Crash Statistics & Epidemiology

Population-level measurement of crash and injury rates: how often crashes happen, per unit of exposure, and how rates differ across groups and countries.

What's here. Counts, rates, and trends work — incidence prevalence, fatality injury trends, demographic disparities, and comparative international comparisons — together with the denominator methods those rates depend on, namely exposure measurement (distance, trips, time at risk) and induced exposure using at-fault versus not-at-fault control samples. Also here is telematics crash prediction, which models crash and claim risk from naturalistic driving, insurance telematics, and other fleet-scale datasets.

What isn't here.

• why a crash happened — contributing factors, crash typology, in-depth case investigation, and naturalistic crash/near-crash coding — Crash Causation & Contributing Factors. This is the closest neighbour and the most common wrong landing. Work here counts crashes and estimates rates over a population; work that attributes a crash to a cause or reconstructs the pre-crash sequence sits there. Note that induced_exposure here is a rate-estimation trick that borrows fault information, not a causal analysis.
• workload_measurement — how driver mental demand is instrumented and scored — Cognitive Workload. exposure_measurement here means quantifying how much driving was done and by whom, so a crash count has a denominator. It has nothing to do with measuring a driver state; measurement of workload, including its physiological indices, is in the attention and workload area.
• what happens after the impact — post-crash driver behavior, secondary collisions, egress, and delays in reaching care — Post-Crash Behavior & Response. Same research area, different shelf. Fatality and injury counts and their trends over time are here; the response and outcome side of the crash sequence is there.