The U.S. Census Bureau has been counting Americans since before the ink dried on the Constitution. Here's everything you need to know to start using this treasure trove of data.
The census has mirrored every chapter of American history — from slavery and Reconstruction to the civil rights era and the rise of multiracial identities.
Not all Census data is the same. The Bureau runs several distinct surveys with different methods, sample sizes, and ideal uses. Choose the right one for your question.
The official population count. Required by the Constitution.
The decennial census attempts to count every person living in the United States on April 1st of the census year. It asks only a handful of questions: name, age, sex, race, Hispanic origin, and housing tenure.
It is the authoritative source for total population figures used to apportion congressional seats, redraw legislative districts, and allocate hundreds of billions in federal funding. It is not a sample — it aims for a complete enumeration.
The workhorse of Census data. Used by researchers worldwide.
The ACS 5-year estimates pool five years of monthly survey responses (e.g., 2019–2023) to produce statistically reliable estimates. About 3.5 million addresses are sampled each year, and results are released annually with a one-year lag.
It covers every topic the old long form covered: income, poverty, education, housing costs, commuting, language, disability, citizenship, and more — across 20,000+ variables. This is the dataset used by most academic researchers and the backbone of CensusGo.
The most current data — but only for large places.
The ACS 1-year estimates use a single year of survey data. Because the sample is smaller, estimates are only statistically reliable for geographies with 65,000 or more people — states, large counties, and big cities.
The advantage is timeliness: the 2023 1-year estimates reflect conditions in 2023, making them useful for tracking rapid economic changes like the COVID-19 recession or post-pandemic income shifts.
The legally authoritative count of eligible voters, by race.
CVAP is a special tabulation of ACS data produced by the Census Bureau specifically for Voting Rights Act litigation. It counts U.S. citizens aged 18 and over, broken down by race and Hispanic origin, at small geographies including voting tabulation districts (VTDs) and census tracts.
Unlike total population, CVAP focuses on those legally eligible to vote. Courts use CVAP — not total population — to determine whether a minority group is large enough to constitute a majority in a proposed district under Section 2 of the VRA.
Census data is organized in a strict hierarchy. Smaller units nest perfectly inside larger ones — a census tract is always inside a county, which is always inside a state. Choose your unit based on your question.
The highest level. Used for national benchmarks — "What is the U.S. median household income?" Most ACS tables are available at this level.
The most commonly used unit for policy research and journalism. Both 1-year and 5-year ACS estimates are available. Every variable in CensusGo is available at the state level.
📰 Journalists' most common geographyThe primary unit of local government outside New England. ACS 5-year estimates cover all counties; 1-year only covers those with 65,000+. The County Deep Dive tool on the Journalists page uses this level.
⚖️ Key for VRA litigation and local reportingCovers incorporated municipalities (cities, towns, villages) and unincorporated census-designated places (CDPs). The City & Place Profiles tool on the Journalists page uses this level. Note: cities and counties don't always align — New York City spans 5 counties.
🏘️ Best for hyperlocal storiesThe standard unit for neighborhood-level analysis. Tracts are designed to be relatively homogeneous in population, economic status, and living conditions. ACS 5-year estimates are available for all tracts. Essential for mapping income inequality, segregation, and health disparities.
🔬 The researcher's workhorse geographySubdivisions of census tracts, used when tracts are too large to capture micro-level variation. ACS 5-year estimates are available for block groups for many (but not all) variables. Margins of error get large at this level — interpret with caution.
The smallest geographic unit. Census blocks correspond roughly to city blocks bounded by streets, waterways, or other features. Only decennial census data is available at the block level — ACS data is not published here due to confidentiality and statistical reliability concerns.
🗳️ Used exclusively for redistrictingYou now know the landscape. Start with a search or try the AI explorer.