TY - JOUR PY - 2019// TI - Accounting for mediation in cyclist-vehicle crash models: a Bayesian mediation analysis approach JO - Accident analysis and prevention A1 - Kamel, Mohamed Bayoumi A1 - Sayed, Tarek SP - 122 EP - 130 VL - 131 IS - N2 - Cyclist safety is affected by many factors on the zonal level. Previous studies have found associations between cyclist-vehicle crashes and vehicle and bike exposures, network configuration, land use, road facility, and the built environment. In addition, the network configuration, land use, and road facility were found to affect bike exposure levels. The association of zonal characteristics with both exposure and crashes may bias the development of macro-level bike safety models. This paper aims to explain these associations simultaneously using a form of Structural Equation Modelling approach. The analysis assesses the mediated effects that some variables have on crashes through their effects on bike exposure (by setting bike exposure as a mediator). Data from 134 traffic analysis zones (TAZ's) in the City of Vancouver, Canada is used as a case study. The indirect effect of network configuration, land use, and road facility on cyclist-vehicle crashes was assessed through Bayesian mediation analysis. Mediation analysis is an approach used to estimate how one variable transmits its effects to another variable through a certain mediator. These effects could be direct only, indirect only (through a certain mediator), or both direct and indirect. The results showed that the bike kilometers travelled (BKT) was a mediator of the relationship between network configuration, land use, and road facility and cyclist-vehicle crashes. The mediation analysis showed that some variables have different direct and indirect effect on cyclist-vehicle crashes. This indicates that while some variables may have negative direct association with crashes, their total crash effect can be positive after accounting for their effect through exposure. For example, bike network coverage and recreational density have negative direct association with cyclist-vehicle crashes, and positive indirect association leading to positive total effect on cyclist-vehicle crashes.

Copyright © 2019 Elsevier Ltd. All rights reserved.

Language: en

LA - en SN - 0001-4575 UR - http://dx.doi.org/10.1016/j.aap.2019.06.009 ID - ref1 ER -