Accessibility Quick Reference

A plain-language desk guide · Target: WCAG 2.2 AA, the standard our DHCS contracts are held to
Everything from today: slides, tools, demos lbohn12.github.io/remediation-workshop

Start here: the words that matter

Screen reader: software that reads the screen aloud for blind / low-vision users. NVDA is the free Windows one; Mac has VoiceOver. If it can't make sense of your file, the file isn't accessible.

Tag: an invisible label that tells a screen reader what each thing is: a heading, a list, a table, an image. A PDF with no tags is just a picture of text. Getting tags right is most of the job.

Real vs. fake heading: a real heading uses an actual Heading style; a fake one is just big bold text. Screen readers only see real ones. Bold ≠ heading.

Alt text: a short written description of an image so a screen reader can say what it shows. Decorative images get empty alt text so they're skipped.

Contrast ratio: how different two colors are in brightness (like 4.5:1; higher = easier to read). A tool gives you the number; you never do the math.

The manual pass: the part no checker can do: Tab through with the keyboard, then listen with a screen reader. A clean automated scan is necessary but not enough.

⭐ Checking color contrast

Acrobat's checker does NOT test contrast; it says “needs manual check.” A clean report ≠ contrast passed.

You have…Use
The colors / hexWebAIM Contrast Checker: paste two colors, get Pass/Fail
Only the PDFColour Contrast Analyser (CCA) eyedropper; click the text, then the background; zoom ~200%, sample a thick letter

AA: normal text 4.5:1 · large (18pt / 14pt bold) 3:1 · buttons & icons 3:1.

UCLA colors: Gold #FFD100 on white ≈1.5:1 never for text · Blue #2774AE ≈5:1 passes · body text Gray-80 #333, not pure black.

Word / PowerPoint → accessible PDF

The easiest accessible PDF is one you built right in the source, so the tags carry over.

  1. Use real heading styles, alt text, link text that says where it goes, real tables with a header row.
  2. Run the built-in checker: Review → Check Accessibility.
  3. Export: Save As → PDF → Options → “Document structure tags for accessibility.”
  4. Never “Print to PDF”: it throws away the tags.
PowerPoint's one weird trap: reading order lives in the Selection Pane (Alt+F10) and reads bottom to top: the bottom item is read first. Give every slide a unique title, too.

⭐ Complex tables: the one that bites us

Merged cells are 90% of the pain in our reports. A screen reader reads a number together with the headers that own it; merge a header across two columns and it can't tell which one owns which number. The table still looks right; that's why it slips through.

Instead of…Do this
Two header rows, “Q1” merged across two columnsFlatten to one header row: “Q1 referrals,” “Q1 completed”
One clever table with merges everywhereSplit it: several simple tables, one per quarter, each with its own caption
A table that runs over a page breakTable Properties → Row → “Repeat as header row”
A picture of a tableNever. Can't be read, searched, zoomed, or copied; this is the unfixable version
Fighting a table that won't behaveShip the accessible Excel alongside the PDF · or call DCP

Run Word's checker before you export (Review → Check Accessibility); it catches merged cells and a missing header row. It can't tell you the table reads correctly; that part is a person.

Do I have to screen-read every PDF? No. The manual pass is over a sample, not every file. Our reports come off a template; validate the pattern once, then check what changed. In practice: arrow across one row of one table and confirm the number announces its header.

See it: the Complex tables page on the site.

The 60-second manual test

Keyboard only; put the mouse down:

Zoom 200% (Ctrl/Cmd +): text reflows, no side-scroll, nothing cut off.

Screen reader (NVDA / VoiceOver): everything read in page order; what you hear matches what you see; headings and buttons announced as such.

NVDA keys: read Caps+↓ · stop Ctrl · headings list Caps+F7 · next heading H

Scanned-PDF check (do first): try to highlight the text. Won't highlight = it's a picture (“page appears empty”) → run OCR in Acrobat first.

“Is it a real heading?”; 5-sec test

Fix in this order

  1. Interactive forms people have to submit
  2. High-traffic pages + shared header / menu / footer templates (fix once, fixes many)
  3. Documents people are required to use
  4. The rest; truly archived / unused: consider removing it

Other content types, briefly

Tools: and what each is

All recommended by UCLA; nothing here is unapproved software.

ToolWhat it is · cost
SiteimproveSite-wide web scanner · free to UCLA. Trust it for pages, not PDFs.
WAVEBrowser add-on, flags issues on a page · free
axe DevToolsDeeper page testing, for developers · free
CCAEyedropper contrast on anything on screen · free
WebAIM CheckerPaste two colors → Pass/Fail · free
Acrobat ProTag PDFs, reading order, OCR · free to UCLA staff
NVDAWindows screen reader (Mac: VoiceOver) · free
AllyIn Bruin Learn; scores course files · free

The 3 answers people ask for

Contrast in a PDF? → CCA eyedropper (Acrobat doesn't check it).

Word / PPT → PDF? → Save As, tick “Document structure tags,” never Print to PDF.

What reads first on a slide? → the bottom item in the Selection Pane.