Most online IQ tests ask ten or fifteen questions of one kind and hand you a number. That is enough for a first impression, but it measures a narrow slice of thinking and carries a wide margin of error. Prism is our answer to that problem: a sixty-question, sixty-minute assessment that measures three different kinds of reasoning and reports each of them separately, together with a single Full Scale IQ. It is free while we collect the results that calibrate its norms.
This article explains what is inside Prism, why it is built the way it is, and how it turns your answers into a score you can trust more than a short quiz.
Why a longer test gives a better answer
Every test score has measurement error. With ten questions, one lucky guess or one careless slip moves the result by several points. With sixty questions that are calibrated by difficulty, the same slip barely matters, and the estimate settles much closer to your real ability.
Length is only half of it. Professional batteries such as the WAIS-IV do not rely on one kind of puzzle. They combine several index scores, including verbal comprehension, perceptual reasoning, working memory and processing speed, because people are rarely equally strong in every area. A person can be quick with spatial problems and slower with words, or the reverse. Our article on verbal, non-verbal and culture-fair tests explains these differences in more detail. Prism follows the same idea in a form that works in a browser.
Three sections, thirteen puzzle types
Prism is split into three sections of twenty questions. Within each section the puzzle type changes every few questions, and every type starts easy and ends hard. That keeps you from getting used to one format and measures reasoning rather than routine.
| Section | Puzzle types | What it measures |
|---|---|---|
| Perception | Twin Fields, Turntable, Signal, Fold, Echo | Visual search, mental rotation, rule finding in motion, spatial visualization, working memory |
| Quantity | Machines, Orbits, Towers, Balance | Inductive reasoning with numbers, relational structure, multi-step deduction |
| Language | Analogies, Outlier, Deduction, Sequences | Word relations, categories, formal logic, ordered meaning |
Perception: puzzles you solve with your eyes and hands
The first section is fully interactive and uses no words or numbers at all.
- Twin Fields shows two small night towns that look identical. They are not. You tap every difference you can find, and each tap is checked on our server, so the positions of the differences are never sent to your browser.
- Turntable is a modern version of the mental rotation task introduced by Roger Shepard and Jacqueline Metzler in 1971. You see a 3D object made of cubes and choose which of four options is the same object turned. The impostors are mirror images or slightly different builds.
- Signal plays four frames of shapes moving across a grid. Each shape follows its own rule, such as bouncing off walls, turning around the center or following another shape. You place every shape where it will be in frame five.
- Fold is based on the paper folding tests that psychologists have used for decades to measure spatial visualization. A sheet folds in 3D, a punch goes through every layer, and you mark where the holes appear once the sheet is opened again.
- Echo is our take on the Corsi block-tapping task from 1972, a classic measure of visual working memory. Tiles light up in sequence and you repeat them, sometimes in reverse order.
Some of these puzzles also look at speed. In clinical batteries, a few subtests reward fast and correct work, and Prism does the same in a small way. Accuracy always counts for far more than speed.
Quantity: number patterns written from scratch
The number section uses patterns that we wrote for Prism and that are not published anywhere else. Machines apply a hidden rule to every input. Orbits place numbers on rings where the rule runs around, across or between rings. Towers build each block from the two beneath it. Balance asks you to work out weights from scales where identical shapes always weigh the same.
None of this requires formulas or school mathematics beyond basic arithmetic. The hardest questions need several careful steps, but each has exactly one answer. You type your answer on a keypad designed for speed, and the empty square in the puzzle fills in as you type.
Language: reasoning in the language you think in
The third section measures verbal reasoning, and you choose its language right before it starts. Prism currently supports twenty languages, including English, German, French, Spanish, Italian, Portuguese, Dutch, Swedish, Norwegian, Danish, Finnish, Icelandic, Polish, Turkish, Russian, Arabic, Hindi, Chinese, Japanese and Korean.
Translating a verbal test is harder than it sounds. A question that works in English can give the answer away in another language. Our first version asked what a constellation is made of, and in German, Dutch, Chinese and several other languages the word for constellation contains the word for star. We replaced that question everywhere. Each language version was checked item by item so that every question keeps exactly one correct answer.
Typed answers are judged by meaning, not by formatting. Capital letters, accents, extra spaces, articles such as "the" or "der", script variants in Japanese and Chinese, and small typing slips in longer words are all forgiven.
How Prism turns answers into an IQ score
Prism does not simply count correct answers. It uses an item response model of the Rasch family, the approach behind many modern standardized tests.
- Every question has a difficulty. It starts from the difficulty we designed and moves toward what the data shows as more people answer it.
- Your ability is the value that best explains your pattern of answers. Solving a hard question tells the model more than solving an easy one, and a lucky guess on one hard question does not outweigh a consistent pattern.
- Every score comes with a confidence range. Alongside your Full Scale IQ you see a 90 percent range, the span in which your score would most likely fall if you took Prism again under the same conditions.
- Scores are compared with real people. The final step places your result on the standard IQ scale, where 100 is the average of test takers and 15 points is one standard deviation.
If you want the background on percentiles and standard deviations, our guide to how IQ tests are scored covers it step by step.
Fair conditions for everyone
A score only means something if everyone takes the test under the same rules.
- The sixty-minute clock runs on our server, not in your browser, and it keeps running if you close the tab.
- Questions come one at a time and every answer is final. You can skip a question, but you cannot go back to it.
- Many puzzles are generated individually for each session, so answers shared online do not help the next person.
- No account is needed. Your session is tied to the browser you started it in.
What you get at the end
Your result appears the moment you finish. It includes your Full Scale IQ with its confidence range, three index scores for perceptual, quantitative and verbal reasoning, a short profile of your strongest band, your pace compared with other test takers, and a breakdown of every puzzle type question by question.
You can download the complete result as a four-page PDF report, and create a certificate with your name. Each certificate carries a unique number that anyone can check on our verification page. Your name is stored encrypted and can only be read through your private certificate link.
Why Prism is free right now
Norms are only as good as the sample behind them. Every completed assessment refines the difficulty of each question and the reference distribution that turns ability into an IQ score. For the first results, Prism relies on carefully estimated starting values, and early scores show a wider confidence range. After the first results come in, the baseline is set from real data and the norms keep updating as more people take the test.
That is why Prism is free while we build its norms. Taking it now gives you a full result at no cost and helps make every future score more precise.
How to get your most accurate result
- Set aside a full hour in a quiet place. The clock does not stop.
- Answer every question. A skipped question counts the same as a wrong one, so a considered guess never hurts.
- Do not rush the early questions. They are easy on purpose, and careless mistakes there cost more than they seem to.
- Use one device from start to finish. A computer and a phone both work well.
Like every online assessment taken without supervision, Prism gives an estimate of cognitive ability, not a clinical diagnosis. Our article on validity and reliability of online IQ tests explains what that means in practice.
When you are ready, take Prism for free. One hour, three kinds of thinking, and a result that shows not just how high you score, but where your strengths lie.