Study
DeGarmo and colleagues (2009) conducted a follow-up study concentrating on long-term outcomes of the fifth graders from the Reid and colleagues 1999 study. Initial 2-year follow-up data indicated almost no substance use among first grade youths, but the data indicated high levels of initiation and growth in substance use among fifth grade youths. Therefore, the focus of the follow-up study was placed on the fifth-grade sample. They collected self-report data of substance use from the students for 7 years until they reached 12th grade, and then analyzed the data for the entire period. Students were included in the final sample if they participated fully in the baseline assessment or joined the study within the first 2 years of follow-up. The total sample was 351 students within 17 different classrooms in 6 randomized schools.
The study looked at substance use initiation and substance use growth over time. Substance use outcome measures included a simple frequency count of any tobacco use (smoking or chewing), any alcohol use (beer, wine, liquor), and any illicit drug use (marijuana, amphetamines, heroin, cocaine). The use of each substance was rated based on responses to the following question: "How many times have you used in the last 6 months?"
Latent variable growth models were used to analyze the data. The authors conducted subgroup analyses on gender.
Study
Reid and colleagues (1999) conducted a population-based, randomized, intervention trial to assess the immediate effectiveness of LIFT. Twelve schools from areas of high juvenile crime were selected at random to participate as either an intervention (n = 6) or control (n = 6) school. These schools were in neighborhoods with higher-than-average rates of juvenile crime from the Eugene/Springfield, Ore. region. All first and fifth grade classrooms within each school (32 classrooms total) participated. Participants consisted of 671 first and fifth graders, their families, and their teachers. Families were recruited by trained recruiters through letters, phone calls, and home visits.
The intervention school participants (n = 382) received LIFT. The control school participants (n = 289) joined in the assessment-only phase of the program. Participants were largely from white, lower- to middle-socioeconomic backgrounds. Participating students were 51 percent female, and 84 percent European American, 5 percent Hispanic, 3 percent American Indian, 3 percent Asian/Pacific Islander, 2 percent African American, and 3 percent “other.” The two groups shared similar backgrounds, though some differences were noted between the groups (the control group had significantly more fathers age 50 or older, more fathers with some graduate education, more mothers with a college education, and more mothers of Asian/Pacific Islander descent). These differences were due to the clustering of demographic characteristics within neighborhoods, such as the inclusion of a graduate student housing complex of a local university.
The intervention took place during the fall. Participants were assessed preceding intervention and again during the spring following intervention. During each assessment, children, parents, and teachers were interviewed and completed a variety of paper-and-pencil questionnaires. Additionally, school and court records were collected, children were observed in the classroom and on the playground, and parents and children were observed during family problem-solving discussions at home or at the research center.
The hypothesis of the study was that the LIFT intervention would have a significant impact on three measures: child physical aggression on the playground, mother aversive behavior during mother-child interactions, and teacher ratings of child positive behaviors with peers during the year following the program. Data was collected from multiple reporting agents (teachers, parents, child, and assessment staff) using multiple methods (observation, questionnaire, and interview) across multiple settings (school, home, and laboratory). The Interpersonal Process Code (IPC) was used to index rates of maternal aversive behavior during mother-child interactions. The IPC was also used to measure rates of child physical aggression towards peers on the playground during recess. The Walker–McConnell Scale of Social Competence and School Adjustment was used to measure social skills of students. Teachers rated students on the Peer-Preferred Social Behavior subscale, which included items such as “shares laughter with peers” and “makes friends easily with other children.”
The analyses involved examination of univariate distributions of the preintervention, postintervention, and follow-up variables. Variables that deviated significantly from normality were transformed using log or square root transformations. Change scores were also calculated (i.e. postintervention minus preintervention, or follow-up minus preintervention). The change score of each outcome variable was regressed on group, grade, sex, and the initial status on that variable, as well as all possible interactions. Random regression analyses included school as a clustering variable and accounted for group, grade, sex, and possible interactions among variables. The authors conducted subgroup analyses on gender and grade level.
