Effect of critical incidents on public transport satisfaction and loyalty: an Ordinal Probit SEM-MIMIC approach

Abstract

Supplying public transport systems with high levels of service quality is fundamental for retaining users and attracting new ones. Policies that improve transit service quality will ultimately lead to more sustainable travel patterns. Measuring overall service quality implies measuring the quality of several specific attributes and is prevalently evaluated through the perceptions of users, using satisfaction rates. In this study, we demonstrate that there is a further element that can influence users’ perceptions, the so-called critical incidents (CI), defined as encounters that are particularly satisfying or dissatisfying. The concept is not restricted to ratings of the predefined product or service attributes, because customers who experience CI remember them well and can usually describe the experience. We implement a framework that includes CI and is innovative for several reasons. Firstly, we introduce attribute-specific (e.g. reliability, safety, comfort) CI to explain attribute-specific satisfaction levels, and then we model these with latent constructs allowing for measurement error in recalling the CI. We also demonstrate that using an Ordinal-Probit approach leads to more accurate results than its numerical counterpart, the latter possibly presenting biased results. Finally, we present a full Structural Equation Multiple Cause Multiple Indicator (SEM-MIMIC) model, which corrects for heterogeneity in the perceptions of users regarding satisfaction with the various service attributes, with the overall service, and with loyalty. For these purposes, we analyse an extensive database (96,763 interviewed passengers) derived from Customer Satisfaction Surveys in the railway services offered in the hinterland of Milan. Our main contribution to the literature is that we show that the occurrence of a CI has a substantial negative impact on passenger satisfaction for all service attributes. As it is a policy-related variable, it can be managed directly by the public transport (PT) administrators. To better plan and improve PT services, avoiding CI in specific items should be the strategy to follow. On the other hand, reliability, and added-value services are the primary service attributes that have a positive effect on satisfaction with the overall service and, in turn, on loyalty. Our model can be useful for PT administrators as it sheds light on how to improve the service according to users’ preferences, and by considering the differences among user categories.