Reconstruction of road accidents combines objective and subjective action. The former concerns science, the latter assessment of human behavior in the context of objective findings. It is not uncommon for experts equipped with an arsenal of tools to obtain similar results of calculations, but to present radically different conclusions about the cause of the accident. The use of sophisticated methods of uncertainty analysis does not guarantee improvement in quality of reconstruction, because, increasingly, the most serious source of reduced reliability of reconstruction is problems in logical inference. In the article the structure of uncertainty and reliability of accident reconstruction was described. A definition of reliability of road accident reconstruction based on the theory of conditional probability and Bayesian network, as a function of modeling, data and expert reliability (defined in the text) was proposed. The uncertainty of reconstruction was made dependent only on the uncertainty of the data. This separation makes it possible to conduct a qualitative and quantitative analysis of reconstruction reliability and to analyze its sensitivity to component parameters, independently of the uncertainty analysis. An example of calculation was presented. The proposed formalism constitutes a tool helpful to explain, among other things, the paradox of reliable reconstruction despite its uncertain results or unreliable reconstruction despite high precision of results. This approach is of great importance in the reconstruction of road accidents, which goes far beyond the analysis of a single, homogeneous subsystem. (Author/publisher)
Samenvatting