Analysis of prior accident history can be used to guide decisions on the deployment of safety-related ITS systems – systems primarily targeted at reducing accident numbers and accident severity. The analysis can help identify what systems to install and where to install them. There is little benefit to road network operations from deploying systems and technologies where they are not needed - and where they will have little impact. For example in many countries the location of speed cameras is determined by considering the relationship between the number of speed violations and excessive numbers of accidents. A similar approach, focused on the relationship between accidents and violations, can be used to determine the best location for red light cameras.
High-quality accident databases and appropriate tools for data extraction and analysis are required to identify problem sites. Proper procedures need to be applied to site identification to avoid selecting sites that are not fundamentally unsafe but may be subject to random fluctuations in accident numbers from one year to another. This problem is known as “bias by selection” where the resulting observed “improvements” in performance are in large part the result of random variation (“regression-to-the-mean”).
Before deciding that an ITS solution is needed, a proper assessment needs to be made of the alternatives. Standard tools are cost benefit analysis (CBA) and cost effectiveness analysis (CEA). CBA evaluates whether the predicted monetised benefits outweigh the costs. CEA measures alternative interventions against success criteria - such as lives saved or improvement in quality-adjusted life years (QALY). The QALY measurement represents the gap between an ideal scenario where everyone lives into old age free of disease and disability and the real-life situation in the population. (See Project Appraisal)
Some cooperative systems, such as intersection collision avoidance systems, are particularly targeted at blackspots. (See New and Emerging Applications)
There is considerable literature on problem site identification in road safety. The standard technique is to apply the Empirical Bayes (EB) method, which considers both the regression-to-the-mean effect and the expected safety performance of a particular site, based on the safety performance of similar locations.
Practitioners in developing countries need to consider the reliability and consistent quality of the available accident data. Where toll roads exist, the data may be fairly complete, in contrast to other parts of the road network. In some countries, data may be recorded where the police have relatively easy access to the accident locations but coverage may be poor elsewhere.
A useful introduction to the Empirical Bayes (EB) method is provided by Hauer, Harwood, Council and Griffith (2002), Estimating Safety by the Empirical Bayes Method: A Tutorial, Transportation Research Record 1784. A preprint version of this paper can be found at http://ezrahauer.files.wordpress.com/2012/08/trbpaper.pdf.