Source choices are part of the finding.
We inspected observations, definitions and original Census values. A variable’s existence in a catalog does not establish that it has usable data.
What this release contains
| Source and product | Geographies | Periods | Uncertainty |
|---|---|---|---|
| Original Census ACS one-year B12001 | U.S.; 50 states and D.C.; 4 counties; 4 cities | 2021–2024; 4 annual estimates | Source 90% MOEs, with calculated share MOEs |
| Data Commons ACS five-year detailed tables | U.S.; 50 states and D.C.; 4 counties; 4 cities | Available windows: 2007–2011 through 2020–2024. U.S. history starts at 2018–2022. | Not returned; remains unavailable |
The one-year cohort is deliberately held at the verified 2024 endpoint. It is not labelled as the newest available Census release. Census published the 2020–2024 five-year product on January 29, 2026; Data Commons did not supply its own import date. Retrieval and page review on October 8 do not change the observation periods.
Data Commons, with an explicit facet
We queried the supported V2 observation API with dates and values selected. The publication uses detailed-table facet 10169881228856630405, measurement method CensusACS5yrSurvey, provenance dc/base/CensusACS5YearSurvey. It excludes automatic national aggregates and subject-table facets with different population or calculation methods.
The five status variables are never married, married and not separated, separated, widowed and divorced. Each uses the matching population aged 15+ denominator.
Why other plausible variables were excluded
The generic-looking Count_Person_NowMarried node has a female population constraint. We did not treat it as an all-population count. Another national facet aggregates to a different denominator; subject-table facets produce fractional derived counts. Neither is substituted for our selected detailed-table facet.
Some sex-specific divorced, never-married and widowed observations were found, but the complete sex-by-status partition was not supported in the chosen Data Commons facet. We use the verified original Census product for that breakdown. The sampled age-specific and cohabitation candidates did not establish publishable matching observations. Household counts need their own table review and are not inferred from people counts.
Access and refresh status
This release was imported through the supported Data Commons API using configured, authenticated access. Readers receive cached, reviewed data without contacting Data Commons. Credentials stay in the import environment. New observations are staged, checked against Census and reviewed before publication; a successful API request does not automatically change a page.
Original Census Summary Files provide the one-year publication and independent five-year checks without an API key. The existing Census API adapter remains optional. No source failure deletes the last good publication.
Commercial reuse and attribution
The Data Commons FAQ licenses its compilation under CC BY 4.0, subject to underlying source terms. We retain adjacent links back to the exact Data Commons variables and identify original Census provenance. Census-produced statistical data are public-domain material; source acknowledgments remain in exports. See Census citation guidance.
Only the reviewed Census-backed cohort is included. Public visibility of another Data Commons source is not permission to redistribute it. No source logos, photos, microdata or third-party graphics are republished here, and no endorsement is implied.
Verification and revisions
All 829 Data Commons place-period partitions sum to the selected population denominator. 300 latest-period status counts and their denominators match original ACS five-year B12001 exactly. Census one-year source cells, table labels, populations and margins of error are retained. The snapshot contains hashes, a metric registry and a geographic crosswalk.
Download the reviewed snapshot · How calculations work · Other CouplesData sources