This paper identifies and describes the elements that nust form part of a framework for predicting degrees of negative feedback to safety measures that are being scrutinized for implementation. These are (1) the availability of an "engineering" estimate of a measure's effect, i.e. the effect the measure would have if there were no feedback; (2) knowledge of the degree of penetration of the measure in the relevant population; and (3) the availability of a behavioral model of sufficient strength. The paper provides an illustration by estimating what degree of negative feedback to expect when, in one way or another, authorities succeed in raising the seat belt wearing rate in their jurisdiction. The illustration makes use of data sets that meet the requirements set forth above. The behavioral model is a decision-making model as proposed by Janssen and Tenkink (1988). The results are encouraging in that the degrees of negative feedback occurring in the data sets are postdicted with a satisfactory level of accuracy by applying the steps specified in the framework. Selective recruitment of parts of the relevant population could provide an alternative explanation for feedback-like phenomena. However, the selective recruitment hypothesis appears to be incompatible with certain essential pieces of empirical evidence. (Author/publisher) For the covering abstract see IRRD 870346.
Abstract