The Market for Safety Regulation and the Effect of Regulation on Fatalities: The Case of Motorcycle Helmet Laws
指出，现有关于摩托车头盔法效果的计量研究因忽略法律的内生性而可能产生偏差。通过自选择模型修正后，发现若法律随机分配，有头盔法的州死亡率平均仅比无法的州低不到1%。
Existing econometric studies of the efficacy of motorcycle helmet laws assume that such laws are exogenously determined and may therefore yield biased results. To correct for possible selectivity bias, we specify and estimate a self-selection model in which the choice of having a helmet law is endogenous. Our results indicate selectivity bias exists and that if laws were randomly assigned, fatality rates for states with helmet laws would on average be less than 1% lower than for states without laws.