Before and after surveys are a common method for measuring the effect of specific policies and projects designed to cause changes in travel behavior. Design issues of before and after surveys used to evaluate projects designed to change travel behavior are addressed. A quantitative understanding of the underlying variability of the parameters to be measured was needed--in particular, the variation over time in car travel by individuals and households. Since no data were available locally, a detailed analysis was made of the 6-week travel diaries from the MobiDrive survey conducted in Germany to estimate the coefficients of variation in key travel parameters. These results were then adapted for use in the presented project. Several features emerged from this analysis. First, in general, larger sample sizes are needed to detect changes in either distance traveled or travel time than for trips undertaken. Second, larger sample sizes are needed to detect changes from repeated cross-sectional surveys than from a panel survey. Third, larger sample sizes are needed to detect changes when using a daily travel diary, compared with use of a weekly travel diary (although this difference can be substantially reduced in a panel survey by maintaining the same day of the week for each household in later waves of the panel). Finally, larger sample sizes are needed to detect changes from person travel data than from household travel data. However, some of the design parameters that allow smaller sample sizes also make the survey more difficult to conduct.
Abstract