Analysing the macro-level determinants of traffic safety for pedestrians and cyclists in England

Abstract

Walking and cycling are often seen as ways to contribute to the future of greener, less congested, more active transport. It is common for governments to encourage people to walk or ride their journeys rather than using motor vehicles, however crash rates per kilometre travelled are notably higher for pedestrians and cyclists than cars or similar modes of travel. Bike sharing schemes in particular are a relatively new concept in the world of transport and are often hailed by policy makers as an efficient way to travel and yet the safety effects of their use is thus far under-researched compared to other modes.

This research conducted a macroscopic investigation to identify area-level factors which have a significant association with active travel crashes. The data used concerns England and Greater London with crashes in the period 2018-2019 as well as the post-COVID period of 2022. The principal method of analysis is a Bayesian approach capable of addressing a number of statistical issues. The complexity of the models involves accounting for the following: the endogeneity of a bike share scheme’s presence on crash frequency to more accurately identify the true impact of the presence of these schemes on safety; unobserved heterogeneity which biases parameter estimates; and the temporal instability of explanatory factors which can mean they affect safety differently across different time periods.

A number of factors were significant across the models such as those related to deprivation, including crime and unemployment, being associated with a higher crash frequency. Also, most models showed a safety in numbers effect where an increase in the number of pedestrians or cyclists was associated with an increase in crashes that was smaller than would be expected if the two were proportional. This indicates that more road users means greater individual safety. For the majority of the modelled variables, the direction of the association (i.e., increase or decrease of crashes) was consistent across England and Greater London suggesting that these findings might be applicable to other cities. The crash frequencies of active travellers before and after the COVID period was found to be significantly different, but the association with bike share presence appeared to be temporally stable.

Bike sharing schemes on the other hand appear to have a complex association with crashes. In Greater London, bike share presence generally appeared to be associated with a reduction in safety, but this was not the case once endogeneity was accounted for; whereas across the rest of England the opposite was mostly found. This would seem to suggest that in Greater London bike share schemes are introduced in inherently ‘unsafe’ locations i.e., those where crash frequency is relatively higher generally due to area characteristics, but in other parts of England they are introduced in those areas which are inherently ‘safe’. The effect of the bicycle share schemes themselves on safety remains ambiguous.

Overall, this research builds a better understanding of active travel safety by providing more robust analysis which addresses endogeneity and unobserved heterogeneity. Policy recommendations can also be given such as a need to target safety interventions in more deprived areas for maximum effectiveness, as well as the importance of practitioners considering the confounding effect of safety infrastructure.

More information

Identifiers

ORCID for Michael George Forrest: orcid.org/0000-0003-2424-4037

ORCID for Tom Cherrett: orcid.org/0000-0003-0394-5459

Catalogue record

Export record