Methodology
Sources are part of the data
The technical foundation is designed so a public claim can be checked, dated, corrected, and reproduced.
Source retention
Published factual values retain the source organization, dataset or report, URL, publication context, and retrieval date.
Defined measures
Metric definitions are separate from values so units, categories, periods, and meaning remain consistent.
Visible uncertainty
ACS cards retain 90% margins of error. Published MOEs are stored directly; derived sums and percentages use documented Census approximation formulas.
Auditable updates
Automated imports will record run status, versions, counts, errors, and source relationships before publication.
Comparable ACS context
Local, county, and state values are compared only when metric, ACS product, vintage, and survey period match. ACS 1-year and 5-year estimates are not mixed.
Rolling-period trends
Each trend point is an ACS 5-year estimate. Adjacent vintages overlap in four survey years, and dollar vintages are not shown as constant-dollar annual changes.
Honest missing data
Suppressed, sentinel, absent, not-filed, or unsupported values remain unavailable. Zero is retained only when a source reports a valid zero; another geography is never substituted.
Municipal fiscal scope
Comptroller revenue, expenditure, and debt values reproduce each municipal Level 1 filing. They are not per-capita values, budgets, or a New Day New York financial rating.
Fiscal stress labels
Fiscal and environmental stress scores and designations are reproduced from the Comptroller monitoring system. A missing score remains missing, including when a municipality did not file.
Scope labels
Education cards describe resident attainment, not school performance. ACS labor-force measures are not labeled as the official current BLS unemployment rate.