What the numbers on this site are, where they come from, and what they cannot tell you.
birthrate.io publishes no primary statistics. Every figure is sourced from a statistical agency or research institution, is attributed on the chart where it appears, and can be traced back to the original release. Where we transform data, the transformation is described here or on the chart itself.
Three rules govern everything on the site: label the provenance of every number, never present a model as an observation, and keep the download that a reader gets identical to the data behind the chart they are looking at.
Series are pulled directly from provider APIs and bulk releases on a weekly schedule, then written into a single indicator table keyed by subject, indicator, and year. Each indicator has one canonical definition, unit, and source, listed in the glossary.
Nothing is hand-edited in the database. When a provider revises a series, the next ingestion overwrites the affected values, so the site always reflects the current vintage rather than a frozen snapshot. Page headers show when the underlying indicators were last written.
Every value carries one of three labels, and the distinction matters more than any single number on the site.
Cross-country comparison is only as good as the weakest definition involved. We prefer a single provider covering all countries over stitching national sources together, because consistency of definition usually matters more than precision in any one country.
Some categories cannot be harmonised at all. Ethnicity, race, and migration background are defined differently in every country that measures them, so those pages are presented per country and are not comparable across borders. Recorded crime other than homicide reflects reporting and policing practice as much as underlying offending.
Aggregates such as regions, income groups, and world totals are excluded from country rankings and maps.
A few features generate numbers rather than report them: the population simulator, the live clock, and the population calculator. These extrapolate from official inputs under assumptions the reader controls, and are labelled as modelled wherever they appear. They are illustrative tools, not official projections, and should not be cited as observations.
Demographic data is revised routinely. Provisional fertility figures move as late birth registrations arrive; population series are rebased after each census; migration estimates are revised most of all. A figure changing is normally a sign the statistical system is working, not that it failed.
Because ingestion overwrites revised values, a chart downloaded today may differ from the same chart downloaded next month. Cite the access date, which the citation block on each page includes.
Coverage is uneven. Countries without complete vital registration appear with modelled or interpolated values from their provider, and the site does not currently distinguish those from directly measured figures. Migration data is the least complete of any domain here. Subnational and city data comes from national offices using their own geographies, so regions are not comparable across countries.
Confidence intervals are not yet displayed, even where the provider publishes them.
If a figure looks wrong, it may well be. Report it with the page and the series, and we will trace it back to the source. Corrections to ingestion logic are applied for every country at once rather than patched for a single value.
