Interesting Results from the ACS - ctpp-news

6 Mar 2004

The email shown reproduce below was posted by Thabet Zakaria.  Because
of the volume of spam/viruses in the last few weeks it inadvertently got
snagged by our filtering system.   I have reposted it below along with
its original attachment..

MEMORANDUM
Date:  February 27, 2004
To:  ctpp_news(a)chrispy.net
From:  Thabet Zakaria, Ph.D.
Subject:  Interesting Results from the ACS

The recent e-mails to the cttp network are interesting.  Dr. Edward L.
Hillsman and his colleague have found out that the average 2002 commute
time to work from the ACS for King County, WA was 5.7% lower than that
obtained from the 2000 census.  They also indicated that the average
commute time over the two year period is down 10% in Washington, DC.  Ed
Christopher of the FHWA tells us that the Census Bureau is going to have
a news conference to discuss new ACS data which show that Albany County
residents have higher levels of education and household income than the
average resident of New York state.  These conclusions are based on the
results of the ACS that was conducted nation-wide in the last several
years based on a small sample of households.

Last year, I reviewed the ACS estimates for the basic variables needed
for transportation planning in the DVRPC region, which includes five
counties in Pennsylvania and four in New Jersey and found the following:

1. The errors in the county estimates of population and employment are
very large and cannot be compared to the Year 2000 census (about 6%).
For example, the ACS underestimated the population of the City of
Philadelphia, PA by 54,731 people.

2. The errors in the means of transportation to work estimates are even
larger than those for population and employment (about 25%).  For
example, the standard error in the estimates of means of transportation
for Mercer County, NJ ranges from 4.54 to 30.69% for workers who drove
alone or took buses to commute to work, respectively.

These erroneous findings are expected since the sample size is too small
to produce comparable estimates to the Year 2000 census.  Based on these
results, we concluded that ACS estimates cannot be used for any of
DVRPCs transportation planning studies.  The ACS estimates cannot be
compared to the decennial census because the sampling errors in the ACS
results are much higher.  In summary, the current ACS results cannot be
used to make rational conclusions for transportation planning analysis.

TZ:ms