Using Geographically Weighted Regression (GWR) to explore spatial variations in the relationship between public transport accessibility and car use : a case study in Lund and Malmö, Sweden

Abstract: In Sweden the number of cars per person has increased since the mid-20th century. With negative impacts on both health and the environment, private ownership of vehicles represents one of the major challenges in urban transport. To move travelers from privately owned vehicles to public transport has shown to be beneficial in reducing carbon emissions. However, in order to create policies that attract people towards public transport, data of factors influencing transit choice is crucial due to its validation of planning and investments. Previous studies have shown that physical proximity to public transport stations is one of the critical factors when considering transport choice. Consequently, the aim of this thesis is to analyze novel GPS data to investigate the relationship between public transport accessibility and car use in Lund and Malmö, Sweden. By modelling this relationship with the spatial regression model of Geographically Weighted Regression (GWR), regional variations are allowed and investigated. The results in Lund imply a negative association between public transport accessibility and car use, thus suggesting that car use decreases with a higher public transport accessibility. Furthermore, results in Lund indicate that the spatial regression model of GWR is a better fit to the data than the non-spatial regression model of Ordinary Least Squares (OLS). In Malmö, on the other hand, results imply that public transport accessibility does not have significant impact on car use, and suggests that the GWR model is not a better fit to the data than the OLS model. Consequently, the results in Lund and Malmö do not coincide. Nevertheless, in Lund, where model performance is the highest, results imply that car use decreases with a higher public transport accessibility. This study is one of the first to use individual GPS data together with spatial analysis to investigate the relationship between public transport accessibility and car use. Consequently, this study contributes to the literature on the effects of public transport accessibility on car use and on the use of local spatial analyses in accessibility studies. Such knowledge can be utilized in transport planning to reduce car usage.