How small is small ? Thoughts on TAZs - ctpp-news

11 Mar 2011

Many of you will be working on TAZ and TAD delineations, starting on
March 16.  March 16 is the day the Census Bureau Geography office hopes
to have the software and User Guide available, and for many states to
have the geographic files for downloading.   If you are part of the TAZ
delineation program, you will get a email directly from the Census
Bureau with the password information.  

I have been reading the NCHRP 08-79 Task 6 "validation phase" report.
(The August 2010 CTPP Status Report includes an article about this
project http://www.fhwa.dot.gov/ctpp/sr1008.htm ) You will recall that
this project is developing a method to create partially synthetic
microdata ACS records to use in the CTPP 5-year (2006-2010) tabulation.
This is necessary otherwise, most of the tables with the variable "means
of transportation to work" would be suppressed.  That is, for this round
of the CTPP,  "partially synthetic data" was determined to be preferred
to data suppression for small geography. 

Only SOME of the CTPP tables are deemed to be risky for individual
disclosure, so only SOME of the CTPP tables will have to use the
synthetic microdata records.  The other CTPP tables will use the actual
ACS microdata records.  

If you attended the TAZ training webinars (2/25 and 2/28), and if you
reviewed the TAZ software specifications, you know that the software
WILL allow for very small TAZs, but the software will warn you if the
resident or estimated worker counts are below 600.  In the NCHRP 08-79
validation phase, the actual ACS data was compared to the synthetic ACS
in the protected environment at the Census Bureau.  For a medium sized
MPO in the test, the MPO's new TAZ system for their regional model has
very small TAZs (on average only 300 residents per TAZ), and in
CTPP2000, the average was 1000 per TAZ.   Although the synthetic data
performed as well as the actual ACS in model output runs, Westat found
that the "ACS cell sample size will create data usability issues for
transportation planners at fine levels of geography, e.g. TAZs, for
cross-tabulations of key variables with means of transportation." 

So, you have a choice.  For defining TAZs for CTPP, you can make larger
TAZs than for your model and have less data perturbation, or you can
make very small TAZs and have a lot of data perturbation.  

Elaine Murakami

FHWA Office of Planning

206-220-4460 