本研究是針對臺灣地區30日內(A1類與A2類)汽車(含小客、貨車)交通事故進行探討。我們使用警政署民國九十七年道路交通事故之統計資料，並應用卡方獨立性檢定與Cramér's V關聯係數找出和傷亡程度有關聯的變數。由於所選出變數過多，故採用卡方互動自動檢視法選出重要的解釋變數。然後再加入重要解釋變數與其兩兩交互作用項建立比例勝算模型。由配適後最佳比例勝算模型我們可得知影響30日內事故傷亡的變數有：保護裝備、飲酒情形、日夜、速限、事故類型、速限與事故類型之交互作用。 This thesis studies the problem for the automobile traffic accident within 30 days in Taiwan area. First, we analysis the data of the Road Traffic Accident from National Police Administration at Ministryof the Interior in 2008, by the chi-square independent test and Cramér's V association coefficient to find some correlated explanatory variables. Because many variables are selected, we use the chi-square automatic interaction detection(CHAID) to select more important explanatory variables. Then we add the first-order interactions among those important explanatory variables to establish proportion odds model. From the fitted proportion odds model, we know the effects for the injury and death of accident within 30 days include the protection equipment, the status of drinking wine, exceeding the limit of driving speed, the accident pattern, the interaction of exceeding the limit of driving speed and the accident pattern.
