How FAIQ actually works
Most ability tests ask you to trust a number. We’d rather show our work — where the questions come from, how the score is built, and exactly what it does and doesn’t mean.
What we measure
FAIQ estimates broad cognitive ability across two families. The reasoning core (the g-loaded part) covers abstract matrices, figural analogies, number and figure series, classification, spatial visualization, and deductive logic. The second family — working memory, symbol-pair learning, rule-switching under interference, numerical intuition, and tonal memory — captures the breadth a pattern-only IQ test leaves out. Together they go a step further than a single matrix section.
Every task is culture-free and knowledge-free: no vocabulary, no trivia, no maths facts. Only thinking and perceiving.
Where the questions come from
Items aren’t hand-written or pulled from a fixed bank — they’re generated by code, freshly seeded each time. That means the test is effectively endless and never-repeating, so it’s practice-resistant: memorising answers does nothing.
Each generated item is machine-verified to have exactly one correct answer — a symbolic solver checks every option — and screened so it can’t be cracked by a shortcut (counting, picking the odd colour, or guessing from the answer set alone). If an item fails either check, it’s thrown away and regenerated.
How your score is built
Each subtest becomes a normalized sub-score, and these combine into your FAIQ on the familiar mean-100, SD-15 scale — via a proper area transform (percentile → z → standard score), never a naive rescale of a lopsided distribution.
We always show a 95% confidence band, not a single falsely-precise number, because no one’s true score is a point. The reasoning core drives roughly 75% of the composite, with memory, control, and intuition making up the rest; efficiency is reported as its own axis, never folded into the headline.
your score sits in a band on the curve — the shaded region is roughly where your true score lies
Speed is a bonus, never a penalty
Time matters, but gently and in one direction only. A fast, correct answer can earn a small bonus; a slow, correct answer is never punished — careful thinking on hard problems is exactly what high ability looks like. We report efficiency as a separate axis so a calm, accurate solver is never mistaken for a weak one.
How the norm is earned
A score only means something against a reference group. We earn that norm from real, anonymized data rather than inventing a bell curve. At launch the norm is provisional and labelled as such; as more people take the test it tightens, calibrated with item-response-theory models. We’ll publish the sample size, the model, and the fit.
What it is — and isn't
FAIQ is an honest, broad estimate of cognitive ability. It is engaging, defensible, and yours to keep. It is not a clinical or official IQ, not a diagnosis, and not a verdict on your worth. We will never inflate your score to flatter you, hide it behind a paywall, or bump the paid number above the free one.
Your privacy
Responses are anonymized at the moment they’re recorded: a random session id and coarse device buckets only — never your IP, full browser string, or any fingerprint. No account is needed. We log per-item timing from day one purely to keep the norm clean (filtering rushed guesses). GDPR-clean by design, for everyone.
See your own profile
The free score is genuinely useful on its own. The one-time report adds the depth — and the price is shown before you start.