Satisfaction with Public Transport Trips

6.1 Future research 
Future studies would benefit from issues raised by this thesis.
First, the impacts and knowledge that may be drawn from the inclusion of PT sub-modes in the 
segmentation process would enable unveiling segments that relate to particular modes and focus on 
mode-specific market analysis.
Second, conducting a longitudinal study of the attribute importance would be useful to infer causal 
relations, the plausible impacts of the newly implemented policy, and the behavioral change process of 
the travelers over a long period of time.
Third, the inclusion of interaction effects between regions, segments and attributes in the models would 
allow the identification of QoSAs that may have a larger impact in different regions and segments within 
a given region. Moreover, it would be interesting to contrast PT objective performance indicators with PT 
travelers’ satisfaction, as a way to evaluate the degree to which travelers’ satisfaction valuations are at 

---

32 
random.
Fourth, the analysis of cross-sectional data does not allow following the same individual over time. The 
use of this type of data does not allow assessing whether changes in prevalence reflect a trend or simply 
differences between different groups of participants sampled from the population. In addition, the use of 
this type data may produce cohort effects which also alter the results. Therefore, it would be interesting to 
analyze panel data as a way to determine whether the overall change in QoSAs is driven by a systematic 
change in individual perceptions. For example, it could be determined whether greater importance is 
attached to a QoSA with which an individual is less satisfied or to variations in PT traveler groups. 
Fifth, it would be recommended that future research base their studies in geographical units with a similar 
size. Differences in size may cause travelers living in the smallest areas experimenting not only the 
characteristics of their postcode area but also the ones of the neighboring surrounding postcode areas. 
Given the smaller size of postcode areas belonging to the city we would assume that this issue has a larger 
impact on the city models. 
Sixth, the collection of instant utilities during the travel experience or the use of methods that have been 
proven to elicit recall bias-free instant utilities (e.g. ecological momentary assessment, daily 
reconstruction) could improve the recollection of retrospective experiences. In addition, measuring 
waiting and transferring times’ satisfaction as separate episodes within the multi-modal trip would allow 
to better single out their impact.
Seventh, future studies should control for important aspects such as PT disruptions and subjective well-
being variables since they are very likely to mediate the effect of, for instance, weather impact. Moreover, 
other studies should incorporate travel attitudes and preferences in the models so to be able to better 
explain the impact of built-environment characteristics on travel satisfaction. A final recommendation is 
the examination of the speed and the extent to which QoSAs switch from one-factor category (basic, 
performance or exciting) to another.