Incorporating systems thinking approach in a multilevel framework for human-centered crash analysis

Variations that exist in the frequency and severities of crashes across regions may be due to differences in road user behaviors or indirectly due to differences in regional characteristics. Regional strategies towards “vision zero” road fatalities, consisting of appropriate safety policies and laws, supported with public education and backed by appropriate sanctions, have the ability to shape road user behaviors in the long term. In this paper, certain human-centered crash factors are viewed as the outcome of a hierarchical system made up of road users nested in regions, in a way that regional characteristics like policies and punitive measures influence road user behaviors. Hence, we propose a multilevel framework that captures driver characteristics and regional attributes that directly and indirectly affect crash outcomes. The concept was applied to crash data analysis for the state of Alabama, where it was found that the probability of a fatal crash involving a typical driver is 0.115. About 6.19% of the variability in the fatal crash rate involving drivers from the state is accounted for by the city and 3.84% is accounted for by the county of residence of the causal driver, leaving 89.97% of the variability to be accounted for by driver attributes or other crash contributing factors. Fatal crash rates varied significantly across the state and some crash factors were more pronounced among drivers from particular cities and counties. In view of these findings, specific countermeasures and structural adjustments may be targeted in locations with the highest proportions of risky driver behaviors.