Principal Investigator: John Reid
Level of Intervention: Universal
Target Population: First and fifth grade elementary children and their families living in high-risk neighborhoods
References: Reid, Eddy, Fetrow & Stoolmiller, in press
Theory (Risk & Protective Factors Targeted):
LIFT is based on the developmental model of conduct problems, recognizing the coercive impact of social agents in the family and peer domains on children at risk for conduct problems. Based on this model, LIFT attempts to lower the probability of oppositional/antisocial behavior within the school, peer, and home domains, lower the probability that members within each domain would retaliate coercively to such behaviors, and to increase the probability that prosocial behaviors would be supported. The overarching focus of LIFT is to modify the reactions of members in each domain to childrens prosocial and antisocial behaviors. Specifically, in the family domain, LIFT promotes calm and consistent limit setting and parental involvement in the childs social domains (especially school). In the peer domain, LIFT targets physical aggression in unstructured settings (i.e. the playground). In the classroom, LIFT promotes developmentally appropriate social skills.
Description of Intervention:
LIFT is a 10 week intervention consisting of parent training, a classroom based social skills program, a playground behavioral program, and systematic communication between teachers and parents.
The school component consists of twenty, 1-hour sessions provided over a 10 week period which include four parts: (1) developmentally appropriate (for 1st or 5th grade students) classroom instruction on social and problem-solving skills, (2) opportunities to practice these skills in large and small group settings, (3) free play in the context of a group cooperation game, adapted from the Good Behavior Game (Dolan, et al., 1993; also see elsewhere in this report), and (4) skills review and presentation or rewards.
The school-parent communication component consists of a telephone and answering machine for each classroom on which teachers leave daily messages about class activities, homework assignments and special events. Parents can call any time to learn about activities or assignments or leave messages regarding their child. A weekly newsletter keeps parents informed and provides suggestions for home activities that compliment those at school.
Finally, a parent intervention focuses on teaches parents to foster a home environment marked by consistent and effective discipline practice and close and appropriate supervision. Parents meet in groups of 10 to 15 families once a week for six weeks to participate in topical presentations, view videotaped scenarios to illustrate new skills, engage in role plays, and receive supplemental reading activities and home practice activities. Parent sessions are held in the school (to foster parent-school connectedness) and are offered each weekday evening and one weekday afternoon to accommodate families schedules. To further encourage attendance, free child care is provided and prize drawings are held. Whenever a family cannot attend the weekly group, a LIFT staff members attempts to visit the home to review the same material. As a last resort, if a home visit cannot be arranged, a home packet of materials covering session content is delivered to the home.
The study was conducted with 671 first and fifth grade students and their families from 32 classrooms in 12 elementary schools in an urban area of the pacific northwest United States. Schools were selected from a catchment area with higher than average juvenile arrest rates. Schools in the study had an average subsidized or free lunch rate of 49.6%, and 2-3 classrooms per assigned grade (first or fifth) with an average of 21 students per classroom.
The 671 students in the final sample represented 85% of the eligible students in the selected classrooms and consisted of 51% female and 11% minority students. Participants were predominantly from lower to middle SES, with parents having complete high school or attended some college.
The study design was a randomized clinical trial with randomization taking place in each of three successive intervention years. Of the eligible schools, two were randomly selected to the treatment group, two to the control group, and two as alternates. One school from each group was then randomly assigned as a first grade school or a fifth grade school. In subsequent years, all schools that did not participate as intervention or control schools in previous years were eligible to particpate.
After randomization, treatment and control groups were generally well matched on socioeconomic and demographic characteristics. One significant difference between the groups was the mothers ethnicity, with fewer control group mothers being white (although the difference was not significant for fathers or childrens ethnicity).
Data were collected during the fall (pre-intervention) and spring (post intervention) of each year, with followup data collected in the winter of the following year. A reduced assessment battery was used for followup. Family/child assessment included child behavior problems, academic skills, peer relations, and family management skills (monitoring/supervision, discipline, problem solving). Within each domain, multiple data sources were used, including parent, teacher, and child reports and observations. Behavior in the home was assessed by parents and children visiting the researchers center for two hours to participate in an interview, complete a questionnaire about parenting practices, child behavior, parental involvement, and child/peer relationship, and to participate in a laboratory task. Parents also completed four brief telephone interviews to assess child and parenting behavior.
School behavior was assessed through a teacher interview and teacher questionnaires which provided information about the academic and social adjustment of each child. Peer nominations were collected from classmates and multiple observations were conducted on the playground for each student by professional, blinded observers. Each student was observed during recess for 10 minutes on three separate days. Finally, end of year academic and discipline data was collected from school administrative records.
Primary analysis of the effectiveness of the LIFT program was conducted based on the Interpersonal Process Code (IPC; Rusty, Estes & Dishion, 1991) and the Peer Preferred Social Behavior subscale of the Walker-McConnell Scale of Social Competence and School Adjustment (Walker & McConnell, 1995). The IPC indexes rates of maternal aversive behavior as well as rates of child physical aggression towards peers, and consists of concurrently coded variables related to behavior, context, and affect.
To examine the immediate impacts of the program, univariate distributions of the pre-, post-, and followup variables were first assessed. Following any necessary transformation of variables that deviated significantly from the norm, change scores were calculated, with multiple imputation procedures used to reduce bias due to missing data. Based on an a priori assumption that program effects would be related to pre-intervention levels on the relevant measures, change and initial-status were z-scored to standardize the parameter estimates in the final regressions. Change scores for each outcome variable were then regressed on group, gender, grade, and initial status, as well as possible interactions. Random regression by school was used since participants were clustered by school. Given the hypothesis that the impact of the intervention would be at the middle of the distribution of the initial status of each antecedent variable, the standardized regression weights reflect tests within the middle range of the initial status.
For each analysis, an effect was found for group in the predicted direction, with the effect interacting with pre-intervention scores for two of the three variables. For teacher ratings of peer preferred behavior, there was only a main effect, with the intervention students social skills viewed more favorably by their teachers the following year than control students. For child physical aggression, there was a significant difference between intervention and control groups for both first and fifth grade students at the pre-intervention mean. For first graders, this effect increased as pre-intervention scores increased, and for fifth graders the effect remained the same across pre-intervention scores. Effect sizes (Cohens d) across these variables ranged from .12 to .57.
The authors also conducted descriptive analyses of implementation fidelity (including program delivery and integrity of control condition), participant utilization of all components, consumer satisfaction, and impact of attrition. As rated by teachers, interventionists, and independent observers, the program material in both the classroom setting and the parent groups was covered quite thoroughly (all ratings between 91% and 97%).
Participation in the parent group meetings was somewhat problematic (as is common with parent interventions). An average group session was attended by 59% of the families, with 23% receiving materials through the mail, 13% receiving a home visit, and 5% not participating. For both the first and fifth grade groups, attendance at the parent group sessions decreased slightly across the intervention year. 93% of all families received all of the intervention materials in some manner, with 53% receiving all materials face-to-face (either in group sessions or through home visits).
The LIFT line was utilized by at least 78% of the families (based on calls where the families identified themselves) with families averaging 11 calls each. A total of 8128 calls were made to the line. In terms of utilization by students, 90% of the children attended 17 or more of the 20 classroom sessions (including the Good Behavior Game). On average, a child attended 18 of the 20 sessions.
Missing data at post intervention due to attrition ranged from a low of 6% for peer nominations and recess observations to 12% for teacher ratings (which were collected at one year followup). It is unclear how this may have biased outcomes. No data was reported on observation measures or peer sociometrics.
Strengths & Limitations:
LIFT is a comprehensive, multicomponent program that targets multiple domains to reduce the antecedents for conduct disorder in the individual child, school and classroom, peer group, and home. The study described is a randomized clinical trial which addressed not only the effectiveness of the intervention, issues of fidelity, and utilization but also tested for differential effects as a function of pre-intervention levels of the antecedents in question. The data used for the analyses were derived from multiple sources within multiple contexts or domains using measurements with moderate to high reliability. The sample size and demographics of the sample population in this study limit to a minor degree the generalizability of the findings. Measures of fidelity taken from multiple sources indicate the program was implemented thoroughly and utilized to a great extent by the sample population. It is unclear how sample bias or attrition may have affected the outcomes.