An expert judgment model applied to estimating the safety effect of a bicycle facility

https://doi.org/10.1016/S0001-4575(99)00090-1Get rights and content

Abstract

This paper presents a risk index model that can be used for assessing the safety effect of countermeasures. The model estimates risk in a multiplicative way, which makes it possible to analyze the impact of different factors separately. Expert judgments are incorporated through a Bayesian error model. The variance of the risk estimate is determined by Monte-Carlo simulation. The model was applied to assess the safety effect of a new design of a bicycle crossing. The intent was to gain safety by raising the crossings to reduce vehicle speeds and by making the crossings more visible by painting them in a bright color. Before the implementations, bicyclists were riding on bicycle crossings of conventional Swedish type, i.e. similar to crosswalks but delineated by white squares rather than solid lines or zebra markings. Automobile speeds were reduced as anticipated. However, it seems as if the positive effect of this was more or less canceled out by increased bicycle speeds. The safety per bicyclist was still improved by approximately 20%. This improvement was primarily caused by an increase in bicycle flow, since the data show that more bicyclists at a given location seem to benefit their safety. The increase in bicycle flow was probably caused by the new layout of the crossings since bicyclists perceived them as safer and causing less delay. Some future development work is suggested. Pros and cons with the used methodology are discussed. The most crucial parameter to be added is probably a model describing the interaction between motorists and bicyclists, for example, how risk is influenced by the lateral position of the bicyclist in relation to the motorist. It is concluded that the interaction seems to be optimal when both groups share the roadway.

Section snippets

Purpose and methods

The primary purpose of this study has been to develop a model for estimating the changes in risk of modifications to the geometrical layout of streets. The model consists of three submodels: a risk index model, an expert assessment model and a risk attribute model. The model is here applied to the safety of bicyclists. A reason for this application is that bicyclists have more injury accidents per kilometer traveled than almost all other road-user categories (Gårder, 1994). On the other hand,

Illustrative example — raised bicycle crossing

In this example, we apply the model using automobile speed and bicycle speed as attributes. The model describes the combined effect of these attributes on collision risk.

The experts were asked to draw curves representing the relative risk of collision between a bicyclist and a right-turning car when the initial speed of the car and the speed of the bicyclist were varied. The experts were also asked to assess the relative injury risk of the bicyclist for different initial speeds of the

Effect on bicycle volumes and the safety effect of changes in this volume

Bicycle flows increased on one of the streets by 75% on one side and by 79% on the other side, and by 100% on the second of the two streets where extensive measurements were taken before and after the reconstruction. Measurements of bicycle flows on the two control sections indicate a general growth in bicycle flows of around 20%. The extra increase on our experimental sections would then be at least 50%. This is probably the result of the ‘better’ layout.

An analysis of the relationship between

Concluding remarks

The experts’ assessments of risk were made separately for each attribute. The exponential model was used to approximate the assessments of the experts. The experts were asked to give their opinions in the form of relative risk curves, from which the parameters of the exponential distribution were estimated. Based on the experts’ estimates, we determined the posterior distribution of the parameter by using a Bayesian multiplicative error model. Although there are obvious difficulties in the

Acknowledgements

The authors would like to express their sincere appreciation to all those who contributed in various ways to the development of this paper, including Mr Rein Schandersson, SweRoad, for contributing his knowledge about ‘knowledge engineering’, and Mr Per Näsman, Prof. Torbjörn Thedéen, Prof. Åke Claesson and Mr Björn Bergman, Royal Institute of Technology, Stockholm, Sweden and Mr Lennart Andersson, City of Gothenburg, Sweden for aiding in the planning and execution of this study. We would also

References (18)

  • K. Mizuno et al.

    Compatibility problems in frontal, side, single car collisions and car-to-pedestrian accidents in Japan

    Accident Analysis and Prevention

    (1999)
  • Anon, 1989. NUREG-1150, Reactor risk reference document. U.S. Nuclear Regulatory Commission, Office of Nuclear...
  • Ashton, S.J., 1982. Vehicle Design and Pedestrian Injuries. I: Pedestrian Accidents. In: Chapman A.J., Wade, F.M.,...
  • Carré J.-R. 1997. Mobilité urbaine et déplacements non-motorisés: situation actuelle, évolutions, pratiques et choix...
  • Clarotti, C.A., Lindley, D.V. (Eds.), 1988. Accelerated Life Testing and Experts’ Opinions in Reliability. Amsterdam,...
  • R.M. Cooke

    Experts in Uncertainty. Opinion and Subjective Probability in Science

    (1991)
  • Ekman, L., 1996. On the Treatment of Flow in Traffic Safety Analysis: A Non-Parametric Approach Applied on Vulnerable...
  • Elvik, R., Mysen, A., Vaa, T., 1997. Trafikksikkerhedshåndbok (Traffic Safety Manual). Institute for Traffic Economics,...
  • P. Gårder

    Bicycle accidents in Maine: an analysis

    Transportation Research Record

    (1994)
There are more references available in the full text version of this article.

Cited by (0)

View full text