Ar Theory

Process

Step 1: Survey Theoretical Landscape

Based on the literature map and scoping report, identify which theoretical frameworks are relevant to this research question. Use WebSearch if needed:

• "{research_question}" theoretical framework
• "{domain}" theory {discipline}

Step 2: Propose 2-4 Candidate Frameworks

For each framework, provide:

• Name: the framework's established name (e.g., "Institutional Theory", "Critical Discourse Analysis", "Actor-Network Theory")
• Intellectual origin: which discipline or tradition it comes from
• Core concepts: the key ideas and vocabulary it introduces
• What it emphasises: what aspects of the research question this framework foregrounds
• What it de-emphasises: what aspects it tends to overlook or marginalise
• How it connects: how it helps answer the specific research question
• Key authors: 2-4 seminal authors associated with this framework
• Rival theories: which other frameworks would lead to different conclusions and why

Step 3: Rank by Fit

Rank the frameworks by:

1. Evidence fit: how well the available corpus supports analysis through this lens
2. Question fit: how directly the framework addresses the research question
3. Methodological compatibility: whether the framework works with the chosen research track
4. Novelty: whether applying this framework offers fresh insight (not just repeating existing studies)

Assign a fit score (high / medium / low) for each criterion.

Step 4: Identify Debate Need

If two or more frameworks score equally well, or if the top-ranked framework has significant weaknesses, flag this for debate:

• Write a brief debate_proposal field explaining why debate is needed
• The orchestrator will invoke ar-debate with debate_type: theory_selection

Step 5: Write Theory Proposals

Write {workspace}/analysis/literature_map/theory_proposals.json:

{
  "proposals": [
    {
      "name": "string",
      "intellectual_origin": "string",
      "core_concepts": ["string"],
      "what_it_emphasises": "string",
      "what_it_deemphasises": "string",
      "connection_to_rq": "string",
      "key_authors": ["string"],
      "rival_theories": ["string"],
      "fit_scores": {
        "evidence_fit": "high|medium|low",
        "question_fit": "high|medium|low",
        "methodological_compatibility": "high|medium|low",
        "novelty": "high|medium|low"
      },
      "rank": 1,
      "recommended": true
    }
  ],
  "debate_needed": false,
  "debate_proposal": "string or null"
}

Quality Rules

• Propose only frameworks that are established in the relevant discipline — do not invent new theories
• Each framework must be genuinely different, not a minor variation of another
• Be honest about what each framework cannot do
• The recommended framework should have the strongest combined fit, not just the most popular one
• Use British English in all output text