Census data glossary: survey and population terms explained

Survey estimates and uncertainty

American Community Survey (ACS). A continuous Census Bureau survey of addresses across the country. Its five-year estimates pool 60 months of responses, so the 2020–2024 figures on this site describe that whole period rather than any single year. Five-year data are the only ACS product published for every county, including the smallest.

Estimate. A figure produced from a sample of responses rather than a full count. Every ACS value on this site is an estimate, including counts such as households and housing units.

Margin of error (MOE). The ± figure next to a survey estimate. At the 90% confidence level used by the Census Bureau, the range from the estimate minus the MOE to the estimate plus the MOE is expected to contain the value a complete count would have produced, 90% of the time. A smaller MOE means a more precise estimate.

Derived MOE. A margin of error this site calculates for a percentage from the published MOEs of its numerator and denominator, using the Census Bureau’s approximation formulas. The methodology gives the formulas.

Percentage points (pp). The unit of difference between two percentages. A change from 10% to 12% is 2 percentage points, which is also a 20% relative increase; the two descriptions are not interchangeable.

Statistically significant difference. On this site, two estimates are called higher or lower only when their difference exceeds the square root of the sum of their squared MOEs, a 90% test. When a difference does not pass, profiles say there is no statistically significant difference; that is not proof the two values are equal.

Coefficient of variation (CV). The standard error (MOE divided by 1.645) as a share of the estimate. A CV above 30% marks an estimate as “use with caution.” Small counties and small subgroups are the most affected.

Population estimates and components of change

Population Estimates Program (PEP). The Census Bureau’s annual estimates of the resident population on July 1. They start from the 2020 Census and are carried forward with records of births, deaths and migration. They are modeled figures without a published sampling margin of error.

Vintage. The release year of a PEP series. Each new vintage revises the whole series back to 2020, so annual figures from different vintages should not be mixed. This site uses Vintage 2025 throughout.

Estimates base. The April 1, 2020 starting population that PEP derives from the 2020 Census. The first component period on profiles runs from April 1 to July 1, 2020.

Natural change. Births minus deaths. A negative value, sometimes called natural decrease, means more residents died than were born.

Net international migration. Moves into an area from abroad minus moves out of the United States, including movement of the armed forces and people between the states and Puerto Rico.

Net domestic migration. Moves into an area from elsewhere in the United States minus moves out to other parts of the country.

Residual. Change the Census Bureau cannot attribute to a specific component. It arises from the way county, state and national estimates are controlled to one another.

Who is counted

Universe. The group a measure describes, such as households, workers 16 and older or the population 25 and older. Two percentages with different universes cannot be compared as if they shared a denominator. Every profile row states its universe.

Household and householder. A household is everyone who occupies a housing unit as their usual residence, related or not. The householder is the person, or one of the people, in whose name the home is owned or rented. Group quarters such as dormitories, prisons and nursing homes are not households.

Family and married-couple household. A family household has a householder and at least one other member related by birth, marriage or adoption. A married-couple household is a family household in which the householder lives with a spouse.

Civilian noninstitutionalized population. The universe for health insurance coverage. It excludes active-duty armed forces and people living in institutions such as correctional facilities and nursing homes.

Labor force. People 16 and older who are working or are unemployed and looking for work. The unemployment rate divides unemployed people by the civilian labor force; it is a five-year survey measure, not the monthly rate published by the Bureau of Labor Statistics.

Foreign-born and naturalized citizen. Foreign-born residents are people who were not U.S. citizens at birth. Naturalized citizens are foreign-born residents who later became citizens. People born in Puerto Rico, the U.S. island areas, or abroad to a U.S.-citizen parent are counted as native.

Income, poverty and housing costs

Median. The middle value when every household, home or person is ordered from lowest to highest. Medians for different places cannot be averaged or added to produce a regional median, which is why this site does not combine them.

Per-capita income. Total income received by residents divided by the total population, including children. It is a mean, so a small number of very high incomes can raise it noticeably.

Inflation-adjusted dollars. ACS five-year dollar figures are expressed in the prices of the final survey year, here 2024. The adjustment uses a national price index; it does not account for differences in living costs between places.

Poverty threshold. The income level, varying by family size and the number of children, below which the Census Bureau classifies a family and its members as in poverty. The thresholds are the same everywhere in the country and are not adjusted for local prices.

Gross rent. Contract rent plus the estimated monthly cost of utilities and fuels the renter pays. It is not an advertised asking rent and includes long-term tenants as well as recent movers.

Rent burden. Gross rent as a share of household income. Households paying at least 30% are commonly described as cost-burdened, and at least 50% as severely cost-burdened. Households whose ratio cannot be computed, such as those reporting no income, are excluded from the denominator.

Home value. The owner’s estimate of how much the property, including the house and lot, would sell for. It is a self-reported survey answer, not a sale price or assessed value.

Vacant housing unit. A unit with no one living in it as a usual residence at the time of the survey. The vacancy share includes homes for rent or sale, seasonal and recreational homes, and units held vacant for other reasons.

Geography

County equivalent. An area the Census Bureau treats like a county where no county exists: parishes in Louisiana, boroughs and census areas in Alaska, independent cities in Virginia and a few other states, and the District of Columbia. Since 2022, Connecticut’s nine planning regions replace its eight former counties in Census statistics.

GEOID. The numeric Census identifier for an area. A county’s five-digit code combines the two-digit state code and a three-digit county code; Franklin County, Ohio, is 39049. Downloads also carry the full form with its summary-level prefix, such as 0500000US39049.

County versus city. A county profile describes everyone living in the county, including those outside any city. It cannot stand in for a city, school district, ZIP Code area or service region that shares its name.

Positions and rankings

Point estimate. The single published value of an estimate, without its margin of error. Rankings order point estimates.

Position. Where an area falls when areas are ordered from the highest point estimate (1st) down. A position describes an estimate; it is not a judgment that higher is better.

Supported range. The span of positions that pairwise 90% tests cannot rule out: from one plus the number of areas statistically higher to the total minus the number statistically lower. A state shown as “3rd–9th” cannot be called third with confidence. See the state rankings for examples.