Despite the uncertainty of 2006 American Community Survey (ACS) budgeting and
implementation, I am raising this query on future CTPP-like products using ACS.
For the CTPP2000, there were 2 FLOW (part 3) tables using income:
Table 3-5 Income (8 classes + total)
Table 3-7 Income (4 classes +total ) by Means of Transportation (4 classes + total)
Both of these tables were subject to the Census Bureau's Disclosure Review Board (DRB)
requirement of 3 unweighted records, resulting in very high suppression of
Origin/Destination pairs at the tract-to-tract (about 70 percent), and TAZ-to-TAZ (often
over 80 percent) level. Because the sample size in ACS (even after 5 years of data
accumulation) are expected to be perhaps 40-60 percent that of the 2000 decennial census
long form sample, it will be even more difficult to reach the threshold of 3 unweighted
records for flow tabulation. The CB's DRB has told us that for special tabulations
(like the CTPP) "rounding and thresholds are here to stay." Right now, the next
level of geography is Place (city or town), which is often too large to be meaningful. We
understand that the CB plans to use PUMA as a standard tabulation unit, which could result
in a very useful PUMA-to-PUMA flow table for large metropolitan areas, but since PUMAs
require 100,000 population, this would be not very useful for smaller metropolitan areas.
Here are some alternatives to consider:
1. Using MEDIAN INCOME instead of a distribution. At least one person at the CB thinks
that the DRB would be less concerned about this. While this might work for Table 3-5, I
don't think it would work for Table 3-7, because there would be a significant
probability for one record for a particular mode, e.g. transit, so that the median would
be the same as the individual record.
2. Having a SuperTAZ or SuperTract new geographic unit for flow tabulations that include
an income variable. How big would a SuperTAZ or SuperTract need to be? I don't know,
but my guess is that it would need to be on the order of 4-5 census tracts combined. We
know that if we combine only 2 tracts, it will result in approximately 70 percent of
suppression, because the unweighted records would be similar in number to the CTPP 2000
3. Omit the income tables from an ACS Flow tabulation, and find another data source.
Please let me know what you think, and especially if you have ideas for alternatives.
AASHTO SCOP has convened a Census Data Working Group. It is chaired by Jonette Kreideweis
of MN DOT. Information about this group was included in the August 2005 CTPP Status
. AMPO and APTA have now been included as
FHWA Office of Planning