Vienna, March 2012
The text presented here advocates the systematic use of so called accident surrogates – critical incidents, traffic conflicts, near accidents etc., – with reference to the latest papers stemming among others from Canada (see reference list). Over the past decades, learning from road traffic accidents was found necessary to develop road safety interventions in what has been termed the “corrective” approach. Accident and injury data are collected, recorded, analyzed, interpreted and discussed.
Typically, road safety diagnoses are performed by classifying accidents (according to location, groups of road users involved, etc.), trying to figure out why they occurred, and trying to identify the causal factors which may be eliminated through counter measures. Some of these factors may be systematically underestimated due to poor quality of available data – for instance, the most frequent cause of accidents is speed, a fact now well-proven though multi-disciplinary research but which often does not appear in accident diagnoses.
What the train traffic and air traffic authorities do is quite the opposite: there, accidents are not expected to happen at all, but in the rare cases when they do, each and every accident is painstakingly documented and analyzed; accidents happen seldom enough that statistical assessment is not needed, and indeed is not possible. In order to learn from experience, every incident, light or serious that may have led to a crash is thus recorded and analyzed. Now this will be difficult to achieve in road traffic, at least in the short run; and, of course, as long as there are accident and injury statistics, it is obvious that thorough analysis work has to be strived for. However, a statement often read in master or doctorate theses in the area of road traffic safety is that the accidents numbers were “too low” for a thorough examination of certain phenomena. Such a statement would never be acceptable in rail and air transport research where it goes without saying that accident numbers should be near zero so that another way to find out where there might be a problem has to be found. This could also, and indeed should, be the philosophy adopted for road traffic safety, at least on the meso (?) and micro level. Especially, it should be possible to perform safety analysis work when past accident numbers are low or zero or when deep changes brought to the road traffic system make accident history irrelevant.
Head note #1 would state: Our main objective should not only be to do thorough and methodologically clean evaluation of accident data and statistics because the goal has to be that accident numbers become very low or zero, at least at the meso- and micro level; instead, accident prevention must be the declared objective so that accident statistics generally lose their importance for traffic safety work at the meso and micro level (which follows the Zero-Vision objective closely). Head note #2 brings the scientific character of the road safety research on the plan. In natural sciences it is necessary to link theoretical aspects with empiricism. For instance, millions of dollars are spent to demonstrate the evidence of theoretically assumed particles in CERN. Until now, the accident as the major, unsolicited event related to road safety research has been hardly or only poorly empirically depicted. Generally, road safety work draws only insufficiently from empirically collected data. If there are problems with a „new“ group of road users, like for instance the bicyclists,
under to-days conditions society will rather speculate about the character of these problems for months and years than systematically look for empiric (= observation) data of this phenomenon. Problematic interactions on intersections which may sometimes and up in accidents, are only rarely – or never – empirically evaluated. The explanation given in this respect is often that the money is scarce for this kind of research. But, on the other hand, if accidents could be prevented the monetary gain would certainly be much bigger than the money spent on research. Consequently, the Head note #3 is as follows: the money that the national economy loses due to accidents is hardly visible and not missing in the books of specific institutions, while the money spent on research is clearly visible in the books. To make it plain – it is easier to wait for accidents to happen. Under this perspective it is quite comfortable to rely on the argument of many “experts” that validity of surrogates, like traffic conflicts, is insufficient or not proved. Unfortunately, this is a conservative approach, which supports and preserves the existing system, because Head note #4 is: correlative validity referring to accidents as “true” criteria does not make much sense because of their lacking prognostic validity on the meso- and especially on the micro level, viz. because of their bad stochastic characteristics; on the local level, the accident data do not allow any valid prediction of potential accidents in the future. And, without in-depth analysis, they do not provide much information concerning what goes wrong in practice, and thus ends up in accidents in some cases. The systematic use of safety indicators, instead, provide a range of empiric information as a more reliable basis for prevention measures.
Thus we have the following recommendation:
• On chosen spots regular traffic conflict or other types of behaviour and interaction investigations must be conducted systematically;
• the choice of critical spots must be refined and optimised step-wise;
• it has to be made sure that such spots are to be designed, or improved, based on such investigations, so that accidents, or certain types of accidents, do not occur there anymore; the savings for national economy should be pointed out and they should be given a practical meaning in the national budget.
If all these steps are taken the importance of such empirical investigations will be acknowledged soon.
|Traffic Conflicts||Traffic Conflicts|