ACS 5-year Race (block Group)

Overview

The ACS 5-Year Race dataset reports population counts by single-race category and "two or more races" for every Census block group in the United States. Published annually by the U.S. Census Bureau as part of the ACS 5-year estimates program, which pools five years of survey responses to provide statistically reliable estimates at small geographies.

Data Description

Each row carries the geoid, state/county FIPS, tract, block_group, name, year_end, vintage, total population (race_total), and seven race-category counts: white alone, Black or African American alone, American Indian and Alaska Native alone, Asian alone, Native Hawaiian and Other Pacific Islander alone, some other race alone, and two or more races.

Each category has a paired margin-of-error column at the 90% confidence level. The seven category counts sum to race_total (modulo rounding).

Hispanic/Latino is NOT included here — Census treats Hispanic origin as a separate ethnicity question independent of race; a person of any race may also be Hispanic.

Coverage

Geographic Coverage: All Census block groups in the 50 U.S. states, the District of Columbia, and Puerto Rico. Approximately 220,000-242,000 block groups per release year depending on Census's BG redraws.

Temporal Coverage: ACS 5-year releases ending 2013 through 2024 (12 release years).

Observation Level: One row per Census block group per ACS 5-year release year per geometry vintage (keyed by geoid + year_end + vintage).

Vintage Information: Two block-group geographic vintages appear across these datasets because the U.S. Census Bureau redrew block-group boundaries after the 2020 Decennial Census. A vintage column distinguishes the two: '2010' for releases ending 2013-2021 (using 2010 Census BG boundaries) and '2020' for releases ending 2022 and later (using 2020 Census BG boundaries). Block-group GEOIDs do not align 1:1 between vintages — the same physical land area can have a different GEOID before and after the 2020 redraw. Cross-vintage joins should use a Census-published block-group relationship file.

Methodology

Pulled from the U.S. Census Bureau's American Community Survey 5-Year Estimates via the public Census Data API.

Each annual release is the average of five years of survey responses. Block-group estimates are computed by Census using its standard survey-weight methodology.

Additional Notes

All Census Bureau data is in the public domain.

This dataset shares the same geographic key (geoid) and temporal columns (year_end, vintage) as the other Census datasets, so they can be joined row-for-row within a vintage.

For trend analyses crossing the 2021→2022 vintage boundary, restrict the cohort to one vintage or apply a Census-published block-group relationship crosswalk.

Licensing and redistribution. U.S. Census Bureau data products are works of the United States Government and are not subject to copyright: "Copyright protection under this title is not available for any work of the United States Government" (17 U.S.C. § 105). Dewey redistributes this processed extract on that public-domain basis. The Census Bureau requests (but does not require) citation, and prohibits using the data to "identify any individual person, household, business or other entity" — a confidentiality obligation that passes through to users of this dataset. This product uses the Census Bureau Data API but is not endorsed or certified by the Census Bureau.