Accounting for heterogeneity in travel episode satisfaction using a random parameters panel effects regression model

Rasouli & Timmermans1 suggested a model of travel episode satisfaction that includes the degree and nature of multitasking, activity envelope, transport mode, travel party, duration and a set of contextual and socio-economic variables. In this sequel, the focus of attention shifts to the analysis of unobserved heterogeneity in the satisfaction ratings. To that end, some variables included in the original model are randomized, assuming normal distributed random effects The model is estimated using data of a sample of respondents, who judged their satisfaction of every travel episode they experienced during a multi-week data collection effort. GPS devices were used to record their travel and impute activity and travel episodes. A Web-based prompted recall instrument was used to verify and if needed rectify the imputed activity-travel agendas, provide information about the degree and nature of multitasking behavior whilst travelling, and judge the degree of satisfaction associated with the travel episodes. Results of the random parameters regression model evidence a substantial amount of unobserved heterogeneity in the travel episode satisfaction ratings.