Project purpose

Build two research assets that brands can cite:

1. an Accessibility & Invisible Disability Experience Index (ADX) for Canadian brand sites/apps;
2. an AI-Equity Audit for marketing workflows and outputs.

Business objectives (what success looks like for Le Code Switch)

• Publishable findings: Produce a short report (8–10 pages) and 1-page executive summary that we can release publicly and use in sales conversations.
• Actionable scoring: Create an ADX rubric with clear scoring rules and pilot scores for 10–15 brands; target Cohen’s κ ≥ 0.70 for inter-rater reliability.
• AI risk view: Deliver an AI-Equity checklist and complete 6–8 sample audits across text, image, or video content.
• Market signals: Generate 5–7 evidence-backed insights and 3–5 recommendations per asset that brands can implement within 90 days.
• Lead generation: Package outputs as a gated download + “book a 30-min debrief” CTA to drive qualified leads.

Student learning objectives (what participants will practice and demonstrate)

• Research design: Plan and execute mixed-methods research (survey + audits + light qual).
• Accessibility auditing: Apply WCAG 2.2 and hands-on assistive-tech testing (keyboard nav, screen readers, captions/alt).
• AI equity assessment: Evaluate GenAI outputs and workflows for bias, representation, safety, and accessibility signals.
• Data literacy: Clean data, document a codebook, run descriptive stats and basic tests, and visualize results clearly.
• Reliability & rigor: Calculate inter-rater reliability; articulate assumptions and limitations.
• Communication: Translate findings into plain-language insights, charts, and social-ready slides.

Measurable targets & KPIs

• Brands audited: 10–15 (ADX) and 6–8 (AI-Equity).
• Survey pilot: ~100 respondents with demographic cuts.
• Reliability: Cohen’s κ ≥ 0.70 on the ADX pilot.
• Outputs: 1 report, 1 executive summary, 1 slide carousel, clean dataset + documentation.
• Actionability: Minimum 3 prioritized recommendations per brand category.

Guardrails

• Follow ethical research practices (consent, de-identification, secure handling).
• Ensure all deliverables meet basic accessibility (alt-text, captions, readable contrast).

Expected project outcomes

• Publishable insights: A concise, evidence-based report that positions Le Code Switch as a go-to source on accessibility and AI-equity in Canadian marketing.
• Actionable scoring framework: A validated Accessibility & Invisible Disability Experience Index (ADX) rubric that brands can apply and that we can scale into future benchmarks.
• Risk-mitigation toolkit: A practical AI-Equity Audit checklist that helps teams find and fix representational, bias, and accessibility issues in content and workflows.
• Lead generation: Gated assets (report + scorecard) and a 30-min debrief CTA ready for the website/newsletter to convert readers into qualified prospects.
• Reusable research system: Clean datasets, codebooks, and a repeatable workflow to expand from pilot to annual/quarterly studies.

Core deliverables (what students will hand in)

1. ADX Rubric v1.0

• Clear scoring rules, examples of “pass/fail,” severity tags.
• Pilot scores for 10–15 brands with inter-rater reliability (Cohen’s κ ≥ 0.70).

1. AI-Equity Audit Checklist v1.0

• Criteria covering prompts, outputs, and governance (bias, stereotyping, toxicity, captioning/alt-text signals).
• 6–8 sample audits (text/image/video) with prioritized fixes.

1. Survey Package

• 15–20 item questionnaire (brand trust, accessibility experience, expectations).
• Pilot results (≈100 respondents) with cross-tabs and a documented codebook.

1. Insights Report (8–10 pages)

• Executive summary (1 page), 5–7 headline insights, 3–5 practical recommendations per asset.
• Charts and callouts formatted for web/PDF.

1. Social/Comms Assets

• 5–7 slide LinkedIn/IG carousel (plain-language takeaways).
• 2–3 suggested post captions + a short pitch paragraph for media/outreach.

1. Data & Reproducibility Handoff

• Cleaned dataset(s), data dictionary/codebook, and analysis notebook (R/Python or SPSS/Stata).
• Read-me with replication steps and limitations/assumptions.

1. Final Presentation

• 10–12 slide deck walking through method, findings, recommendations, and next steps.
• Q&A notes and a backlog of “future improvements.”

Format & quality standards

• Accessibility: All visuals include alt-text; slides and PDFs meet basic contrast and font-size standards; captions for any demo video.
• Citations: Sources listed; any external data properly referenced.
• Clarity: Plain-language summaries; jargon minimized; bilingual EN/FR labels where feasible.
• Ethics & privacy: Consent templates, anonymization of any personal data, and secure file handling.

Success metrics (acceptance criteria)

• Reliability: ADX κ ≥ 0.70 on pilot scoring.
• Coverage: 10–15 brands (ADX) and 6–8 content/workflow audits (AI-Equity).
• Rigor: Documented method, codebook, and reproducible analysis.
• Actionability: ≥3 prioritized recommendations per brand category; quick-win and longer-term fixes identified.
• Readiness: Report, deck, and carousel are client- and media-ready with no further editing required beyond brand styling.