Model-Based Approach to Synthesize Household Travel Characteristics across Neighborhood Types and Geographic Areas

Household travel survey data are crucial in regional travel demand analysis. However, good quality data are not always available owing to financial constraints, privacy concerns, poorly designed sampling schemes, and/or low response rates. Thus, various data synthesis techniques have been proposed in the past. In this paper, we identify the limitations of the existing data updating/synthesis methods and propose a two-level random coefficient model to synthesis household travel characteristics across geographic areas. Then the two-level structure was applied to the sampled households in the 2001 National Household Travel Survey across (consolidated) metropolitan statistical areas of various population sizes. One particular travel characteristic, journey to work vehicle trip rate, is investigated. The study findings confirm the effect of neighborhood (defined at the census tract level) attributes (e.g., intersection density, average auto mobile work trip travel time) on household number of journey to work vehicle trips. This effect is significant on the urban households of study, whereas the suburban counterparts across the country do not seem to be affected by their living environments after controlling for neighborhood type. In general, the two-level structure is shown statistically superior to the one level.