Appendix A Methodology
Source and Accuracy
The findings from the 2001 NHTS survey are based on travel data collected from a random digit dial sample of telephone interviews conducted with over 60,000 individuals in approximately 26,000 nationally representative households. Interviews were conducted between March 2001 and May 2002. Individuals in the NHTS sample were asked to complete a travel diary for a specified day, known as the travel day, and were also asked to report on the characteristics of long-distance trips of 50 miles or more from home made during a 4-week period, known as the travel period.
Estimates reported here are based on weighted data to account for selection probabilities at the household and individual level, and are further adjusted for household and individual nonresponse. Comparisons made in this report are statistically significant at a 0.05 level.
Creation of Pre-9/11 and Post-9/11 Data Files
The 2001 NHTS was conducted from March 2001 through May 2002. The NHTS person-level dataset was divided into two parts: pre-9/11 data and post-9/11 data. The pre-9/11 dataset includes trips taken in the period from March 2001 to September 2001, a period of 5 1/2 months, and includes the summer season in which a large proportion of long-distance trips are taken. There were approximately 22,000 persons responding about travel prior to 9/11. The post-9/11 dataset includes trips taken in the period from September 2001 to May 2002, a period of roughly 8 months, and includes Thanksgiving and Christmas—a traditionally heavy season for long-distance trips. The survey had responses from approximately 38,000 persons on their long-dis-tance trips after 9/11. The composition of persons who took long-distance trips prior to 9/11 and those that took long-distance trips following 9/11 was not the same. A more detailed description of how the pre-9/11 and post 9/11 data files were created is contained in Chapter 1 of “National Household Travel Survey—Pre- and Post-Data Documentation” found on www.bts.gov.
Additional steps were taken to make each of the pre-9/11 and post-9/11 groups a nationally representative sample. This is achieved by constructing new weights using statistically sound methods brieﬂ y described below.
Reweighting of Pre-9/11 and Post-9/11 Data Files
Weights are needed to produce valid popula-tion-level estimates so that the results of a survey of the population are representative of the population as a whole. For example, if in the survey 47 percent of the respondents were male and in the U.S. population 49 percent of the population was known to be male, then the male survey respondents would be weighted stronger so that their data and travel information would count for 49 percent of the population’s travel patterns. Adjustment and post-stratification are performed on collected data to reduce bias of estimates. Post-stratification reweights the data so that the characteristics of the respondents are the same as the characteristics of the population.
To see whether the pre- and post- 9/11 groups were representative samples of the national population it was necessary to construct population control totals for key survey variables using Census 2000 numbers. Tolerance levels were used in the final reweighting program to determine which combination of variables to keep. An example of a tolerance level would be for Hispanics to be within 4 persons of a census generated control total that would represent the total Hispanics in the United States for a designated timeframe. A more detailed description of how the pre-9/11 and post-9/11 data files were reweighted is contained in chapter 2 of “National Household Travel Survey- Pre- and Post-Data Documentation” found on www.bts.gov.