Research Methods Guide: Comparative Policy Analysis
A methods guide walking research students through designing and running a defensible comparative policy analysis.
Demonstration writing sample. A condensed methods guide illustrating my research-training writing.
When to use this method
Use comparative policy analysis when your question is about how different institutions regulate the same thing — not about how policies are experienced (interview instead) or whether they work (evaluate instead).
The five design decisions
1. Define the comparison set
Decide which institutions are in scope and why, before you look at any policy. A defensible set is defined by criteria (e.g. “all public universities in one jurisdiction”), not by availability.
2. Define the document universe
Specify which document types count: formal policy, procedures, student guidance? Institutions distribute the same rules differently, so comparing “the policy” of each institution usually means comparing document sets.
3. Build the framework before coding
Derive your coding fields from your research question and pilot them on two or three documents. Every field needs a definition, permitted values and a worked example. Expect the pilot to break your framework; that is its job.
4. Code with an audit trail
Code from archived, dated captures. Record a clause reference for every value.
Use an explicit not-addressed value — silence is a finding, not a gap in
your data.
5. Analyse patterns, not paragraphs
The analysis compares coded values across institutions: where is there convergence, divergence, silence? Return to the document text to interpret patterns — the matrix tells you where to look, the documents tell you what it means.
Common failure modes
- Availability sampling — comparing whichever policies were easy to find.
- Framework drift — coding late documents with a subtly different understanding than early ones. Re-code the earliest documents last.
- Over-claiming — policy text supports claims about provisions, not about practice, enforcement or effect.
Reporting
Report the comparison set criteria, document universe, capture dates, the full coding framework, and reliability procedure. A reader should be able to reproduce your matrix from your appendix.