Sustainable tourist mobility in Island destinations: A mixed logit analysis of mode choice, time and cost

This paper analyses tourists’ reported main transport mode for intra-destination trips in Madeira using post-pandemic airport-exit survey data collected between 2021 and 2023. The analysis focuses on main-mode choice and does not reconstruct complete multimodal trip chains, transfer sequences or door-to-door Mobility as a Service journeys. A panel Mixed Logit model is estimated with mode-specific cost coefficients and error components that capture persistent correlation across repeated choices. The model is benchmarked against a standard Multinomial Logit specification. An extended panel model then allows travel-time sensitivity to vary normally across respondents. This extension substantially improves model fit: the standard deviation of the travel-time coefficient is strongly significant, and 98.73% of the estimated coefficient distribution has the expected negative sign. Travel time is a robust determinant of mode choice, whereas monetary cost sensitivity is most clearly identified for rented-car and taxi use. Values of Time derived from the fixed-time specification and retained as comparative benchmarks, range from approximately €12.6/h for public bus to €30.8/h for rented car. Under the preferred random-time specification, a 10% reduction in bus travel time increases the predicted bus share by 2.47 percentage points, while a 20% increase in rented-car generalised cost reduces the rented-car share by 4.29 points. The combined scenario produces the largest policy-relevant reallocation: bus share rises by 3.96 points and rented-car share falls by 5.68 points. Continuous sensitivity analysis confirms that the direction of these responses remains stable across a wider range of attribute changes, although their magnitude is nonlinear. The analysis concerns the reported main mode used for each trip rather than complete multimodal trip chains, transfer sequences or door-to-door Mobility-as-a-Service journeys. Within this scope, the findings provide a transparent basis for ex-ante transport-policy appraisal in island and tourism-dependent destinations, particularly for coordinated packages combining public-transport time improvements with calibrated rented-car demand-management measures.