Comparisons of Predictive Power for Traffic Accident Involvement; Celeration Behaviour versus Age, Sex, Ethnic Origin, and Experience

Abstract

Driver celeration behaviour theory (DCBT) assumes that risk for a driver of causing a road crash is linearly related to speed change in any given moment and that the speed change variable (celeration) captures all risk (all vehicle control movements can be measured as acceleration). When sampling driver behaviour, the celeration variable is calculated as the average of all absolute values of acceleration when the vehicle is moving. DCBT predicts that no other variable can be a stronger predictor of (the same set of) traffic accident involvements than celeration, given equal reliability of the predictors. Also, other predictors, regardless of which ones, should associate with celeration in ways that are similar to how they correlate with accidents. Predictions were tested in a sample of bus drivers, against variables with reliabilities close to 1 (age, sex, experience, ethnic origin), which are not necessarily optimal predictors for testing but were the only predictors available. The results were largely as predicted from theory. The principles for testing the kind of predictions made from celeration theory were discussed, outlining the importance of a larger number of variables, preferably with repeated measurements.