Subcommittee on Census Data for Transportation Planning--A1D08(1)
 Annual Meeting Minutes
January 12, 1998
 Washington, DC

1.0 Statement from the Chair
The meeting of the Census Data for Transportation Planning Subcommittee A1D08(1) was called to order at 7:40 p.m. by the Chair, Ed Christopher, Chicago Area Transportation Study.

The attendees were:

Carlos H. Arce  NuStats International
Charles Baber  Baltimore Metropolitan Council
Gene Bandy   Baltimore Metropolitan Council
Todd Blair   Bureau of the Census
Deborah Buchasz  AASHTO
Carolee Bush  Bureau of Transportation Statistics, DOT
Mark Bryam   Ohio Dept. of Transportation
Ed Christopher  Chicago Area Transportation Study
Nathan Erlbaum  New York State Dept. of Transportation
Jerry Everett   FHWA
Leslie Goldenberg  AMPG
Ed Herlihy   MCV Assoc.
Pat Hu   Oak Ridge National Lab
Joe Huegy   Triangle Transit Authority
Ho-Ling Hwang  Oak Ridge National Lab
Pat Johnson   Bureau of the Census
Robert LaMacchia  Bureau of the Census
Edward Limoges  Southeast Michigan Council of Governments
Tom Mank   FHWA
Rick Miller   Kansas Dept. of Transportation
Elaine Murakami  FHWA
David Pearson   Texas Transportation Institute, TexasA&M Univ
Chuck Purvis  Metropolitan Transportation Commission
Michael Ratcliffe  Bureau of the Census
Phil Salopek   Bureau of the Census
Ashish Sen   Urban Transportation Center, U. of Ill. at Chicago
Robert Sicko  Puget Sound Regional Council
Siim Sööt   Urban Transportation Center, U. of Ill. at Chicago
Todd A. Steiss  Richmond Regional PDC
Darwin Stuart  Chicago Transit Authority
Nancy Torrieri  Bureau of the Census
Cheng-I Tsai  OKI Regional Council of Governments
Ron Tweedie  New York State Dept. of Transportation
Anant Vyas   Argonne National Lab
Jay Waite   Bureau of the Census
George Wickstrom  Transportation Consultant
George Wiggers  Transportation Consultant
Ernest Wilson, Jr.  Bureau of Transportation Statistics, DOT
Thabet Zakaria  Delaware Valley Regional Planning Commission

Ed welcomed the attendees to this the inaugural meeting of the subcommittee.  He addressed the purposes of the subcommittee, and stated that we face many pressing issues.  He said we don’t have an official membership, and that we will probably form subgroups to deal with individual issues.  The following handouts were distributed:

· Subcommittee meeting agenda [1.1]
· Subcommittee web site pages [1.2]
· TRB transportation data committees and task force folder [1.3]
· "The Cliffhanger Census" editorial, American Demographics, January 1998 [1.4]

Chuck Purvis, Metropolitan Transportation Commission, Chair of the Subcommittee’s parent Committee on Urban Transportation Data and Information Systems A1D08, welcomed the new subcommittee.  He said that it represented a continuation of a long string of efforts on census data, and commented on Ed Christopher’s interest and enthusiasm, including the creation of a webpage before the first meeting.  Regarding the next two agenda items, 2.0 and 3.0, Ed Christopher said that Alice Watland of TRB was unable to attend due to illness, and then asked the attendees to introduce themselves.

4.0 Census 2000 Initiatives
Jay Waite said that what he sees is pretty scary.  The Census Bureau works for Congress, and there are members who believe that we do the decennial census only to count people.  At this point in the census cycle, the number one concern of Bureau staff is how to use our resources to do the count.  The focus is on getting out there, finding everyone, and counting them.

Historically, the decennial census has produced an undercount of the population.  Undercount occurs at different rates, depending upon population characteristics.  In 1990, while the overall undercount rate was about 1.6 percent, it was 0.7 percent in white middle class areas, compared to around 5.0 percent in black inner city areas.

For the 2000 census, the Bureau wants to use a type of sampling designed to counteract the problem of undercount.  But because the undercount rate varies by population group, there is concern within Congress that statistical adjustment will favor one state, group, or political party at the expense of another, by differentially affecting the allocation of post-census representation and revenues.  The 1998 appropriation specifies that by February 1999 the Bureau will be prepared to do either a sampling (for undercount) census or a nonsampling census.  At that time, the decision on 2000 census sampling for undercount adjustment will be made.

The Census Dress Rehearsal has been delayed until April 18, due to the dispute over sampling.  The three locations for the Dress Rehearsal are Sacramento CA, several counties around Columbia SC, and the Menominee Indian Reservation, WS.  In the Columbia SC area, there will be "traditional census", with a 100 percent follow-up of nonresponding households.  In Sacramento and the Menominee reservation, the Census Bureau’s proposed sampling methods (non-response sampling, or NRFU, and integrated coverage measurement, or ICM, will be used.)

The Dress Rehearsal experience will:

1) test the Bureau’s ability to conduct an enumeration in April, and then use ICM to complete the adjustment of those numbers by December.  (For the 2000 census, final population numbers must be delivered by December 31, 2000);

2) on the nonsampling side, indicate the possibility of hiring enough workers to do the enumeration with follow-up during a six to eight week period;

3) evaluate other related issues.

The Bureau wants to hire 150 programmers and statisticians nationwide, but expects difficulty in finding such people.  Fiscal year 1999 funding is still in dispute.  The Bureau faces a real problem in preparing to do both the most expensive parts of sampling, i.e. ICM, and nonsampling, i.e. 100 percent follow-up for nonrespondents.

Concerning the creation of a Master Address File (MAF), the frame from which the census will be done, the Bureau had hoped to use solely the Postal Service’s delivery sequence file.  Unfortunately, the quality of that file varies, and so a 100 percent address canvas must be done.  The Program for Address List Supplementation (PALS) that allowed governmental units and regional and metropolitan agencies to submit address lists has been a failure.  Another program, Local Update of Census Address program (LUCA), is in operation.  It asks each community to review the Bureau’s updated housing unit address list and inform the Bureau about problems, to be checked and corrected by Bureau staff.  LUCA takes the place of post-census local review.

There will be a long form questionnaire in 2000.  In addition, the American Community Survey (ACS) using Continuous Measurement, is alive and well.  To encourage retired persons to be census workers, the Bureau is in the process of getting approval that income from census work would not be counted against Social Security and military pensions.

5.0 1990 Processes
Elaine Murakami, Federal Highway Administration, distributed and described a handout [5.1] comparing 1990 process and steps related to the Census Transportation  Planning Package (CTPP) to the current status of equivalent 2000 census procedures.  For example, MPOs defined 1990 census TAZs in 1992 or 1993, but for the 2000 census that would be done Fall 1998, thereby giving TAZs a status more like other census statistical areas.  Regarding workplace coding, the possibility of two-stage coding was discussed, the second stage being TAZ-level, including local assistance for problems addresses in late 2000 and early 2001.

Discussion followed on 2000 census data availability by traffic analysis zone (TAZ) geography, as compared to the availability by the 2000 equivalent of Summary Tape File (STF) 3 census statistical areas.  Nathan Erlbaum wanted to be able to link the two data sets, and advocated that both TAZs and block groups (BGs) be made standard geographic areas of the 2000 CTPP.

Elaine said that if TAZs are demarcated and incorporated into TIGER prior to 2000, we should be able to get STF 3-like tables by TAZ.  Phil Salopek said that if a special tabulation of 2000 data by TAZ is required, it will probably be done by the Data Access and Delivery System (DADS), but for a fee.

6.0 Questionnaire Content:
Phil Salopek, Census Bureau, distributed and described a handout [6.1] comparing 1990 and 2000 census journey-to-work questions.  There are very few changes.  The handout also explains that the only housing item that will be collected from all households, and therefore will be available by block, is tenure (owner/renter status).

Nathan expressed interest in knowing a more specific workplace location for U.S. commuters to Canada or Mexico. Phil said that there were currently no plans to code more than the country name, but that he and Bob LaMacchia would discuss it.  Responding to a question about counts of total housing units (occupied plus vacant) in 2000, Jay Waite said that we are not sure that there will be an actual total housing unit count.  Clarification is needed regarding how sampling for nonresponse would impact both household and housing unit totals.

7.0 Coding Issues

7.1 TAZs and TIGER
Nancy Torrieri, Census Bureau, distributed and summarized a handout [7.1(1)] on the traffic analysis zones (TAZs) program for 2000.  The program will start late Fall 1998.  Participants will be providing their TAZ boundaries either as delineations or as equivalency files, using TIGER Line ‘98.  Subsequently, participants will also have the opportunity to verify the boundaries they submitted.  TAZ boundaries submitted in advance of the census will make possible the use of TAZs for DADS tabulations.   Nancy showed a copy of the reference map.  The reference maps will include the 1990 TAZs.  Features that shouldn’t be used as TAZ boundaries, or for which address ranges are needed, will be shown in red.  The maps will incorporate the results of the 1998 Boundary and Annexation Survey.  Several attendees said that it was critical that the agencies responsible for TAZ revision be informed soon that this task is coming by Fall of this year, and what they need to do to prepare for it.

7.2 Geocoding
Phil Salopek began by saying that there is not unanimity concerning how geocoding of workplace location responses should be conducted.  As currently scheduled, the geocoding period will run from May 1, 2000 to December 15, 2000.  There will be an automated phase, followed by a clerical phase.  Concerning the possibility of involving MPOs and similar agencies in resolving cases that Bureau staff cannot geocode, there are two questions:

1) Given the above schedule, is there time for such outside participation?

2) Is it legal to refer actual problem cases to non-Bureau persons?  Responding, Bob LaMacchia, Census Bureau, said he felt that the Bureau could pass the cases on to MPOs given guarantees that the MPOs would protect the confidentiality of the information.  Concerning the two-stage geocoding described in Elaine’s handout, there are implications, both funding and scheduling, that need to be explored.  The matter of employer and establishment name and address files was brought up.  In 1990, just about every MPO responds to the Bureau program that requested employer files.  This was a lot of work for MPOs, but the format diversity of the lists made them difficult to use.  For the 2000 census the development of an employer file from one or two references (e.g. ZIP+4, Electronic Yellow Pages) is being considered.  The resulting employer file would be sent to MPOs for their review.

7.3 Industrial Reclassification
Patricia Johnson, Census Bureau, noted that the 2000 census is the first one involving simultaneous major revisions to both the industrial and occupational classification systems.  The North American Industry Classification System (NAICS), the successor to the Standard Industrial Classification (SIC) system, is now the official U.S. system and must be used in the 2000 census.  Several Census Bureau committees are currently working on ways to link NAICS and SIC.  The Standard Occupational Classification (SOC) system is also undergoing major revision, and it will probably be another month before changes are finalized.  Data users should send any concerns about employment/worker classification to the Bureau in writing.

Ed Limoges stated that transportation planners need to have 2000 census small-areal employment data tabulated according to three separate classification systems:  the SIC-based codes used in the 1990 census, so that 1990 to 2000 trends can be identified; NAICS-based codes, because we need to relate 2000 census data to other NAICS-based employment data series; and land use classes, so that we can link data about workers to the economic activity and traffic generation characteristics of individual establishments.  NAICS could facilitate the land use classification of employment. However, several major NAICS classes present real challenges in coding type of industry responses on the 2000 census questionnaire.

8.0 Local Data Review
Chuck Purvis, Metropolitan Transportation Commission, discussed the suggestion that there be a review of data tabulations by larger geographic areas, e.g. counties, so that major geocoding errors could be spotted and corrected before the release of smaller area data.  The review would be done by MPOs/COGs.

Chuck also referred to the Census Bureau’s Local Update of Census Address (LUCA) program.  He handed out a document 8.1 that evaluates the LUCA process conducted in the Sacramento area, one of the Dress Rehearsal sites.

9.0 Data Dissemination
Ron Tweedie gave out a sheet describing the 1990 Census Transportation Planning Package (CTPP) chronology [9.1].  He summarized the events that extended from September 1988 to Spring/Summer 1993.  He said that we need to look at how many MPOs used the data and for what purposes.  Ed Christopher suggested a survey to go to states and MPOs.  Ed also asked if we want a CTPP, or do we want to go the DADS (Data Access Dissemination System) route.  Nathan emphasized the value of having data on a CD-ROM, as was done for the 1990 CTPP, and Phil counseled caution regarding DADS.

10.0 Continuous Measurement/American Community Survey
Jay Waite distinguished between the process, Continuous Measurement, and the program, the American Community Survey.  Phase I, 1996 to 1998, is the test phase, involving seven counties.  Phase 2, 1999 to 2001, is the benchmark phase.  The 40 counties involved are listed in the handout [10.1].  This phase will allow the comparison of 2000 census data with three years of ACS aggregated data.  Phase 3 is expected to begin in 2003, with a three percent sample every year thereafter, on a five-year cycle.  The assumption is that the ACS questions will be the same as those of the 2000 census.  In 2003 the Census Bureau might be willing to ask other questions e.g., a travel day.  The Bureau will ask for our guidance on this.

11.0 Future Business and General Announcements
Two other topics were mentioned.  One was the archiving of 1980 CTPP tapes.  The other referred to continuous household surveys that are being conducted in Melbourne and Sydney, Australia, and asked what could we learn from them.  The meeting was adjourned at 9:55 p.m.

Respectfully submitted,
Edward Limoges
Subcommittee Secretary

Back Home

Want to add something?  Have a comment? Questions? Interested in the subject? Write the web host Ed C.