Linking Response Quality to Survey Engagement: a Combined Random Scale and Latent Variable Approach
S Hess, ITS University of Leeds, UK; A Stathopoulosy, University of Trieste, IT
Random utility modelling bases its estimates of the systematic component of utility on the attributes presented in a choice set along with measurable respondent characteristics.
Random utility modelling bases its estimates of the systematic component of utility on the attributes presented in a choice set along with measurable respondent characteristics. This focus on ?pure preferences? is gradually changing. An important body of research indicates that factors beyond what is typically accounted for in measuring taste variations, influence choices. More than other fields, the study of transportation behaviour has progressively been informed by findings, drawing on cognitive sciences, suggesting that previously unobserved effects can explain choice outcomes on a par with the observable contents of a choice experiment. One significant contribution is the recognition that norms and attitudes can be determinant in guiding decisions. Another important strand of literature focuses on individual heterogeneity in the decision rules that people apply to solve decision problems. These include lexicographic rules, asymmetric attendance of attributes and disproportionate preferences for status quo options. Several findings to date illustrate that accounting for a broader set of factors may have important implications for econometric estimates. However, an important shortcoming in the literature is the isolated modelling and quantification of these innovative approaches. As a step towards a better understanding of transport decision making, in this paper a unifying modelling framework is put forward where these two strands of research are accounted for jointly. Using data from stated commuting mode choices of bus and train users we analyse the role of several attitude measures and choice strategy heuristics. The design of the study anchors attribute values around individual reference points to ensure realism. Apart from the attributes describing commuting scenarios, attitude constructs (such as mode satisfaction) and indicators of non-standard processing strategies (like attribute attendance), are accounted for. Along with socio-demographics and measures of perceived task complexity these indicators are incorporated in different econometric models. The results show that including non instrumental attitude constructs and information on how choices are processed lead to better model performance, as well as more reliable willingness to pay measures for commuter decisions.
Association for European Transport