Abstract
In recent years, spurred by possible regulatory action in the US, researchers have been applied increasingly sophisticated methods in their efforts to understand the relative importance of factors that can contribute to vehicle rollover. Prior statistical work is extended by applying path analysis to police-reported data on crashes involving large sets of diverse passenger vehicles. Results of this study help elucidate causal relationships among the many pre-crash and at-crash factors associated with vehicle rollover as an outcome of motor vehicle collisions. In this context, how and where people drive appear much more important than characteristics of vehicles they operate. (A)