What AbilityBench is allowed to tell you

A screening questionnaire and a measurement are different objects, and the difference decides what a page may say afterwards. If a page asks you to rate how often you lose your train of thought and adds the ratings up, the number it produces is a summary of your ratings; it moves when your mood moves, and it cannot be checked against anything you did. If a page shows you the word green printed in red ink and times how much longer you take to name the ink, the number is an interval you actually produced.

Every page here is built on the second kind wherever the second kind exists. Where it does not — imagery vividness, sensory sensitivity, the trigger sounds in the misophonia test — the page uses the questionnaire the research uses, says that it is one, and pairs it with a behavioral item where a behavioral item is possible. The aphantasia test is the clearest case: the imagery scale on its own is self-report, so the page adds a task whose answer depends on whether you can hold a picture, and reports the two halves separately rather than blending them into one score.

What this rules out is the sentence that would make several of these pages much more popular. No page tells you that you have ADHD, dyslexia, prosopagnosia, amusia, autism or anything else, and none implies it by writing “your results suggest”. That is not caution about liability. A screening exercise administered by a website, to somebody who chose to take it because they already suspected the answer, has a base rate nothing like the population one, and a confident label from it would be wrong far more often than a visitor would ever find out.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish any condition, on any page here. Only a qualified professional, working with more than a browser, can make that judgment.

This site carries 48 reference figures. 39 of them name an author, a year and a title; the rest are figures that are nobody’s finding — a chance level that is true by construction, a scale’s defined range, an organization’s published admission criterion — and each of those states the basis you would check it against. There is no fourth category. A number cannot be printed beside a score on this site without its provenance traveling with it, because the file the figures live in makes the source a required field with no default and refuses to hand over the value on its own.

The reason for the mechanism rather than a house rule: an invented percentile is the worst defect this domain could ship, precisely because it is the one a visitor believes and repeats. Somebody told they read at the thirtieth percentile will say so, out loud, for years. “Studies show”, “research suggests” and “the average person” are not sources; an author, a year and a title are.

The corollary is the useful half. 43 figures that a page here would want are recorded as gaps rather than supplied, each with the reason and with what the page must do instead:

• 11 — the norm table belongs to a publisher
• 10 — no defensible figure exists
• 6 — collected under conditions a browser cannot reproduce
• 5 — published estimates disagree too widely
• 4 — could not be confirmed well enough to print
• 3 — widely quoted and false
• 2 — it is a formula, and lives in the page code
• 2 — another site measures it

Three of them are worth knowing even if you never take a test here. The eight-second human attention span, quoted everywhere and attributed to a national statistics agency that has never published it, has no primary source and never did — so the attention span test reports how far your own performance fell across a sustained block instead of comparing you to a number nobody can find. Trail-making norms cannot cross into a browser, because in the published administration the examiner interrupts you the moment you touch a wrong circle and the time spent being corrected is part of the score; remove the examiner and you have a different task with the old task’s norms attached. And a situational judgement test has no general scoring key at all, because a real one is scored against the consensus of experienced people in one job at one employer — which is what makes it a valid selection tool and exactly why a generic version cannot mark you right or wrong. That page teaches the format and says out loud that it is not marking you.

Which brings us to the thing this site cannot do, stated plainly rather than buried: no page here prints a percentile, and not one of the 48 reference figures is a mean-and-standard-deviation distribution. Every candidate for one either belongs to a copyrighted instrument or was collected with a pencil, a stopwatch and an examiner across the table, and a percentile computed against the second kind would be arithmetically correct and factually false. So the result panels report a raw score with a reference range beside it, and where a comparison would need a distribution the page says there is none.

The Wonderlic and CCAT item banks, Raven’s actual matrices, the Wechsler subtests, the TAS-20, the Cambridge Face Memory Test, the imagery questionnaire’s published form: all of these are copyrighted, and several are controlled for a reason that has nothing to do with copyright. An item bank whose contents circulate stops measuring anything, because the next candidate has seen the items. A site that leaked them would be damaging the instrument for the people who have to sit it.

So every item on this site was written for it, in the same format as the real thing and never copied from it. The trade is stated on the pages where it matters: a score from the Wonderlic practice test is not comparable to a score from the real Wonderlic, and the results panel says so rather than the footer. What does transfer is the format and the pace — fifty items in twelve minutes is what makes that test hard, not the items — and the pace is the thing a candidate can actually rehearse.

Wonderlic, Predictive Index, Hogan, Caliper, Korn Ferry, SHL, CCAT, Mensa, Watson-Glaser and VARK are their owners’ marks. They appear here only to name the assessment a page prepares for, never in the site name, the logo or the front of a page title, and each of those pages carries an explicit line saying there is no affiliation, no endorsement and no reproduction of the published test. If you are here for the unbranded formats instead, the aptitude test practice page samples all of them at once.

Reaction time is a real measurement in a browser and a much smaller one than the numbers usually printed alongside it suggest. A trial here is timed from the frame that painted the stimulus to the timestamp the browser attached to your keypress, both read off the same monotonic clock, so the subtraction is legal. Two popular alternatives are not: a timer callback fires when the browser gets round to it and measures the timer queue rather than the screen, and a frame callback runs before its own frame is painted, so an onset taken there is a whole frame early — about 16 ms, in the direction that flatters the visitor.

What none of that removes is the display pipeline. The compositor, the cable and the panel each add latency no web API exposes, and frame quantization alone is around 16 ms on a 60 Hz screen before you are involved at all. Every page here that prints milliseconds states that floor, quotes no precision the platform cannot support, and leads with the within-visitor contrast rather than the absolute number: the Stroop test reports the difference between your incongruent and congruent blocks, because that difference cancels your monitor, your keyboard and how tired you are, while the raw reaction time is dominated by all three.

Two rules follow from the same place. A response faster than about 150 ms after the stimulus is an anticipation rather than a reaction — the key was already moving — and it is discarded with the rule stated. And a trial during which the tab went to the background is void rather than slow: browsers throttle animation frames in a hidden tab to roughly one a second, so a run that kept counting would record a stimulus you never saw and a forty-second response to it. Every page here drops those trials, says how many it dropped and why, and where it can, runs them again.

Every number on this site is worked out by JavaScript in the tab you are reading it in. Your answers, your response times and your score are never uploaded, logged or kept — there is no server-side code here to keep them on, because the pages are static files and there is no API for a test to call. That is also why the tests keep working after you disconnect from the network.

This site asks your browser for nothing. No microphone, no camera, no location, no permission prompt of any kind: the capture layer was deliberately left out, and a page that would otherwise need to hear you — a spoken fluency task, a sung pitch-matching task — takes typed or tapped input instead and says why on the page. A result you want to keep is copied as plain text or carried in the address bar, and what travels is the handful of numbers the page printed plus the seed that rebuilds the run, never the responses.

And no result is ever held back. There is no registration, no interstitial, no partial score with the rest behind a form and no upsell. The corner of the internet these queries live in is built on taking someone through forty questions and then asking for nineteen dollars or a mailing address before showing them their own answer; being the site that simply shows the number is the whole of the position here.

This describes what AbilityBench does. It says nothing about what a browser extension, a corporate proxy or the machine itself might see — only that this site adds nothing to that list. The pages carry advertising slots and an analytics tag, which are the ordinary third parties of a free site and are not given anything you answered.

One domain for 98 tests is deliberate. Each of these questions is too small to carry a domain, a deploy and a search ramp of its own, and together they compound: the links between them pass authority around, one name accumulates trust, and somebody who came to find out whether they have absolute pitch finds the interval training they were going to want next. The cross-links at the foot of every page are that structure made visible.

Underneath, five shared modules do the work that must not be decided ninety-eight different ways: the clock and the frame loop, Web Audio, the statistics, the trial runner that owns what happens between “start” and “here is a number”, and the file the reference figures live in. A page supplies the stimulus and the copy; it does not get to invent its own definition of a valid trial. Pages are generated ahead of time and served as static files, so what a search engine or an AI crawler reads is the same text you are reading now, with the test itself as the only part that needs JavaScript.

Sources are printed on the pages that use them, in full. As an example of the format, the reading-rate figure this site quotes is Brysbaert (2019), How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language.