Modelling interactions between rail operational delays and passenger satisfaction

Since the privatisation of British Rail in 1994/1995, the number of passengers on Britain’s railways has more than doubled. This growth however has not been without problems. More passengers have resulted in overcrowding on peak time services in metropolitan areas, whilst many more train runs on a constrained network has affected reliability, punctuality and ultimately passenger satisfaction, which has been trending downwards since 2012 for commuters in particular. Whilst investments in improving capacity and reliability can help alleviate these issues, such interventions are expensive and take time to introduce and are not always appropriate even if they are approved in a limited resource context. Instead, creative solutions are required to improve the handling of delays.

The aim of this PhD, therefore, is to investigate how railway performance impacts on passenger satisfaction. This will hopefully give railway companies the insights needed to formulate hands on solutions that could improve passenger satisfaction, within the constraints that the railways face.

In this thesis, the first step was to identify how incidents impact on the railway, in terms of incident duration, trains affected, seasonality and location/timetable related aspects. Historic delay attribution data was enriched with real time train movements data, obtained from Network Rail. Survival models were used to identify which factors had the highest impact on incident duration. Infrastructure related delays have the highest impact, in terms of incident duration, number of trains impacted and the average delay per train. Infrastructure related incidents pose most problems, whilst a higher number of train movements is associated with lower incident duration, though this is because of a high number of sub-threshold delays. On busier railway routes, the probability that an incident ends within the next minute decreases after 9-11 minutes, whilst on quieter locations this is between 17-20 minutes.

The second step was to identify how rail passengers perceive the UK railways, and the service aspects that are most important to them in determining their overall satisfaction. To this end, 32,000 observations were selected from the 37th and 38th survey wave of the National Rail Passenger Survey (NRPS). A latent class ordered logit was estimated, which accounts for taste variation via three homogeneous groups that represent differences amongst the respondents. Punctuality and the frequency of trains are found to be core attributes for respondents, followed by upkeep and repair/comfort. Respondents also place a high value on the ticket price. Through the latent class ordered logit it becomes clear however, that there is significant taste variation amongst the three groups, which has not properly been captured amongst earlier studies. The first group, which on average represents 15% of the respondents, is more likely to represent commuters and/or respondents on delayed trains, with a high value placed on punctuality/reliability. They are likely to be unsatisfied with the railway service. The second group, which on average represents 31% of the respondents, is likely to represent commuters/business travellers that have high expectations from the railway service. For them, comfort and upkeep/repair of the train has the highest value, followed by the length of time scheduled for the journey, punctuality/reliability, and the frequency of trains on the route. They are likely to be neutral or satisfied with the service, indicating that this is a group of respondents with high expectations. The final and largest group, which on average represents 54% of the respondents, is more likely to represent leisure and personal business travellers, and the ticket price is the main factor in determining their overall satisfaction. Punctuality plays a much smaller role for this group, perhaps also because they are less likely to indicate that their train was late. Overall, they seem to be satisfied or very satisfied with the railway service offered to them. From this objective, it is clear that punctuality/reliability is the most important aspect amongst the respondents, but there is significant variation amongst the groups of passengers, with commuters and business passengers placing a higher value on punctuality/reliability compared to other groups.

The third step was to identify how respondents respond to real time train performance. Passenger satisfaction data from the 36th and 37th NRPS survey waves were matched with real time operations data and historic delay attribution data obtained from Network Rail. This resulted in a dataset where each NRPS response was enriched with information about the exact operational performance (e.g., train punctuality, service frequency, delay cause, magnitude of delay). From this, the 7,000 or so respondents that had experienced a delay were selected to understand how passenger satisfaction was affected by rail delays. Using a mixed ordered logit model, it was found that respondents reacted heterogeneously, but very negatively to an increase in real-time delay minutes, with more than 50 per cent of passengers on average predicted to be dissatisfied after 30 minutes of lateness. This dissatisfaction was exacerbated when passengers had to stand during the journey and/or received poor information and was particularly high when the train was cancelled. Interestingly, the cause of the delay did not seem to have an impact on passenger satisfaction, however, incidents that impact on a large number of trains result in worse satisfaction outcomes. Further findings indicated that perceptions vary somewhat across gender, age, and trip purpose, with daily commuters most likely to be negatively impacted by delays. These findings could help with formulating passenger focused delay recovery tactics. For instance, by setting delay thresholds for certain trains, or by avoiding cancellations of certain trains, in combination with better information provision, it might be feasible to improve passenger satisfaction during adverse service conditions.

To conclude, based on this study, it becomes clear that infrastructure incidents, as well as incidents on busy locations are most problematic in terms of their impact. This matters, as punctual rail services are a core requirement for railway passengers, even though the importance varies amongst passenger groups. In particular, commuters and business passengers seem to value punctuality much higher relative to other respondents. The study further quantifies how satisfaction is impacted by unreliability events, where it becomes clear that the main factor is the duration of a delay, and cancellations are particularly dissatisfying. These findings may help railway companies to formulate or improve on their delay recovery procedures, whilst keeping in mind how to minimise passenger discomfort.