How ProofCritic Rates What Works
ProofCritic exists so that complex evidence becomes understandable — without hiding uncertainty. This page explains the conceptual process behind every rating.
The process
- Define the claim. Each treatment is evaluated against a specific claim — e.g. "minoxidil slows hair loss in androgenetic alopecia" — not a vague promise.
- Identify relevant research. Scientific literature, systematic reviews and clinical guidelines are collected for the claim.
- Prioritise high-quality evidence. Study design and rigour determine how much weight each source receives.
- Extract outcomes. What did the study actually measure — hair count, patient ratings, adverse events?
- Evaluate consistency. Do high-quality studies agree? Agreement matters more than any single result.
- Assess effectiveness. The size and clinical relevance of measured effects.
- Assess safety separately. Common and important risks are rated independently of effectiveness.
- Analyse real-world experience separately. User sentiment is a valuable signal — and is never blended into the scientific evidence rating.
- Assign confidence. How sure are we, overall, in the verdict?
- Publish sources and update dates. Every rating names its inputs and when it was last reviewed.
The evidence hierarchy
In general, higher-quality study designs receive more weight. A treatment's verdict is built primarily from the top of this list.
- Systematic reviews & meta-analyses Strongest weight
- Randomized controlled trials High weight
- Controlled clinical studies Moderate weight
- Observational studies Moderate–low weight
- Case reports Low weight
- Expert opinion Low weight
- Anecdotal evidence Context only
Core scoring principles
These principles are non-negotiable and appear throughout the product.
Evidence ≠ effectiveness
Strong evidence and large effects are different things. Both are shown, separately.
Sentiment ≠ science
User scores reflect experience, never evidence strength. The two are never combined.
Absence ≠ impossibility
"Insufficient evidence" means we do not know — not that a treatment does not work.
Commerce never influences ratings
Affiliate relationships have no influence on verdicts, grades or rankings.
Traceability
Every important conclusion should eventually be traceable to underlying evidence.
Visible uncertainty
Confidence is always stated. Where evidence is weak, we say so plainly.
Methodology v0.1
This MVP uses methodology version 0.1 — a prototype framework. The scoring system is being developed and will evolve before production ratings are published. All current scores are demonstration data created to validate information architecture and design, and must not be read as validated medical conclusions.
As the methodology matures, each change will be versioned and documented here, and every published rating will state which methodology version produced it.