Delivering sustainable mobility targets : investigating the characteristics of car users most likely to switch to sustainable modes

Abstract

Dependence on car use generates adverse effects on the sustainability of our urban environment due to air and noise pollution and global warming. Along with technological and infrastructure improvements, an overall solution to this problem would require dramatic change to the way we chose to travel. Behavioural change of car users to have them move to other sustainable travel options making less impact to our environment is crucial. Behavioural change of car users is not straightforward given the huge range of factors governing their travel decisions. Local authorities with constrained budgets require evidence to help them introduce policies that will have significant impact on car use. Therefore, we need to understand which specific groups of car users are more inclined to change to sustainable modes so that ultimately targets can be met. The global aim of this research is to characterise target groups of car users (as a driver or passenger) who are more likely to switch from private transport to sustainable modes. A comprehensive analysis of the British Social Attitudes (BSA) survey data collected during 2011 to 2014 was used in this research. The research identified: important attitudinal factors through dimension reduction analysis; groups of car users’ travel mode choice decisions based on socio-demographic variables; relationships between car users’ groups and attitudinal factors which relate to modal shift potential and finally deriving a mathematical model to predict the likely uptake of sustainable modes. The statistical analysis techniques included Descriptive Analysis (DA), Factor Analysis (FA), Cluster Analysis (CA), Multiple Correspondence Analysis (MCA), and Multinomial Logistic Regression (MLR) coupled with a Multivariate Probit Model (MPM). When developing MPM, Bayesian inference was taken into consideration because it allows the uncertainty in the parameters to be incorporated in the model. Finally, the MPM was used to investigate the relationships between the responses of car users to the different attitudinal questions. The FA is particularly useful to reduce a wide range of variables into a smaller number of factors and three main factors labelled as Attitudes to transport and the environment; Traffic awareness; and Modal shift potential emerged from 14 attitudinal variables. The CA is a method to group the car users based on their socio-demographics and travel related variables. Five clusters emerged namely middle–aged (35-44), female, full–time iv employee (Cluster 1); middle–aged (35-44), male, full–time employee (Cluster 2); mature adults (45-54), male, full–time employee (Cluster 3); older–aged (65+), male, retired (Cluster 4); and middle–aged (35-44), female, looking after the home (Cluster 5). Whilst the majority of respondents are strongly car-orientated either as a driver or as a passenger, the car users associated with Cluster 2 were found to be more likely to cycle once a week already (29%) and travel by train less often than once per month (39%) compared to other car users in other clusters. The MLR investigates the relationships between the factors and the clusters exploring the details of the change of attitudes over the years, for instance from 2011 to 2014. The outcome of the results show that Cluster 2 has considerably higher environmental awareness compared to other groups of respondents. Therefore, this group is likely to have potential to switch travel modes. The MPM was developed in this study specifically for ordinal responses and enabled responses to several questions, which can be correlated to be considered in a single model. This approach is different to MLR, which does not consider these correlations. The MPM suggested that younger and older cohorts are the least likely to be susceptible to change whilst the middle-aged population is more likely to mode shift to cycling or public transport. However, the reverse is true for respondents with larger household size. The willingness to switch from the car to walking and cycling for short journeys of less than 2 miles appears to increase depending on the increasing number of people living in the household. Females will be less likely to switch mode from cars to cycling for short journeys. However, switching to walking and going by bus were more or less equally acceptable for both males and females. In addition, there was a greater tendency to agree to use cars with lower CO2 emissions for the sake of the environment among respondents with one car per household compared to respondents with four or more. This research demonstrated that fitting MPM using Bayesian inference is both a practical and effective way to analyse ordinal survey data and is a novel aspect in this study.