Analysing bike crash reasons and predicting occurrence in Izmir city: insights from a logit regression model

Abstract

This study aims to analyse the occurrence and causes of bicycle crashes in Izmir, Turkey, with a focus on socio-demographic factors, cycling behaviours, and urban infrastructure. It seeks to identify key determinants of crash probability and reasons using regression models. Data were collected via face-to-face surveys from 1,796 cyclists across Izmir. 25% of surveyed cyclists reported having experienced a bicycle crash. Two statistical models were applied: a binary logistic regression to predict the probability of crash occurrence and a multinomial logit model to determine crash causes. Spatial data were also analysed to identify high-risk urban areas using GIS tools. The binary logistic model showed that higher age, education, income, and use of safety equipment reduced the likelihood of a crash. The multinomial logit model revealed that cyclist distraction, poor road infrastructure, and the sudden appearance of vehicles and pedestrians were the main crash causes. Younger cyclists were more vulnerable to sudden appearances. Spatial analysis identified Konak as the highest risk area, with Bornova and Karsiyaka also showing significant crash densities. Findings highlight the critical role of dedicated cycling infrastructure, especially in urban centres with mixed traffic. Policymakers should prioritize expanding bike lanes, improve road conditions, and promote cyclist education, particularly among youth and low-income populations.