Managed-lanes stated preference survey in Atlanta, Georgia: measuring effects of different experimental designs and survey administration methods

With the vast majority of choice modelling studies making use of data collected using Stated Preference (SP) surveys, there is considerable interest in the benefit of advanced survey techniques, such as efficient designs. However, while there is a growing body of statistical evidence of the advantages of such advanced methods, there is still relatively little evidence from actual real-world studies. In this paper, we present the findings of a large study making use of SP surveys in the context of road tolling initiatives. The study has the rather special characteristic that the data collection made use of survey questionnaires using various different experimental design approaches. This allows us to directly compare the stated choice behaviour across designs. We observe significant differences in a number of crucial indicators, such as the value of time and elasticities with respect to income and trip distance, depending on what design is used. The analysis further shows that superior performance is obtained when making use of an efficient design, while also highlighting the disadvantages of random blocking approaches.