Center for Family and Demographic Research
American Community Survey (ACS)
What is the American Community Survey?
The American Community Survey (ACS) is a nationwide survey providing yearly data on social, economic, demographic, and housing characteristics of the U.S. population. The ACS is designed to replace the Decennial Census long form, providing information on a yearly basis as opposed to every ten years. The Census Bureau began full implementation of the ACS in 2005 with a sample of three million addresses each year. The ACS includes people living in housing units and group quarters throughout the United States and Puerto Rico.
How do I access ACS data?
The easiest way to obtain estimates from the American Community Survey (ACS) is through American FactFinder http://factfinder2.census.gov/main.html. The U.S. Census Bureau makes available several ACS data products via American FactFinder, including detailed tables, data profiles, subject tables, ranking products, geographic comparison tables, thematic maps, and selected population profiles. Help using American FactFinder can be found here: http://www.bgsu.edu/downloads/cas/file52384.pdf.
Often researchers need information that is not available in the American Fact Finder tables. In this case, researchers need to use the Public Use Microdata Sample files (PUMS) for their analysis. Data files are available through the U.S. Census Bureau: http://www2.census.gov/.
What preparation is needed to use ACS data?
These datasets are very large, so it is very important to select the appropriate data before downloading. You must choose between the 1-year, 3-year, and 5-year estimates and between the housing and population records. For data on populations of 65,000 or more, 1-year, 3-year, and 5-year estimates are available. Data are available for populations of 20,000 or more in 3-year and 5-year estimates. For small geographic areas with populations less than 20,000, only 5-year estimates are available. In general, the single-year estimates are preferred, as they will be more relevant to the current conditions. For queries about the attributes of geographic areas such as metropolitan areas, counties, places, minor civil divisions, tracts, or block groups the population records should be used. For queries about the attributes of individual housing units, households, and people the housing records should be used. The population and housing records can be merged. Data files are available for the entire U.S. or may be downloaded for individual states.
Since ACS files are so large, they must be downloaded as separate files and then appended into one file for analysis. See the “How do I append ACS files in SAS” or “How do I append ACS files in Stata” documents that are available online if you need assistance appending the files: http://www.bgsu.edu/organizations/cfdr/page97321.html.
How do I weight ACS data?
Since the ACS uses a complex sample design, all standard errors related to the estimates and significance tests need to take into account the design. When using American FactFinder to obtain tables, researchers will find that these tables already provide the estimates of “margin of error” that were adjusted for sampling design. With these estimates, researchers can examine whether the point estimate of the relations between two variables are significantly different from zero. If researchers want to construct custom tables from American FactFinder and calculate “margin of error” for the new tables, the instructions for doing so can be found here: http://www.census.gov/acs/www/Downloads/handbooks/ACSGeneralHandbook.pdf.
When using PUMS data for analysis, researchers need to use replicate weights to calculate accurate standard errors of parameter estimates. These standard errors describe the confidence intervals around the parameter estimate and are essential for hypothesis testing. Replicate weights are available for the ACS starting in 2005. ACS data provide 80 replicate weight variables at the household and population levels that allow users to derive accurate standard errors of parameter estimates. Information about how to correct standard errors using replicate weights in Stata and SAS can be found here: http://usa.ipums.org/usa/repwt.shtml.
Where can I learn more about the ACS?
The U.S. Census Bureau releases informative handbooks on the ACS called the Compass Series which can help determine if the research question should be answered with 1-year estimates, 3-year estimates, or 5-year estimates. Several handbooks are available, including one called, “What General Data Users Need to Know” found here: http://www.census.gov/acs/www/guidance_for_data_users/handbooks/.
Additional tools for using the ACS can be found on the Center for Family and Demographic Research’s website: http://www.bgsu.edu/organizations/cfdr/page97321.html. For further help using the American Community Survey or questions about this handout please stop by the Center for Family and Demographic Research at 5 Williams Hall or email us at CFDR@bgsu.edu.