No valid historical comparisons
One limitation to the American Community Survey (ACS) is that since
it is so new, it's not yet possible to use it to make valid comparisons
over time. ACS data cannot accurately be compared to data from either
previous long-form Censuses or from the Current Population Survey-ASEC.
There are ACS data going back to 2000, but these are not perfectly comparable either. The ACS
went national for the first time in 2000, when it was conducted
alongside the long-form in order to allow for comparisons between the
two. The sample size was much smaller than the three million
needed to replace the long form and it was called the Census
Supplementary Survey. The survey remained at that size through 2004,
although it switched names to the ACS in 2002. Results from the
2005 data are based on a much bigger sample size than previous years and it is quite
possible that differences between previous surveys and the 2005 survey
could be related to this change in the survey .
Reference year for income data
One of the most important set of
questions in the ACS ask about income for the "past 12 months." Since the survey
is administered throughout the year, the income questions ask
different respondents about different time periods. In 2005, the
households who received the survey in January 2005 were asked about
February 2004 through January 2005, while the December respondents were
asked about January 2005 through December 2005. While the results of
the survey are reported as 2005, they are a mix of 2004 and 2005. This figure shows
how this works. The black line shows what percentage of the three
million 2005 respondents were asked about each month from February 2004
through December 2005. The grey line shows the same for the 2006
respondents. There is considerable overlap between the two periods,
meaning that the ACS income data is a sort of rolling average, which
would tend to smooth data between years and dampen year-to-year changes.
Simplicity of income questions
The Current Population Survey's Annual Social and Economic Supplement (CPS-ASEC) asks dozens and dozens of detailed
questions and follow up questions over the phone. By comparison,
the ACS questionnaire asks only
eight questions about income on a mail-in form. The more complex
questions of the CPS-ASEC as well as the fact that all respondents are asked
about the same time period (the previous calendaryear) may mean that the CPS-ASEC income data are more accurate.
2008 rewrite
The Census Bureau plans a
significant reconstruction and improvement of the questionnaire in 2008
which will cause a break in the data, exarcerbating the problem of
having no valid historical comparison.
Return to main ACS/CPS-ASEC page.
|
