Modelling consumer perceptions of service quality for urban public transport systems using statistical models for ordinal data

Abstract

The paper deals with ordinal response models to evaluate urban public transport systems with the purpose of interpreting consumers’ responses with reference to their profiles. New methodological developments provide opportunities for a more thorough and accurate analysis of perceived service quality. The evaluation of the uncertainty component accounts for accuracy in the assessments. Diagnostic procedures allow to evaluate model specification, with respect to the proportional odds assumption, the adequacy of the mean structure and the occurrence of heterogeneity. The impact of the covariates on the discrete distribution of the observed response is appraised through their marginal effects. The selection of the appropriate covariates leads to the identification of clusters of users, which are compared through ordinal superiority measures. Consequently critical situations are detected.