How Our Tests Are Built
Eight cognitive tests, each modelled on the structure of an established instrument and built from our own original items. This page explains which scales we measure on, where our norms come from and what we do not claim.
Established Structures, Original Items
Our tests follow the architecture of the published instruments they are named after — the Stanford-Binet, the Wechsler adult scales, Cattell-Horn-Carroll batteries, the MoCA and the Wonderlic — using our own items. None of them is the licensed instrument, and we do not claim to be. What we take from the literature is the structure: which abilities to sample, how to route and stop, and how to scale and report.
The same rules apply to all of them. Every item is ours, and most are generated fresh for each session, so the answer key never leaves the server. Every score is computed, never invented. And every report says what it is: an estimate with a confidence interval, not a clinical result.
Take the Main IQ Test →Scale and Confidence Intervals
All IQ results use the deviation IQ scale with a mean of 100 and a standard deviation of 15, the reference of the major professional tests; inside the batteries, task scores are scaled scores with a mean of 10 and a standard deviation of 3. A confidence interval is shown beside the score, wider for the Quick versions than for Full, because no single sitting measures anyone exactly.
Where the Norms Come From
Percentiles come from our own completed tests, refitted as more runs come in, and are provisional until each test has enough of them per age group. If you give your age group, your score is compared with your peers. Because online test takers report more education than the adult population, we place 100 four points below our takers' average. We do not have a nationally representative standardisation sample and will not claim one.
Timing and Integrity Checks
Every test is timed: per question in the IQ family, per section in the WAIS-style test and with one overall clock in the others. A question that times out counts as wrong. Runs that are far too fast or score below chance level are flagged instead of scored, and the random seed of each session keeps two people from seeing the same test.
What Each Test Is Modelled On
Each test measures on the scale of the instrument it follows. Below is the structure of each one; the test's own page explains it in more detail.
Main IQ Test
Our main test, built on the Cattell-Horn-Carroll model: five domains — logical reasoning, pattern recognition, spatial, verbal and memory — sampling fluid reasoning, visual processing, crystallised knowledge and working memory. The picture items are generated per session and selected adaptively, with 45 seconds per question. Scoring uses item response theory, so a correct answer on a hard item counts for more than one on an easy item. Quick (15 items), Normal (25) or Full (40), all on the same scale.
Fast & Accurate IQ Test
A fixed-length picture test across the same five domains, with 60 seconds per item: Quick (15 items), Normal (30) or Full (45). Items are weighted by difficulty, the Quick version reports a range rather than a single number, and the Full version adds a score per domain. Norms are provisional until each length has enough completed runs.
Mensa-Style IQ Test
Thirty-five matrix puzzles in 25 minutes, pictures only, in the high-ceiling fluid-reasoning format of Mensa admission tests. The items are built to discriminate in the 120–160 range, and the report says whether your score reaches the top-2% level that Mensa societies usually require. We are not affiliated with Mensa.
Stanford-Binet-Style Test
Follows the structure of the Stanford-Binet Fifth Edition: two routing tasks set your starting level, then testlets of six items with basal and ceiling rules across five factors — fluid reasoning, knowledge, quantitative reasoning, visual-spatial processing and working memory — in a nonverbal and a verbal domain. It reports full-scale, nonverbal, verbal and factor scores. The verbal tasks are available in English only for now; in other languages you take the nonverbal half.
WAIS-Style Test
Nine original tasks in the structure of the Wechsler adult intelligence scales, with practice items, start points and stop rules. Task scores roll up into five index scores — verbal comprehension, visual-spatial, fluid reasoning, working memory and processing speed — and a full-scale IQ estimate. Quick takes three tasks, Normal five and Full all nine.
CHC Cognitive Test
Up to twelve tasks measuring six broad abilities of the Cattell-Horn-Carroll model: fluid reasoning, comprehension-knowledge, visual processing, working memory, learning efficiency and processing speed. The Full version has two tasks per ability, as professional CHC batteries do; Normal has one per ability and Quick has three tasks for one overall score. Tasks start easy and stop by themselves when the items get too hard.
MoCA-Style Cognitive Screening
Modelled on the Montreal Cognitive Assessment: the same kinds of tasks — drawing, tapping, typing and remembering words over several minutes — rebuilt as original online tasks and scored out of 30, with points per area and a memory index. Quick (5 tasks, about 5 minutes), Normal (about 10) or Full (about 20). It is a screening, not a diagnosis, and this online version has not been clinically validated.
Wonderlic-Style Test
The speeded format of the Wonderlic: mixed verbal, numerical, logical and spatial items at roughly 14 seconds each — Quick (30 items in 8 minutes), Normal (40 in 10) or Full (50 in 12). The items are original and generated for every run. The raw score out of 50 is compared with publicly available population and occupational norms.
What We Do Not Claim
No online test replaces an individually administered assessment by a psychologist, and none of our tests gives a clinical diagnosis. Our norms are our own and provisional, we have not published validity studies, and the verbal tasks of the Stanford-Binet-style and CHC tests exist in English only for now. The tests are available in 20 languages, with the picture items identical across languages and the word items translated and reviewed per language.