This report uses signal detection theory (SDT) to model motorists’ decisionmaking strategies at grade crossings in order to understand the factors that influence such decisions and to establish a framework for evaluating the impact of proposed countermeasures. This report is intended to update and expand the original analysis conducted by Raslear (1996), which examined the effectiveness of grade crossing warning devices and determined whether their effectiveness was due to variations in the signal-to-noise ratio (sensitivity), bias to stop, or a combination of these two components of signal detection theory. This report documents the results of four empirical and theoretical tests of the SDT model to understand how different warning devices and countermeasures influenced drivers’ decisions at grade crossings in the 21 years from 1986 (as reported by Raslear) to 2007 (the most current year available when this effort began). In the first analysis, we compare accident data from 2007 with that from 1986 and describe the necessary adjustments to our assumptions in setting up the model. In the second analysis, we apply this revised framework to a more detailed historical analysis of driver decisionmaking at grade crossings. The third analysis describes our test of the robustness of the SDT model and our application of SDT to predict the effect of proposed countermeasures and safety factors. The fourth analysis was based on a theoretical model to test the predictive abilities of the SDT framework through performance by an ideal observer. While the previous analyses examine the empirical changes in sensitivity and bias over time and with changes in the grade crossing environment, the analysis of the ideal observer posits theoretical mechanisms for those changes and compares theoretical outcomes with actual outcomes. (Author/publisher)
Abstract