Background: We evaluated the effects of driving under the influence (DUI) mandatory preconviction and postconviction drivers' license suspension laws in each of 46 U.S. states using 1 to 2 decades of long-term follow-up data on fatal car crashes. State-specific results were combined using meta-analytic techniques, and provide a direct test of the concept of celerity--time between offending behavior and consequent punishment--from deterrence theory.
Methods: Four key outcome measures of monthly fatal alcohol-related crash involvement were examined using data from January 1976 through December 2002: single-vehicle nighttime, breath or blood alcohol concentration (BAC)=0.01 to 0.07, BAC=0.08 to 0.14, and BAC>or=0.15 g/dL. Missing BAC test data for some individual cases were filled using multiple imputation methods, and consequent increases in standard errors were incorporated into subsequent analyses. Separate ARIMA models were estimated for each state, including controls for state-specific levels of crash involvement over time due to other factors and effects of other major DUI countermeasures. Estimates were pooled across states using inverse variance weighting methods.
Results: Administrative or preconviction drivers license suspension policies have statistically significant and substantively important effects in reducing alcohol-related fatal crash involvement by 5%, representing at least 800 lives saved per year in the United States. Moreover, these laws have similar effects on drivers at all drinking levels--from lower-risk drivers below the legal alcohol limit to drivers at extreme levels of intoxication. In clear contrast, postconviction license suspension policies have no discernable effects.
Conclusions: The effectiveness of a deterrence policy appears to be more strongly affected by celerity-the speed by which punishment is applied after the offending behavior--than by the high severity of the penalty. This finding could be fruitfully applied to other areas of alcohol control policy and laws and regulations in general.