View/hide extracted text (may have errors)
SAFETY PERFORMANCE FUNCTIONS FOR BICYCLE CRASHES IN NEW ZEALAND AND AUSTRALIA Shane Turner Beca Infrastructure Ltd 119 Armagh Street, Christchurch, New Zealand Phone +64 3 374 3188; Fax +64 3 366 3188; E-mail: shane.turner@beca.com Graham Wood Department of Statistics, Macquarie University NSW 2109, Australia Phone: +61 2 9850 8553; Fax: +61 2 9850 7669; E-mail: graham.wood@mq.edu.au Tim Hughes New Zealand Transport Agency 129 Hereford Street, New Zealand Phone: +64 3 964 2843; Fax: +64 3 964 2855; E-mail: tim.hughes@nzta.govt.nz Rohit Singh Beca Infrastructure Ltd 119 Armagh Street, Christchurch, New Zealand Phone: +64 3 366 3521; Fax +64 3 366 3188; E-mail: rohit.singh@beca.com ABSTRACT After decades of decline, recreational and commuter cycling is becoming more popular in many Australasian cities. While this is encouraging from a sustainable transport and public health perspective, a major concern to national, state and local governments is the higher crash risk faced by cyclists compared with drivers or passengers in motor-vehicles, particularly when cycling on roads. It is important that transport professionals understand the level of risk faced by cyclists within various parts of the road network and the measures they can employ to mitigate that risk. This paper presents research findings from three main safety studies undertaken in New Zealand using data from New Zealand cities and Adelaide in Australia. Research has been undertaken using both generalized linear modelling and before-after control-impact methods. Over the various studies, crash, traffic and cycle volumes and layout data has been collected for urban road links, traffic signals and roundabouts. Flow-only models have demonstrated a ‘safety in numbers’ effect; with crash risk per cyclist shown to be lower as cycle volumes increase. By adding other variables to the models, it is been possible to gain a level of understanding of the impact that road section length, motor-vehicle speed, visibility, presence and type of cycle facilities and lane and road width have on various crash types. Before and after analysis has been employed to help understand whether there is any bias in the sites that have received cycle facilities The research findings concerning the effect of cycle facilities in improving safety are mixed. Well designed facilities, including those of adequate width and painted with colour appear to perform the best. Key words: Safety performance functions, Crash prediction models, Generalized linear models, Before and after analysis, Cycle safety Word count: 6,868 Words + 4 Figures + 6 Tables TRB 2011 Annual Meeting Paper revised from original submittal.
