Effectiveness of safety belts and Hierarchical Bayesian analysis of their relative use

Abstract

Seat belts are effective devices for both preventing serious injury and reducing the likelihood of a fatality. Police crash data made available from the Emirate of Dubai in the United Arab Emirates are utilized to evaluate the effectiveness of safety seat belts and to propose a methodology to assess levels of the device use based on demographic factors. Analysis is based on rates and standardized rates. However, rates based on data disaggregated to small levels tend to be unreliable due extreme values resulting from small counts. A Hierarchical Bayesian approach is employed to smooth the standardized safety belt use rates (SBR). Extreme SBRs are shrunken towards an overall mean. Results of the analysis indicate that restrained occupants are less likely to incur fatal or serious injuries compared to the unrestrained. There is no significant difference between ethnic groups (local/expatriate) living in Dubai in terms of level of compliance with belt use.