摘要：AbstractRoad traffic injuries are one of the leading causes of deaths around the world. Due to lack of available resources and due to unplanned growth, predominant portion of these fatalities occur in middle-income and low-income countries. For the improvement of traffic safety, it is important to understand various characteristics of traffic risks both temporally and spatially. Lack of detailed data in developing countries is a serious challenge in addressing traffic safety. In India, number of traffic related fatalities is the only reliable traffic safety related data available. Multiple regression models, which are generally used in these countries analyze the fatality data, do not consider the effect of omitted variables on the dependent variable. This study is an attempt to develop a model based on readily available data for India to estimate traffic safety risk while considering the individual specific effect of various regions. Panel data analysis is commonly being used for various studies due to the advantages of panel data. Panel data can accommodate data with more information, more variability, less collinearity among variables, more efficiency and more degree of freedom. Using cross sectional time series panel data for 28 states of India over the period of 2004-08, panel model is developed on the number of fatalities. Data on total number of deaths due to road crashes, road infrastructure, population, area and vehicle registration are used in this study. Most of the data are retrieved from reliable sources such as publications of National Crime Records Bureau (NCRB), Ministry of Road Transport and Highways (MORTH), Census Bureau of India, and Ministry of statistics and programme implementation (MOSPI). The coefficients for the variables included into the model are assumed to be fixed based on the analysis of covariance test. This assumption is unrealistic. However, with the availability of data for more years, the variation of the coefficients of variables can be taken into account.
