A Two-Step Model of Satisfaction with Public Transportation

Abstract

The purpose of this paper is to present a model of consumer satisfaction with urban trips. This model was first developed for the Paris Public Transportation Authority (RATP) and was later applied to other modes of transportation (railway, automobile traffic, pedestrian walk). The objective of the model is to measure and analyze satisfaction with a specific consumption experience (one trip from its beginning - leaving home for instance -to its end- arriving destination) and not a general level of satisfaction with the city transportation systems. The research process included a qualitative survey, a first test on a small sample (100 interviews), a full-scale survey (1,000 interviews: 100 x 10 strata based on transportation modes and combination of modes). The model consists of a two-step structuration of the evaluation of a specific ride:
1. a perceived performance model in which elementary events (pleasant or unpleasant feelings that may occur during the ride) are first structured into aggregate (or "synthetic") dimensions (through multidimensional scaling and cluster analysis) ; these events correspond to the level of managerial action ;
2. a satisfaction model in which these dimensions play the role of intermediary variables between elementary impressions and overall satisfaction with the ride ; they are integrated into a linear additive model which also includes measurements of disconfirmation and image variables.
Segmentation studies have been conducted to investigate the differences between strata: different satisfaction levels or distinct satisfaction processes? In the first part of the paper, the general framework of the model and its developmental process are presented. The second part is devoted to methodological problems encountered in the estimation phase of the second step of the model (linear additive): interactions among explanatory variables. A in-depth analysis has been conducted on two strata; methods used include: multitrait--multimethod matrices, factor analysis, two-way ANOVA with interaction, LISREL. Results show that:
1. perceived performance and disconfirmation have additive effects on satisfaction level, and
2. image of transportation modes does not reach discriminant validity from performance evaluation and consequently cannot be used as a generalized expectation level.