The American Community Survey (ACS) replaced the decennial long form beginning in 2005. While the subjects are very similar between the ACS and the decennial long form, the survey methodology is substantially different.
The American Community Survey was created to gather detailed trend data for planning purposes on a more frequent basis than the decennial censuses. It is run every year and uses a quite different methodology from the decennial census long form. For this reason, while the subjects included in the ACS are very similar to the decennial long form subjects, data from the two surveys are often not directly comparable. Subjects are quite varied and include data on income, education (enrollment and attainment), employment and occupation, transportation to work, health insurance coverage, detailed housing characteristics like prevalence of complete kitchen and bathroom facilities and heating fuel, and more. Subjects are subject to revision (health insurance coverage was added to the 2009 questionnaire); check the technical documentation for specific subjects.
The ACS began full implementation in 2005 but group quarters were not included until 2006. The questionnaire is administered every month and the overall sample is accumulated over multiple years: geographies with more than 65,000 people are represented in the one-year (annual) data; geographies with between 20,000 and 65,000 people are represented in the three-year data; geographies with fewer than 20,000 people are represented in the five-year data. Data in the ACS, like the decennial long form, are available down to the Census block group level but estimates at this level may not be reliable given the smaller sample size of the ACS (12.5% before 2011; about 13.5% after 2011). Users should pay close attention to the Margins of Error published with the data—they can be quite large for small groups of population and small geographies.
The Population and Housing Census is required by the United States Constitution decennially (every ten years) to enumerate the population for purposes of representation, that is, for the redistricting process. It is conducted in years ending in zero. Its questions are asked of all residents and for this reason it is sometimes referred to as 100% data; it is also known as short form or Summary (Tape) File 1 data.
From 1940 to 2000, a long form questionnaire was included in the Population Census to collect more detailed data from a sample of the population (about 17%). It was referred to as long form, sample or Summary (Tape) File 3 data. In 2010, the long form was dropped from the decennial census. Those data are now collected via the annual American Community Survey.
The subjects covered by the decennial census now are relatively basic counts of individuals by age, race & ethnicity, sex, households and housing units, but also includes description of household relationships, occupancy status and tenure (whether the residence is owned or rented). Data are available down to the smallest Census geography, the Census block.
Maki images courtesy of Mapbox
American FactFinder has only the most recent years of its surveys; Social Explorer and NHGIS have all the population ones back to the first one in 1790.
American FactFinder has ALL available geographies (except block groups for ACS data prior to 2013); Social Explorer and NHGIS have block groups for all ACS years. NHGIS has all geographies. Social Explorer may not have some less common geographies.
American FactFinder has all available variables; Social Explorer doesn’t have variables that involve some suppression. NHGIS is working to include all but is a work in progress.
Social Explorer and American FactFinder both have this capability; American FactFinder is harder but has some variables (from current surveys!) that Social Explorer does not.
Only NHGIS allows download of shapefiles for use with ESRI mapping products.
Only DataFerrett or IPUMS* provide access to public use microdata sample (PUMS) data (but the only available geographies are public use microdata areas [PUMAs], super-PUMAs or higher levels of geography, e.g., states).
Only DataFerrett* allows you to create your own, but only with PUMS data, so only with PUMS geographies.
American FactFinder has most of the (current) Bureau surveys; Social Explorer has all of the decennial population surveys, 1790 to present plus a few other things; DataFerrett* has decennial and ACS data, the Current Population Survey and a few other Bureau surveys (SIPP, SAIPE, et al.), plus non-Bureau data like the Home Mortgage Disclosure Act data; the National Survey of Fishing, Hunting and Wildlife-Associated Recreation; and much more.
Social Explorer doesn’t provide margins of error (MOEs) with the data. If you need to be extremely precise, get data from NHGIS or AFF to be able to evaluate the MOEs.
*If you need help, see contact information above at right for the Librarian for Numeric Data Services and Data Management.