Scoring methodology
How we calculate the score — complete transparency.
Current methodology: v2.0.0 (since 10/21/2026)
What is a methodology version?
Science is built on results. But not all results are equal.
Publi-Score evaluates the methodological quality of scientific publications — that is, the rigor of the process that produced the results. It is not a judgment on the truth of conclusions or the 'absolute scientific quality' of a publication. This distinction is fundamental and intentional.
"We evaluate the quality of the thermometer, not the temperature it shows."
Publi-Score evaluates the process (methodological rigor, transparency, integrity), never the direction of the result (positive, negative, significant or not). Principle of directional neutrality — the authors' conclusion does not influence the score.
The formula
The 7 categories of methodological quality
Click on a category to see the sub-criteria details with a concrete example.
Integrity coefficient
The integrity coefficient is a multiplier (0 to 1) that penalizes weaknesses in transparency and independence. The lowest of the two sub-coefficients dominates.
A. Authors' integrity
- • Conflict of interest declaration
- • Funding independence
Coeff. A = average of both
B. Editorial process
- • Peer review quality
- • Submission → acceptance delay
Coeff. B = average of both
Scale — Funding independence
| Score | Situation |
|---|---|
| 1.0 | Public or academic funding (NIH, ERC, foundations, universities) |
| 0.7 | Mixed with real analytical independence — public co-funding and statistical analysis by independent third-party CRO or academic unit |
| 0.3 | 100% industry without analytical independence — includes: sponsor = manufacturer of tested device/molecule, data controlled by sponsor, lead authors paid by manufacturer |
Blocking alert signals
| Alert signal | Forced coefficient | Detection |
|---|---|---|
| Retraction | 0.0 | 100% API (PubMed + Retraction Watch) |
| Predatory journal | ≤ 0.2 | 100% API (DOAJ + Beall's) |
| Expression of Concern | ≤ 0.3 | 100% API (PubMed) |
| Editor = Author | ≤ 0.5 | PDF + manual verification |
Interpretation grid
| Tier | Score | Label | Interpretation |
|---|---|---|---|
| A | 76–100 | High confidence | You can trust the result, whether positive or negative: the approach is rigorous and the publication provides the means to verify it. |
| B | 56–75 | Fairly high confidence | You can trust the result, whether positive or negative: the approach is rigorous and the publication provides the means to verify it. |
| C | 46–55 | Moderate confidence | A result to consider with caution: some weaknesses limit confidence. |
| D | 26–45 | Limited confidence | The result is not necessarily wrong, but the publication does not provide the means to trust it. |
| E | 1–25 | Insufficient confidence | The result is not necessarily wrong, but the publication does not provide the means to trust it. |
| X | 0 | Invalidated | Retracted or proven fraud — do not cite |
Example of tier X: Raoult et al. (2020) — rétracté (PMID 32205204)
Three scoring modes
Quick score
- • Input: PMID or DOI
- • Duration: ~2–5 seconds
- • Coverage: ~55/100 evaluable pts ("Metadata only")
- • 100% automatic (9 APIs)
- • Alert signals detected immediately
- • Published to catalogue
Free · Unlimited · No account
Full manual
- • Input: PMID + PDF upload
- • Duration: ~10–30 minutes
- • Coverage: 100/100 pts
- • Guided form + PDF extraction
- • All sub-criteria evaluated
- • Not published to catalogue (human bias)
Free · Unlimited · Account required
🤖 Full AI
- • Input: PMID + PDF upload
- • Duration: ~30–60 seconds
- • Coverage: 100/100 pts
- • Direct AI analysis (objective)
- • All sub-criteria evaluated
- • Published to catalogue (reproducible)
5/month · Free account · Unlimited on Premium
Scope of application
Publi-Score only evaluates empirical studies conducted on human populations and published in peer-reviewed journals.
✓ Within scope
- Randomized controlled trials (RCT)
- Cohort and case-control studies
- Cross-sectional studies on human populations
- Meta-analyses and systematic reviews
- Prospective and retrospective observational studies
⊘ Out of scope
- Preclinical — in vitro, animal models, C. elegans…
- Editorial, letter, commentary — no primary data
- Preprint — not submitted to peer review
- Narrative review — without systematic protocol (PROSPERO/PRISMA)
- Erratum — correction of an existing publication
Applying the Publi-Score grid to a preclinical study would produce a misleading score. The methodological rigor of a C. elegans trial cannot be compared to that of a multicenter RCT. This exclusion is an assumed positioning decision, documented on our transparency page. transparency page.
Frequently asked questions
Answers to the most common objections and misunderstandings about the Publi-Score methodology.
For more technical objections and our detailed responses → transparency page
Scoring modes
Publi-Score offers three analysis modes, depending on available data.
Understand the differences →The trade-offs we assume
Our methodology has real limits. We document them: 12 argued trade-offs and 7 empirical validation protocols.
Suggest an improvement
Do you have a suggestion to improve the scoring grid? Share it — every proposal is reviewed by our team.