What is the relation between crashes from crash databases and near crashes from naturalistic data?

Journal article, 2020

Naturalistic cycling data are increasingly available worldwide and promise ground-breaking insights into road-user behaviour and crash-causation mechanisms. Because few, low-severity crashes are available, safety analyses of naturalistic data often rely on near-crashes. Nevertheless, the relation between near-crashes and crashes is still unknown, and the debate on whether it is legitimate to use near-crashes as a proxy for crashes is still open. This paper exemplifies a methodology that combines crashes from a crash database and near-crashes from naturalistic studies to explore their potential relation. Using exposure to attribute a risk level to individual crashes and near-crashes depending on their temporal and spatial distribution, this methodology proposes an alternative to blackspots for crash analysis and compares crash risk with near-crash risk. The novelty of this methodology is to use exposure with high time and space resolution to estimate the risk for specific crashes and near-crashes.