Prove It

Choose one approved UConn page and review it using WAVE, Chrome Lighthouse, manual testing, and one generative AI assistant.

Your goal is not to collect as many errors as possible. Explore what the tools report, check whether selected findings are actually supported, look for things automation may have missed, and use WCAG evidence to support your conclusions.

Explore → Validate → Manually test → Ask AI → Verify → Cite evidence

No paid tools, additional browser extensions, advanced coding, or remediation implementation are required.

• UConn Student Activities
• UConn Recreation
• UConn Dining Services
• UConn Library

Review one linked page, not an entire website. Record its URL, review date, browser, and relevant page state so your observations can be understood later. These are live pages and may change; none is assigned because it is known to contain a specific problem.

Run WAVE on your selected page. Explore Errors, Contrast Errors, Alerts, Features, Structure, and ARIA where present. You do not need to explain every result.

Record:

• Several results that seem important
• At least one item that appears to require human judgment
• Anything interesting in Structure, Features, or ARIA

Show what you explored beyond the error count. If no alert or error is reported, inspect a feature or structural item for contextual judgment. See Session 14’s WAVE instructions or WAVE Help.

On the same page, open Chrome DevTools → Lighthouse → Accessibility and run the analysis.

Record the Accessibility score, several audit results, anything Lighthouse asks you to inspect manually, and anything useful or confusing about the report. You do not need to find a difference between the two tools.

The Lighthouse accessibility score is not a WCAG conformance score.

See Chrome’s accessibility scoring documentation.

Choose three findings from WAVE, Lighthouse, or both to investigate. Findings can include alerts, features, or passed audits; you are not required to find three failures.

For each, briefly document:

• Tool result: What did the tool report?
• What I observed: Inspect the actual page and reproduce or examine the condition.
• WCAG evidence: Where applicable, give the relevant criterion number, name, conformance level, and a link or citation to the W3C source.
• Conclusion: Supported; Partially supported; or Not supported / requires more context. Explain why.

You do not have to agree with the tool. A well-supported qualification or rejection is acceptable.

Perform at least two manual accessibility checks appropriate to this page. Options include keyboard navigation, visible focus, focus order, headings, landmarks, accessible names, alt text quality in context, zoom/reflow, forms and errors, screen-reader testing, or unexpected interaction behavior.

Did you find anything important that WAVE and Lighthouse did not establish?

Document the methods, steps, and results of both checks, including at least one manual observation. That observation may be a real problem, something requiring more investigation, or a correctly implemented behavior that automation did not explain. Do not manufacture a failure.

The Field Guide testing workflow can help you choose an appropriate check.

Use ChatGPT, Microsoft Copilot, or another comparable assistant you already use as a second opinion. Give it a concise summary of your findings, without private or sensitive information. A free option is sufficient.

Recommended prompt:

I am reviewing this webpage for accessibility. Here are several findings from WAVE, Lighthouse, and my manual testing. For each, tell me whether the evidence appears to support an accessibility issue, identify the relevant WCAG success criterion and level where applicable, and suggest what I should verify before making the claim.

Choose three AI claims or recommendations to check. For each, document:

• What the AI said: A brief excerpt or summary
• How I checked it: W3C/WCAG, direct testing, browser inspection, or course resources
• My conclusion: Supported; Partially supported; or Unsupported, with a brief explanation

If the response contains fewer than three claims or recommendations, ask the assistant to consider additional observations from your review. Keep the follow-up in your AI Use Note. The AI may be correct; you do not need to find a mistake. What matters is verification.

Use at least three relevant WCAG success criteria somewhere in your submission. For each, include its number, name, conformance level, and W3C source. Explain how it relates to what you actually observed; unrelated citations do not support a finding.

Use the WCAG 2.2 Quick Reference to locate criteria and their Understanding pages. You choose the criteria. Relevant evidence can support a correctly implemented behavior or a qualified conclusion as well as a failure.

End with brief answers to:

1. What did WAVE and Lighthouse help you establish?
2. What still required human judgment or manual testing?
3. Did you find anything the automated tools missed or could not establish?
4. How useful was the AI assistant?
5. What can you confidently say about this one page?
6. What can you not claim from this review?

Reviewing one page does not establish accessibility or conformance of an entire UConn website.

A tool report is not a conformance determination.

Screenshots are encouraged where useful. You do not need a screenshot of every result, and there is no prescribed page count.

Include:

• Selected UConn page, including URL and review context
• WAVE notes
• Lighthouse notes
• Three validated automated findings
• At least two manual tests and one manual observation
• Three checked AI claims or recommendations
• At least three relevant WCAG citations, with number, name, level, and W3C source
• Short final reflection
• Brief AI Use Note

AI output is not evidence.

Name your assistant and briefly describe how you used it, including any follow-up prompts. Distinguish its suggestions from your verified conclusions.

• Tool exploration - 2 points

  Meaningfully uses and documents both WAVE and Lighthouse.

• Validate findings - 3 points

  Investigates three automated findings and supports conclusions with appropriate evidence.

• Manual testing - 2 points

  Uses appropriate manual testing and recognizes something automation does not establish.

• AI verification and WCAG evidence - 2 points

  Checks AI claims and uses relevant WCAG criteria accurately.

• Limits, communication, and AI Use Note - 1 points

  Keeps conclusions within the available evidence and communicates clearly.

The official submission deadline and assignment submission are in HuskyCT/Blackboard.