. Journal of Open Psychology Data, 3(1), e3.
CC BY 4.0Stimuli used unmodified apart from splitting each pair and trimming the background. - Number and letter series
Written for this test to the same specifications. The generating rule for every item is published in the item explanations.
— - Verbal reasoning
Written for this test — analogy, categorisation, and syllogistic reasoning items.
— - Classification bands
Wechsler scale convention (mean 100, standard deviation 15), as used for the classification bands throughout this site.
—
One set we deliberately do not use: the International Cognitive Ability Resource (ICAR). Its items are licensed for scientific use only, and its maintainers decline commercial requests in order to protect the integrity of the item material. A free public test is exactly the case that licence exists to prevent, so we built on openly licensed sources instead.
Common questions
Is this a real IQ test?
It is a real screener built on validated items and scored on the standard IQ scale. It is not a clinical assessment, which requires a licensed professional.
How accurate is the score?
For a ten-minute online test it is a solid estimate, because the items are research-validated. A proctored clinical test is still more precise. Expect a few points of movement between attempts.
What scale do you use?
The standard scale: mean of 100, standard deviation of 15.
Where do the questions come from?
The matrix items come from the Sandia Matrices (Matzen et al., 2010, BSD-3-Clause) and the rotation items from the Ganis & Kievit (2015) stimulus set (CC BY 4.0). Series and verbal items were written for this test.
Why does the score stop at 145?
Because sixteen items cannot reliably distinguish scores beyond about three standard deviations from the mean. A test that claimed to place you at 160 off sixteen questions would be overselling itself.
Can I cheat?
Easily. The test runs entirely in your browser, so the answers are in the page. We chose that over sending your result to a server. Cheating only costs you a real number.