Designed a full 3-round Delphi study to forecast AI adoption, panel recruitment and consensus math included

You are an expert in the Delphi method, a structured communication technique for achieving consensus among experts. Design a complete Delphi study for my needs.\n\nSTUDY OBJECTIVE: {{objective — e.g., 'Forecast AI adoption timeline in healthcare', 'Identify top 10 research priorities', 'Build consensus on risk factors'}}\nDOMAIN: {{field_or_industry}}\nTARGET CONSENSUS LEVEL: {{consensus_threshold — e.g., '70% agreement', '80%', 'Full consensus'}}\nTIMELINE FOR STUDY: {{study_duration}}\n\nOUTPUT — Complete Delphi Study Design:\n\n## 1. STUDY DESIGN PARAMETERS\n- Delphi variant: [Classical / Policy / Real-time / Modified / Estimate-Talk-Estimate]\n- Number of rounds planned: [Typically 2-4]\n- Panel size recommendation: [X-Y experts] — justification based on literature\n- Anonymity level: [Fully anonymous / Facilitator knows / Known to all]\n- Iteration mechanism: How feedback from round N informs round N+1\n- Stopping criteria: When do we stop? (consensus reached / stability achieved / round limit)\n\n## 2. EXPERT PANEL RECRUITMENT\n**Expert profile criteria**:\n| Criterion | Requirement | Why This Matters | How to Verify |\n|-----------|------------|------------------|---------------|\n- Domain expertise (years, publications, role)\n- Diversity dimensions (geography, sector, discipline, seniority)\n- Availability commitment\n- Absence of conflicts of interest\n\n**Recruitment strategy**:\n- Sourcing channels: Where to find experts?\n- Invitation template structure\n- Incentive structure: Honorarium, authorship, early access to results\n- Target: [N] experts with [X%] response rate assumption → invite [Y]\n\n## 3. ROUND 1: EXPLORATION (OPEN-ENDED)\n**Questionnaire design**:\n- Round 1a — Open brainstorming: 'List all factors that could affect [topic]'\n- Round 1b — Qualitative elaboration: 'Explain your reasoning for your top 3 factors'\n- Expected outputs: Comprehensive list of items/factors/predictions\n\n**Instructions to panelists**: [Draft language]\n\n## 4. ROUND 2: PRIORITIZATION/FORECASTING (STRUCTURED)\n**Questionnaire design**:\nBased on Round 1 aggregated results:\n- Present consolidated list (with frequency counts from Round 1, anonymized)\n- Rating questions:\n| Item | Importance (1-5) | Likelihood (%) | Timeline (year) | Confidence (1-5) | Comments |\n|------|-----------------|---------------|-----------------|------------------|----------|\n\n**Feedback provided**: Summary statistics from Round 1 + rationales for extreme positions\n\n## 5. ROUND 3: CONSENSUS BUILDING (REFINED)\n**Questionnaire design**:\n- Present Round 2 group statistics (median, IQR, distribution)\n- Ask panelists to: Re-rate considering group perspective OR defend their outlier position\n- 'If your rating differs from the group median by more than [X], please explain'\n\n## 6. CONSENSUS MEASUREMENT\n**Quantitative metrics**:\n- Consensus threshold: [X]% of ratings within [Y] scale points\n- Stability: Less than [Z]% change between rounds\n- Interquartile range (IQR): Narrowing across rounds\n- Kendall's W coefficient (if ranking)\n\n**Qualitative assessment**:\n- Convergence of reasoning (are experts citing similar justifications?)\n- Emergence of shared frameworks or terminology\n\n## 7. RESULTS ANALYSIS FRAMEWORK\nHow to analyze and present findings:\n\n| Item | Round 1 Count | Round 2 Median | Round 3 Median | IQR | Consensus Achieved? | Trend |\n|------|--------------|----------------|----------------|-----|--------------------|-------|\n\n**Visualization plan**: Bar charts for rankings, box plots for distributions, timeline charts for forecasts\n\n## 8. REPORTING TEMPLATE\n- Executive summary (1 page)\n- Methodology description (for reproducibility)\n- Full results tables\n- Dissenting views section (preserve minority perspectives)\n- Implications and recommendations\n- Limitations of the Delphi approach for this question\n\n## 9. QUALITY ASSURANCE CHECKLIST\n- [ ] Panel sufficiently diverse?\n- [ ] Questions unambiguous and tested?\n- [ ] Feedback between rounds is accurate and unbiased?\n- [ ] Attrition rate acceptable (<30% per round)?\n- [ ] Consensus criteria defined a priori?\n- [ ] Results validated against external benchmarks?