「酒後駕車行為決策之研究」主要目的在於建立一個實證模型，找出並判定影響個人酒後駕車決策之因素，並驗證Viscusi 的「貝式學習過程(Bayesian learning process) 」是否成立；即實證理性個人是否能利用其環境資訊，修正其先驗信念 (prior belief) ，以形成其事後認知 (posterior perception)， 進而影響其個人之行為或決策。文中採用遞迴的聯立方程計量模型，利用民國90 年10 月間針對臺北縣、市地區、18 歲以上具有汽機車駕照之受訪者進行電話訪問的原始資料，作為本文實證分析之依據。我們的實證結果顯示，Viscusi 的「貝式學習過程」的確存在，而且個人酒後駕車與否，亦受事後的兩個主觀認知─即警察取締酒後駕車的執法強度與酒後駕車發生意外事故風險機率−−的影響。

This paper attempts to build an empirical model to find and judge the factors that affect a rational person’s decision making of drinking driving. To verify the Bayesian learning process of Visusi is true or not, we would like to prove a rational person whether modify their prior belief by using the environmental information to construct their posterior perception and affect their behavior and decision making. This paper adopts the recursive model and uses the telephone survey data of Taipei County and Taipei City during October of 2001. The results indicate that the Bayesian learning process of Viscusi indeed exists. Furthermore, personal behavior of drinking driving is indeed affected by two posterior subjective perceptions, that is, the enforcement perception and the accident risk perception.