ABSTRACT
We tested the generalizability of a science-based community prevention design to reduce DUI crashes. Previous researcher-led studies have confirmed the effects of an intervention design of visible enforcement coupled with heightened public awareness of enforcement to increase driver perception of likely detection for drinking and driving. A community coalition based the project on a prevention intervention model that included two key intermediate variables: levels of visible enforcement and of public awareness of enforcement. We evaluated the project using community-specific monthly time-series measures of DUI crashes and state level trends in DUI crashes, indicators of enforcement, and public attention to enforcement. We devised the evaluation design to determine if an observed trend in DUI crashes declined and to verify if key intermediate variables increased, as stimulated by local efforts. DUI crash analysis documented an upward trend during a pre-trial period from July 2010-December 2011, which matched the upward trend in state DUI crashes. After the local intervention began in January 2012, local DUI crashes began a clear downward trend (average 2013 crashes were 23% lower than in 2012 and a 5-month post-intervention average from 2013 was lower than the equivalent 5-month pre-intervention average). This contrasted with the continued upward state DUI crash trend, with a 2-year increase of 16%. The downward trend in local crashes was associated with an increase in DUI enforcement as well as news stories concerning DUI enforcement that were stimulated by the efforts of the community prevention project. These results confirm the generalizability of two previous community research trials that were conducted with limited or no research resources or leadership. We discuss the importance of controlling for external factors in attributing causation in a local prevention evaluation by confirming both sufficient local prevention efforts and a decline in DUI crashes.
