Print Page   |   Contact Us   |   Report Abuse   |   Sign In   |   Register
TL v58n1 The American Community Survey: Your Source for Current Data about Your Community
Share |

Tennessee Libraries 

Volume 58 Number 1


The American Community Survey:
Your Source for Current Data about Your Community 


Eleanor J. Read
Social Science Data Services Librarian
University of Tennessee

Program Abstract: The U.S. Census Bureau’s American Community Survey (ACS) is the new survey program that provides U.S. demographic, social, economic, and housing data on an annual basis.  Learn about the differences between the decennial census data and the ACS data, and the variety of ACS data products that are available.


Since 1790, the United States Census Bureau has been responsible for enumerating the population of the country and collecting other information about various demographic, social, economic, and housing characteristics of the population.  The process by which this was accomplished has evolved over the years to meet requirements of the law and the needs of policy makers and researchers.

The decennial census is the primary data collection program that is familiar to most people.  The decennial census program has, in the past, had two components.  Most households in the United States received the short form questionnaire, which included seven basic questions about the age, sex, race, Hispanic or Latino origin, and household relationships of all members of the household, and the tenure (whether owned or rented) and vacancy status of the housing unit.  About one in six households received the long form questionnaire, which included the aforementioned questions plus numerous others covering the population and housing subjects listed below.

  • Population: Ancestry; Disability; Grandparents as caregivers; Income; Labor force status; Language spoken at home and ability to speak English; Marital status; Migration; Occupation, industry, and class of worker; Place of birth, citizenship, and year of entry; Place of work and journey to work; School enrollment and educational attainment; and Veteran status.
  • Housing: Farm residence; Heating fuel; Number of rooms and number of bedrooms; Plumbing and kitchen facilities; Telephone service; Units in structure; Utilities, mortgage, taxes; insurance, and fuel costs; Value of home or monthly rent paid; Vehicles available; Year moved into residence; and Year structure built.

The collection of the data for the seven core questions on the short and long forms constituted a census of the population.  That is, data were collected from every person and housing unit in the United States, and this provided an enumeration of the population, as required by law, to reapportion legislative districts every ten years.  In years other than the decennial census years, the Census Bureau relies on its Population Estimates Program to produce official annual population estimates.  Subject data collected on the long form, on the other hand, were collected only once every ten years, so as the decade progressed, the data became more and more dated and were less likely to be representative of the current status of the population.  Fortunately, the Census Bureau has implemented a new survey program, the American Community Survey, to simplify the decennial census and provide more timely access to data describing the characteristics of the population.

The American Community Survey

The American Community Survey (ACS) is part of the U.S. Census Bureau’s 10-year census program.  The survey, which contains about sixty questions on the long form subjects listed in the introduction, is replacing the long form (sample) survey in the 10-year census program.  In the 2010 census, there will be no long form survey sent to a sample of the population.  All households in the United States will receive the same short form survey to produce the decennial population count.  The 2010 ACS will provide estimates for the demographic, social, economic, and housing characteristics of the population for that year.

The ACS began as a demonstration project in 1996 with a small sample of communities in the United States.  It has expanded in scope over the years and reached full implementation in 2005.  Individuals living in group quarters, however, were not included in the 2005 sample due to funding issues.  Group quarters include institutional and non-institutional quarters such as nursing homes, correctional facilities, mental hospitals, college dormitories, military barracks, group homes, missions, and shelters.  Group quarters have been included in the survey since 2006.

The ACS is an annual survey conducted monthly.  That is, a sample of households is surveyed each month and all of the data collected during the year are accumulated for calculating estimates.  The sample includes approximately one in forty addresses per year, which is about 2.5% of the housing units.  Over a five year period this adds up to about 12.5% of the population, which is less than the one in six sample (about 17%) for the former long form survey.  The smaller sample in the ACS has ramifications for the release of estimates for smaller geographic areas.

Data Release Schedule

The release schedule for the ACS data is a bit more complicated than for the decennial census.  Because the data collection is sample-based, and the sample is smaller than in the decennial census, it will take up to five years to accumulate enough data to calculate estimates for some geographic areas.  Table 1 shows the release schedule for the next few years.  In any given year, the data are disseminated in several releases, usually starting in August (see the ACS Release Schedule for specific dates and products).

Data Product

Population Threshold

Year of Data Release











Year(s) of Data Collection

1-year Estimates










3-year Estimates










5-year Estimates

All Areas









Table 1.  ACS Data Release Schedule

Three types of estimates will be released, based on the population threshold of the geographic area.  One-year estimates will be released for all geographic areas with a population of at least 65,000.  Three-year estimates will be released for geographic areas with at least 20,000 population.  Five-year estimates will be released for all areas, including census tracts and block groups.  Note that the ACS will not produce any estimates at the census block level.

Since 2005 was the first year of full implementation, the only data available at this writing are 1?year estimates for areas with at least 65,000 population for the 2005 and 2006 surveys.  The first release of 3-year estimates will occur in August and September 2008 for areas with at least 20,000 population.  The first release of 5-year estimates will be in 2010 and will cover all geographic areas regardless of size.  The multi-year estimates include all of the data collected in the three or five year period.  For example, the 2005 3-year estimates to be released in 2008 will be calculated with the data collected in 2005, 2006, and 2007.  Each successive year the multi-year estimates will be recalculated dropping the first year and adding the most recent year.

A consequence of this release schedule in the early stages of full implementation is that all areas at a particular level of geography may not have data available yet. Tennessee, for example, has 95 counties, but only 19 of the counties currently meet the population threshold of at least 65,000.  Thus, at this time, county-level information about the characteristics of the Tennessee population is limited to just 20% of the counties.  When 3-year estimates become available in August, more counties (in the 20,000+ group) will have data available.

Data Estimates and Measures of Error

The types of estimates or statistics that are presented in the ACS data products include the following (with an example in parentheses).

  • Totals (total foreign-born population)
  • Proportions/Percentages (percentage of population enrolled in public school)
  • Means/Averages (mean hours worked)
  • Medians (median family income)
  • Ratios (average household size)

It is important to pay attention to the universe for which an estimate has been calculated so that the meaning of the data may be interpreted properly.  Estimates may be calculated for population (individuals), households (groups of individuals living together, whether related or not), or housing units (living quarters), and various subsets of these (e.g., total population 25 and over, people who are two or more races, occupied housing units).

Because the ACS is sample-based, all estimates are subject to sampling error.  The standard error is a measure of the variability of an estimate due to sampling.  The margin of error (MoE) is an indicator of the precision of an estimate at a given level of confidence.  The Census Bureau has calculated the margin of error at a 90% confidence level for all estimates and provides those numbers along with the estimates.  The margin of error is important if the user wants to make statistical comparisons of ACS data, say from one year to another or between one or more geographic areas.  Note that the long form data estimates from the decennial censuses also were subject to sampling error, but the Census Bureau did not typically release the margin of error numbers in the various data products.

Data Products

The Census Bureau provides a variety of data products to present the ACS estimates, thus allowing users flexibility in accessing the data they need.  The ACS data products currently available include tables, profiles, thematic maps, and microdata files that provide data for the user’s selected geographic area.  The geography and table selections and the data product type may be changed easily by clicking on the breadcrumb links at the top of the page or other links in the left margin of most data product pages.  The notes below the tables explain if/why some data may not be displayed, and if any of the selected geographic areas have insufficient data to meet nondisclosure rules.  The products may be accessed from the ACS website or through American FactFinder.  The examples that follow were obtained from American FactFinder. 

Base/Detailed Tables

These tables present the most detailed data and generally are cross-tabulations of two or more variables (e.g., Sex by Age by Veteran Status).  These tables are analogous to the detailed tables for the decennial census.  In the ACS, the tables may be referred to as base tables or detailed tables.  Some detailed tables have two versions with prefixes B and C in the table number.  The B table is the basic table with all the levels of the variables shown.  The C tables present the same data, but collapsed into fewer levels or categories.

To select a detailed table in American FactFinder, the user must first select the level of geography (e.g., nation, state, metropolitan statistical area, county).  The page will refresh to show the available options after each selection.  For example, if interested in educational attainment for Knox County, Tennessee, and several of its neighboring counties, the user would 1) select County from the geographic type list, 2) select Tennessee from the state list, 3) select Anderson, Blount, Knox, and Sevier counties from the Tennessee county list (one at a time or together using the Ctrl key), 4) click on the Add button to put the county selections in the Current Geography Selections box, and 5) click on the Next button to move to the table selection page.

On the table selection page, the user may select the desired table from the list of tables if he or she knows the table number or wants to scroll to find a table.  Two finding aids are available to help users unfamiliar with the table options to identify appropriate tables.  The By Keyword search brings up all tables that have the entered keyword and the By Subject search allows one to browse a list of subjects, then get a list of tables in the selected subject area.  After selecting one or more tables click on the Add button to put the table selections in the Current Table Selections box, then click on the Show Result button.  The tables will be displayed one after the other if more than one table was selected. 

Figure 1 displays Table B15002: Sex by Educational Attainment for the Population 25 Years or Older.  Note the columns for each county, and within those columns are a column for the estimate and a column for the corresponding margin of error, presented in a plus/minus format.  Data are shown for sixteen levels of education for males and for females.  A collapsed table for this data, C15002, has reduced the number of levels of education to just seven for each sex. 

Figure 1.  Detailed Table Example 

Figure 1.  Detailed Table Example

Data Profiles and Narrative Profiles

The Data Profiles include a variety of characteristics of the population for a geographic area.  There are four profiles covering social, economic, housing, and demographic characteristics.  These are excellent products for quickly finding data on basic characteristics of the population.  If the particular information needed is not available in the profiles, the user should then go to the detailed tables.

The first step is to select the desired geographic area, then the social characteristics profile will be displayed.  Links to the economic, housing, demographic, and narrative profiles are at the top of the page so it is easy to move from one profile to another for the same geographic area.  The estimates in each profile are arranged by subject area (e.g., educational attainment, place of birth, income, house heating fuel, number of bedrooms in housing unit, race, age).  This table has two data columns, the estimate and its corresponding margin of error.  Figure 2 shows the selected social characteristics profile for Knox County, Tennessee.

Figure 2.  Social Characteristics Data Profile Example 

Figure 2.  Social Characteristics Data Profile Example

The Narrative Profile presents data from the data profiles in a text and graphics format rather than with tables.  This is a nice alternative format for users who may not be comfortable dealing with large tables of data.  Figure 3 illustrates the Knox County narrative profile.

Figure 3.  Narrative Profile Example 

Figure 3.  Narrative Profile Example


Selected Population Profiles

The Selected Population Profiles are similar to the data profiles discussed above, but they allow the user to select a specific population group based on race, Hispanic or Latino origin, or ancestry group.  The profile includes selected demographic, social, economic, and housing characteristics.  Figure 4 shows a population profile for the state of Arizona focusing on the American Indian Alone population group.  The table shows the estimates and margins of error for the selected geographic area as a whole in addition to the chosen population group within that geographic area.

Figure 4.  Selected Population Profile Example

 Figure 4.  Selected Population Profile Example


Subject Tables

Subject Tables provide topic-specific summaries for the selected geographic area.  The indicators that are presented in a subject table vary based on the subject chosen.  For example, Subject Table S2201: Food Stamps (Figure 5) includes data on households with older people and children, poverty status, disability status, race and Hispanic origin, household income, and work status, all of which would be relevant to a study of participation in the food stamp program.  Estimates are shown for the whole geographic area and the subsets of households receiving food stamps and households not receiving food stamps.  As with the indicators, the particular population subsets (columns) shown will vary with the subjects.

Figure 5.  Subject Table Example 

Figure 5.  Subject Table Example


Geographic Ranking Tables, Geographic Comparison Tables, and Thematic Maps

The Geographic Ranking Tables and Geographic Comparison Tables provide a quick way to compare data on a particular subject across all available geographic areas within a larger geographic level.  Comparing all states within the United States or all counties within a state are common, but other levels of comparison are also available, including all metropolitan and micropolitan statistical areas within the United States and all cities within a state.

The Geographic Ranking Table ranks the states based on the values of their estimates for the selected subject.  There is no need to select a geographic unit because the ranking is already set to be all states and the United States.  The user selects the subject of interest from a list organized by broad subject group.  The table displays the states in descending order of their estimates.  The table can be resorted in alphabetical order of the states by clicking on the small down arrow next to the column heading for State.  From links on the left side of the page, the user can view the data as a chart, which graphically shows the estimate from each state with its confidence interval (estimate plus or minus the margin of error).  This enables a visual determination of statistically significant differences between the states.  Another link at the left allows the user to highlight statistical differences in the table.  Figure 6 shows an example of a ranking table for the Percent of People Born in Mexico.


Figure 6.  Geographic Ranking Table Example 

Figure 6.  Geographic Ranking Table Example

The Geographic Comparison Tables are similar in layout to the Ranking Tables, but they do not order the geographic areas by magnitude of the estimates.  They do, however, provide more options for comparison of areas within a larger geographic area, a few of which were noted earlier.  The user must select either Nation or States to indicate the level of comparison.  If State is selected, then the state of interest is selected from a pull-down list.  Next, the user selects the  table format, followed by the subject table to be displayed.  Figure 7 shows GCT0801: Mean Travel Time to Work for the counties in Tennessee.  Note that only the 19 counties that currently have estimates available (those with at least 65,000 population) are shown.

Figure 7.  Geographic Comparison Table Example 

Figure 7.  Geographic Comparison Table Example


The Thematic Maps display the data in the geographic comparison tables graphically in a map.  A map can be made from the table display page by clicking on the “view as a thematic map” link at the left side of the page.  The map for the comparison table noted above is very sparse at this point since 80% of the counties do not have estimates yet.

Public Use Microdata Sample

All of the data products discussed so far provide aggregate data.  That is, the reported totals, averages, percentages, and such are based on aggregated responses from the survey respondents.  A Public Use Microdata Sample (PUMS) is a dataset that contains the original data (microdata from individual persons or housing units) from a sample of the housing units that responded to the survey.  The PUMS file may be used to create custom tabulations if the standard ACS data products do not provide data in the desired format, or to do statistical analysis or modeling on the microdata.  Using PUMS files requires analytical skills and the ability to deal with a large data set and a statistical program.  PUMS files are more likely to be used by advanced researchers.


The American Community Survey offers students, researchers, policy makers, and government officials access to current data on the characteristics of the U.S. population to help prepare school papers, conduct research, make decisions, prepare grant proposals, and provide social services, among other things.  Although it will be a few more years before a full set of data products is available that covers all geographic areas of the United States and Puerto Rico, the American Community Survey already provides more relevant and timely information about the United States than ever before this late in the decade.  The American Community Survey is a significant and valuable resource from the U.S. Census Bureau, and is a welcome addition to the 10-year census program. 


Appendix: Key Web Pages and Helpful Documents


Membership Software Powered by®  ::  Legal