| name | evidence-grading |
| description | Use when the user wants to assess how strong the evidence is behind scientific claims based on design, replication, sample size, venue, and recency. |
Evidence Grading
Overview
Evaluates the strength of evidence behind scientific claims based on study design, sample size, replication status, venue quality, and recency. Produces structured evidence grades that help teams know how much weight to put on any given finding.
When to Use
- User asks "how strong is the evidence for [claim]?"
- Lab needs to decide how confidently to build on a published result
- Writing a manuscript and need to calibrate hedging language
- Grant reviewers will scrutinize the strength of the preliminary evidence
- User wants to rank a set of papers by evidentiary weight
Key Capabilities
- Grade individual papers on study design quality (RCT > observational > case study, etc.)
- Assess replication status: single study, replicated, widely replicated, meta-analyzed
- Factor in sample size, effect size, and statistical rigor
- Consider venue quality (Nature/Science/Cell vs. preprint vs. workshop)
- Produce Oxford CEBM-style evidence levels or custom grade schema
- Summarize overall evidence strength for a specific claim across multiple papers
- Generate hedging language calibrated to evidence grade
Usage Examples
Grade a specific paper
evidence_grading.grade_paper(
doi="10.1038/s41586-024-00001-0",
domain="machine_learning",
include_rationale=True
)
Grade evidence for a claim across all supporting papers
evidence_grading.grade_claim(
claim="Scaling laws hold for code generation models",
supporting_papers=claim_tracker.get_papers("claim_0099"),
schema="custom"
)
Generate calibrated hedging language
evidence_grading.suggest_language(
claim="Our method outperforms baselines on protein fitness prediction",
grade="preliminary",
context="manuscript_results_section"
)
Output Format
Returns grade, level label, confidence score, and rationale. For claim-level grading, returns aggregate grade with per-paper breakdown. Hedging language output is ready-to-use prose.
Best Paired With
claim_auditor for calibrating claim strength
experiment_skeptic when the paper's conclusion seems stronger than its evidence
Notes
- Grade rubrics are domain-aware: ML/CS standards differ from clinical/biomedical standards
- Preprints are graded conservatively until peer review is confirmed
- Combine with
contradiction-detection — contradicted claims automatically receive a grade penalty