Are the speed-crash models applicable for low speeds?

Abstract Several speed-crash models have been developed, enabling proactive estimations of road crash changes in response to driving speed changes. The models have been based primarily on the macro- level high-speed data and have gained popularity; however, their suitability to the localised (micro- level) low-speed conditions is uncertain. To fill this gap, the present study collected a number of speed-crash data samples to study the relationship between the expected crash changes based on the Exponential model and real crash changes, using the specific categories based on speed management measure characteristics. Although there were some positive results, the overall findings were mixed: approximately half of the categories did exhibit a statistically significant relationship between the model prediction and reality, but half did not. While the Exponential model is a valuable speed-crash estimation tool, it may not be fully valid for the micro-scale evaluations, including the low-speed environment and speed management measures with small speed reductions.