Predicting DUI decisions in different legal environments: investigating deterrence with a conjoint experiment

Driving under the influence (DUI) enforcement practices and sanctions contribute differentially to the certainty, swiftness, and severity of punishment, which are the key components of general deterrence theory. This study used a conjoint experiment to understand the decision-making process of potential DUI offenders and tested how variation in enforcement and legal punishment affects drinking and driving decisions. It sought to verify and quantify the unique deterrent effects of certainty, severity, and swiftness and to predict the rates of drinking and driving in different legal environments.

One hundred twenty-one college seniors and graduate students at the University of Maryland participated in the Web-based conjoint experiment. They were randomly assigned to 4 blocks, each of which included 9 hypothetical scenarios composed of different levels of DUI enforcement and penalties. Respondents were asked to state their likelihood of drinking and driving under each scenario, as well as their estimated chance of being caught by the police for DUI.

Intensified enforcement, harsh jail penalty, and immediate long license suspension were found to be the strongest deterrents to drinking and driving. Alternative ways to get home were also important in reducing people's willingness to drive. These factors accounted for most of the attribute effect on the DUI decision, whereas delayed punishment due to judicial processing, fine penalty, and legal blood alcohol concentration (BAC) limit had negligible effects. For the personal characteristics, college seniors and those who had previously driven after drinking were more likely to choose to drink and drive, whereas those who expect a jail penalty for a DUI offense were less likely to drive.

Our research confirmed and quantified certainty of punishment as the greatest deterrent to DUI, but it also indicated the equally important effect of a severe jail penalty. It provides evidence on the feasibility of using a conjoint experiment in future studies to understand the general driver population and, with the help of a simulation tool, to predict DUI decisions in different legal environments. Such predictions can be used to better inform policy decisions on developing targeted general deterrence programs in different communities.