實證方法使用長期追蹤資料固定效果與負二項迴歸固定效果模型，以鄉鎮市區層級的分析結果，降低酒測值門檻與加重刑事責任能顯著降低台灣本島每日酒駕事故率；若個別觀看六都之間的每日酒駕事故率，則修法顯著效果參差不齊，顯示除了全國性修改法規更需要因地制宜訂定與落實酒駕政策。至於每日酒駕死亡與受傷人數，歷次法規修法之影響效果有限，對死傷人數的降幅分別逐次減少或不顯著，但以社會福利角度出發，每年拯救因交通事故傷亡所減少的社會成本絕不容小覷。以個人層級分析整體駕駛之酒駕肇事特性方面，年齡增加與性別為男性愈容易酒駕肇事，機車族群酒駕肇事率多於小客車車種，而歷次降低酒測值門檻與提高刑法罰則皆能顯著減少駕駛酒駕肇事率。整體而言，修法對小客車駕駛的嚇阻效果又大於機車駕駛。 ;The study is to investigate the impact of amending laws targeting drunk driving accidents and compare the effects between motorcyclist and car drivers by using the A1 and A2 road traffic accident investigation report of National Policy Agency from 2003 to 2013. The laws for driving under the influence (DUI) include Article 114 of Road Traffic Safety Regulations, Article 185-3 of Criminal Code of the Republic of China, and Article 37 of Road Traffic Regulations on Administrative Penalties.
The empirical analysis used panel data and negative binomial regression model, fixing the regional effects respectively. As the results shown on township-level, lowering the blood alcohol concentration (BAC) limit and harsher penalties reduced the daily accident rates of drunk driving, but had little effects on six metropolitan cities. It might indicate the DUI policies should be formulated and executed upon local conditions. As for drunk driving casualties, amending DUI laws had a little effects, but reducing the numbers of traffic casualties could save lots of social costs each year. As the results shown on individual-level, age increased, male and motorcyclists were more easily committing drunk driving. Lowering the BAC limit and harsher penalties could reduce drunk driving accident rates in driver’s samples. In general, the deterrence of strict laws was more effective in car drivers than motorcyclists.