Safety ratings by car model are published by various organisations, research institutes, and car magazines, all using different rating methods. Not seldom, the results from different studies are contradictory - and maybe there is a reason to question the reliability of certain results and rating procedures. This paper suggests some criteria for a fair evaluation system and gives a brief overview of the three main safety rating procedures: (1) ratings based on accident data, (2) ratings based on laboratory test data, and (3) ratings based on expert assessments. Pros and cons and possible pitfalls when using these methods are discussed. Multi-severity laboratory testing in various crash modes, connecting laboratory test data to real world injury data, is suggested as the best way to predict the overall real world safety of a car model in traffic, before it is actually out on the roads. (A)
Samenvatting