Evidence audit · Updated 2025-12-01
How Well Our Articles Show Their Sources
A public, reproducible audit of external evidence links in every Doctors & Miracles blog article.
Executive Summary
- Source-link coverage is 100%. 43 of 43 articles link to at least one qualifying external source; 0 do not.
- The visible evidence base is narrow. Across the library, the audit found 146 qualifying links from 22 source domain and 7 DOI links.
- This is a quality baseline, not a credibility score. A link can be weak or misapplied, and an uncited statement can still be accurate. Claim-level review remains necessary.
Articles audited
43
With sources
43
Without sources
0
Coverage
100%
The next standard is claim-level support
Article-level link counts are only the first layer. The companion claim-to-source map exposes each cited passage, its section, source class, and review status so a reader or reviewer can inspect whether the neighboring source carries the claim.
Recommended next steps
- Prioritize physician wellness and safety-related claims because they can affect real health decisions.
- Verify that every important numerical and medical claim is connected to the exact primary study or authoritative guidance.
- Add medical review only where a named qualified reviewer actually reviewed the page.
- Record corrections and re-run both datasets after each sourcing batch.
Articles by editorial category
| Group | Items | Share |
|---|---|---|
| Physician Wellness | 20 | 46.5% |
| Ghost Stories | 6 | 14% |
| Miraculous Recoveries | 5 | 11.6% |
| Faith Medicine | 4 | 9.3% |
| Near Death Experiences | 3 | 7% |
| Publishing | 3 | 7% |
| Research | 2 | 4.7% |
Download the article-level audit
Each row includes the article, category, author label, editorial status, source-link count, and detected source URLs. Amazon purchase links are excluded because they are commercial links, not evidence.
Further questions
- Do the cited sources support the precise claim beside them?
- Are systematic reviews, professional guidance, and primary studies used in the right contexts?
- Which uncited pages receive enough search traffic or reader engagement to justify expert review first?
Caveats and assumptions
The detector counts Markdown links beginning with HTTP or HTTPS, excluding Amazon and amzn.to links. It does not inspect unlinked publication names, validate study quality, or test whether a citation supports its neighboring claim. Results describe the repository as of 2025-12-01.
