The American Community Survey is the ambitious new national survey from the U.S. Census Bureau that is replacing the long form portion of the decennial census for the new millenium. While some
version of this survey has been in the field since 1999, it was not fully implemented in terms of coverage until 2006. In 2005 it was expanded to cover all counties in the country and the 1-in-40 households sampling rate was first applied. However, persons living in group quarters (such as nursing homes, dormitories and prisons) were not added to the survey until 2006. (The original plan was to begin GQ coverage in 2005 but last-minute budget reductions delayed it for a year.)
The full implementation of the (household) sampling strategy for ACS entails having the survey mailed to about 250,000 households nationwide every month of every year and was begun in January 2005. In January 2006 sampling of group
quarters was added to complete the sample as planned (albeit several years later than originally planned.) In any given year about 2.5% (1 in 40) of U.S. households will receive the survey.
Over any 5-year period about 1 in 8 households should receive the survey (as compared to about 1 in 6 that received the census long form in the 2000 census). Unfortunately, receiving the survey is not the same as responding to it, since the Bureau has adopted a strategy of sampling for non-response. This has resulted in something closer to 1 in 11 households actually participating in the survey over any 5-year period.
Data Release Plan Based on Population Size of Geographic Area
Data based on the ACS surveys for any calendar year will be published in the late summer of the following year for geographic areas with a minimum of 65,000 population. For smaller areas the Bureau will only publish data based on surveys for multiple consecutive years as follows:
- For geographic areas of 20,000 or more population data will be published based on 3 consecutive years of survey data. Thus, for example, the first time we saw data tabulated for Jefferson City, MO (population around 40,000) was in December of 2008 and it was based upon the surveys done in 2005 through 2007. In late October, 2009 data were released based upon the surveys taken in 2006-2008, etc.
- For geographic areas with populations under 20,000 (down to the block group level, but NOT to the block level) data will be published based on 5 consecutive years of survey data. Thus, for example, data for the majority (60 out of 115) of counties in Missouri, and for 46 out of 53 counties in the state of North Dakota, and for all census tracts and block groups everywhere, data will first appear some time late in 2010 and will be based upon the combined survey data of 2005 through 2009. New data for these areas should then be published each year, based upon the most recent 5 years of surveys. Even if a census tract should happen to meet the population threshold of 20,000 (which is rare) no data will be published for it other than the 5-year "period estimates". Similarly, data at the ZIP code/ZCTA and state legislative district levels will only be published as 5-year period estimates, even though there are many ZIP codes and LD's that meet the 20,000 threshold.
- View this Map of Missouri by County (pdf file) showing the three population categories (<20K, 20K to < 65K and 65K+) relevant to ACS data products.
In addition to the population threshold rules that are used to limit the publication of data for geographic areas the Bureau also applies their "data release rules" for each
table for each geographic area (that passes the total population threshold filter). Basically they analyze the cells of a table and assign a
measure of the statistical reliability of each cell based on the margin of error. We have indicated our dismay at the algorithm elsewhere (see "Thing 6" in
our Ten_Things to Know... page). The link there to the document that describes
the algorithm is broken, but we found it (or at least something quite similar) at http://www.census.gov/acs/www/Downloads/survey_methodology/acs_design_methodology_ch13.pdf .
The following excerpt from the 8-page document outlines the method. Be sure to note the final sentence. :
Data Release Rules
Even with the population size thresholds described earlier, in certain geographic areas some very
detailed tables might include estimates with unacceptable reliability. Data release rules, based on
the statistical reliability of the survey estimates, were first applied in the 2005 ACS. These release
rules apply only to the 1- and 3-year data products.
The main data release rule for the ACS tables works as follows. Every detailed table consists of a
series of estimates. Each estimate is subject to sampling variability that can be summarized by its
standard error. If more than half of the estimates in the table are not statistically different from 0
(at a 90 percent confidence level), then the table fails. Dividing the standard error by the estimate
yields the coefficient of variation (CV) for each estimate. (If the estimate is 0, a CV of 100 percent
is assigned.) To implement this requirement for each table at a given geographic area, CVs are cal-
culated for each table’s estimates, and the median CV value is determined. If the median CV value
for the table is less than or equal to 61 percent, the table passes for that geographic area and is
published; if it is greater than 61 percent, the table fails and is not published.
Whenever a table fails, a simpler table that collapses some of the detailed lines together can be
substituted for the original. If the simpler table passes, it is released. If it fails, none of the esti-
mates for that table and geographic area are released. These release rules are applied to single-
and multiyear period estimates based on 3 years of sample data. Current plans are not to apply
data release rules to the estimates based on 5 years of sample data.