Evidence quality metadata is a practical confidence scale on what we can trust from studies and real-world signal strength.
Why it matters
Confidence affects how strongly a suggestion should influence your plan versus requiring more confirmation.
Reader-facing quality tiers
- Study design quality and consistency.
- Population relevance.
- Reproducibility and effect direction stability.
This is not one-dimensional; one outcome can have stronger evidence than another outcome.
How many sources updates confidence
As source count rises and sources align, confidence score can rise.
If new sources conflict, confidence may hold or drop until trends are validated.
“Direction, not diagnosis” principle
Evidence supports directional behavior, not personal diagnosis.
Use these insights as a planning layer, not a substitution for professional care.
Cross-site references
Uncertainty
- Evidence is limited for strong causal attribution in mixed-stack real-world behavior.
- Evidence is limited for outcomes in short monitoring windows.
How this appears in Unfair
Unfair attaches confidence labels and explanation copy to recommendations to prevent over-indexing on weak signals.
Clinical safety note
If high-impact symptoms appear, do not let “weak evidence confidence” delay escalation to clinical review.