Accessibility

We treat WCAG 2.1 Level AA as the target we build toward. We are not claiming to meet it. No one has audited us against it, we have not measured ourselves against every success criterion, and the specific failures we do know about are listed further down this page.

A student sets their own accommodations under Settings → Test accommodations. A teacher or school administrator cannot set or lock them on a student's behalf.

• Extended time — standard, +50% (1.5×), or double (2×). The multiplier is applied to every timed module of a full-length SAT, an SAT Checkpoint, and the ACT test and ACT Checkpoint.
• Reader — shows a read-aloud control on the SAT practice screens, and marks it as an enabled accommodation in the SAT test runner, where the control is available to every student. It speaks the question sentence by sentence and, in browsers that support it, highlights the sentence being read. It uses the speech voices built into the student's browser or device; we do not send the text or the audio to our servers or to a speech service of our own choosing. Which voice is used is up to the browser, and some browsers offer voices that run on the browser maker's online service. In a browser with no built-in speech support the control does not appear at all, and there is currently no message explaining why.
• Large print — starts each module at the largest text size.

Reader and large print apply in the SAT test runner and the SAT practice screens. The ACT screens do not have them; only extended time reaches the ACT.

The time multiplier is frozen onto an attempt when the test starts, so changing the setting cannot alter a test already in progress. Where an SAT score report — the results page or the downloadable PDF — shows a percentile, a readiness benchmark, or a college score range for an extended-time result, it also says the test was taken with extended time, because those comparisons come from standard-time populations and are not like-for-like. A shared SAT score card omits the percentile rather than disclosing accommodation status.

Limits worth knowing: there is no way to add time to a module already running. The full-length SAT runner does not display the student's extended-time status during the test, so a student cannot confirm mid-test that it applied. And the ACT score report shows a national percentile and readiness benchmarks without saying the test was taken with extended time.

These describe what our code does, as checked by reading it and, for some items, by automated tests. They are not results of testing with assistive technology, and they are not aspirations. Unless an item says otherwise, it describes the SAT test runner — the screen used for full-length SATs and SAT Checkpoints, and the surface a student spends the most time in.

• Every page declares its language and carries a keyboard-reachable "skip to main content" link.
• Answer choices are marked up as a radio group: each choice has the radio role and a checked state, and a choice crossed out with the eliminator has "(eliminated)" in its label, so the selection and eliminations are in the markup and not conveyed by styling alone. They are buttons given the radio role rather than native radio buttons: each choice is its own Tab stop, and when the single-key shortcuts are switched off the arrow keys do not move between choices. Flagging a question is announced. The question number and total are announced as they change.
• The core test tasks can be done from the keyboard: selecting an answer, moving between questions, flagging, opening the question map, reaching the review screen, and submitting. Not every tool can. Making a passage highlight or note needs a text selection, which in most browsers means a mouse or touch unless the browser's caret-browsing mode is on. We have not checked keyboard access to the embedded Desmos calculator. There are shortcuts (N, P, F, 1–4, and the up and down arrows), listed in the app under More → Keyboard shortcuts.
• In the two-pane Reading and Writing layout the passage scrolls in its own box. That box is a labelled region and its own Tab stop, so a student reading with the keyboard can reach it and scroll it. The arrow keys move an answer selection that already exists; on a question with nothing selected yet they leave the answer alone and scroll the page as they normally would.
• The directions, the keyboard shortcuts list, the question map, the math reference sheet, the pause screen, and the time's-up notice all keep the keyboard inside themselves while they are open. Focus is moved onto each one as it opens and put back where it was when it closes — except the time's-up notice, which does not give focus back to a test that has ended. The pause screen is announced as a dialog with a name and a description, so a screen reader is told the test has paused. The time's-up notice draws above the question map rather than behind it.
• The single-key shortcuts can be switched off in that same Keyboard shortcuts dialog. Stray single-key presses are a known problem for people who use speech input or switch devices, but we have not tested the product with either. The on-screen buttons keep working, and the choice is remembered in that browser on that device. The shortcuts do nothing while the directions, the keyboard shortcuts list, the question map, or the math reference sheet is open, and the answer keys do nothing while the review screen is showing.
• Moving to another question moves keyboard focus to the header just above the new question, so reading continues from the new question rather than from the navigation buttons. That header is a named group, announced as "Question 4 of 27" when focus arrives, so a student is told which question they have landed on.
• When the countdown reaches zero, the test runner freezes the question area, usually within a second: the answer choices and the grid-in box are disabled and the keyboard shortcuts stop working. Our server enforces the same deadline independently, and refuses an answer that arrives more than a minute after it — so a second browser tab, or a page left open, cannot keep changing a scored answer after time is up. The extra minute is deliberate: an answer given before the buzzer that is still retrying on a slow connection still counts.
• If the automatic submit at time-up cannot reach our servers, the student is not left at a silent dead end: the screen says what went wrong, distinguishes a temporary problem from an expired sign-in, and offers a Submit button they can press themselves. A module whose time ran out more than a day ago asks the student whether to submit the answers they gave or start again, instead of submitting on their behalf.
• The countdown clock is deliberately not an announcing region, so it does not read itself out every second. Screen-reader warnings fire at five minutes, one minute, and thirty seconds, and the clock can be hidden.
• Exponents written in caret notation (such as x^2 or x^-1), which many of our Math questions use, are shown as superscripts and given a spoken form — "x squared", "x to the power of negative 1" — in the question text, in the answer choices, and in the written explanation afterwards. This applies in the SAT test runner, SAT practice and answer review, the ACT test, checkpoint, drill and practice screens, the free diagnostic test and its review, the daily challenge on the dashboard, and the drill screens, including a drill built from a test's results page. The read-aloud control speaks those exponents as words too. Exponents written any other way are a gap; see below.
• Charts and diagrams in questions carry a text description built from the figure's data instead of a bare "image". A bar chart, line chart or histogram reads out its values, including when it plots more than one series, and a scatter plot reads out its points, grouped by category where it has them. Where there are too many numbers to listen to, the description gives the shape of each series instead — where it starts, where it ends, and its lowest and highest values, each with the point it occurs at — rather than falling silent. Some figures carry their full data: a box plot gives all five of its values, a number line its range and shaded interval. Measurements and angle labels drawn on a polygon, and the angle labels around a transversal cutting two lines, are read out rather than counted. Question text cannot contain an unlabelled image. How far those descriptions go is still limited; see below.
• In Rapid Fire and in topic practice, an answer marked right or wrong is labelled in words — "Correct", "Incorrect", "Correct answer" — and not only by turning green or red. Rapid Fire replaces the question in about a second, so it also announces the result to a screen reader, naming the right answer when the student got it wrong.
• A high-contrast mode (Settings → Accessibility) raises text contrast, draws borders in the text colour and thickens them on cards and answer choices, underlines links, and adds a strong focus ring. Once turned on, it is switched back on as each page loads, so it stays on across reloads and new visits (the full-page error screen and the email unsubscribe confirmation are the exceptions). Inside the test and answer-review screens it also darkens the text, while those screens keep their light background. Its limit: the setting is stored only in that browser, so it does not follow the student to another device or browser.
• CSS animations and transitions respect the operating system's reduce-motion setting. This is site-wide.

This is the section we would want to read first if we were reviewing us. Each item was confirmed in the code, and we would rather a school know before it buys.

Mathematical notation is only partly readable by a screen reader. This is the most serious gap we have. The spoken form described above covers exponents written in caret notation, and nothing else. Many questions store exponents as ready-made superscript characters instead (such as x² or x⁻¹); those reach a screen reader as raw characters, and we have not tested how any screen reader or read-aloud voice pronounces them. Question text that contains formatting markup — a table, a bold run — skips the conversion as well, so a caret exponent inside it stays a literal caret. Some questions store mathematical expressions in a markup we do not render on the test and practice screens, and those appear as literal symbols. Screens that merely list questions rather than ask them — the saved-questions list, for instance, which shows the opening of each stem — show the stored text as it is, so a caret exponent there is read out as a subtraction. The practical consequence: a student who depends on a screen reader cannot independently take our Math sections, and we do not recommend them for that use. Lessons, flashcards, and the AI tutor render math with a library that includes MathML for assistive technology, but we have not tested those with a screen reader either.

Some charts and graphs do not carry enough data to answer the question. A chart reads out every value only while there are few enough of them: about a dozen for a single series, and about two dozen numbers in total across a multi-series chart. Past that the description falls back to the shape of each series — first, last, lowest and highest — so a question that turns on one of the values in between is not answerable from it. Unless the question's author wrote a description, a graph of a function is described only by the kind of curve it shows — "a parabola", for example — not by its equation or its points, and a dot plot by how many values it has and the range they span, not by how they are distributed. Geometry figures are partly the same: a solid is described with its measurements only when every one of them was stored, so a cylinder with a radius but no height, or a prism missing one of its three dimensions, is read out as just "Cylinder" or "Rectangular prism". Any figure that carries information in the drawing which was never stored as data reaches a screen reader without it.

On the ACT screens, the chosen answer is shown by colour alone. In the ACT test, ACT Checkpoint, ACT drill and ACT practice screens, the answer choices are plain buttons that do not tell a screen reader which one is selected. Our ACT product is not part of what a school would be buying today: it is unlinked and kept out of search while we verify it. We are listing its state here so that nothing on this page reads better than it is.

Keyboard problems in the SAT test runner. The single-key shortcuts are on by default and cannot be remapped. Once a question has an answer selected, the up and down arrow keys move that selection instead of scrolling the page, so a student re-reading a long passage after answering cannot scroll with the arrows unless they switch the shortcuts off.

Focus is only partly managed. Focus moves to the new question when a student moves between questions, but it is not moved at all when the review screen opens or when a new module begins, including when one module hands off to the next: after dismissing the module directions, a student is put back where they were rather than on the first question.

The test and answer-review surfaces are always light. The rest of the site follows the system dark-mode setting and has a theme toggle, but the exam paper, the answer review, and the practice answer cells are deliberately fixed to a light background. A student who needs a dark or inverted display does not get one where they spend the most time. High-contrast mode darkens the text there, but it does not make the background dark.

Smaller gaps. The in-app text-size control stops at 150%, short of the 200% WCAG asks for — browser zoom still works, and is the better tool for now. There is no dyslexia-friendly font, line-spacing, or letter-spacing option. Printable PDF exports are untagged: they carry no document structure and no text alternatives for their charts or diagrams, and are not screen-reader accessible. Many pages, including the SAT test runner, the ACT screens, and the admin and school pages, contain a second "main" landmark nested inside the site's own, which makes landmark navigation noisier than it should be. The embedded Desmos graphing calculator is Desmos's own product: whatever accessibility it has is theirs, we pass it no accessibility configuration, and we have not tested it with a screen reader.

We have no automated accessibility testing. There is no axe or similar rule scanner in our test suite or in our build, and nothing fails a release for an accessibility regression. The SAT test runner has a real, rendered accessibility test suite covering landmarks, the radio-group semantics of answer choices, the timer, keyboard shortcuts and switching them off, what the arrow keys do and do not do, focus between questions, the passage's own keyboard stop, the question map, the grid-in input, the spoken form of math in answer choices, the focus trap on the time's-up notice, and the freeze when time runs out; that suite covers one component. Some other pieces have accessibility tests of their own — the focus trap the dialogs share, the pause screen, the worded right-and-wrong labels in Rapid Fire, the descriptions generated for charts and diagrams, and the math rendering on the practice, drill and dashboard screens — but no other screen has a suite like the runner's.

Two of those files used to assert nothing about the code they named — they checked a copy of the behaviour written inside the test, and would have passed with the real code deleted. Both now drive the real thing, and each case was checked by breaking the code and confirming the test failed.

We have not done systematic testing with JAWS, NVDA, or VoiceOver, and we have not tested with students who use assistive technology. Everything on this page is the result of reading our own code, which is why it is specific — and also why it is not a substitute for an audit.

We do not have a VPAT. We have not had an independent accessibility audit or assessment, and we do not claim conformance with WCAG 2.1 at any level, with Section 508, or with any state accessibility standard. We are a small company and it would be misleading to imply otherwise. If a district needs a completed VPAT or an independent audit as a condition of purchase, we do not have one. Contact us before relying on this page for a purchasing decision.

We have not set a remediation timeline for the gaps above, so we are not publishing one. The math-notation gap is the one we consider most serious.

Write to support@unlimitedtests.com or use our contact form. Tell us the page, what you were trying to do, and the assistive technology and browser you were using, if you know them.

A person reads these messages. We have not agreed a response-time commitment with anyone, so we are not claiming one here. If a student in your school is blocked right now, please say so in the message so we can see that first.

Related: security and privacy.