Every factual claim on project-diada.org is traceable to a primary source. This page explains the checks and what happens when a claim fails them.
The rule
We do not write articles from press releases, and we do not write from other outlets’ coverage. A news write-up is a lead, not a source. The article is written from the paper.
Finding the paper
Every story begins by locating the study itself. We use the DOI where one is given, and search Crossref, Europe PMC and PubMed where one is not. The full text is read where it is openly available; where it is not, we work from the abstract and say in the article that method detail was limited.
If the paper cannot be located, the story does not run. A study we cannot read is a study we cannot report.
What we check on every paper
- Is it peer reviewed? Preprints are labelled as such in the article text.
- Has it been retracted? We check the publisher record for retraction and correction notices before writing. A retracted paper is never the basis of an article.
- What was the design? This determines what the article is permitted to claim. See how we write.
- Who was studied, and how many? Sample size, age range, country, and how participants were recruited.
- How large was the effect? The actual number, translated into plain language.
- Correlation or causation? Almost always the former, and the verbs follow.
- Who funded it? Funding and competing-interest statements. Industry funding is not disqualifying, but it is disclosable and we disclose it.
- Does it stand alone? Whether the finding replicates, contradicts or extends existing work. A single novel result is reported as a single novel result.
Statistics and official data
Prevalence figures, trends and population data are taken from the originating body — the statistical agency, health department or international organisation that produced them — not from a secondary citation. Each is published with its data year and the population it describes.
This matters more than it sounds. Several widely-quoted mental health figures rest on surveys conducted two decades ago and are routinely presented as current. Where we use one, we date it.
Quotations
A quotation is published only when we have the source where that person said those exact words, and it is attributed with name, credential, affiliation and setting. We do not construct composite quotes, we do not quote unnamed experts, and we do not put a paraphrase inside quotation marks.
What happens when a claim fails
Three outcomes:
- Dropped. The paper cannot be found, has been retracted, or the headline claim is not in the study. Nothing is written.
- Reframed. The finding is real but the news framing overstates it. We write the honest version, which is often a more interesting article.
- Published with caveats. The claim survives, and its limits go in the article.
After publication
Checking does not stop at publication. Where a paper we have covered is later retracted, corrected, or contradicted by stronger evidence, we update or withdraw the article and say so. Our corrections policy covers the mechanics.
Found an error?
Send it to [email protected] with the URL and the specific claim. We would rather be corrected by a reader than leave something wrong on the site.