Study 1
Kilmer and colleagues (2013) studied the effects of South Dakota’s 24/7 Sobriety Project by comparing changes in public health outcomes between 2001 and 2010 in counties that implemented the program with control counties that did not implement the program. Data from the South Dakota Attorney General’s Office was used to determine when the program was implemented in counties within the state. The database included information on participant-level data (such as demographic characteristics, county of residence, and dates of participation) as well as information about every test (such as date and time of each test and the result) for all individuals who participated in the program since it began in 2005.
Between 2005 and 2010, 63 percent of new program participants entered for a DUI offense, 6 percent entered for community corrections violations, 5 percent for domestic violence, 5 percent for assault, 5 percent for drug possession, and 17 percent for other offenses. Half of new program participants were aged 18 to 40 years when they entered the program. The majority of participants in the program during this time were white. No information was provided on other races/ethnicities or gender.
The program was considered operational in a county once the number of county residents participating in the program for a given month equaled or exceeded a quarter of the number of DUI arrests in the county. DUI arrests were defined as the county’s moving monthly average during the previous year, to address any seasonality. In 2005, five counties were piloting the program. By the end of 2006, 19 counties were administering tests for the program.
The primary outcomes of interest included arrests for driving while under the influence of alcohol (DUI), arrests for domestic violence, and traffic crashes. Data on DUI and domestic violence arrests from 2001 to 2010 came from the South Dakota Department of Criminal Investigation. Arrests for DUI were examined separately for people arrest for the first time and people repeatedly arrested for DUI. Data on traffic crashes reported to the police from 2004 to 2010 came from the South Dakota Office of Highway Safety.
A number of time-varying county characteristics were controlled for in the analyses, including unemployment rate, snowfall, per capita police officers, per capita on- and off-premises alcohol outlets, an indicator variable for the Sturgis Motorcycle Rally, and an indicator variable for whether college was in session (for the four counties with substantial student populations). Because the outcome measures involved count data and included zeroes, Poisson regression was used to analyze the data.