Multinomial probability assessment of motorcycle injury severities

The analysis of motorcycle crashes that occurred in Florida has shown an average increase of 11.9% from 1999 to 2004. During that time period, 48.5% resulted in severe injury while 2.6% were fatal. Various studies have evaluated contributing causes of motorcycle crashes and factors determining injury severity. No specific conclusions have been reached on factors directly connected to injury severity. In view of these statistics and findings from previous researches, this study applies the Multinomial Logit (MNL) and the Multinomial Probit (MNP) models to analyze factors associated with motorcycle crash injury severities. The multinomial models are used instead of the ordered models since the latter are not flexible in quantifying the effect of the independent variables for each injury severity category. The MNL and MNP models yield the similar results in terms of the sign and magnitude of the coefficients. However the standard errors of the coefficients in MNP are tighter than those in MNL, implying that MNP is a more efficient estimator than MNL. It was found that, increase in number of lanes, alcohol and drug use, high posted speed limit, curved areas, turning movements, ramps, and driving with no adequate daylight increase probability of severe injury. Collision with truck and buses and more than two vehicles in the crash also increase the probability of severe injury.