U.S. Public Data Sources: Crime, Food, Work & More

Justice and victimization

FBI UCR measures police-reported offenses; BJS NCVS estimates victimization through a survey. Arrests, offenses, cases, convictions and incarceration are different events or populations. Missing agency reports are not zero crime, and reporting-system transitions can break a trend.

A future rate must use the reporting jurisdiction’s denominator. National survey estimates cannot be relabeled county measurements. The site excludes named-offender records, background checks and identifiable justice records.

Food, agriculture and material resources

USDA food insecurity, retailer access, prices, agricultural production and program participation describe distinct concepts. The Food Access Research Atlas now distinguishes SNAP-authorized Retailer Access and Large Retailer Access; their retailer definitions and vintages cannot be silently merged.

NASS Quick Stats requires commodity, unit, period, geography and suppression fields. Annual surveys differ from agricultural censuses. Program eligibility, enrollment, monthly caseload, annual recipients and expenditures are not interchangeable or additive counts of unique people.

Healthcare and work

CMS provider data and National Health Expenditure products require exact facilities, cohorts, denominators and reporting periods. Registered providers do not establish active practices, capacity, appointments or patients treated. Historical spending and projections must remain separate; provider-location spending differs from spending for residents.

County Business Patterns counts establishments with paid employees, not all businesses. BLS occupational wages and BEA regional price parities need their own universes and units. Residents with jobs differ from jobs located in an area; regional price differences are not local inflation rates.

Transportation and digital access

NHTSA FARS and BTS products distinguish crashes, fatalities, locations and exposure-based denominators. Transit trips are not necessarily unique riders. FCC availability is provider-reported and separate from household subscriptions and measured performance; address fabric redistribution terms require review.

The current infrastructure measures use ACS resident characteristics. They do not count all trips or establish service hours, operating capacity or household accessibility.

Veterans, civics and extensions

VA veteran populations, enrolled patients and benefit recipients have different universes. Census survey-reported voting is not a certified ballot count. Immigration sources need stocks, admissions, applications, encounters and repeat events kept separate; language or ethnicity cannot identify immigration status.

Science and research, food/program, immigration and other proposed endpoints remain candidates. No observations from these products are integrated or downloadable locally. Producer access, schemas, dates, rights and geographic coverage require further validation before publication.

U.S. public data sources: crime, food, work and more