No abstract

No abstract

Long-distance and overnight travel are growing in importance in the United States, yet the data and methods needed to model these activity patterns are lacking. One important factor for activity simulation is the interval between overnight trips away from home. This study used a unique 1-year panel data set of overnight trips for the estimation of a negative binomial regression model of intertrip time intervals. Most respondents indicated that they took between two and seven overnight trips per year, but time between trips varied widely regardless of the total number of trips. This time difference suggests some clustering of trips for some but not all people. Model results indicated that both regional and household attributes affected intertrip time intervals, but that both prior and next trip factors were significant. The distance from home on previous trips and income were not factors in intertrip interval times. The results of this study demonstrate the range of factors needed to model the time interval between trips. The model format lends itself well to activity simulation models that can expand beyond 1 day and the local region to include 1 year of travel and, potentially, a global landscape. In this way, long-distance and overnight travel can be integrated more appropriately into transportation system planning.

Research on key aspects of long-distance and overnight travel is limited, and knowledge about the household life cycle is even more limited. This work used data from the 2013 Longitudinal Survey of Overnight Travel (conducted as part of this effort) to ( a) identify groups of respondents with statistically significant differences according to their annual long-distance overnight travel schedules and ( b) summarize the life-cycle characteristics of the respondents that belonged to each group. The panel was not representative of the general population but, rather, consisted of more frequent travelers of higher income who mostly worked full time; many members of the panel traveled for work. K-means clustering was used to partition the travelers into six distinct groups on the basis of their annual travel. The largest group that was formed put a heavy emphasis on longer-distance travel. One group was dominated by travel with children. The other groups formed placed approximately even levels of emphasis on the number of tours of various distances but placed the very highest level of emphasis on work tours or a high level of emphasis on personal tours. The six groups showed distinct differences in demographic, neighborhood, and regional airport factors. Although additional work to refine these groups will be possible with this data set, to create fully representative groups for long-distance travel, more robust, larger data sets are needed. The findings of this research provide a solid expectation that the use of groups with statistically significant differences is a viable tool within long-distance travel modeling.

No abstract