Chris-- No survey is perfect but I wouldn't call the Census numbers "flawed." I believe that what you are saying is that unlinked transit trips are a different measure than worker flows from the Census questionnaire. In areas with wide transit choices such as the Washington, DC region, it is important for other data to be combined with Census data to check and verify results. For example, regional household travel surveys, on-board transit surveys, and park-and-ride surveys, are necessary to provide complete information about daily travel behavior. Because the Census Long Form goes to 1 in 8 addresses (on average in urban areas), the high sampling rate and high response rates result in data that is considered very high quality. While the Census questionnaire limits people to choose only one mode to work, and asks for "usual mode last week," the mode differences occur in both directions. People who usually drive alone may take the bus occasionally, but also people who usually take transit to work may drive alone or carpool to work. The 1995 NPTS data show that over 20 percent of workers who say they "usually" use transit either drove alone or carpooled on an assigned travel diary day. For workers who say they usually use a private vehicle to work, 0.3 percent say they used transit on an assigned travel diary day. (http://www.fhwa.dot.gov/ohim/trvpatns.pdf, page 79 (p. 101 of the Acrobat pdf file) One story from the field collection of the 1995 NPTS was a woman who called because she was very upset that on her assigned travel day she had driven to school and then drove to the airport to pick up a friend. Her usual mode to school was by bicycle. She gave the telephone interviewer what her actual behavior had been, although she could have told us her "usual" commute and we would not have been the wiser. Some of us count ourselves "lucky" that the Journey to Work Questions were retained in the Census 2000, and are in the process of justifying the inclusion of these questions in the American Community Survey. The Census Bureau is recommending that the American Community Survey replace the Census Long Form, so the Census 2000 Long Form is considered the "last one." There have been suggestions to change the question on the Census surveys (decennial long form and the ACS) to allow for multiple modes, and to allow for different modes on different days, however, the chances of getting more "real estate" on the questionnaire are "slim to none." For capturing variability in daily travel behavior, I have been advocating multi-day diaries, but adding response burden to surveys that already have a low response rate is a difficult problem. Elaine Murakami Federal Highway Administration ....So aren't the Census numbers ALWAYS going to be biased toward the "dominant" mode (unless everyone does the same thing every single day)? As others have posted, there are other reasons for disagreement between Census travel trends and all other sources, but this is clearly one. .... Using these flawed Census numbers to argue for further depriving people of transportation choices is completely outdated thinking. Chris. ------------------------------------ Christopher V. Forinash U.S. EPA: Office of Policy, Economics and Innovation 1200 Pennsylvania Ave NW (MC 1808), Washington DC 20460 (Delivery: 401 M St. SW, #WT-1013D) 202-260-5044 vox 0174 fax; forinash.christopher(a)epa.gov ------------------------------------ Development, Community & Environment Division: http://www.epa.gov/smartgrowth A partner in the Smart Growth Network: http://www.smartgrowth.org