# American Community Survey The Census Bureau's continuous household survey. It replaced the decennial census long form after 2000 and has been fielded nationwide since 2005, sampling roughly 3.5 million addresses a year. It publishes one-year estimates for places of 65,000 people or more and five-year estimates down to the block group, covering income, education, housing, commuting, employment, disability, language, ancestry, veteran status and citizenship. ## The census people actually use The decennial count supplies a headcount. The ACS supplies almost everything else attributed to "census data": - the inputs to federal formula funding, - the citizen voting-age population tables used in [[Voting Rights Act of 1965]] litigation, - congressional-district demographic profiles, and - the local labor, housing and income data that planners, businesses and journalists have no substitute for. It plays **no part in apportionment**. Seats come from the decennial enumeration alone ([[How the Census Apportions Congress]]). ## The citizenship question already lives here The ACS asks the citizenship of every sampled person, as the long form did before it. That is why Census Bureau researchers argued in 2018 that a decennial citizenship question was unnecessary: ACS data plus administrative records produce better citizenship estimates than a census question would. It is also why substituting the ACS instrument in the 2026 Census Test put a citizenship question in front of about 155,000 households in a *decennial* rehearsal, without anyone deciding to ask it on the decennial form. See [[Census Citizenship Question]]. ## The exposed tier Unlike the apportionment and redistricting files, the ACS has **no statutory release deadline**. It exists because the Bureau publishes it ([[Census Data Products Tree]]). That makes it the easiest census product to reshape without a vote, and 2025–2026 pressed on it from three directions. **Mandatory response.** Answering is compulsory under [[Title 13]], backed by a $100 refusal fine, and response is what makes the data usable. The House Appropriations Committee's FY2026 Commerce-Justice-Science bill carried a § 605 barring funds "to enforce involuntary compliance" with any Census Bureau survey, and standalone House and Senate bills to make the ACS voluntary were introduced in the 119th Congress. The cautionary precedent is Canada: when it replaced its mandatory long form with a voluntary survey in 2011, national response fell from 94% to 69%, with steeper drops in rural provinces, and the mandatory form came back five years later. **Disclosure method.** [[Disclosure Avoidance for Statistical Products]] (DAO 216-26, effective June 4, 2026) bars noise infusion. On **August 4, 2026** the Bureau postponed the 2025 ACS one-year estimates pending a compliant method. On **September 4** more than 100 organizations wrote to Commerce asking that the delay end. As of late September 2026 there is no release date. **Race and ethnicity.** The [[2030 Census Residence Rule]] would bar race and ethnicity questions from the decennial form while keeping them on the ACS (§ 70.4). That would leave a sample survey, one with no release deadline and a postponed release, as the census program's main source of race data for [[Voting Rights Act of 1965]] enforcement. Taken together, these moves make the survey that carries the analytical weight also the one with the fewest legal protections. ## See also - [[Census]] · [[Census Bureau]] · [[Census Data Products Tree]] · [[Census Data Access Crosswalk]] - [[Disclosure Avoidance for Statistical Products]] · [[Differential Privacy]] · [[Public Use Microdata Sample]] - [[Census Citizenship Question]] · [[2030 Census]] · [[2030 Census Residence Rule]] - [[Federal Statistical System Tree]] · [[The Politics of Official Statistics]]