American Community Survey (ACS) data files:
The ACS is comprised of 3 different data files:
- ACS 1-Year Estimates
- ACS 3-Year Estimates
- ACS 5-Year Estimates
In previous decades the Long Form (SF3) was released by the US Census Bureau (USCB). That is no longer the case. They have now opted to do a continually running survey. In order to collect enough data so as to model meaningful and valid statistics about small locations the USCB has agglomerated data from multiple years.
Advantages of each year of the ACS.
The methodology and variables are almost identical between the 2009, 2010 and the 2011 data files. The
difference is the geographic boundaries.
2009 ACS is still in the 2000 boundaries which makes it ideal when comparing it to the 2000 Long Form.
2010 ACS (and all subsequent years) are in the 2010 boundaries (as established with the 2010 Redistricting)
2011 ACS (and subsequent years) has added the Zip Code Tabulation Area as a geography for which you can access 5-year estimates
2012 ACS is almost identical to the 2011 except that there are some newly added tables, and of course the 2007 data has been dropped and instead the 2012 added.
The Geographic Differences Between the Different Durations of the Estimates
The 1-year estimates have data just from the survey year and are available for geographic areas with populations of 65,000 or more.
The ACS 3-year estimates were released for geographic areas with populations of 20,000 or more and were collected over 3 years from January two years prior to the survey year through December of the survey year. For example for the 2010 ACS 3-year estimates they will run from January 2008 – December 2010.
Using the ACS 5-year estimates you can analyze small areas (with a population less than 20,000) such
as Block Groups and Tracts (and starting with 2011 the zip code). The problem is that it took 5 years
of surveying 2 million people a year to accumulate a large enough sample to provide estimates with
accuracy similar to the decennial census. This data file will include data collected starting in January five
years prior to the survey year and running for 5 years through December of the survey year (using 2010
as the example January 2006 – December 2010). The data does not show information for individual years
but rather only a single number for the average of the area for the 5 years.