PREDICTING NONAGGRESSIVE NEGATIVE LIFE OUTCOMES FROM EARLY AGGRESSIVE BEHAVIOR TRAJECTORIES: DIFFERENTIAL MULTIFINALITY AMONG GIRLS AND BOYS. Cindy Schaeffer1, Hanno Petras2, Nicholas S. Ialongo2, Jennifer Matvya1, Jeanne Poduska3, Sheppard Kellam2, 1University of Maryland Baltimore County, Baltimore, MD United States; 2Johns Hopkins University, Baltimore, MD United States; 3American Institutes for Research, Washington, DC United States
Several longitudinal studies have identified pathways of aggressive behavior development across childhood and have linked high risk pathways to later antisocial behavior (e.g., Broidy et al., 2003; Nagin & Tremblay, 1999; Schaeffer et al., 2003). However, the extent to which different aggressive behavior trajectories place children at risk for other negative life outcomes has remained largely unexplored. Chronic aggressive behavior in childhood interferes with the successful completion of many developmental tasks, such as complying with the academic demands of school and establishing positive peer relationships. Thus, it seems likely that youth with patterns of chronic aggression during childhood would be at higher risk for negative young adult outcomes, such as dropping out of high school or engaging in early sexual behavior, than youth with nonaggessive trajectories. Because early aggression trajectories are less predictive of later antisocial behavior for girls than for boys (Schaeffer et al., 2004), it is particularly important to understand multifinality in outcomes of aggressive behavior trajectories for girls. The present study used multiple groups analysis within a general growth mixture modeling (GGMM) framework to explore the risk of early aggression trajectories on 5 negative outcomes in young adulthood. Participants included 664 girls and 675 boys in the control condition of an evaluation of two school-based universal preventive interventions. Participants were assessed annually in 1st-5th grades, and a large portion (76%) participated in a follow-up interview regarding life outcomes at age 19-20. The software package Mplus Version 3.0 was used in all analyses. Three aggression trajectory classes were derived for girls and boys in previous research with this sample. In the present study, girls with chronic high aggression (CHA) were more likely than girls with stable low aggression (SLA) to drop out of high school and to be unemployed. Boys with increasing aggression (IA) over time were more likely to drop out of high school, engage in early sexual intercourse, and to be unemployed than boys with an SLA trajectory. In a comparison of girls and boys in analogous trajectory classes, CHA girls were more likely to experience a teenage pregnancy than CHA boys. Boys with IA were more likely to drop out of school, engage in early intercourse, and have an arrest for a nonviolent crime than girls with moderate aggression (MA); MA girls were more likely than IA boys to have a teenage pregnancy.
A POPULATION-BASED ANALYSIS OF ANTECEDENTS AND CONSEQUENCES OF EMPIRICALLY DERIVED JUVENILE CRIMINAL TRAJECTORY PROFILES. Frauke Kreuter1, Bengt Muthen2, 1University of Maryland College Park, College Park, MD United States; 2University of California, Los Angeles, Los Angeles, CA United States
Taking a longitudinal approach in studying juvenile criminal behavior enables the study of empirically-based profiles of criminal behavior defined by behavior patterns over time that can characterize the development, persistence, and desistence of such behavior across several developmental stages. There are challenges, however, when attempting the application of growth modeling technique to an epidemiologic sample where the prevalence of the non-normative behavior, such as juvenile arrest, is low at any given age. Understanding the prevalence of different juvenile crime profiles requires the use of a modeling technique that allows the identification of the less prevalent but more problematic profiles of criminal activity. Once these subgroups of non-normative juvenile offenders are better understood, further analysis can be done to investigate the predictive power of various covariates related to membership in one of the problematic profile groups and the consequences of such membership for adult criminal behavior. Using a well-known data set—Farrington & West, 1990—this talk will evaluate new extensions to Growth Mixture Modeling introduced by Asparouhov and Muthén (2004). These models consider individual differences in development over time with outcomes whose distributions are inherently non-normal, e.g., count data such as the number of juvenile arrests for each subject at each age. For this data, we focus on the application of different versions of zero-inflated Poisson (ZIP) mixture models with and without random effects. The ZIP aspect of the models addresses the relevant question: "Is an absence of an arrest record for a given age period equivalent to an absence of criminal behavior in that period?" The ZIP allows for two latent classes of individuals: 1) those who during a given time period are not engaged in the behavior at all and 2) those who are engaged in the behavior but happen to have zero outcome at the time of measurement. These models also allow flexibility in testing hypotheses related to covariate influences on the trajectory classes. For example, specifying direct effects from covariates to the outcome counts permits testing whether variables such as up-bringing and parental monitoring have diminishing effects on the rate of criminal behavior over time. Other extensions to the model will be discussed along with implications for future research.
PREDICTIVE RELATIONS BETWEEN TEMPERAMENT TYPES IN EARLY CHILDHOOD AND BEHAVIOR PROBLEMS PATHWAYS THROUGH ADOLESCENCE. Julien Morizot1, 1University of California, Los Angeles, Los Angeles, CA United States
In this paper, using data from the National Longitudinal Study of Youth (NLSY79), I demonstrate how different temperament types identified during the first years of life can be used to predict behavior problems pathways during the elementary and high school years. In the first stage, a temperament typology is identified using latent profile modeling on a series of mother-rated temperament traits of children aged 1-2 years. A number of theoretically meaningful covariates associated to the family background (family structure, nature and quality of home environment), to the mother (age at child´s birth, SES, cognitive abilities, postnatal healthcare, psychotropic substance use, parental attitudes), and to the children (cognitive abilities, motor and social development) are also considered in the models. In the second stage, a behavior problems pathways typology is identified using multiple-process latent growth mixture modeling on externalizing and internalizing behavior problems scales rated by mothers from age 5 to 14 years. Finally, in the third stage, using a generalized latent variable mixture model linking the models identified in the two previous stages, the children´s developmental pathways are predicted by the temperament types. This generalized model showed in a probabilistic manner how early childhood temperament types are differentially related to the unfolding of behavior problems pathways through adolescence. Theoretically, this study provides support to psychological theories postulating the fundamental role of early temperament in the causal chain leading to the development of adjustment problems. For practice, this study suggests that an empirically derived temperament typology during the preschool years can be a valuable tool for screening children at risk of developing behavior problems during adolescence and who could participate in an indicated preventive intervention.
CONCURRENT 9, EFFICACY TRIALS, Grouped papers
EFFICACY OF SCHOOL BASED INTERVENTIONS
Chair: Denise Gottfredson
PROJECT NORTHLAND FOR URBAN YOUTH: PRELIMINARY RESULTS AFTER TWO YEARS OF INTERVENTION. Kelli Komro1, Cheryl Perry2, Sara Veblen-Mortenson2, Kianoosh Farbakhsh2, Karen Munson2, Kari Kugler2, 1University of Florida, Gainesville, FL United States; 2University of Minnesota, Minneapolis, MN United States
The purpose of this presentation is to describe the implementation and outcomes of the 7th grade intervention of the urban adaptation of Project Northland, a multiple component alcohol use preventive intervention. An adapted and enhanced Project Northland is currently being evaluated using a randomized trial of 60 schools and surrounding neighborhoods in Chicago. The intervention group is being exposed to three years of curricula, family interventions, and community-wide activities. The intervention is being evaluated with a cohort of 6th-8th grade students. The cohort sample is primarily African American (37%), Hispanic (32%) and low-income (70%). The main outcome of youth alcohol use is measured via classroom-based surveys of students each year. The 7th grade intervention included: 1) a nine-session classroom program entitled “Amazing Alternatives!”; 2) a four-session home-based program; 3) five parent postcards with prevention tips mailed directly to parents; and 4) community organizing. The curriculum was fully implemented by 70 trained classroom teachers and 434 peer leaders in each of the 28 intervention schools. Seventy-five percent of families completed at least one of the four home-based activities and 55% completed at least three. Postcards were mailed directly to parents and approximately 5% of them were returned because of an incorrect address. Returned postcards were then delivered to the child in the classroom to take home to their parents. Ten full-time community organizers work with their designated schools and community areas to develop action teams made up of adult volunteers to initiate community-wide strategies to prevent youth drinking. Preliminary outcome analyses have been performed on the cohort who did not report drinking at baseline, which represents 83.4% of the full sample [N= 1,115 boys (66.7% of baseline sample) and N = 1,176 girls (66.8% of baseline sample)]. Among baseline nonusers (83.4% of the sample), boys in the intervention group compared with the control group had a lower rate of alcohol use in the past year (21% and 29%, respectively, F (1,20) = 4.11, p = .06) and ever been drunk (5% and 8%, respectively, F (1,20) = 5.63, p = .03). Three other alcohol use items were also lower in the intervention group, but were not statistically significant. There were no significant differences among girls in the intervention group compared with the control group. Given the lack of positive results for girls, we have made modifications to the final year of interventions for eighth grade students.
BUILDING ADOLESCENT COMPETENCE IN PREVENTION MEDIA LITERACY: EARLY EFFECTS OF MEDIA BUZZ IN PROJECT STEP. Mary Ann Pentz1, Chih-Ping Chou1, Maykami McClure1, Peter Bunce2, Eric Y.I. Wang1, Cornelia Pechmann3, 1University of Southern California, Alhambra, CA United States; 2One Great River, Inc., Shreveport, LA United States; 3University of California, Irvine, Irvine, CA United States
Media literacy programs teach youth how to critically analyze media influences and environments around them, with the intention of building competency in both critical thinking as well as in making decisions about media influences on their lives. Media literacy programs are typically open-ended rather than applied to specific decisions about health behavior. As part of a larger drug abuse prevention diffusion trial, STEP, 52 schools in five states were randomly assigned to an applied drug prevention media literacy program, Media Buzz, or a control condition. Participants were all 6th, 7th, and 8th grade students in the middle schools (N=5,780). Teachers from program schools were trained via interactive satellite television training (N=72) to deliver a four session program that taught students how to recognize and counteract pro-tobacco, alcohol, and drug use media images, understand media driven industry objectives, look for and analyze anti-use media messages, and improve upon anti-use messages by creating their own. A self-report survey was used to evaluate Media Buzz effects, from baseline to one year follow-up. Outcomes included six composite scores generated from rotated factor analysis and representing the constructs addressed in the program: belief in media ads, exposure to drug use ads and images, exposure to anti-use ads and images, beliefs that media influence kids to use drugs, awareness of ad message intent to glamorize use, and awareness of indirect or implicit media influences on drug use. Regression analyses, including using school as the unit, and proc.mixed analyses with individual as the unit controlling for school, were conducted with grade and program group as predictors, and each of the six composite media scores as outcomes, as well as participation in prevention activities outside of Media Buzz, and intentions to use cigarettes, alcohol, or marijuana in the next two months. Results showed that Media Buzz had significant effects on beliefs, watching and comparing use and anti-use ads, extra participation in prevention activities, and a marginal effect on intentions to use drugs (p=.11). Findings suggest that a brief prevention media literacy program can increase student competencies in critical thinking about media, participation in prosocial prevention activities, and decisions about future drug use.
STEERING TOWARD SAFETY: RESULTS OF TWO DEVELOPMENTAL BOOSTER SESSIONS TO PROMOTE HEALTHY DRIVING BEHAVIOR. Kevin Haggerty1, Charles Fleming1, Tracy Harachi1, Richard Catalano1, 1University of Washington, Seattle, WA United States
Sixteen- to nineteen year-olds have the highest incidence of motor vehicle deaths among licensed drivers (Miller et al., 1998). Motor vehicle crashes among teenage drivers have been linked to several factors: insufficient driving experience, adolescent risk taking behavior, lack of clear parental expectations and guidelines, peer influence, and substance use ( e.g. Jelalian et al., 2000). Recently, interventions promoting safe driving practices have targeted teens (Griffin et al., 2004) and their parents (Simons,-Morton et al., 2004). The purpose of this study was to evaluate the specific impact of two targeted family booster sessions designed to promote safe driving practices as part of a social development intervention. The first session consisted of a pre-driving home visit with families when students were 15 years old to promote the development of clear driving standards and expectations. This was followed by a visit at the time of licensure to develop a written family contract to promote positive driving behaviors (e.g. drinking and driving, contingent use of car, car care and costs).
Booster session materials were developed out of focus groups that explored families´ communication regarding driving and related risks. The focus groups yielded an understanding that families seek information and skills regarding safe driving, family communication, and setting family guidelines about driving. Although driver education classes and graduated driving laws address some of these risks, parents and teens both desire a family based approach.
Data reported here were collected on the longitudinal panel (n=924) of students who began the Raising Healthy Children project in the fall of 1993 (while in the first and second grades) in the spring of 2004. The average age of the students was 17.6, 54% were in the experimental group and 46% in the control group. Logistic regression analyses revealed that students in the intervention group were more likely to report having a written driving contract (p=.000, OR=4.86) and participating in making driving rules in the family (p=.007 OR=1.67) than students in the control group. Further, students in the intervention group reported less driving with someone who has been drinking, driving under the influence of alcohol, and driving under the influence of drugs. There were no significant differences between students in the intervention and control groups with respect to number of traffic tickets or accidents. Findings will be discussed in light of using developmentally sequenced “booster” visits to promote well being and reduce risks during times of adolescent transitions.
RESULTS FROM AN INDEPENDENT EVALUATION OF PROJECT ALERT DELIVERED VIA SCHOOL-EXTENSION COLLABORATAIONS
Chair: Tena L. St. Pierre
RESULTS FROM AN INDEPENDENT EVALUATION OF PROJECT ALERT DELIVERED VIA SCHOOL-EXTENSION COLLABORATAIONS. Tena L. St. Pierre1, Richard Spoth2, 1Pennsylvania State University, University Park, PA United States; 2Iowa State University, Ames, IA United States
Strides made in the identification of empirically supported school-based drug prevention programs have produced new challenges for prevention research: (1) strengthening existing evidence by conducting effectiveness studies, ideally through independent evaluations, of programs previously designated evidence-based through efficacy trials; (2) creating strategies to maximize implementation fidelity in real world settings; and (3) developing methods to broadly disseminate and sustain programs. This symposium presents findings from an independent multi-site effectiveness study of Project ALERT previously found to have positive results through efficacy and effectiveness trials, both conducted by the developer. Few programs listed on national registries, including Project ALERT, have been independently evaluated. Among the few that have, weaker or even negative results have been reported. One challenge of replicating programs through effectiveness trials is the difficulty of implementing faithfully under real world conditions. Some have suggested that it may be more practical and cost-effective to train qualified outside providers who may be more likely to implement with fidelity. Overworked teachers may appreciate having someone else teach prevention and school administrators may be more enthusiastic about adopting new programs since staff time is a key barrier to implementation. With this rationale in mind, Project ALERT was implemented through a collaborative model called EXSELS (Extension and Schools Enhancing Life Skills). County-based Penn State Cooperative Extension (CE) Educators in 8 counties partnered with middle schools to deliver ALERT by community program leaders hired through CE instead of by teachers. CE has a history of applying and disseminating research-based knowledge across the U.S. and therefore may have potential to broadly disseminate evidence-based prevention programs. Papers in this symposium correspond to the three challenges presented above. The first paper presents outcomes from an independently evaluated effectiveness study of Project ALERT. Eight schools randomly assigned two 7th-grade classrooms to each of three conditions: adult-led; adult-led, teen assisted; and control. Utilizing measures provided by ALERT´s developer, two cohorts (n=1,648) were tested before and after each of the two program years and after one follow-up year. The second paper examines implementation quality and the relationship to outcomes. Several measures assessed implementation from the perspective of classroom observers and students.The third paper describes processes of adoption and maintenance of Project ALERT through CE. The potential for evidence-based programs to be broadly disseminated through CE will be discussed.
ADOPTING PROJECT ALERT VIA SCHOOL-EXTENSION COLLABORATIONS: CE EDUCATOR AND SCHOOL PERSPECTIVES. Claudia C. Mincemoyer1, Tena L. St. Pierre1, D. Lynne Kaltreider1, Tina Kauh1, 1Pennsylvania State University, University Park, PA United States
Cooperative Extension (CE), as part of the land-grant university in each state, employs CE Educators in counties across the U.S. Characterized as the oldest and most successful diffusion system in the U.S. (Rogers, 1995), primarily in production agriculture, CE may have potential to broadly disseminate evidence-based drug prevention programs. CE increasingly has collaborated with schools and community agencies to deliver and evaluate prevention programs. However, little is known about the processes of adoption by CE and community collaborators. How viable is CE as a diffusion system for drug prevention? This paper addresses three questions related to the adoption of Project ALERT through school-extension collaborations. Data were collected within the context of a NIDA-funded study to test the effectiveness of ALERT delivered by outside program leaders hired/supervised by local CE Educators affiliated with Penn State. One, how did the CE Educators who agreed to take part in the project differ from those who were not interested? Before implementation, phone interviews were conducted with 52 CE Educators to ask what would influence them to take part in this type of project and perceptions of their workplace environment. ANOVAs conducted to assess responses from participating (n=8) and nonparticipating (n=44) Educators found that the 8 participating Educators differed significantly in 3 ways. They (1) cited community characteristics more frequently as influencing their decision to participate (i.e., need for prevention, support from schools); (2) reported more positive perception of their workplace environment; and (3) perceived less work pressure.Two, of schools approached by CE to collaborate how many refused, and why? Of 28 schools, 20 declined and 8 adopted. Most frequently given reasons for refusing were (1) simply not interested; (2) could not fit the program into their schedule; and (3) research issues including not wanting a control group and/or student surveys. Three, of the 8 schools that collaborated to implement ALERT, how many sustained the program, and why? One year after the project ended, 3 schools continued ALERT with the CE program leader. Four factors were common to the sustaining schools: (1) only these 3 schools maintained the administrators who had championed the program; (2) all had a slot in the 7th grade curriculum with supportive teachers; (3) all had outside program funding, 2 from CE´s involvement with local coalitions with tobacco settlement funds. To create and sustain school-extension collaborations to deliver evidence-based drug prevention programs, issues within each organizational environment must be addressed. Viability of CE as a diffusion system for evidence-based drug prevention programs will be discussed.
OUTCOMES FROM AN EFFECTIVENESS STUDY OF PROJECT ALERT DELIVERED IN SCHOOLS BY COOPERATIVE EXTENSION. Tena St. Pierre1, D. Wayne Osgood1, D. Lynne Kaltreider1, Claudia Mincemoyer1, Tina Kauh1, Frances Burden1, 1Penn State University, University Park, PA United States
Project ALERT, a universal drug prevention program for middle school students, has been recognized as an exemplary (USDoE) and model (CSAP) program based on a multi-site experimental efficacy trial using trained health educators. Revised Project ALERT, later tested in a large-scale randomized effectiveness study with regular classroom teachers, yielded stronger results in curbing initiation of cigarettes and marijuana, current and regular cigarette use, and alcohol misuse. This paper presents results from an independent multi-site effectiveness study that found no positive effects from revised ALERT despite evidence of high-quality implementation. Through local school-extension collaborations, ALERT was delivered by outside program leaders. Eight Pennsylvania middle schools randomly assigned two 7th-grade classrooms to each of three conditions: adult-led; adult-led, teen-assisted; control. Using the developer´s measures, baseline data were collected from 1,648 students. Retention was high; 73% participated in all five waves and 88% in at least four waves. Because random assignment at each site was among only 6 classrooms instead of roughly 120 students, baseline comparability across conditions was carefully examined. Analyses showed that conditions for both cohorts were comparable at all schools (except for cohort 2 at two schools which were eliminated from the analysis). Comparability was tested using ANOVAs controlling for differences among schools and taking into account nesting of students within classrooms. Analyses produced fewer differences nominally significant at the .05 level than would be expected by chance. ALERT's effectiveness was tested through a hierarchical linear model (Raudenbush & Bryk, 2002) designed to focus on the program's impact on within-individual change (Osgood & Smith, 1995; Esbensen et al., 2001). The three-level model (waves nested within students nested within classrooms) controlled for school differences. The analysis maximized power for detecting program effects by contrasting the pretest with the mean of the four posttests, while including additional contrasts within the posttest period to aid in interpreting any overall effects. Highly skewed variables were dichotomized and analyzed through a logistic version of this model. Analyses of 58 outcome measures targeted by the program yielded no significant evidence of beneficial effects. Further analyses indicated that program effects did not vary across risk groups or schools. Finally, there was no indication of positive effects in a variation of the analysis that attempted to match the previous effectiveness trial of Project ALERT (Ellickson et al., 2003: limited to the 18-month posttest; covariance correction for pretest differences). Implications will be discussed.