Modelling and analysing cyclist road safety performance in Scotland: a safety in numbers perspective

Meade, Margaretha Suzanne

Abstract

Reported cyclist casualties are disproportionally high relative to their modal share. This is a well-documented problem and yet we currently do not have local or national models that can estimate exposure (i.e. intensity of travel by bike). There is therefore is limited capacity among practitioners and professionals to estimate normalised risk at a disaggregate level and there is little evidence available for network level risk factors associated with cyclists’ safety. Simultaneously an increasingly popular vulnerable road user policy development is the safety in numbers effect. The main theory behind safety in numbers is simply that more cyclists, or pedestrian activity, reduces the overall risk of having an accident.

This research investigates whether there is a safety in numbers effect in Scotland; examines if there are wider spatial, demographic and policy differences affecting cyclists; and develops a novel modelling method to estimate cyclist exposure based on open data and open software. A comparison of traditional road safety macro-level global regression models, with local meso-level geographically weighted regression models to investigate safety in numbers was used to explore the nature of the safety in numbers effect in Scotland.

The comparison of the global and local model forms yielded four main results. First, local models’ account of spatial dependence provide a better statistical fit than the traditional global models. Second, both the global and the local models confirm that there is a safety in numbers effect in Scotland but that the effect is less than reported in the literature and referenced in Scottish policy documents and guidance. Third, the local models confirm that safety in numbers is not static and that the effect varies spatially, depends on local infrastructure factors and the intensity of cycling activity. Finally, a safety in numbers effect can co-exist with hazard in scarcity, weaker safety in numbers effects were found among women and between injury severity levels.

Edinburgh was identified as an urban area with high potential for a safety in numbers effect within Scotland because, unlike across the most of Scotland, cycling doubled between 2001 and the 2011 census and is likely to double again by 2021 given current trends. The results found that there is a safety in numbers effect in Edinburgh for slight casualties but that there is little to no effect for killed or serious injuries (KSIs). The strength of the effect (i.e. less cycling risk) is associated with higher concentrations of some types of cyclist infrastructure but not others. Unprotected on-road cycle lanes, advanced stop lines and bus lanes were not positively associated with improved cyclist safety, however quiet routes, offroad cycle lanes and segregated facilities were found to be safer. Therefore, despite higher cycling activity, Edinburgh does not yet benefit substantially from a safety in numbers effect. This confirms that cycling numbers alone do not produce safety in numbers; and effective and ineffective cycling infrastructure was also identified.

A further finding and benefit of using spatial modelling is the visualisation of safety in numbers in a local context to identify where it does or does not manifest and this also facilitates evaluation of facilities and other policy interactions or factors. Furthermore, the safety in numbers effect can be used as a Safety Performance Function to assess road safety which is a superior metric than rate-based measures. This has not previously been demonstrated in the literature and therefore this research contributes and adds to the understanding of the safety in numbers effect and demonstrates the need to develop cycling flow models to provide evidence based research.

Citation

Meade, M. S. Modelling and analysing cyclist road safety performance in Scotland: a safety in numbers perspective. (Thesis). Edinburgh Napier University. Retrieved from http://researchrepository.napier.ac.uk/Output/2682661

• Thesis Type: Thesis
• Deposit Date: Aug 24, 2020
• Publicly Available Date: Sep 10, 2020
• DOI: https://doi.org/10.17869/enu.2020.2682661
• Keywords: Safety in Numbers, Spatial Model, Exposure, Cyclists, STATS19, Traffic safety
• Public URL: http://researchrepository.napier.ac.uk/Output/2682661
• Award Date: Oct 31, 2019