Sensitivity testing of alternative public transport passenger satisfaction analysis techniques

Abstract

Paper presented at the 30th Annual Southern African Transport Conference 11-14 July 2011 "Africa on the Move", CSIR International Convention Centre, Pretoria, South Africa.

This paper investigates alternative methods of analysing public transport user satisfaction and importance rating data. It reviews criticisms of conventional 'importance-performance analysis', as developed and applied in the field of market research over the past three decades, and reports on sensitivity tests undertaken to assess the impact of alternative techniques upon the robustness of resulting recommendations for priority system improvements from a user perspective. The alternative techniques include: the plotting of satisfaction and importance ratings in a 'quadrant model', versus a 'diagonal model'; alternative methods of estimating the most accurate slope of the iso-priority line in 'diagonal models'; and the use of mean ratings, versus the percentage of dissatisfied respondents, in plotting service attribute performance. The sensitivity tests use a (n=993) passenger satisfaction intercept survey dataset, collected from train, bus and minibus-taxi passengers in Cape Town in 2009 using Likert rating scales. The paper explores which analytical techniques hold greatest promise in further measurements of public transport passenger satisfaction planned in Cape Town. It argues that 'diagonal models' produce more reliable prioritisations than 'quadrant models', but that the alternative methods of determining the slope of the diagonal each have weaknesses. The paper discusses the dynamic relationship between satisfaction rating and changing passenger expectations resulting from the implementation of system improvements, and concludes that changes in public transport service provision are likely to result in changes in passenger expectations and satisfaction, rendering longitudinal comparison problematic.