A BIVARIATE ORDERED PROBIT MODEL WITH TRUNCATION - HELMET USE AND MOTORCYCLE INJURIES

I consider an ordered probit model in which some of the observations i n one of the categories are missing. The model can be estimated becaus e of a second variable with a similar truncation scheme. The second va riable has the effect of increasing the number of categories quadratic ally while only increasing the number of parameters linearly. This mod el arises in the analysis of the effectiveness of helmets in reducing the severity of head and neck injuries in motorcycle accidents. The pr oblem is that accidents in which the rider did not receive a head or n eck injury were observed only if the rider received a body injury, and vice versa. The results show that helmets are effective in reducing t he severity of the worst head or neck injury but, because the worst ov erall injury that a rider receives is often a body injury, their effec t on the severity of the worst overall injury is smaller.