While the US census collects data from all US households every 10 years, the American Community Survey (ACS) draws on a probability sample of these households, providing a timely picture of the nation every year.
After a decade of testing to evaluate the reliability of the survey, the US Census Bureau launched the ACS in 2005. It replaced the census “long form” as the nation’s foremost source of data on population and housing characteristics. Currently, the ACS draws its data from an annual sample of approximately 3.5 million households, reporting on the social, economic, housing, and demographic characteristics of individuals and communities in the US.
ACS data is collected throughout the year and published both annually and in five-year increments, pooling together all responses gathered within that time frame.
An explanation of the differences between the American Community Survey and the decennial census is available here. For more on probability samples and how they allow us to estimate the characteristics of a population and the amount of error associated with our estimates, see Statistics Canada’s explainer.
The ACS 5-Year Public Use Microdata Sample that the US Census Bureau makes available to researchers for custom analyses, like those on which the state profiles are based, contains approximately two thirds of all responses to the survey. It provides data on approximately five percent of the US population.