Supplementing transportation data sources with targeted marketing data: Applications, integration, and internal validation.

Unlike many third-party data sources, targeted marketing (TM) data constitute holistic datasets, with disaggregate variables – ranging from socioeconomic and demographic characteristics to attitudes, propensities, and behaviors – available for most individuals in the population. These qualities, along with ease of accessibility and relatively low acquisition costs, make TM data an attractive source for the supplementation of traditional transportation survey data, which are facing growing threats to quality. This paper develops a typology demonstrating ways in which TM data can aid in the design of transport studies, as well as in the augmentation of modeling efforts and policy scenarios, allowing for improved understanding and forecasting of travel-related attributes. However, challenges associated with integrating, validating, and understanding TM variables have resulted in only a few transportation studies that have used these data thus far. In this paper, we provide a transportation discipline-specific resource for TM data, informed by our integration of an extensive TM database with both the National Household Travel Survey (Georgia subset) and a statewide travel behavior survey conducted in Georgia on behalf of the Georgia Department of Transportation. Using the resultant datasets, we validate TM data by means of several approaches, and find that the TM dataset reports gender, age, tenure, race, marital status, and household size with match rates ranging from 70% to 90% relative to both transportation surveys. However, we also identify biases in favor of population segments that may have more longstanding financial/transactional records (e.g., males, homeowners, non-minorities, and older individuals), biases comparable but not identical to those of survey data. While this work suggests wide-ranging implications for the use of TM data in transportation, we caution that flexible and responsible approaches to using these data are critical for staying abreast of evolving privacy regulations that govern third-party data sources such as these. [ABSTRACT FROM AUTHOR]