Accessibility Statement

Target standard

WCAG 2.2 Level AA. We do not claim full conformance — the limitations further down are the reason — and this page is updated as they close.

What is in place

• Keyboard operation throughout. Every control works from the keyboard, including the whole test engine: answering, flagging, moving between questions and using the question navigator. The focus ring is visible and is never removed.
• A skip link as the first focusable element on every page.
• Native semantics. Answer choices are a real radio group inside a labelled fieldset, so assistive technology announces the question context and arrow keys move between options without custom code.
• Text alternatives generated from the same data as the figure. Instrument dials and block piles are drawn from numbers, and the description is produced from those same numbers — so it cannot drift from the picture. A block-counting description states every column height and where the marked block sits.
• Reduced motion respected, and decorative motion suppressed for everyone during a timed section rather than only for those who request it.
• Light and dark themes as equal implementations, both contrast-checked, with a manual override that beats the system preference in either direction.
• Colour is never the only signal. Correct and incorrect carry text labels as well as colour.
• Large type and generous targets. A 17px base size, and a 44px minimum on anything tappable.
• Responsive down to small screens, with wide tables scrolling inside their own container so the page itself never scrolls sideways.
• Automated checks in continuous integration using axe-core, so a regression fails the build rather than reaching a reader.

Known limitations

• Timed tests are inherently a barrier. Several subtests give ten to twelve seconds per question because the real exam does. We do not currently offer an untimed or extended-time practice mode, and we should. This is the most significant gap on the site, and it disadvantages exactly the candidates who most need to rehearse the material without also fighting the clock.
• Automated testing is not real testing. Our checks catch structural failures. They cannot tell us how the test engine feels with a screen reader across a long session, and this site has not yet been tested by people who use one daily.
• Block Counting is irreducibly visual-spatial. The text alternative describes the pile accurately, but holding a three-dimensional structure in mind from a description is a materially harder task than seeing it. We do not have a good answer for this, and describing it as equivalent would be dishonest.
• Time remaining is not announced continuously. Announcing every second would be intolerable, so we currently announce only when a section is nearly over. We are not confident that is the right balance.
• No formal audit. There has been no independent accessibility audit of this site.

The tension between accessibility and fidelity

Most of this site’s accessibility work is uncomplicated: label things properly, keep contrast high, make everything reachable from a keyboard. Nobody loses when a page is easier to read.

The test engine is different, because part of what it exists to reproduce is difficulty. The real exam gives you ten seconds a question and does not describe its figures. Removing the clock or restructuring the visual items would make our practice more accessible and less honest about what you are walking into — and a candidate who practises under conditions unlike the exam is being set up to fail in the room.

We do not think that tension resolves cleanly, so we are stating our current position rather than pretending it does. Timed mode stays timed because that is the product. But an untimed mode should also exist, for learning the material separately from rehearsing the pressure, and its absence is a gap rather than a principle.

Where a text alternative genuinely conveys the same information — the instrument dials, where bank, pitch and heading are just three numbers — we consider that solved. Where it does not, as with holding a three-dimensional block stack in mind from a description, we say so rather than calling it equivalent.

What we will not do

We will not claim conformance we have not verified, and we will not describe a text alternative as “equivalent” where the underlying task genuinely differs. Overstating accessibility wastes the time of the people it is supposed to help, which is worse than admitting a gap.

Report a barrier

If something here does not work for you, please tell us through the contact page. Include the page, what you were trying to do, and — if you are comfortable sharing it — the browser and assistive technology you were using.

Accessibility reports are treated as defects rather than feature requests, and they take priority over new content. If you would be willing to test the test engine properly with assistive technology, we would particularly like to hear from you: that is the limitation above we most want to close.