Why road accident data correlations do not reveal causations?

Road accident analysis is focussing too much on accident characteristics and too little on accident causes. Very often, we look where the data are and not where the problems are. Consequently, there is imperative need for data not only on casualties but also on: –    exposure to risk (traffic kilometres of vehicles and pedestrians) to highlight the real dimension of the problem, and –    intermediate indicators on the performance of the transport system components: driver behaviour, infrastructure and vehicle performance (including for example percentage of speeding, seat belt, helmet and cell phone use, but also existence of sidewalks, pedestrian crossings, etc.) in order to shed light to the accident causes and proceed better from correlations to causations. The data needed to support road safety improvement can be either microscopic or macroscopic depending the level of road safety solutions examined (interventions, measures, programmes, policies). Microscopic data are mostly needed for more operational safety measures and interventions and concern the safety behaviour and performance of drivers, passengers and pedestrians, the safety performance of junctions, road segments and specific local road networks and the safety performance of the different types of vehicles. Macroscopic data are mostly needed to support road safety policies and programmes and concern safety performance of the whole population, infrastructure and vehicles of a city, region or country. Detailed crash data are meaningful only if they are combined with exposure data (crashes per km driven, per traffic characteristics, per time, etc.) and crash causalities are revealed when crashes are correlated with safety performance indicators (behaviour, infrastructure, traffic, vehicles). Furthermore, even when accident causes are identified, the selection of corresponding countermeasures requires appropriate data and knowledge on measures implementation and effectiveness. Analysis of high resolution data coming from both public and private entities has a great potential to reveal hidden and critical crash properties and causes, especially when analysed over time and thus leading to identifying the appropriate solutions. Contribution at the Together for Safer Roads Roundtable on New Trends and Opportunities in Road Safety, Washington, June 2017  ΠΗΓΗ: Ε.Μ.Π