Research Refiner GPT – Post-analysis Model Refinement Assistant

A post-analysis refiner that aligns empirical results with the research model, refines the conceptual model and hypotheses using provided gap/literature cues, and recommends the next step via an updated YAML protocol.

Overview
Version
v1.0.0
Created
2025-12-16
Updated
2025-12-16
research-assistantconceptual-modelhypothesisgap-analysispost-analysisyaml-protocol
research-refinermodel-refiner
Key functions
  • Check consistency between empirical results and the initial research model (RQ/variables/hypotheses/paths)
  • Refine conceptual model and hypotheses using provided gap analysis and literature suggestions (add/remove mediators/moderators, adjust paths)
  • Diagnose discrepancies (theory/measurement/sampling/modeling/trivial) and document changes
  • Select exactly one next step (analysis, data_collection, literature, writing, stop) and output a single updated YAML protocol
Technical details
_id
g-6917d1f5ee008191bb3add10c2bc633b-research-refiner-gpt
gpt_id
g-6917d1f5ee008191bb3add10c2bc633b-research-refiner-gpt
viz1
public
viz2
show_url
language
en
Other fields
additional_features
["Automatic version bumping and reconciliation_notes (summary + bullet change log)"]
example_commands
["Given the YAML below (RQ/nodes/edges/hypotheses/results/gap_analysis/literature_suggestions), update it to a refined_model YAML and choose exactly one next_step.", "H2 and H4 are direction/significance mismatches—use the provided literature suggestions to add a moderator and revise the hypotheses accordingly."]
gpt_id
g-6917d1f5ee008191bb3add10c2bc633b-research-refiner-gpt
ideal_use_cases
["Feed a results+gap+literature-suggestions YAML and get a coherent refined model/hypotheses for reporting", "When key hypotheses are unsupported, revise structure (mediators/moderators/paths) and decide the single best next step"]
limitations
["Does not run new statistical tests; refines only based on the provided YAML (results/gaps/literature suggestions)", "Does not alter numeric results or fabricate new literature beyond what the user provides", "Outputs exactly one YAML document (protocol) per run"]
target_users
["Researchers drafting theses/papers who need post-analysis model/hypothesis refinement", "Analysts/RAs who need to reconcile empirical findings with theory and decide next steps"]