S. Dissanayake, J.J. Lu, U. Gamgamuwa, M.S. Shaheed

Pages: 95-106

Abstract
This research was conducted to identify critical factors that are important in increasing trends on motorcycle rider fatalities in the United States, with particular attention on the helmet law. In addition to fatality crash data, demographic, policy, weather, and traffic related data were collected for each of the fifty states and the Washington DC from 2012 to 2014. Modeling was carried out by considering motorcycle rider fatalities per 10,000 registered motorcycles and motorcycle rider fatalities per 100,000 population as response variables. Dataset was divided into two subsets of two-thirds and one-third to be used in model development and model validation. Generalized linear regression models were used to develop the two models and maximum likelihood method was used to estimate the regression parameters. Mean of the residuals and mean squared error (MSE) were calculated using model validation dataset to validate the developed models. Results of the model developed for motorcycle rider fatalities per 10,000 registered motorcycles shows that having mandatory helmet law in the state reduced the fatality risk by 12%. The second model developed for predicting motorcycle rider fatalities per 100,000 population showed that having mandatory helmet law in the state reduced the motorcycle fatality risk by 18%. Results of model validation showed that both developed models have MSE which are close to the model MSE and the mean of the residuals are close to zero which indicated that the developed models have a good fit. Finally, quantile-quantile plots and the studentized residual plots were developed for both models to check whether the basic assumptions of multiple linear regression are satisfied. According to the results obtained from both models, it can be seen that among other critical factors, having mandatory helmet law will have a considerable effect on reducing motorcycle rider fatalities.
Keywords: helmet law; motorcycle rider fatality; motorcycle rider safety; generalized linear models