How we check.

Every post on Mostly Proven follows the same three steps. Here is what they mean, where the evidence comes from, and how we handle mistakes.

Claim, evidence, verdict

1. Claim. We pick one claim, worded the way it circulates. Not a person, not a product.
2. Evidence. We read the studies behind it and put the numbers next to the claim, with the study type, who was studied, and how large the evidence is.
3. Verdict. We rate the claim on a five-step scale and say why it stops where it does. The verdict comes with a source, or not at all.

Every check ends with what we don’t know. “Not shown yet” is not the same as “shown not to work”.

How we choose sources

• Peer-reviewed first. We use published, peer-reviewed research, and we link every source by DOI and PubMed so you can open it yourself.
• Systematic reviews and meta-analyses are preferred, then large cohort studies and randomized trials. A single study is a hint, not a verdict.
• We say what a source is. Narrative reviews, expert consensus statements and animal or mechanism studies are labeled as such, and they carry less weight.
• We name conflicts of interest where the authors report them and where they matter for the claim. If a source has a known correction (erratum), we say so on the page.
• A person reads the key numbers against the original paper before a post goes out.

Evidence levels

Each check shows the level of the evidence behind the verdict.

What the verdicts mean

A verdict rates the claim as it circulates. It never rates a person, and it is not a verdict on a single study.

• Proven

  Very well supported and consistent. Study design can separate the effect.

  Typical evidence: Several randomized trials or meta-analyses of trials that agree. Rarely given.

• Mostly proven

  The core holds, but part of the claim goes beyond the data.

  Typical evidence: Consistent cohorts or trials, but dose, cause or group is still open.

• Partly proven

  One part is supported, an important part is not.

  Typical evidence: Mixed findings, stand-in measures, observational data.

• Not proven

  The claim has not been shown (yet).

  Typical evidence: No fitting study, or only mechanism or animal data.

• Not true

  The claim contradicts the data.

  Typical evidence: Only with strong evidence against it. Very rare.

“Mostly proven” is never a throwaway phrase. We always say why only “mostly”. We don’t give verdicts on treatments or medications, and we don’t make predictions.

What we don’t do

• No personal advice, no diagnosis, no promises that something heals or prevents a disease.
• We don’t sell supplements or products and we don’t run a shop.
• No faces, no “experts” without sources.

Corrections

We make mistakes, and we fix them in the open. If you spot one, write to hello@mostlyproven.com with the subject “Correction” and the page or post concerned. We check it, correct the page and the post, and add a visible note with the date. We don’t edit silently.