Case Study: Lessons Learned & Best Practices from Likert-Scale Analysis in a Public Health Dissertation

---

1. Project Overview a Likert scale analysis case study

Likert scale analysis case study in a Public Health PhD dissertation, 413 healthcare workers across urban and rural hospitals rated their perceptions of operational protocols using a 10-item, 5-point Likert scale. The study collapsed responses to a 3-point scale, aggregated related items into composites, and conducted chi-square analyses on a derived “high agreement” category. Throughout this process, several operational challenges emerged—and with them, key lessons and best practices.

---

2. Challenge & Lesson #1: Scale Transformation Bias

• Issue: Directly collapsing 5 → 3 without safeguards risked misclassifying slight disagreements as neutrality.
• Lesson: Implement tiered collapsing with clear cut-points (e.g., 4–5 → Agree; 3 → Neutral; 1–2 → Disagree) and validate by inspecting the distribution before and after transformation.

---

3. Challenge & Lesson #2: Composite Aggregation Pitfalls

• Issue: Averaging items with differing variance inflated some composites, leading to unstable category assignments at Level 2.
• Lesson:
  • Standardize variance across items (e.g., z-score them) before averaging, or
  • Use weighted composites reflective of item reliability (Cronbach’s α) to ensure each item’s contribution matches its internal consistency.

---

4. Challenge & Lesson #3: Threshold Sensitivity

• Issue: Small shifts in the “high agreement” threshold (e.g., including neutral as agreement) markedly changed subgroup effect sizes and p-values.
• Lesson:
  • Perform sensitivity analyses (“What If” scenarios) to document how alternative definitions impact results.
  • Pre-register threshold rules in the dissertation protocol to avoid post-hoc bias.

---

5. Challenge & Lesson #4: Workflow Fragmentation

• Issue: Spreadsheet users and R users maintained separate versions of the “masterchart,” causing drift.
• Lesson:
  • Centrally store a single masterchart CSV and script all transformations in R (or, if using VBA macros, invoke them via R’s system() calls) to maintain a unified pipeline.
  • Automate schema checks in R to flag unexpected column names or missing-value codes before analyses run.

---

6. Challenge & Lesson #5: Stakeholder Interpretation

• Issue: Stakeholders found categorical outcomes (1/2/3) less intuitive than raw mean scores or visual distributions.
• Lesson:
  • Provide both categorical summaries and continuous visualizations (density plots, boxplots) for transparency.
  • Accompany every chi-square table with an executive snapshot: e.g., “X% ‘agree’ vs. Y% ‘disagree’,” avoiding jargon.

---

7. Best Practices Checklist from a Likert scale analysis case study

1. Transparent Transformation
   • Document each collapse rule with code comments and distribution histograms.
2. Psychometric Validation
   • Assess composite reliability (Cronbach’s α) before aggregation; consider item weights.
3. Sensitivity Logging
   • Automate “What If” scripts to output results for all threshold scenarios.
4. Unified Pipeline
   • Maintain one master dataset; script all steps from import to final table.
5. Dual Reporting
   • Present both categorical and continuous results; use R Markdown to weave narrative, tables, and figures.
6. Audit Readiness
   • Generate a final Data Audit Report detailing transformation logs, assumption checks, and version stamps.

---

8. Outcome & Impact of a Likert scale analysis case study

By incorporating these lessons and best practices:

• Reproducibility improved, with zero divergence between spreadsheet and script-based datasets.
• Reliability of composites increased (α > 0.80), ensuring robust category assignments.
• Stakeholder Engagement rose through clearer visuals and succinct executive summaries.
• Credibility was bolstered by pre-registered transformations and comprehensive sensitivity analyses.

---

This Lessons Learned & Best Practices case study showcases refined protocols for transforming, aggregating, and reporting Likert-scale data—ensuring that Public Health dissertations yield transparent, robust, and stakeholder-friendly insights.

---

Want to explore more PhD-level case studies? Check out our Comprehensive Case Studies on PhD Statistical Analysis guide page.