Census 2010 will not include a long form.
The socio-economic data previously collected on the
long-form every ten years
has been replaced with the annual American Community Survey (ACS).
American Community Survey (ACS) from the Census
ACS functions as an ongoing perennial survey of population, housing, economic and income statistics. ACS will provide yearly updated socio-economic data for communities with a population of 65,000 and larger. Communities with populations between 20,000 and 65,000 will obtain updated data every 3 years. The remaining communities, with population less than 20,000, will obtain updated data every 5 years.
Having updated socio-economic data on a 1-, 3-, and 5-year period is an improvement over the long-established 10-year interval. However, there is a considerable drawback to ACS data.
On August 28th, ACS released 2006 personal income data for Connecticut’s statewide population and the eight towns with populations over 65,000: Bridgeport, Danbury, Hartford, New Britain, New Haven, Norwalk, Stamford, and Waterbury. Click on any chart to get a Printable view.
1. Missing Statewide Household Income
|
Chart 1 compares IRS Adjusted Gross Income (AGI) and CT AGI to household income from Census 2000 and the ACS.
This chart illustrates that ACS statewide totals consistently understate household income in CT.
This is due to income derived from capital gains, recorded by the IRS, but which the Census/ACS does not include .
See the report |
2. The Average (or Mean) is Misleading in Connecticut
ACS reports that the mean household income in Connecticut in 2006 was $87,856 – while the median was lower at $63,422. Comparison of these statistics shows that income in Connecticut is highly skewed. With the average almost 20% above the median provides evidence of a substantial component of upper income households that is not typical of many statewide averages. As a result, income statistics based on statewide averages in Connecticut mis-represent the "average citizen" and should not be used in public policy or socio-economic analysis.
A preferable alternative could be to use median household income statistics that will more truthfully represent the typical household in Connecticut. The median represents the household that is in the middle – 50% of households are above/below the median.
|
Chart 2 tracks median household income in Connecticut from 1999 to 2006. These income statistics are inflation adjusted to 1999 dollars allowing for a real-dollar comparison. From this chart it is clear that the typical household in Connecticut had a higher income in 1999 than in 2006. This is counter to the conclusion that would be made if average income were tracked. As previously stated, average income statistics are misleading for Connecticut. |
3. High Margin-of-Error for Town Statistics
In their March 2007 report, Using the American Community Survey: Benefits and Challenges the National Research Council (NRC) reported the following concerns:
- ACS estimates will have higher sampling error than the 2000 census long-form.
- 1- and 3-year period estimates may have a margin-of-error as high as +/- 24%.
- 5-year period estimates may have a margin-of-error as high as +/- 45%.
|
Chart 3 tracks the aggregate household income from 2004 to 2006 for the eight Connecticut towns with populations greater than 65,000. In 2005, Danbury had the highest margin-of-error at +/- 13.4%. In 2006, Stamford had the highest margin-of-error at +/- 11.5% or +/- $635,000,000. |
Our Concern
It is questionable whether ACS data, which is more frequent but with significant error margins, is of greater value than the traditional decennial census long-form data, which was less frequent but with less error.
The impact of deficient ACS data on public policymaking will be unpredictable.
| Contacts |
|
Download Adobe Reader 8,
or earlier version, at no charge.