Exploring Targeted Long-Distance Travel Survey Sampling Frame Approaches to Support Travel Survey Collection

Abstract

Long-distance travel continues to grow in economic, environmental, and infrastructural importance. To inform policy, infrastructural funding, and travel demand forecasting, detailed travel survey findings are needed at the national-level. However, the costs of running a full-scale national long-distance travel survey, both fiscally and temporally, has limited recent attempts. As such, survey data users have had to use older national survey data and then optimize their findings based on more recent, smaller-scale state travel surveys. These smaller-scale surveys have sample framing limitations, but while these limitations may be allowable given the immediate needs and scope associated with its original purpose, their utility for extrapolation and aggregation for broader use compared to the scope of national-level annual panel surveys is still fairly unknown. This dissertation aims to identify how targeted sampling frame approaches can be used by national long-distance travel surveys. Having a full-sized, population-proportioned sample is always a tenant of good survey design, but the associated costs, especially for more niche subjects, can lead to an unwillingness to fully commit resources to survey deployment. This can lead to patchwork solutions to reduce costs such as with asking respondents for their long-distance travel habits over a very small timeframe or using regional survey findings to update national travel models. While these patchwork solutions offer potentially valid solutions, the actual impacts on data validity are practically unknown. This dissertation aims to fully explore these aspects, by not only exploring how targeted sampling frames might affect long-distance travel survey data accuracy, but also the best approaches for creating a targeted sample frame for the purposes of capturing nation-wide US long-distance travel. Results suggest that the long-distance travel survey sampling frame can be targeted to reduce both fiscal and temporal costs considering seasonal variability trends, using targeted sociodemographic sampling, or using targeted geo-economic sampling. At the same time, these reductions can still provide statistically viable samples for sociodemographic groupings, travel volumes, mode splits, and purpose splits comparative to full-scale national surveys like the 1995 American Travel Survey, 2001 National Household Travel Survey, and 2013 Longitudinal Survey of Overnight Travel.