How we verify every statistic
Every number on statfern passes the same four gates. A statistic that fails any gate does not get published, no matter how good it would look in a headline.
Gate 1: primary source
Each figure must trace to the institution that produced the data: a government agency, a peer-reviewed paper, an established research firm or a company's own published dataset. We do not cite other blogs, "listicles" or news summaries as sources.
Gate 2: the number must exist at the source
An editor opens the source page and confirms the figure appears there, with the same scope and year we quote. If a page says "37% of respondents" and a blog claims "most workers", we quote the 37% version.
Gate 3: context travels with the number
Every entry records the sample size, population and year. A survey of 400 gig drivers is not evidence about the whole workforce, and we say so on the page.
Gate 4: random re-check
Before a report is published, a random sample of at least 10% of its entries is re-opened at the source. If any sampled entry fails, the whole report goes back for re-checking.
Dates
Each report shows the date its data was last verified. When we update a report, the date changes because the data changed. We never touch a date just to look fresh.
What we do not do
- We do not estimate figures and present them as measured data.
- We do not mix survey years inside one claim.
- We do not accept payment for changing a number. Paid placements, when they exist, are labelled.
Need numbers we haven't published?
Custom research, built on the same discipline
Bespoke statistics reports for teams and publications. The same rule applies to every project: each number carries a primary source.