The Census Bureau plans to replace the decennial census (long form) with a continuous survey approach, now called the American Community Survey. In this approach, a five year accumulation of data will be used in place of the snapshot of data from the decennial census. In this project, ACS data for 1999, 2000, and 2001, from five or six counties will be compared directly with the results of the Census 2000. Data comparisons will be made at the county, place and tract or TAZ level. Statistical tests will determine if there are significant variations between the two data sources, and if patterns of variability such as by population characteristics, such as by race or income, or variability by geographic location can be identified.Research findings: Travel time distribution and average travel time, and carpool as means of transportation to work were found to have statistically significant differences between decennial 2000 and 1999-2001 ACS results at the county level. Tract level analysis was pursued for San Francisco, CA and Broward County, FL to determine if a pattern could be identified contributing to these differences. Did not find spatial correlation, and did not identify any specific demographic characteristics that correlated with the difference.
Main benefits to the research:
1.The ACS sample will be smaller than the decennial census long form. This means that there will be fewer O/D pairs in the flow tabulation, with each response having a higher weight. Currently, the ACS plan under “full implementation” is 12.5 percent of housing units in 5 years of ACS surveys, compared to 17 percent of housing units for decennial. This suggests that TAZ definition will need to be re-thought as the transportation community has wanted finer and finer grained geography reported from the Census Bureau, but this will be less appropriate with smaller sample sizes.
2.For any specific O/D pair, at small geographic tabulation (tract or TAZ), there is a high probability of missing data in any specific year. By having this project, we educated the ACS programming staff about the importance of the O/D pair tabulation and worked out methods for tabulation of flow data for the NCHRP 08-48 project, and future ACS tabulations.
To compare the results of the pilot American Community Survey with decennial Census 2000 Long Form data. This will help inform the transportation community on how to use the ACS data, should the CB implement it beyond the current test.
Murakami
Ongoing