Hawkins and colleagues (2008) started The Community Youth Development Study (CYDS) in 2003 as the first community-randomized trial of the Communities That Care (CTC) system. Twenty-four communities across seven states in the United States were selected for this five-year experimental study. The communities selected to be in the study were asked to target their interventions on children ages 10 to 14 and their families. Interventions available for selection and implementation could be found in the “Communities That Care Prevention Strategies Guide.” Each intervention within this guide has demonstrated positive effects in rigorous evaluations with middle school–aged children, and contains training, materials, and assistance.
The 12 communities selected 13 different prevention programs to implement during the 2004–05 school year, and 16 programs for the 2005–06 school year. The following programs were selected and implemented: All-Stars, Life Skills Training, Lion’s Quest Skills for Adolescence, Program Development Evaluation Training, Participate and Learn Skills, Big Brothers/Big Sisters, Stay Smart, Strengthening Families 10-14, Guiding Good Choices, Parents Who Care, and Family Matters. Close to half of these programs were selected by multiple communities and offered multiple times a year. For example, Guiding Good Choices was administered 38 times across six communities.
Communities were drawn from a larger pool, and within the seven states, matched pairs were created. Matching was done on key characteristics, such as total population, poverty, racial composition, unemployment, and crime. Twelve matched pairs of communities were then randomly assigned to receive the CTC system or the control condition, where one community was the treatment and its counterpart, the control. From these communities, a panel of students was selected to be followed from fifth grade to eighth grade.
After obtaining consent, the resulting sample of adolescents was 4,404. This sample was evenly split between male and female students. The majority of participants were white (70 percent), followed by Hispanic (20 percent), Native American (nine percent), and African American (four percent). Just more than half of the sample (55 percent) came from treatment communities; the remainder comprised the control communities.
The Youth Development Survey was used to gather data on risk factors, substance use, delinquency, and basic demographics. This is a self-report survey students fill out to disclose their own behavior. The three outcomes examined in this study were risk factors, substance use, and delinquency. Risk factors were measured by a scale of composite items. Delinquent behavior was measured as the first occurrence, in the past year, of any of nine behaviors. Questions included “How many times in the past year did you attack someone with the idea of seriously hurting them? Bring a handgun to school? Sold illegal drugs?” Onset of substance abuse was measured as the first reported lifetime use of alcohol, marijuana, cigarettes, or other illegal drugs.
Analysis of covariance (or ANCOVA) was used to determine differences in average levels of risk factors between CTC and control communities. The general linear mixed model accounted for the hierarchical data structure of students within communities, communities within matched pairs of communities, and the variability of interventions across communities. A multilevel discrete-time survival analysis was used to assess the effects of the CTC system on preventing the initiation of substance use and delinquent behavior between grades Five and Seven. Students who had not used alcohol, cigarettes, marijuana, or inhalants, or engaged in delinquent behavior were examined to determine when, if ever, they would engage in these behaviors. Each wave of data was analyzed, and students that initiated in any of the measured behaviors were excluded.
Hawkins and colleagues (2009) used a subset of the CYDS to examine the effects of the Communities That Care system on alcohol, tobacco, and other drug use for a panel of students followed from grade Five through grade Eight. This study period was four years after the implementation of CTC and almost three years after CTC communities started implementing proven prevention programs.
Recruitment of the student panel started in the fall of 2003. From the 24 communities in CYDS, 88 schools agreed to participate. All fifth-grade classrooms in these schools were eligible for participation. After obtaining consent from parents, the resulting longitudinal panel was 4,407 students (2,194 girls and 2,213 boys). A little more than half of the panel (55 percent) came from intervention communities. Just as with the CYDS communities, treatment and control students did not differ significantly in demographics or socioeconomic indicators.
Data on drug use and delinquent behavior was collected with annual surveys of students. The first wave of data collection in spring of 2004 was a preintervention baseline measure. Scientific-based prevention programs started in schools and communities the following fall of 2004. The last data collection was in the spring of 2007, when students should have advanced to the eighth grade. Accounting for attrition and additional recruitment procedures, 96.7 percent of the sample completed at least three out of the four waves of data collection.
A multilevel discrete-time survival analysis was used to determine the effect of the CTC intervention in preventing the incidence of drug use and delinquency between grades Five and Eight. Students who had not used alcohol, cigarettes, marijuana, or inhalants, or engaged in delinquent behavior were examined to determine when, if ever, they would engage in these behaviors. Each wave of data was analyzed, and students that initiated any of the measured behaviors were excluded. A mixed-model analysis of covariance was used to determine the effect of the CTC intervention on reducing the prevalence in the eighth grade of heavy drinking within the past two weeks, drug use in the last month, and delinquency in the last year.