Using Logistic Regression For Estimating The Influence of Some Accident Factors on Severity

This study applied logistic regression to accident-related data collected from traffic-police records in order to examine the contributing factors to accident severity. A total of 560 subjects involved in severe accidents was sampled. The accident severity (dependent variable) in this study is a dichotomous variable with two categories, Fatal or Injury. Therefore, each of the subjects sampled is classified as a fatal accident or an injury accident. Due to the binary nature of this dependent variable -accident severitya logistic regression approach is suitable. Among nine variables obtained from police-accident reports, tw independent variables found most significantly associated to accident severity; namely, location and cause of accident. This paper gives a statistical interpretation of the modeldeveloped estimates in terms of odds ratio concept. The findings show that the logistic regression used in this research is a powerful tool in providing meaningful interpretations that can be used in future safety improvements in Riyadh. INTRODUCTION Accident severity is of special concern in traffic safety, as many efforts address accidents tend to be measures not only to prevent accidents but also to reduce the severity of accident. One way to do so is to identify the most probable contributing factors that affect accident severity. This study aims at examining not all factors, but some believed to have a higher potential for serious injury or death, such as accident location, type, and time; collision type; age and, nationality of driver at fault, and his licensing status; and vehicle type. The reason for not examining more factors was due to substantial limitations of data obtained from accident reports. Logistic regression was used in this study to estimate the effect of the statically significant factors on severity. Logistic regression and other related-categorical-data regression have often been used to assess risk factors for various diseases. However, it has been also used in transportation studies. Following is a brief literature review for the use of this type of regression in traffic safety. Regression methods have been become an integral component of any data analysis concerned with describing the relationship between a response variable and one or more explanatory variables. The most common regression method is conventional regression analysis (CRA), either linear or nonlinear when the response variable is continuous (iid). However, it is often the case that the outcome variable (response) is discrete. The conventional regression analysis is not appropriate. Among several reasons, the following two are the most significant: 1. The response variable in CRA must be continuous. 2. The response variable in CRA can take non-negative values. These two primary assumptions are not satisfied when the response variable is categorical. Jovanis and Chang (1986) found a number of problems with the use of linear regression in their study applying Poisson regression as a means to predict accidents. For example, they discovered that as vehicle-kilometers traveled increases, so does the variance of the accident frequency. Thus, this analysis violates the homoscedasticity assumption of linear regression. In a well-summarized review of models predicting accident frequency, Milton and Mannering (1997) state that “the use of linear regression models is inappropriate for making probabilistic statements about the occurrences of vehicle accidents on the road”. They showed that the negative binomial regression is a powerful predictive tool and one that should be increasingly