A common method of normalizing crash fatality data for comparing subgroups of drivers has been the estimated vehicle miles traveled (VMT). Unfortunately, the VMT method fails to provide for exposure to risks such as those related to alcohol consumption (among others). Recently, the "crash incidence ratio" (CIR) has been introduced to address some of these limitations. The goals of this study are first, to show that the CIR method is intrinsically similar to an increasingly popular quasi-induced method: the relative accident involvement ratio (RAIR); second, to compare the VMT-based, the CIR, and the RAIR methods when applied to the evaluation of alcohol-related crash fatalities across racial/ethnic groups. We use the 1990-1996 Fatal Accident Reporting System (FARS) with information on the drivers' race/ethnicity and alcohol involvement (BAC). Descriptive and statistical ratio tests were applied. The RAIR and CIR are indeed closely related measures that, when used for comparisons against a reference group, yield exactly the same numerical estimates. Strikingly different outcomes were obtained depending on using the VMT or the CIR/RAIR. Choosing one measure over another should depend on the questions to be answered. The implication of this finding for researchers and policy makers is discussed. (Author/publisher).
Samenvatting