„Competition“ between insurance frauds and experts generates evolution of methods on both sides. Theoretical and experimental works are developed the aim of which is to improve the accuracy of calculation results and reduction of errors. Research capabilities have become so refined that very narrow specialisation is necessary. This, however does not eliminate the possibility of errors. The article presents criteria for differentiation between real accidents from simulated ones and the effect of errors on expert opinion preparation. Methodological principles on how to avoid errors have been suggested. Particular attention has been paid to the problems of sound and unsound data.

This paper is unfortunately in the moment only in german language available.