Travel survey methodology: advantages, disadvantages, and unintended side-effects of survey design choices

Abstract

Travel surveys collect information on travel behavior and are important data sources in transportation planning and modeling. Naturally, the data quality affects the quality of the transport models and analysis, and further, policy and planning; low-quality travel survey data can, at worst, lead to wrong prioritizations in transportation planning and policy. High-quality travel survey data can, at best, provide transportation planners, modelers, and policymakers with a robust basis for decision-making. Thus, it is important to aim for high-quality travel surveys. This thesis aims to contribute to the methodological development of travel surveys. In the thesis, I provide enough information on travel survey methodology so policymakers, planners, students, and others interested in conducting a travel survey can create a survey design based on best practices within the field.

Both qualitative and quantitative methods were used in this thesis. The thesis is based on a document study of national travel survey methodology in Norway, Sweden, Denmark, England, France, and Germany, quantitative analysis of Norwegian National Travel Survey data for the years 2016-2019, an enquête (short survey), multiple data collections with the smartphone app TravelVu, a focus group study and personal interviews.

Data quality was a challenge in all the national travel surveys studied. The countries have attempted to mediate them differently, and all have discussed and/or tested GPS or smartphone solutions. The methodologies vary between countries and are affected by local traditions. An emerging trend is combining data collection tools (i.e., multi-mode solutions). However, mode effects (i.e., data collection tool affecting the data) have been identified in several of the countries that have chosen a multi-mode solution. There are advantages and disadvantages to using different methods in travel surveys. For example, multi-mode solutions can make it possible to get information on more population segments (advantage) but introduce data comparison issues (disadvantage). On the other hand, using one data collection tool reaches fewer population segments (disadvantage), but eliminates data comparison issues (advantage). The choice of method affects the data quality and the nature of the data. Self-reporting methodology gives another type of data than a smartphone-based travel survey, and the best method depends on the segment one wishes to study. Although a smartphone-based travel survey gives detailed data material, results indicate that smartphone-based travel surveys show a lower response rate than traditional methods. The underlying value dimensions of fear of risk and technology interest can explain high nonresponse in smartphone-based travel surveys.

Since different methods give different data types, comparing data across a time where other methods are used should be done with utmost care. However, this does not mean that one should not change the survey design if the current survey design is unsatisfactory. The wish to continue the time series should not be used as an argument for using a survey design that results in low-quality data. To mediate the NTS challenges identified, it is necessary to understand how survey design, data collection tools, and structural changes can affect the data material. If necessary, develop tools and mechanics to reduce potential errors. This demands methodological awareness among those responsible for the NTS, from survey design to data collection and processing. The presented NTS advice can be used to improve the NTS survey design and is based on survey and transportation methodology research.