`For and against' arguments are presented in this article. The `for' arguments are as follows. The nature of human risk perception is described. Evidence for risk homeostasis in road crashes is outlined with reference to countries that changed from left to right hand traffic. Drivers were initially more cautious until they perceived the risks to be lower than they had thought, consequently there was a temporary reduction in the number of accidents. The effectiveness of highway improvements, and the likelihood that as one location is made safer another location experiences more accidents, are debated. The importance of reducing the target level of risk for drivers is stressed. The `against' arguments include that driver behaviour cannot be based on achieving constancy of risk, particularly since that risk is not based on frequent experience. Widespread participation in lotteries and casino gambling reveals how poorly the public understands the laws of probability. The human capacity for self-delusion in relation to risk and the likelihood of human error is explored. Flaws in some research studies due to inappropriate aggregation of data are exposed. On the contrary, observations on the behaviour of drivers before and after the introduction of safety belt laws in Newfoundland found no changes in risk related driving behaviours. If drivers worked to constancy of risk, it could be argued that they would fail to fasten safety belts in the presence of air bags, but this does not occur. The most compelling argument against risk homeostasis is the observation that occupant death rates in passenger cars per distance travelled decreased by nearly two-thirds in the USA from 1964 to 1990. Ninety percent of this reduction is due to vehicle modifications.
Samenvatting