A number should be easy to check.
We publish the definition, period, source, and calculation alongside the finding. Here is how the first Autism Stats release was prepared.
Start with the original record
Our reports use published government data, federal audit reports, and credential-board records. A third-party statistics list is a research lead, not evidence for a number. Source links point to the issuing agency or original report.
The first release uses research files retrieved on October 12, 2026. Each report records the source publication period and the version used. A new date on this website does not make old observations new.
Keep the populations separate
Eight-year-olds at CDC monitoring sites are not all US children. Students served under an education category are not all people with a diagnosis. Credential counts do not establish how many therapists have appointments available.
For calculations, we identify the numerator and denominator. We retain missing values and distinguish them from zero. We avoid comparisons across changed age definitions, different service scopes, and incomplete baseline reporting.
Check the exported data
Our release checks match CMS national and geographic amounts to the original workbook, reconcile CDC counts and rounded rates, recompute BCBA population ratios, and compare the endpoints of the school trend with the original IDEA files. Audit amounts are checked against the saved federal report texts.
Charts, displayed tables, and downloadable CSVs use the same curated observations. The chart export includes its source and scope. Selected comparisons are labelled as selected; the complete approved table stays available.
Separate processing from interpretation
This site was built with AI assistance. Scripts perform data processing and consistency checks. They do not establish clinical expertise or turn a descriptive relationship into proof of a cause.
Research is published by Zechariah Tokar. This release has not been reviewed by a named clinician or an external statistical reviewer. Reports do not claim that it has. Survey-derived estimates requiring further statistical review are withheld from this release.
Preserve corrections and versions
A material change to a value, source definition, or calculation should appear in the correction log and a new dataset version. Old citations need a stable page even when the finding is updated.
Sources retain their own reuse terms. We link readers to the source, publish a curated subset, and do not claim ownership of third-party observations.