CTPP 2000 Status Report

On Feb. 19, 2004 the Census Bureau (CB) finished releasing preliminary Part 2 data in .ivt format to state DOTs and MPOs. The preliminary Part 2 data are also available in flat ASCII files on the BTS website. To download, go to http://transtats.bts.gov, and click on CTPP 2000.

Final Part 1 data in flat ASCII files are available for download at the same BTS site. CB expects to start shipping final Part 1 CDs in .ivt format with software to state DOTs and MPOs around the end of April. Shortly thereafter, BTS will process orders for CDs from the general public at htpp://www.bts.gov (click on Products).

Preliminary Part 3 data in flat ASCII files are expected during April 2004 as well. Download from the BTS Transtats site will be available. CDs with the .ivt version and software are not expected until May.

For questions on the flat ASCII files, contact Clara Reschovsky at 301-763-2454 (e-mail Clara.A.Reschovsky@census.gov).

For questions on the CTPP Access Tool or to comment on the data, please contact Nanda Srinivasan at 202-366-5021 (e-mail: nanda.srinivasan@fhwa.dot.gov).

Journey to Work: Census 2000 Brief

by Elaine Murakami, Federal Highway Administration

The Census Bureau has finally (!) released the Journey to Work: 2000 Brief. This was written by Clara Reschovsky, who has been a key Census Bureau staff in the production and distribution of the CTPP 2000. Many of you have talked to Clara by phone or email for technical questions related to TAZ definition, MPO region definition, determining the number of copies of CDs to be mailed, ad infinitum. We are proud that in addition to her work at the Census Bureau, Clara has been working on her Master's degree, and expects to finish in May 2004 with a degree in Transportation Policy from George Mason University.

Some of the results are:

• The average travel time for all modes of travel is nearly 26 minutes. The typical U.S. commuter drove alone, with an average of 24 minutes to get work. (Note: source is CTPP 2000 Profile).
• Overall, people were leaving home earlier and spending more time traveling than in previous censuses.
• Men continue to commute longer than women, 27.2 minutes compared to 23.6 minutes.

The proportion of the U.S. population living in metropolitan areas increased from 1990 to 2000 by 13.1 million (from 93.1 million to 106.3 million). However, for the first time, more than half of metropolitan area resident workers worked outside of the central city portions of the metropolitan area.

Commuting to Downtown Study

By Ed Christopher (Chair, TRB Committee on Urban Transportation Data and Information Systems) (ABJ30)

The Transportation Research Board (TRB) Committee on Urban Transportation Data and Information Systems (ABJ30 formerly A1D08) wants your help to build "downtown profiles" for the largest cities in the United States.

The "downtown profile" is envisioned as a two-page product: one page with a half-page downtown locator map and a half-page table showing commuters by means of transportation to work. The second profile page would include a narrative summary.

Table 1 shows some of the data to be included. Note that the worker count is from Census workplace results, and is not adjusted for total employment. The value of the profile is that it includes a time dimension in addition to the distribution of travel modes, and total workers and employment.

Figure 1 illustrates the share of transit for downtown commutes. Although transit commute mode share for the entire U.S. is around 5 percent, transit plays a critical role in commuting to downtowns. In several cases, transit shares have increased since 1990, returning to rates similar to those found in 1980.

Chuck Purvis (Metropolitan Transportation Commission, Oakland) is coordinating the effort. For an abstract of the study, please visit http://www.mtc.ca.gov/trb/urban/commute/ and http://www.mtc.ca.gov/datamart/census/ctpp2000/downtown/

Please participate! For more information, please contact Ed Christopher (edc@berwyned.com) or Chuck Purvis (cpurvis@mtc.ca.gov).

Figure 1. Transit Shares to Downtown in 2000

Table 1. Mode Shares to Downtown for Selected Cities

Data Rounding in CTPP 2000

By Nanda Srinivasan, Cambridge Systematics Inc.

All CTPP tables (except mean, median, and standard deviation tables) are subject to the following rounding rules:

• 0 is kept as 0.
• 1-7 is rounded to 4.
• Anything above 7 is rounded to the nearest multiple of 5 (Example: 9 is rounded to 10, 17 is rounded to 15 etc).

Columns for Totals in each table may not match the sum of the categories because Totals are rounded independently of the cells. For example:

• 0 vehicle households = 6 is rounded to 4
• 1 vehicle households = 14 is rounded to 15
• 2 vehicle households= 8 is rounded to 10
• 3 vehicle households= 8 is rounded to 10
• 4 vehicle households= 3 is rounded to 4.

Total households= 39 is rounded to 40 (and NOT 4+15+10+10+4 = 43 rounded to 45).

However, because some variables (Example: Mode to work) are classified in more than one way, the number of possible answers is higher. Chuck Purvis (see Table 1) (Chuck Purvis, Metropolitan Transportation Commission, Oakland, California, e-mail posted to the CTPP-news listserve on February 19, 2004) notes that sometimes, up to 15 answers may be possible!!

Table 1. Different Tables and Different Geographies yield (slightly) Different Answers!

Because each table may provide a slightly different answer, we recommend that where possible; try-obtaining totals from published CTPP tables, rather than aggregating geographies, or individual table categories.

Rounded Values vs. Unrounded Values

A comparison of CTPP 2000 data and Summary File 3 data showed that the CTPP estimate was more likely to be LOWER than the SF 3 value. We noticed that the rounding value of 1-7 to 4 generally provides a lower estimate than the actual value. This occurs because:

• The Census "long form" sampled 1 in 6 persons nationwide, and 1 in 8 persons in urban areas.
• Individual weights (after adjusting for non-response) may vary anywhere from 1 to 20 (or higher). In urban areas, weights are more probably in the order of 5 to 20.

Which value to use: Totals or "Sum of the Parts"?

Rounding does not affect the "statistical" significance of the data. For example, there is no difference between 431,899 and 431,753. The difference of 136 is well within the sampling error. The choice of using published totals versus summing the different parts of the table is left to you. If you are reporting percentage distributions, we would suggest you use the published totals as the denominator. If you are making a pie or a bar chart, you could use the "Sum of the parts" as the total.

Aggregating Geography: How to Make Allowances to Rounding?

When you want to aggregate geography, first check to see if there is a summary level in CTPP 2000 that contains the aggregated geography. For example, there is no need to aggregate data from tracts or TAZs to county geography because county data are published as a summary level in CTPP. Table 2 compares the sum total of tracts, and TAZs with the published totals for Mode to Work (by Place of Work) for Montgomery County, MD. Because the county is divided in to 117 tracts, and 1124 TAZs, the TAZ totals are off more than the tract totals. However, the percent of transit commuters is the same in all three cases.

Table 2. Mode to Work by Place of Work, Montgomery County, MD

Aggregating to an Unpublished Geography

To aggregate data to an unpublished geography (example: Central Business District - CBD), consider the LARGEST geography that can be used to aggregate. For example, if the CBD can be defined using Tract geography, then use Tracts instead of using more finely defined TAZs. This will minimize distortions introduced by rounding.

Figure 1 shows how 2000 TAZ, and 2000 Tract boundaries can be assembled to obtain the 1990 CBD definition for Seattle, WA. 21 TAZs OR 4 Tracts approximately equal the 1990 CBD geography. Table 3 shows that both the TAZ and Tract results are close, even with rounding and geographic differences.

How Bad is Rounding?

Chuck Purvis notes in his post to the CTPP list serve, "The rounding of values inside the CTPP is, right now, a modest, annoying data processing issue. As professional data analysts, we are always on the lookout to make sure our numbers "add up" so that we're not missing anything." Chuck would prefer that the CB use a different approach to disclosure avoidance that results in numbers that add up. Chuck's final recommendation (Chuck Purvis, Metropolitan Transportation Commission, Oakland, California, e-mail posted to the CTPP-news listserve on February 19, 2004) is to "Develop a sense of humor... Don't take these data issues too seriously. And it's kind of funny that the numbers don't add up. Or, as they say: "close enough for government work."

Figure 1. Combined TAZs/Tracts Overlaid on 1990 CBD Definition - Seattle, WA

Table 3. Comparison of Tract, and TAZ based definitions of CBD

Race/Hispanic Origin Tables in the CTPP 2000

By Elaine Murakami, Federal Highway Administration

CTPP 2000 has a wealth of tables that include Race and Hispanic origin. CTPP from previous decades virtually ignored these characteristics. However, at the time of CTPP table design, environmental justice issues were prominent, resulting in the addition of tables using these variables, and also poverty status (Table 1).

Table 1. Race by Hispanic Origin Tables in CTPP Parts 1 and 2

Table 2 (obtained from CTPP 2000 for San Diego County, CA) clearly shows how different Hispanic households are, with over 50 percent of households with 4 or more persons, and over 40% with income below $25,000. White, non-Hispanic households on the other hand were dominated by 1 and 2-person households, with 43 percent with incomes above $60,000.

Source: CTPP 2000, Table 1-82

Similarly, Summary Files 3 and 4, standard products from the Census Bureau have a wealth of tables in race and Hispanic detail. The SF 3 tables are limited to census tract reporting, called PCT and HCT. Among the PCT and HCT tables in SF3 are two of particular interest to transportation planners, PCT65 (Means of Transportation to Work) and HCT33 (Vehicles available). The tables are structured as follows: PCT##

1. White alone, (Hispanic and non Hispanic)
2. Black alone
3. American Indian and Alaska Native alone
4. Asian alone
5. Native Hawaiian and Other Pacific Islander alone
6. Some other race alone
7. Two or more races
8. Hispanic, all races
9. White Alone, non-Hispanic

A large proportion of the Hispanic population selected "other" as their race.

Mode to work in 2000 has important differences by race and Hispanic origin. Many of these differences are correlated with differences in household income, household size, and residential location. Data from Summary File 3 Table PCT65 shows (See Table 3):

• Driving alone is the predominant mode for everyone, however,
• White, non-Hispanic workers are the most likely to drive alone,
• Hispanic workers are much more likely to carpool, and
• African Americans are most likely to use public transit.

Table 3. United States: Mode by Race by Hispanic Origin

Source: Census 2000 Summary File 3, Tables PCT 65 b, h, and i.

Census Data Subcommittee at TRB Annual Meeting

By Ed Limoges, Sabre Systems, Inc.

The Subcommittee on Census Data,

ABJ30 (1), met on January 14, 2004 at the 2004 Annual Meeting of the Transportation Research Board. Over thirty persons attended, representing Federal and State agencies, MPOs, consulting firms, and academia. The meeting was chaired by Bob Sicko. Here are a few highlights from the meeting. The complete notes are posted at http://www.trbcensus.com/notes.html

American Community Survey

Jay Waite, Associate Director for Decennial Census and American Community Survey, presented an update on plans and activities for Census 2000, including ACS.

The nationwide ACS will start in July 2004. The earliest that small geographic reporting, e.g. block group, would be available would be an accumulation of 2005 through 2009 data, with data release toward the end of 2010. The five-year accumulated data will be published in the geography of the last year, so that 2010 data would reflect annexations up to that year. Census staff is currently working on workplace coding options.

There will be a process for modifying the questionnaire content with any changes taking place with the 2008 collection year. Census Bureau will meet and discuss ACS issues with staff from each of 21 Federal agencies.

NCHRP 08-48

Kevin Tierney, Cambridge Systematics, discussed the ACS NCHRP project. The purpose of the project is to develop a guidebook for transportation planners about how to use ACS, adjust for the loss of the long form, and to identify new uses for the data. A draft of the guidebook is expected to be available for the Spring 2005 TRB Census data conference in Irvine, CA.

Elaine Murakami stated that to augment the NCHRP project DOT contracted with the CB for a special tabulation of the 3-year test ACS data. The tabulation is designed as a "mini CTPP" with a Part 1, Part 2 and Part 3 set. Although the data tabulation was developed to support the research associated with the NCHRP project, the data will also be made available to interested researchers.

CTPP 2000

Phil Salopek, Census Bureau, presented an update on the production and distribution for CTPP 2000.

Public Use Microdata Sample (PUMS)

Larry Blain, PSRC, described his use of PUMS data for cross tabulation, and urban simulation. Ken Cervenka, NCTCOG, will do reasonableness checks on the worker flow data, and use PUMS to analyze vehicle availability. Chuck Purvis, MTC, distributed a one-page handout describing travel demand modeling applications and exploratory research.