Estimation of Calibration Functions for Predicting Crashes on Rural Two-Lane Roads in Arizona

This paper describes a study that used data from rural two-lane roads in Arizona to illustrate two issues. One issue is the importance of selecting an appropriate sample size for calibrating the Highway Safety Manual (HSM) predictive models based on the desired accuracy of the calibration factor, instead of relying on the guidance from the HSM (the HSM recommends 30 to 50 sites with at least 100 crashes per year but provides no statistical basis for this guidance). The second issue is the usefulness of estimating calibration functions when individual calibration factors do not provide a proper fit of the local data. On the basis of the outcome of the exploration of these two issues, this study recommends a simple calibration function for predicting total crashes on rural two-lane roads in Arizona. This paper also provides a brief overview of a procedure in Microsoft Excel that can be used by practitioners (after appropriate training) to estimate simple calibration functions.