Accessibility statement
This statement covers the online editions published with Homeroom on this site.
Homeroom builds accessibility into this online edition by construction: the published pages are rendered as semantic HTML (landmarks, a single page heading, a skip link, alt text on every image, a reflowable single-column measure) and are continuously self-audited against a zero-violation automated baseline. This statement is an INTERNAL SELF-ASSESSMENT — it documents what we test and enforce; it does NOT assert WCAG conformance or Section-508 compliance for this edition, and it is not a substitute for an independent audit by a human using assistive technology. We welcome accessibility feedback at the contact below and will work to remove barriers.
How accessibility is built in
| Aspect | Detail |
|---|---|
| What the self-audit checks (continuously) | Semantic structure (landmarks, one page heading, heading order), a skip link, alt-attribute presence on images, accessible link names, page language, and a reflowable measure — gated against a zero-violation automated baseline on every change. |
| What it cannot decide (human + assistive technology only) | Whether alt text is meaningful (not merely present), reading/focus order makes sense, color is never the only signal, error-recovery flows work, and a real screen-reader user can complete a task. Roughly half of WCAG needs a human driving NVDA / JAWS / VoiceOver. |
| What this statement is — and is not | An INTERNAL self-assessment that documents and accelerates accessibility work. It is NOT an independent audit and does NOT assert conformance or certification — those remain a qualified, independent reviewer's act. |
Honest scope of this self-assessment
Accessibility BY CONSTRUCTION: these surfaces are continuously self-audited — a real-browser axe sweep gated against a zero-violation baseline, a structural jsdom audit, and an in-VPC AI reviewer for the non-automatable criteria. This ACR is an INTERNAL SELF-ASSESSMENT that ACCELERATES external validation; it does NOT assert WCAG conformance, Section-508 compliance, or certification. A signed VPAT/ACR and any 'Supports/conforms' representation still require a human manual + assistive-technology audit by a qualified, independent reviewer.
What a clean self-audit does not clear
The following require a human, assistive technology, and an independent reviewer — a green automated self-audit does not discharge them:
- MANUAL + ASSISTIVE-TECHNOLOGY AUDIT — a human driving a screen reader (NVDA / JAWS / VoiceOver) plus keyboard-only must verify meaningful focus order, logical reading sequence, sensible (not merely present) alt text, no keyboard traps across a whole task flow, and screen-reader comprehension. The automated sweep finds the machine-detectable class only (~30–50% of WCAG failures).
- WCAG SUCCESS CRITERIA NO SCANNER CAN DECIDE — e.g. 1.3.2 meaningful sequence, 2.4.3 focus order, 1.1.1 alt-text QUALITY, 1.3.3 sensory characteristics, 1.4.1 use-of-color in situ, 3.3.x error-recovery flows, 1.4.10 reflow on dense data routes. The AI auditor can FLAG these; it cannot DECIDE conformance — that needs human judgement.
- INDEPENDENCE IS IRREDUCIBLE — a self-assessment (even AI-assisted) is BY DEFINITION not an independent audit. Many districts require a third-party firm (TPGi / Deque / Level Access) to observe disabled users with real assistive technology and SIGN a published VPAT/ACR. A 'Supports/conforms' representation to a customer is a counsel decision, not a code output.
- OPERATOR / ENV + 508 OPERATIONAL FACTS — production content (adviser-authored pages, uploaded photos with their alt text, themed palettes) must be re-swept + human-reviewed per real book, because authored content can reintroduce contrast/alt defects the engine cannot pre-empt; and 508 §602/603 needs a monitored accessibility contact + an alternate-format / accommodation process (an operational, human commitment).
- AT-USER FEEDBACK — conformance is not usability. Real assistive-technology users testing real task flows surface barriers no automated pass or single expert review will.
- NAMED HONEST GAPS the audit will still find — first-class placed-photo ALT-TEXT AUTHORING is not built (the canvas derives a name; the printed/edition image alt is not an authored field — a 1.1.1 gap); time-based-media captions/transcripts are unbuilt (1.2.x); tagged-PDF / PDF-UA is not validated. The AI auditor can flag these; it cannot fix the authoring-model gaps.
Report an accessibility barrier
If you encounter a barrier on this edition, please tell us. We monitor this contact and will work to provide the information in an accessible format and remove the barrier.