Methods: 160 college seniors and graduate students from the University of Maryland participated in a Web-based conjoint exercise, which is a widely used survey technique in marketing and transportation research. The conjoint exercise was composed of hypothetical scenarios with different types of legal environments for drinking and driving (i.e., various levels of enforcement, jail, fine, license suspension penalties, etc.). Respondents were asked about their likelihood of driving after drinking under each scenario. Mixed logit modeling (a special type of logistic regression) was employed, which accounts for both sources of impact on the driving decision, i.e., individual characteristics such as demographics, past drinking behaviors, etc., and the levels of legal sanctions presented.
Results: It was found that the highest level of enforcement (i.e., with a sobriety checkpoint and special patrols) and of jail penalty (i.e., 7 days or more in jail) would significantly reduce the likelihood of drinking and driving of college students (p<.05), whereas fine penalty and lowering the legal BAC limit would almost have no effect. On the other hand, providing alternative ways to get home other than driving also plays an important role in preventing students from drinking and driving.
Conclusions: These findings suggest that programs aimed at increasing the certainty of punishment such as increasing the enforcement level, are the most effective in preventing college students from drinking and driving, followed by harsh jail penalties based on increasing the severity of punishment. This exploratory study provides evidence on the feasibility of using conjoint exercise to understand drinking and driving decision making, and can be extended to evaluate the deterrence-based prevention efforts of alcohol-impaired driving among the general public.