MEDIATORS OF THE RELATIONSHIP BETWEEN STRESS AND SUBSTANCE USE DURING ADOLESCENCE. Diana Fishbein1, Mindy Stahl2, Jason Williams3, Diana Eldreth1, Christopher Hyde4, Mallie Paschall5, Scott Hubbard6, Nicholas Ialongo6, 1Research Triangle Institute, Baltimore, MD United States; 2Research Triangle Institute, Durham, NC United States; 3Research Triangle Institute, Research Triangle Park, NC United States; 4Bioassessments, Elkton, MD United States; 5PIRE, Berkeley, CA United States; 6Johns Hopkins University, Baltimore, MD United States
Exposure to stress has been implicated in both the initiation and progression of drug use (e.g., Kaplan et al., 1986; Newcomb & Bentler, 1988; Wills et al., 1996). However, little is known about the mechanisms through which stress operates to affect substance use behaviors (Bremner et al., 1999). Chronic or severe stress has been shown to impair function of cortico-limbic structures resulting in impairments in executive cognitive functioning (ECF) and emotional regulation. Moreover, deficits in ECF and emotional regulatory processes have been directly associated with substance abuse (Fishbein et al., 2003). The purpose of the present study was to examine the direct effect of stress on adolescent substance use as well as the effects of cognitive and affective processes in mediating this relationship. The sample consists of a subset of 120 adolescents from the Baltimore City Public School system who participated in a larger study conducted by the Prevention Intervention Research Center at Johns Hopkins University to assess the effects of first grade interventions on the development of psychopathology (Ialongo et al., 2001; Kellam et al., 1998). For the present study, adolescents were recruited to participate in additional neurocognitive and emotional regulatory testing to assess adolescents´ ability to process prevention curricula. The sample was all male, 88% African American (12% white), and the mean age was 16.1 years. Approximately one-half of the sample had conduct disorder. A battery of tests was used to assess ECF, emotional perception, IQ, conflict resolution skills, social competency, negative life events, and substance use. We used a multiple mediator model to test our research questions. Analyses revealed that risky choices were significantly higher and social competency was significantly lower in participants who reported greater stress in the prior three years. However, only social competency was significantly related to substance use in grade 10. The mediated effect was significant indicating that the effect of stress on use of alcohol, marijuana, and other illicit drugs was significantly mediated by social competency. The direct effect of stress was reduced to nonsignificance when mediation was taken into account. Findings highlight mechanisms through which substance abuse may develop and identify important targets for preventive interventions, especially for highly stressed children and adolescents.
A NEW LOOK AT RISK: DEFINITION AND PROCESS MODEL. Elizabeth Johnson1, Daphna Oyserman1, 1University of Michigan, Ann Arbor, MI United States
The notion of “risk” is fundamental to the field of prevention science. Understanding what makes for a risky environment, the psychological and physiological consequences of living with risk, and what factors diminish risk or promote well-being, is central to the basic mission of prevention science. At the heart of questions about risk are basic questions about how contexts “get under the skin” to influence development and behavior, the extent to which developmental trajectories are malleable, and how they can be altered. Critical to questions about the malleability of developmental trajectories are questions about the malleability of risk factors themselves. Because many risk factors are at least theoretically malleable, how we conceptualize and operationalize risk has important implications for when, where, and how we intervene in the lives of individuals and their environments, the policies and social structures we advocate for, and our understanding of what is universal or basic and what is learned or contextual derived.
Despite the importance of both general and specific notions of risk for prevention science, the field of risk research has been hampered by a number of problems including ambiguity in risk terminology, lack of process models, and atheoretical statistical representations of risk. Lack of precision in risk terminology and underutilization of process-oriented models of risk can create problems in summarizing and integrating research, impede conceptual and empirical progress, undermine the translation of research to intervention, distort policy implications, and set the stage for use of inappropriate statistical methods.
In an attempt to remedy these issues, the current paper defines and operationalizes risk and at-risk, outlines a process model that distinguishes distal from proximal, fixed from malleable, and general from specific risks, proposes psychological and physiological processes as the most immediate mediators of the risk to outcome relationship, and incorporates biological-genetic and developmental phase-time as moderators of the relationship between contextual factors and health and mental health outcomes. By incorporating both biological and psychological processes into a process model of risk, we aim to contribute to the more general integration of these two importance facets of development in prevention research and to enable research to move forward in conceptually and practically meaningful ways.
DO THE FEMINIST BATTERER TREATMENT PROGRAMS WORK?: AN HISTORICAL AND PHILOSOPHICAL EXAMINATION. Hyunkag Cho1, 1Florida State University, Tallahassee, FL United States
Since the late 1970s, numerous treatment programs for batterers have been established around the country. Because the batterer program was originated by battered women´s advocates and feminists, the feminist perspective has influenced most programs. However, the results of numerous evaluation studies as to the batterer treatment programs are inconclusive. Then, how can we know whether the treatment programs work or not? How can we know whether the feminist perspective actually works well aside from its explanatory and political attractiveness, and claims of effectiveness? The purpose of the study is to evaluate the effectiveness of the feminist perspective by reviewing the evaluation studies on batterer treatment programs, as well as to examine the historical development and philosophical position of the feminist perspective.
Hostorically, feminist researchers are explicitly dedicated to women´s advocacy and social change, which is intrinsically political. In order to make women´s experiences visible and to recover, or create, the equilibrium between men and women, feminists concentrate their major efforts on social change-oriented works both academically and politically.
As to the effectiveness, the feminist batterer treatment program shows no meaningful effectiveness, although it seems to be a little more effective than the programs based on other perspectives. In other words, the overall effectiveness of batterer treatment programs is inconclusive regardless of their modality.
Philosophically, there are two explanations as to knowledge development; justification and falsification. Feminism explicitly employs justificationary philosophy of science. Because feminism is devoid of a testable hypothesis, employs induction as its research method, and depends on social constructionism in which individuals and groups construct their own truth based on their unique experiences and their own sociocultural backgrounds.
Feminism has provided both the academic and social communities with unprecedented epistemological insights into the experiences of the oppressed, such as women. However, it must not be used as the direct theoretical background to design and implement interventions for any population. Because interventions based on untested hopes and beliefs rather than scientifically tested theories may do harm to clients rather than good.
As a conclusion, the author argues that batterer treatment programs should be suspended until a sort of new program that has survived rigorous tests appears, and that the resources allocated to the programs should be redirected to other interventions and services related to violence against women, such as services for victims and their children, and economical support for victims.
A PRELIMINARY STUDY OF AN OBSERVATIONAL MEASURE OF IMPLEMENTER COMPETENCE. Wendi Cross1, Peter Wyman1, 1University of Rochester, Rochester, NY United States
One of the primary obstacles to transporting preventive interventions to communities is a lack of precise ways to measure and monitor competent delivery by implementer agents (IAs). Recent research has demonstrated that self-report ratings of IA behaviors are significantly less predictive of clinical change than more objective measures (e.g., Miller & Mount, 2001). Observational ratings, however, provide an effective method of measuring IA competence and skill which trainers and supervisors can use to assess training programs, monitor implementation, and investigate the relationship between IA competence and clinical outcomes. For example, Forgatch and colleagues (Knutson, et al., 2002) developed an observational measure of competent implementation of Parent Management Training – Oregon (PMTO), The Fidelity of Implementation Measurement System (FIMP). The FIMP operationalizes both process and content in the application of PMTO, and IA scores on the FIMP have been found to predict clinical outcomes. The purpose of the current pilot study was to: a) operationalize implementer behaviors in delivering a school-based preventive intervention (the Promoting Resilient Children Initiative, PRCI; Cross et al., 2002), and b) assess inter-rater reliability in an observational rating scale of IA competence. PRCI is a curriculum-driven, skills-based indicated prevention program that is delivered by trained paraprofessionals to individual K-3rd grade children demonstrating early adjustment difficulties. IAs received 4 months of weekly training followed by 4 months of biweekly supervision. Eight skill domains were identified. These domains reflected intervention components as well as common IA skills (e.g., greeting, introducing the skill focus, practicing the skill focus with the child) that may be relevant to other child-focused interventions. Criteria for scores on each dimension were operationalized by descriptions of skills on Likert-scale scores ranging from 1 (poor) to 5 (excellent). IAs were rated on the eight domains by two independent raters who observed 12 complete IA-child contacts of 30 minutes via videotape or observation. Observer ratings showed a substantial range of scores and no ceiling effect. Inter-rater reliabilities (ICC), ranging from .72 to .90, are presented. These preliminary findings indicate that IA competency in delivering the PRCI intervention can be reliably measured using behavioral observations. Findings are discussed in terms of future measure development, implications for training and monitoring IAs, and planned studies of the relationship between IA competence and child outcomes.
ADULT DRUG COURTS: THE EFFECT OF COERCION ON PROGRAM RETENTION RATES. Natasha Williams1, 1Morgan State University/Johns Hopkins Urban Health Institute, Baltimore, MD United States
Overview of Hypothesis Coerced treatment refers to the criminal justice system´s use of legal sanctions to induce drug-abusing offenders to enter substance abuse treatment. Court-mandated treatment provides the incentive for the offender to attend treatment, however, the question remains whether it produces the type of internal motivation that is necessary to change substance-abusing behavior. Drug courts are a major innovation in penal responses to drug crime, and the first successful rehabilitation movement since the mid-1970s. Drug courts, which utilize a sanction/reward system, are judicially supervised programs that place the drug-abusing offender in an intensive community-based drug treatment program that not only provides treatment but other rehabilitative services such as job training, parenting classes, and GED assistance. Upon program completion, the court may dismiss the original charge, reduce the sentence, set aside the original sentence or offer a combination of these remedies. The purpose of this study is to evaluate the effect of sanctions and rewards among adult drug courts on program retention rates.
Methodology The sample was drawn from the Drug Courts 1999 Program Update Survey conducted by the U.S. Department of Justice, Office of Justice Programs Drug Court Clearinghouse and Technical Assistance Project (DCCTAP). The survey was mailed to 210 adult drug courts that were in operation as of December 31, 1999. The sample used for data analysis consisted of 141 adult drug courts that had been in operation for at least twelve months. The data analysis involved factor analysis and regression analysis.
Results Due to limitations of the drug courts´ various theoretical models and outcome measures, the research design was unable to explain the interaction among the structural variables.
Policy Implications This research exposes inadequacies in current research and provides a basis upon which to develop more pragmatically focused evaluation designs built upon theoretical models that better explain drug court functions as well as generating a policy agenda that links research to policy development.
PROBLEMS AND SOLUTIONS IN MEDIATION ANALYSIS WITH CORRELATED MEDIATORS. John Graham1, Allison Olchowski1, 1Pennsylvania State University, University Park, PA United States
Mediation analysis is one of the most important and often used analytic tools in the social sciences, especially in prevention research. The mediation model is
X=>M=>Y where α is the b-weight for the program (X) predicting the mediator (M) and β is the b weight for M predicting the outcome (Y). The mediation effect (the α β product) is significant when the two effects, α and β, are jointly significant (MacKinnon et al., 2002).
When one has multiple mediators, mediators are nearly always correlated. Just as the correlations among predictors can obscure prediction of the outcome in multiple regression, non-trivial correlations among the mediators can obscure the effect of the mediators on the outcome. When the true M=>Y link is significant, but weak, the correlations among the mediators can lead the investigator to conclude erroneously that there is no mediation.
The standard approach to analysis when one has multiple mediators is to estimate mediation without special concern for the correlations among the mediators. Unfortunately, this approach is not a solution, but rather is the cause of the problem. Partial solutions are easy to envision, but a more complete solutions is needed. Below, we describe such a solution, based on dominance analysis.
Dominance analysis was developed to study the relative importance of predictors when a model contains multiple correlated predictors. Budescu (1993) developed a method in which the relative increase in R2, across all possible sub models, is assessed for each predictor variable. The sum of these weighted scores equals R2, and each predictor's general usefulness score is the proportion of outcome variance for which that predictor accounts.
Other approaches to dominance analysis make use of both the original set of correlated predictors and a set of orthogonalized predictors to assess the relative importance each predictor. Johnson (2000), following Green et al. (1978), developed such a weighting scheme. After the outcome variable has been regressed on the uncorrelated orthogonalized predictors, the squared b weights are separated in order to determine the portion of variance in Y attributable to each original predictor. Original predictor importance scores sum to R2. Johnson (2000) found that his weighting scheme, which is easier to implement than that of Budescu (1993), yields importance weights very similar to Budescu's.
The Johnson (2000) importance weighting scheme is easily adapted to mediation analysis, with the result that all of the true mediation can be divided up amongst the multiple mediating variables. Realistic t-values are easily calculated for each mediation effect. Our poster will present simulation results, along with an empirical example.
A MODEL FOR MOTIVATING LICENSED PREMISES TO PERFORM ENVIRONMENTAL CHANGES TO REDUCE DRUG USE.. Johanna Gripenberg1, Sven Andreasson1, 1Karolinska Institute, Stockholm, Sweden
Background. The use of drugs at licensed premises has increased in Stockholm. As a result, a community drug prevention program with the aim of reducing the prevalence of drug use at licensed premises has been initiated. The community drug prevention program includes mobilization of important gatekeepers and authorities, drug-training programs for employees, policy work, increased enforcement, PR campaigns, and changes in the physical environment at licensed premises.
Objective. To evaluate a model for motivating licensed premises to perform environmental changes to reduce drug use.
Design. Pretest (2004) -post-test (2005) design.
Setting. Licensed premises in the Central part of Stockholm, Sweden.
Method. A model for motivating licensed premises to conduct environmental changes has been developed, including a checklist. The checklist covers observation of the physical environment, documentation of house policy work, prevalence of drug-trained doormen and servers, and information to patrons. As part of the intervention, the licensed premises are visited and suggestions for improvements are made.
Results. Baseline observations based on the checklist were made during the fall of 2004. Follow-up, with renewed visits, is planned during the spring of 2005.
USING A BEHAVIOR PROBLEMS CHECKLIST TO ASSESS RISK OF SUBSTANCE USE: AN IRT ANALYSIS OF DIFFERENTIAL ITEM FUNCTIONING ACROSS GENDER. Stephanie Lanza1, E. Foster1, 1Pennsylvania State University*, University Park, PA United States
The identification of children at-risk for developing substance use problems is an important step in prevention efforts. Further, better screening tools are needed to identify those children. One well-established risk factor for subsequent substance use is behavior problems in childhood. The Child and Adolescent Disruptive Behavior Inventory (CADBI) is a behavior problems checklist where parents or teachers code the frequency with which a child engages in a wide range of externalizing problems (including attention problems). The measure has been used widely (e.g. Burns et al., 1997). The CADBI has several advantages over alternative measures, such as Achenbach´s Child Behavior Checklist (CBCL). One example is a greater number of response options (eight response categories for most items), presumably leading to greater specificity of measurement. In addition, the specific behaviors included better map onto diagnostic criteria. Using an 18-item subscale from the CADBI that measures attention and activity level, we will demonstrate that the high specificity of measurement is important for capturing individual differences in the underlying trait among children, particularly those with higher levels of activity and greater attention deficit. We will present preliminary work based on an IRT framework in order to better understand the properties of the CADBI measure and explore measurement invariance across gender using parent-report data from 525 first-grade students from The Success Through the Incredible Years Project (Taylor et al., under review). We will demonstrate how to (1) test the assumptions underlying scales based on classical test theory; (2) identify the range of the underlying construct over which each item and the test as a whole is most informative; and (3) examine differential item functioning by gender.
Burns, G., Walsh, J. A., Owen, S. M., & Snell, J. (1997). Internal validity of attention deficit hyperactivity disorder, oppositional defiant disorder, and overt conduct disorder symptoms in young children: Implications from teacher ratings for a dimensional approach to symptom validity. Journal of Clinical Child Psychology, 26(3), 266-275.
Taylor, T. K., Biglan, T., Metzler, C. M., Rusby, J. C., Johnson-Shelton, D., Foster, E. M., & Blair, J. Outcome research for dissemination of evidence-based practices: The Success Through the Incredible Years Project as an example. Under review at Prevention Science.
MEASUREMENT SENSITIVITY TO CULTURAL DIFFERENCES IN PARENTING BEHAVIOR:. Martie Skinner1, Elizabeth MacKenzie1, Kevin Haggerty1, 1University of Washington, Seattle, WA United States
Substance use prevention programs rely on teen and parent report, and occasionally coded observations, of important parenting constructs to determine the efficacy of interventions and to understand the etiology of substance use in various populations. We currently know very little about how these three different measurement strategies may differ in their reflection of purported differences between parenting behaviors of African and European American parents. Measurement equivalence has been supported in a few studies (Rosay, Gottfredson, Armstrong, & Harmon, 2000; Gottfredson and Koper, 1997) examining teen reports. Catalano, et al. (1992) found differences between African American and European American parents in mean levels and predictive validity using student reports in the 5th grade. Little evidence is available on how cultural differences may or may not be evident depending on the informant. Data on 168 European and 163 African American families participating in an efficacy trial of the Parents Who Care, a substance abuse prevention program for families and young adolescents, were collected to measure a number of family factors believed to influence initiation into substance use. Parent report, teen report, and videotaped interactions were collected prior to intervention (8th grade). We examined 10 constructs related to parenting where we had measures from at least 2 of the 3 informants (parent, teen, observer ratings). Scales were examined for mean differences by race (EA, AA) as well as interactions between scale and race in predicting teen report of initiation of tobacco, alcohol, or marijuana by 8th grade. AA parents rated themselves higher on harsh parenting, number of explicit rules, and monitoring. Observers also rated AA parents as more harsh. Additionally, observers rated EA parents higher on opportunities and rewards for pro-social involvement. No differences by race were detected in teen reports. Separate logistic regressions predicting initiation of substances were estimated for each of the three respondents. No significant interactions between parenting scales and race were detected. This study extends our investigation of cultural differences in parenting to include self-report and observer rated measurement strategies. Consistencies and inconsistencies with previous research using student reports will be discussed.