The traffic conflicts technique is a device for indirect safety measurement. It requires at present the conduct of a field survey to count conflict occurrence. On this basis the rate at which conflicts occur is estimated. This report deals with the accuracy of such estimation and its dependence on the design of the field survey. First, present practice in conflict count duration is reviewed. Next, the relationship between count duration and estimation accuracy is examined. Using data obtained from several sources the daily variability of conflict counts is described. It is concluded that the expected conflict rate varies from day to day. Use of the negative binomial distribution is suggested as appropriate for the representation of the distribution of sample means obtained from conflict studies. On this basis, confidence limits and probabilities of type i and type ii errors in hypothesis testing are obtained and tabulated. Their use in study design is illustrated by numerical examples. The marginal increase in estimation accuracy diminishes rapidly as conflict counting time increases. Thus, there is little to be gained by counting longer than three days. This establishes a practical limit to the accuracy with which expected daily conflict rates can be estimated. (Author/publisher)
Abstract