Defining the effects of traffic violations on crash frequency by applying a spatial panel model

Previous studies have examined the relationship between traffic violations and crash risk, and most have shown that increasing the density of traffic enforcement and issuing more tickets may deter aggressive driving and reduce local crash risk. However, few studies have discussed the spatial-temporal effects. Hence, this research applied Moran's I and a spatial panel model to define the effects of traffic violations on the frequency of crashes. Important environmental factors were used as control variables in order to capture the spatial-temporal relationship. The results show that if the violations are related to driving style, such as driving-under-influence and speeding, increasing the corresponding enforcement and issuing more tickets will decrease number of crash. However, if the violations are related to drivers' daily car use habits, then areas with higher numbers of violations may represent urban areas with more drivers and greater traffic exposure. The final model we suggest is a spatial autoregressive model with autoregressive disturbance and a fixed effect (SARAR_FE), which indicates that while there were differences in crashes by city, there were still some spatial patterns caused by unobserved variables. This model shows this unexplained spatial pattern in the error term, and spillover effects of enforcement in the lag term. Based on our results, police authorities should try increasing their enforcement density (for violations resulting from driving style) in target cities; the corresponding crash frequency could be reduced due to deterrent effects.