In 2005, the Census Bureau changed the process it employs across the United States to gather detailed demographic data.  This new survey, titled the American Community Survey (ACS), involves choosing a statistically valid sample of individuals every year to receive a questionnaire asking for detailed information.  Recipients are requested to complete and mail the ACS form back to the Census Bureau.  Follow-up is conducted by Census Enumerators to enhance response, but this is not as extensive as what is undertaken for the Decennial Censuses.  The ACS is unlike the Decennial Census as responses are requested from only a select number of individuals rather than from everyone. 

The institution of the ACS made the Census Bureau discontinue use of the Decennial Census “long form,” which formerly gathered similar details every ten years from a subset of the entire population.  The annual sample size (number of individuals asked for information) for the ACS is smaller than what had been utilized for distribution of the Decennial Census “long form.” 

Although ACS information is collected annually, it is not necessarily reported every year.  For geographic areas with greater than 65,000 people, summary reports are issued annually.  For areas with 64,999 – 20,000 people, the information is combined into three year findings.  Unfortunately, for areas with under 20,000 residents, such as VMI, the sample information is collected for five years and combined before any data are reported.  The reason the Census Bureau does not release information for smaller geographic areas more frequently is because staff do not have the confidence that data from small sample sizes are statistically correct.  The Census Bureau describes this information gathered over a period of time as “pooled data” and specifies that “Multiyear estimates do not represent any one year or the midpoint of a period; they represent the entire period over which the data were collected.” (Bureau of the Census, “Understanding Multiyear Estimates from the American community survey”, Compass/presentations.html). 

An example of the difference in sample size between the use of the “long form” for previous Decennial Censuses and the ACS can be seen by comparing the number of people included in the first ACS reporting period for VMI, 2005-2009, versus the number sampled using the last “long form” during the 2000 Decennial Census.  For the 2000 Census, 13% of the VMI population (1,354 people) were included in the sample which asked detailed demographic questions.  This information was then reported as relevant to the year 2000.  For the first reported ACS for VMI, less than half that number were sampled: 653 individuals or 6% of the Island population.  This ACS sample of VMI residents was accumulated over a five year span, so that the information reported is not related to any single time period but rather refers to all of 2005-2009.

Additionally, when ACS data are presented, the Census Bureau includes a “margin of error” relative to every reported figure.  This statistical calculation reflects the level of confidence the Census Bureau has in the accuracy of the statistic.  For example, an ACS estimate of the VMI poverty rate of 8.2% for the period 2005-2009 came with a margin of error of 2.9%.  This means that the actual poverty rate is predicted to fall within a range from 11.1% to 5.3%.  Comparatively, the 2000 Decennial Census estimated the poverty rate for VMI during 1999 to be 6.0%.

Unfortunately, the information gathered through the ACS includes topics such as marital status, occupation, education, income level, poverty status, and many other characteristics that have previously been presented in Decennial Census data – some from the beginning of such collection on VMI in 1880.  Due to the loss of the Decennial Census sample “long form” and use only of the ACS, this valuable information will no longer be available every ten years to present a snapshot of VMI residents at one point in time but will only be offered for five-year periods.

Because of lack of confidence in ACS statistics for VMI, no ACS data are reported on this Census Project website.  For those interested, ACS data specific to VMI can be located through the Census Bureau’s searchable site American FactFinder at:  The researchers involved with the Census Project urge any ACS users to be aware of the small sample size for the survey and the fact that data are reported for a combined five-year period when drawing any conclusions concerning VMI from this information.

March 29, 2014