Evaluation of Life Skills Training and Infused-Life Skills Training in a rural setting: outcomes at two years.

Abstract

This study reports on findings from the first two years of a study to compare a standard Life Skill Training (LST) program with an infused (I-LST) approach. Nine small, rural school districts were randomly assigned to LST, I-LST, or control conditions in grade seven. The LST program significantly reduced alcohol use, binge drinking, marijuana use, and inhalant use after one year for females, and the I-LST program significantly reduced smoking, binge drinking, and marijuana use for females. At the end of the second year the I-LST program continued to impact female smoking, but all other results were non-significant. There were no effects on males at either time point.

Key Words: Substance Use, Adolescents, Prevention

INTRODUCTION

Although a variety of substance abuse prevention programs have been implemented in many school districts throughout the U.S. for several decades, the use of alcohol, tobacco and other drugs continues to be a major health and behavior problem for American youth. Many adolescents initiate alcohol consumption at a very young age, and a high proportion of young people drink often and/or heavily (Johnston, O'Malley, & Bachman, 2001). Furthermore, substance use is as prevalent in rural areas as urban locales (Johnston, O'Malley, & Bachman, 2000). The Center for Addiction and Substance Abuse (CASA) reported that rural eighth grade students are more likely to report alcohol use, drunkenness and smoking cigarettes than their urban counterparts (Center on Addiction and Substance Abuse, 2000). While most scientifically supported prevention efforts have been aimed at urban and suburban youth, these recent data indicate that prevention programs are also needed in rural areas (Johnston et al., 2000).

The field of prevention has identified a number of effective prevention programs and an emerging science of prevention. Examples for early adolescents include: Life Skills Training (LST) (Botvin, Baker, Dusenbury, Botvin, & Diaz, 1995); Star and community task forces model (Pentz et al., 1989); the All Stars normative education program (Hansen, 1996); Alert (Ellickson & Bell, 1990); and Project Smart (Hansen, Johnson, Flay, Graham, & Sobel, 1988). Reviews of these and other prevention approaches indicate that, overall, these strategies have shown favorable results in reducing adolescent substance use (Botvin, & Botvin, 1997; Hansen, 1992; Tobler, & Stratton, 1997). Conversely, Gorman (1998) argued that the results of these programs have been exaggerated and have not been sustained over time. Life Skills Training, in particular, may have selected a biased sample in the six year follow up in which only those subjects who had received 60% or more of the program (high fidelity sample) were included (Gorman, 1998). Regardless of the relative merits and long term effectiveness, these "model" programs constituted only one third of all drug prevention programs currently being implemented under one Federal program designed to increase use of science-based programs (Swisher, 2001). Ennett et al. (2002) reported that approximately one-quarter of teachers implementing substance use prevention are using an evidence-based curriculum. A number of reasons account for this lack of diffusion. Included among these reasons is the requirement that LST and other curricula require that a separate time be provided for drug prevention. In face of scholastic demands for increased performance and accountability, this separate time requirement presents obstacles for both school administrators and teachers.

Infusion as an alternative delivery mechanism

Infusion is an approach that integrates prevention objectives and activities into basic academic objectives and activities at the same time. For example, an English teacher might teach vocabulary and integrate knowledge of risk of chewing tobacco. Similarly, a social studies teacher might teach personal decision making steps in relationship to critical decisions made at some point in history, and math teachers can teach charts and graphs with normative education data about non-use trends. The rationale for a decentralized infusion strategy to implement science-based effective substance abuse prevention programs in schools is that the content and skills to be learned are spread across several different teachers thereby reducing the burden on any one teacher. In addition, if the infusion is done in the context of the subject being taught, it is possible to teach two objectives simultaneously. This type of infusion does not require large amounts of additional time for the objective of substance abuse prevention and allows the teacher to accomplish his/her current subject matter objective.

The major strategy for facilitating adoption of Life Skills Training by the Adoption of Drug Abuse Prevention Training (ADAPT) Project has been to assist teachers with the infusion of LST into their normal teaching. This approach has been labeled constructivist (Gatewood, 1998) and also is a multi-disciplinary or interdisciplinary approach. Jacobs (1989) describes infusion as an approach that consciously applies content from more than one discipline to examine a central theme or issue which, in this case, was substance abuse. The ideal form of infusion centers the curriculum on life itself rather than on the mastery of fragmented information with the boundaries of subject areas (Beane, 1995). In the educational literature, the term integration also has been used synonymously with infusion.

In the field of substance abuse prevention, the term infusion is often used to describe the infusion of content and process into the standard basic subjects of a school curriculum. This process requires the active involvement of teachers in the design and delivery of the prevention content and process in their regular classes. Infusion becomes more than using guest speakers or fragmenting a prevention program into blocks spread among several teachers. Rather, infusion requires the teachers to work with others to creatively weave a complete prevention program that fits well within the total curriculum of the school.

Although teachers are often encouraged to develop and implement lessons that integrate or infuse prevention into their classes, many teachers are not aware of specific techniques or a structure for maximizing an infusion approach. To address this potential problem, the ADAPT project staff employed a step-by-step approach to curriculum infusion (Cinelli, Rose-Colley, & Bechtel, 1995) that identifies key objectives in their academic subject and key objectives from a science-based prevention program. ADAPT also assisted teachers with developing skills in using the best interactive teaching approaches (e.g., cooperative learning, coaching, and behavioral rehearsal) for accomplishing these dual objectives.

Differential program effects by gender

Prevention programs often take a broad approach, attempting to effect change in a number of different populations. However, research suggests that risk and protective factors for substance use may be different among various populations. Gender differences, in particular, have often been overlooked when designing and evaluating prevention programs. Research suggests that risk factors for substance abuse may differ for males and females and that gender may play a role in how risk factors influence substance abuse (Amaro, Blake, Schwartz, & Flinchbaugh, 2001).

A review of gender differences among substance abuse prevention programs conducted by Blake, Amaro, Schwartz, and Flinchbaugh (2001) found that although few studies report on differential effects by gender, several studies have found differences to exist. Of those studies reviewed that reported differences separately by gender, 90% reported stronger effects for females (Blake et al., 2001). Consistent with the findings of Amaro et al. (2001) regarding differential risk and protective factors, those interventions that had greater success with females tended to focus on social skills, peer pressure, and normative education. Most pertinent to the current study, differential treatment effects by gender were found at the one year follow-up for adolescents involved in the Life Skills Training program (Botvin, Baker, Filazzola, & Botvin, 1990). Significant positive effects were found for girls in relation to tobacco, alcohol, marijuana use and problem drinking while only problem drinking was significant in boys (Botvin et al., 1990).

METHODS

The interventions

The primary objective of the investigation, funded by the National Institute on Drug Abuse, was to test the effectiveness of an infused approach to teaching the LST components. The LST program has long been touted as one of the most effective ATOD prevention programs, and has been empirically tested in numerous investigations (Botvin et al., 1995; Botvin, Baker, Dusenbury, Tortu, & Botvin, 1990). Most often, the I.ST curriculum is taught by one or two teachers within the framework of a single course. The I-LST condition utilized the LST components, but with an innovative delivery method. During the I-LST training sessions a matrix of L.ST principles was mapped onto the curricular content areas of the participating teachers. This exercise assured that all students were exposed to all LST elements. (For a further discussion of the training procedures see Swisher, Bechtel, Henry, Vicary, & Smith, 2001; Karshin et al., 2000). The I-LST teachers were then asked to generate subject-specific lessons that integrated the LST principals and methods. For example, a normative beliefs clarification lesson might be infused into a math lesson by graphing the percentage of U.S. students that do not use marijuana.

Due to the nature of the infused delivery method, the dosage and implementation mechanisms are different for the two program conditions. The standard LST curriculum, delivered in a planned sequence, consists of 15 lessons in the first year (seventh grade), 10 in the second year (eighth grade), and 5 (or 7 depending on time) in the third year (ninth grade). The core LST components consist of self-image and self-improvement; decision making; smoking, marijuana, and alcohol myths and realities; smoking and biofeedback; advertising awareness; coping with anxiety; communication skills; social skills; and assertiveness. The I-LST curriculum has no set number of lessons. Instead, the matrix as described above is completed to assure that all LST core components are taught to each student in at least one subject area.

Teachers were recruited in both the LST and I-LST conditions to participate as providers of the respective approaches. In the LST schools, teachers were trained in the LST approach by an approved LST trainer. In the I-LST schools teachers were trained by members of the ADAPT staff in LST principles. The comparison schools did minimal drug prevention education during the period of this study.

Implementation fidelity and student dosage was monitored using teacher self-rating forms that were filled out after every lesson delivered. These forms indicated the lesson implemented (LST) or the topic covered (I-LST), as well as the list of students in attendance. Teachers also had at least one lesson videotaped and/or observed, and were provided technical assistance by a member of the research staff on an as-needed basis. These data collected indicated that in the schools in LST condition, an average of 81% of the lessons were delivered. In the I-LST condition, 84% of the LST topics were covered. In all schools, the student-level dosage reflects the average attendance pattern, which averaged approximately 93% attendance per day.

Sample

The investigation utilized a prospective, group-randomized design in nine rural school districts in central Pennsylvania. Two main criteria were used to determine a school's eligibility for the study: (1) low socioeconomic status and (2) relatively small size (enrollment of less than 1,000). During the academic year 1999-2000, nine middle schools were randomly assigned to three conditions, three schools per condition: LST, I-LST, and comparison (C).

The impact of the programs on the students will be assessed through five student surveys. The first survey (T1) represents the pre-intervention survey at the beginning of seventh grade and the second survey (T2) was administered at the end of seventh grade. All subsequent surveys occur at the end of each academic year: 8th grade (T3), 9th grade (T4), and 10th grade (T5).

Sample Characteristics

This investigation utilizes survey data from T1, T2, and T3, and includes the flail sample of students for whom we have active parental and youth consent (N = 732). Sixty-eight percent of the students from the LST condition (n = 234), 71% of the students from the I-LST condition (n = 297), and 72% of the students from the control condition (n = 201) received parental permission and consented to complete the survey. The sample was 54.4% male and 96.6% white. There were some pretest differences across the three groups in free lunch eligibility (an SES indicator), substance use, and the incidence of problem behaviors (Swisher, Shute, & Bibeau, 1984); as a result, these variables were controlled in the analyses.

Protocol

Trained staff members administered all surveys in small classes with no school staff members (including teachers) in attendance. The staff members stressed the confidential nature of the survey to the participants orally at the time of data collection. In addition, unique identification codes were utilized in place of names.

Measures

Substance Use. Students' self-report of cigarette, smokeless tobacco, alcohol (including frequency of use, frequency of drunkenness and frequency of binge drinking), marijuana, and inhalant use was collected at all three surveys. The item responses include 1 (never), 2 (a few times, but not in the last year), 3 (a few times per year, 4 (once a month), 5 (a few times per month), 6 (once per week), 7 (a few times per week), 8 (once per day), and 9 (more than once per day). The questions were taken from measures created by Borvin et al. (1989). For analyses, those students responding with one or two were combined. A value of 1 was subtracted from each subject's response in order to allow the intercept of the tested models to be more meaningful.

The answer categories one and two were combined due to the inconsistent responses that students gave across the three time points. For example, a few students indicated at the end of the 8th grade that they had "never" drunk alcohol, but at the beginning of the 7th grade indicated "a few times but not in the past year." Similarly, a few students indicated at the end of the 8th grade that they had drunk alcohol "a few times but not in the past year," but at the end of the 7th grade indicated "never." In both instances, these answers, if interpreted literally, are illegal responses. These inconsistent responses had to be reconciled by either recoding one of the responses or combining the two answer categories. By combining the two answer categories, these analyses effectively define users of all substances to be those who have used in the past year. The number of students to be recoded in this manner was 17; analyses conducted without these students did not change the results.

Missing Data

Of the 732 students who completed surveys, 712 students participated in the survey at T1,704 students at T2, and 659 at T3. Nineteen students were removed from the data at T3 due to suspicious answering patterns, patterned responses, and inconsistent answering. The majority of other missing data were due to attrition or absence, with little within-survey missing data. In order to obtain unbiased and efficient parameter estimates, multiple imputations were used to create the data sets used to test for program effects. This imputation procedure increases the likelihood that data meet the assumption of mission at random (MAR) and reduces bias due to missing not at random (MNAR) or non-ignorable missing data patterns (Collins, Schafer, & Dam, 2001). Therefore, in addition to the variables of substantive interest, two additional variables (academic performance and religious participation) were included in the missing data model to predict the missing values. In order to model differences by school and the interaction between gender and condition, dummy variables to represent the nine schools and the interaction of the nine schools with gender were included in the missing data model. The imputation was completed using NORM version 2.03 (Schafer, 1997). The EM algorithm took 21 iterations to converge. In total, 10 imputed sets were created and analyzed, allowing 100 iterations between each imputation. Data augmentation diagnostics indicated a successful imputation. All analyses were performed on each of the imputed data sets. The parameter estimates were then combined using the procedures outlined by Rubin (1987), resulting in unbiased estimates of the population values.

Analyses

Combining data across schools within condition raises some analytic issues (Palmer, Graham, White, & Hansen, 1998) but the number of schools (n = 9) in this analysis was inadequate to appropriately assess level two variance. Multilevel analysis with under 10 groups is not recommended as the model parameters tend not to be reliable both in theory (Snijders, & Bosker, 1999) and in simulation studies (Maas, & Hox, 2002). Nevertheless, we initially analyzed our data using multilevel methods and found that the estimate of the intraclass correlation was negligible (potentially due to both low power and a highly homogeneous sample across schools). Given the recent evidence on the minimal numbers of level two units, the analysis for this study utilized a fixed effects analysis of covariance regression model with maximum likelihood estimation procedures.

Path models were specified using Mplus (Muthen, & Muthen, 1998) to represent the main effect of LST and I-LST on the T2 and T3 dependent variables of interest after controlling for baseline substance use, SES, and T1 problem behaviors. Due to the non-normal distribution of the substance use variables, robust standard errors were used. Based on the literature suggesting differential treatment effects by gender, an a priori decision was made to test the models separately for males and females. In addition, a second set of path models were specified for each of the T2 and T3 dependent variables that included two interaction terms representing the differential effects of the treatment programs by gender. All significance tests were two-tailed.

RESULTS

Separate path models for males and females were first performed to assess the effect of each of the treatment groups, compared to the control group, on substance use at the end of year one (T2) and at the end of year two (T3) while controlling for baseline substance use, SES, and problem behavior at T1. Tables 1 and 2 list the coefficients associated with these analyses. In these analyses, the degrees of freedom vary due to differences in the rate of missing information for each of the parameters (Schafer, & Graham, 2002). A higher number of degrees of freedom is indicative of a lower rate of missing information.

At the end of the first year of programming, the treatments demonstrated no significant impact on the males (Table 1). None of the substance use variables were significantly impacted. However, significant results were observed at the end of year one for the females (see bolded values in the rows for LST and I-LST in Table 2). The LST program significantly reduced alcohol use, binge drinking, marijuana use, and inhalant use. Furthermore, separate analyses (see shading) indicated that the LST treatment by gender interaction for smoking (B = .546, t = 1.99), binge drinking, ([beta] = .357, t = 2.65), and marijuana use ([beta] = .390, t = 2.61) was statistically significant (i.e., the effect for females was significantly greater than the effect for males). The I-LST program also resulted in significantly less smoking, binge drinking and marijuana use among the females. In addition, the I-LST treatment by gender interaction was significant for smoking ([beta] = .250, t = 2.43). By the end of year two, the males remained unaffected by the programs. All effects were lost for I-LST females and only one effect remained for the I-LST females. The I-LST females maintained their significantly lower use of cigarettes as compared to the control females. In the context of Cohen's (1988) consideration of effect size, all of the significant effects may be considered as between small (below .20) and medium (below .50) effects.

DISCUSSION

The results presented here reflect only the first two years of this trial. Both the LST and the I-LST approaches had a moderate positive influence on the substance use patterns of participating females by the end of the first year of programming, but these effects attenuate by the end of the eighth grade. Among the females exposed to either LST teaching method, alcohol consumption, binge drinking, and marijuana use are all lower than that experienced by the comparison females. The standard LST approach also influenced the use of inhalants. There was no impact on males in either of the intervention groups at either time point.

These program effects suggest that either approach is moderately successful with white, rural females during the seventh grade. The review conducted by Blake and colleagues (Blake et al., 2001) indicates that this gender finding is not unique, and that most programs have a stronger effect on females. Previous research on the LST program using a non-rural sample also indicated that the early effects of the program on girls were stronger than the effects on boys (Botvin, Baker, Filazzola et al., 1990).

By the end of the second year, program effects on substance use were almost non-existent. There remained no effects on males, and only smoking among the I-LST females was lower than that reported by the comparison females. The reduction in program effects on females is disappointing, as is the lack of any significant effects on males at either the end of Year One or Year Two. Baseline differences across the three conditions, while controlled for in the analyses, do suggest that the males in the I-LST treatment may have been a more problematic group at the beginning, perhaps making it more difficult for the program to influence their behavior. This was not true, however, for the LST males. The small rural schools in this study housed junior high students with high school students and possibly provided pressures for early socialization. Consequently, it is possible that the females who benefited from both approaches to LST had the opportunity and need to use life skills.

This study, like many that deal with interventions in school environments, faces a number of limitations that need to be considered. First, while most studies rely on self-report data, the validity of these reports can always be questioned. Our data collection procedures attempted to maximize the students' perception of the confidentiality of the survey. Second, the requirement of active parental consent increases the probability that the highest risk students did not participate in the survey. In this study, a comparison of the characteristics of participants versus non-participants, using school-based data, indicates that this bias did exist (Henry, Smith, & Hopkins, 2002). This selection bias, however, was consistent across conditions and therefore had a likely minimal effect on the differences observed. A third limitation of the current study is the inability to account for school effects in the analysis. We are aware of the issues posed by pooling student data across schools (see Palmer et al., 1998), however the random coefficient approach in this analysis is not an appropriate choice due to an inadequate number of schools (n = 9) to appropriately assess level 2 variance.

It should be noted that the analyses conducted in this research included all of the students, regardless of substance use status at pretest; as a result, the focus here was on change in use. We have conducted the same analyses reported here and omitted those youth who were users at pretest, following the strategy used with Project Northland (Perry et al., 1996). The rationale for excluding users is that LST is a prevention program and it was not designed as a cessation or intervention program. When the T1 users are omitted, a somewhat more promising picture emerges. While there are still no effects on boys, the effects on girls at the end of the second year indicates a positive LST effect on smoking and a positive I-LST effect on smoking, drunkenness, and marijuana use. However, these three positive findings were out of twenty-six possible findings and should be interpreted conservatively.

The differences in the conclusions reached, based on whether to include or exclude users at pretest, reinforce the need to reanalyze these program effects as these youth get older. A careful analysis of the actual transition patterns suggest that the substance use reported here involves movement that is predominately at the lower ends of the substance use scales. This suggests that the data and program effects (and non-effects) reported here are mostly confined to early experimentation. The real test will be whether or not these programs can effect the rate of substance use in future years, when the onset of substance use begins to escalate.

If the I-LST strategy is able to show some positive effects at the end of this real (data will be collected through the end of the 10th grade), this approach may prove to provide another model for adoption of effective prevention programs. From the perspective of program implementation, our experience with these relatively small rural school systems suggests that either of these approaches to implementation are acceptable. The standard LST approach may be the most efficient path in school districts where a dedicated teacher and curriculum flexibility are available. However, for schools where these conditions are not present, the infused approach provides an alternative delivery mechanism.

Regardless of the approach, our experience, and that of Botvin, is that the provision of these programs requires follow-up training for teachers. We interviewed the seventh grade teachers of both LST and I-LST approximately twenty months after they had received training (i.e., when the study cohort students were in the eighth grade); it was clear that a short refresher was needed. It should also be recognized that the I-LST strategy probably requires more up-front investment, due to the need to coordinate the efforts of multiple teachers. Conversely, however, the loss of one teacher does not mean the loss of the program, which can be the case with LST in rural schools.

Table 1: Main treatment effects for males

                           Cigarettes

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept        -0.08       -0.65       1414
T1 Smoking        0.82*       5.01*     18756*
T1 Drinking      -0.14       -0.61        944
T1 Drunk          0.71        1.44         29
T1 Binge         -0.38       -1.79         49
T1 Marijuana     -0.36       -0.99         29
T1 Inhalants     -0.17       -0.57        167
T1Smk Tob.       -0.08       -0.85       1379
T1 Low SES        0.22        1.64       6011
TI Prb. Beh.      0.66*       2.76*      1661*
LST               0.27#       1.71#     10371#
I-LST             0.16#       1.08#     61205#

Time 3 (End of 8th Grade)
Intercept         0.57        2.63       2595
T1 Smoking        0.69*       2.98*     65193*
T1 Drinking      -0.19       -0.73        263
T1 Drunk          0.68        1.14         54
T1 Binge         -0.42       -1.53         56
T1 Marijuana     -0.07       -0.17         21
T1 Inhalants     -0.85*      -3.38*      1359*
T1Smk. Tob.       0.02        0.13       1890
T1 Low SES        0.36        1.69        610
T1 Prb. Beh.      0.66*       2.26*       664*
LST               0.11        0.39       4282
I-LST             0.17       -0.72       2602

                            Alcohol

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.31        2.96       46018
T1 Smoking        0.10        1.58        4131
T1 Drinking       0.79        2.27       10723
T1 Drunk         -0.21       -0.35         195
T1 Binge          0.17        0.56         444
T1 Marijuana     -0.25       -0.59         145
T1 Inhalants      1.20*       2.18*        176*
T1Smk Tob.        0.13        1.51        1670
T1 Low SES       -0.09       -1.07        4905
TI Prb. Beh.      0.16        1.07        4365
LST              -0.06       -0.47       12853
I-LST            -0.07       -0.61      299833

Time 3 (End of 8th Grade)
Intercept         0.81        4.85        2312
T1 Smoking       -0.06       -0.74         530
T1 Drinking       0.80        2.10          82
T1 Drunk         -0.44       -0.55         482
T1 Binge         -0.02       -0.05         298
T1 Marijuana     -0.07       -0.13         337
T1 Inhalants      0.58        1.67         476
T1Smk. Tob.       0.12        1.12        5034
T1 Low SES        0.09       -0.63         261
T1 Prb. Beh.      0.20        1.01         506
LST              -0.21       -1.28        3180
I-LST             0.03        0.20        1037

                          Drunkenness

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.16        1.53       82136
T1 Smoking        0.12        1.75       17460
T1 Drinking       0.26        1.37       43564
T1 Drunk          0.43        0.92         389
T1 Binge          0.11        0.52         774
T1 Marijuana     -0.26       -0.74         232
T1 Inhalants      0.13        0.44       12978
T1Smk Tob.        0.13        1.52       20270
T1 Low SES       -0.12       -1.67        8051
TI Prb. Beh.      0.10        0.67        6814
LST              -0.03       -0.21      457915
I-LST            -0.08       -0.82      452782

Time 3 (End of 8th Grade)
Intercept         0.28        3.24         912
T1 Smoking        0.03        0.37        2658
T1 Drinking       0.47*       2.23*        396*
T1 Drunk         -0.50       -0.81         131
T1 Binge          0.01        0.04         168
T1 Marijuana      0.29        0.61          92
T1 Inhalants      0.62        1.49       20522
T1Smk. Tob.       0.16        1.60        3778
T1 Low SES       -0.04       -0.39         406
T1 Prb. Beh.     -0.04       -0.32         258
LST               0.04        0.39        8407
I-LST             0.08        0.79        2772

                         Binge Drinking

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.04        1.00         737
T1 Smoking        0.04        0.79        3007
T1 Drinking       0.15        1.02       10459
T1 Drunk          0.43        0.85          82
T1 Binge          0.31        1.12         381
T1 Marijuana     -0.26       -0.70          63
T1 Inhalants      0.19        0.45      319970
T1Smk Tob.        0.11        1.33        3209
T1 Low SES       -0.10*      -2.10*       3512*
TI Prb. Beh.      0.09        0.78         855
LST               0.05#       0.79#      74510#
I-LST             0.00       -0.01       23949

Time 3 (End of 8th Grade)
Intercept         0.24        2.31         820
T1 Smoking        0.05        0.62        1260
T1 Drinking       0.24        1.26         889
T1 Drunk         -0.36       -0.54         182
T1 Binge          0.17        0.58         399
T1 Marijuana      0.25        0.46         225
T1 Inhalants      0.84*       2.16*       4547*
T1Smk. Tob.       0.16        1.55        4812
T1 Low SES        0.08        0.86        1366
T1 Prb. Beh.      0.00       -0.01        1218
LST              -0.06       -0.55        3667
I-LST             0.04        0.32        3480

                            Marijuana

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept        -0.03       -0.52        5600
T1 Smoking        0.12*       2.46*      10851*
T1 Drinking       0.17        0.93       47875
T1 Drunk          0.08       -0.25          78
T1 Binge         -0.24       -1.67         139
T1 Marijuana      0.59*       2.70*         48*
T1 Inhalants     -0.19       -1.17        2168
T1Smk Tob.        0.00       -0.96        3714
T1 Low SES        0.00        0.04       65495
TI Prb. Beh.      0.28        1.93        7257
LST               0.06#       0.69#      99894#
I-LST             0.00        0.00       77927

Time 3 (End of 8th Grade)
Intercept         0.23        1.80        5336
T1 Smoking        0.08        0.91        4590
T1 Drinking      -0.03       -0.21       22143
T1 Drunk         -0.32       -0.67          85
T1 Binge          0.02        0.07         188
T1 Marijuana      0.71*       2.28*         32*
T1 Inhalants      0.09        0.20        1921
T1Smk. Tob.       0.09        0.79        7240
T1 Low SES        0.13        1.16       29328
T1 Prb. Beh.      0.22        1.26        1108
LST              -0.09       -0.65        7367
I-LST             0.05        0.31       40725

                           Inhalants

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.06        1.03      200000
T1 Smoking       -0.01       -0.34        1695
T1 Drinking       0.11        0.70        4975
T1 Drunk         -0.39       -1.45        5809
T1 Binge          0.62*       5.85*       2171*
T1 Marijuana     -0.39*      -2.79*       1363*
T1 Inhalants      0.52        1.35       28140
T1Smk Tob.       -0.06*      -2.32*       2882*
T1 Low SES        0.00       -0.06      131313
TI Prb. Beh.      0.17*       2.06*      25031*
LST               0.01        0.11      427209
I-LST            -0.09       -1.28      330959

Time 3 (End of 8th Grade)
Intercept         0.02        0.54         833
T1 Smoking       -0.02       -0.50       21033
T1 Drinking      -0.02       -0.30        2950
T1 Drunk          0.12        0.70        5457
T1 Binge         -0.04       -0.56         634
T1 Marijuana     -0.15       -1.10        3723
T1 Inhalants     -0.12       -1.87        2209
T1Smk. Tob.       0.09        1.48      190568
T1 Low SES        0.05        1.19       14422
T1 Prb. Beh.      0.05        0.72         643
LST              -0.02       -0.56        1090
I-LST             0.09        1.86        6899

                        Smokeless Tobacco

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.28        1.58        3441
T1 Smoking        0.16        1.35      243658
T1 Drinking       0.21        0.74       11673
T1 Drunk          0.44        0.81         114
T1 Binge         -0.56*      -2.00*        313*
T1 Marijuana     -0.72*      -2.98*         35*
T1 Inhalants     -0.72*      -2.11*       4390*
T1Smk Tob.        0.72*       6.35*       4053
T1 Low SES       -0.04       -0.25        1563
TI Prb. Beh.      0.32        1.40        3108
LST              -0.09       -0.50       16538
I-LST             0.13        0.73       38890

Time 3 (End of 8th Grade)
Intercept         0.70        3.40        1826
T1 Smoking       -0.11       -1.32         788
T1 Drinking      -0.12       -0.47        1691
T1 Drunk         -0.28       -0.32         236
T1 Binge         -0.28       -0.86         246
T1 Marijuana      0.10        0.16         160
T1 Inhalants      0.79        1.04       57894
T1Smk. Tob.       0.88*       8.81*       1303*
T1 Low SES        0.08        0.43       12369
T1 Prb. Beh.      0.05        0.23         151
LST              -0.07       -0.33       39393
I-LST             0.07        0.35        1857

Note: Bolded values are significant at p [less than or equal to] .05
and indicate that the Time 1 variable is a predictor of the outcome
variable at the top of the column for either Time 2 or Time 3. The
degrees of freedom vary due to differences in the rate of missing
information for each of the parameters. Shading denotes a significant
(p [less than or equal to] .05) gender by treatment interaction when
comparing values between Table 1 and Table 2.

Note: Values indicated with * are significant at p [less than or
equal to] .05 and indicate that the Time 1 variable is a predictor
of the outcome variable at the top of the column for either Time 2
or Time 3.

Values indicated with # denotes a significant (p [less than or equal
to] .05) gender by treatment interaction when comparing values
between Table 1 and Table 2.

Table 2: Main treatment effects for females

                             Cigarettes

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.23        1.22       2294
T1 Smoking        0.74*       7.07*       166*
T1 Drinking       0.20        0.74        659
T1 Drunk         -0.02       -0.04         46
T1 Binge         -0.29       -0.95        544
T1 Marijuana     -0.47       -1.31       1411
T1 Inhalants      1.32        1.22       8473
T1 Low SES        0.49*       2.22*       430*
TI Prb. Beh.      1.30*       3.23*      1693*
LST              -0.14#      -0.64#      2598#
I-LST            -0.40#      -2.01#      3744#

Time 3 (End of 8th Grade)
Intercept         0.72        3.11       1777
T1 Smoking        0.49*       3.70*       148*
T1 Drinking       0.85*       2.25*       154*
T1 Drunk         -0.96       -1.72         22
T1 Binge          0.06        0.17         82
T1 Marijuana      0.04        0.14        141
T1 Inhalants     -1.40       -1.12        409
T1 Low SES        0.96*       3.41*      1447*
T1 Prb. Beh.      0.66        1.45        263
LST              -0.46       -1.58        828
I-LST            -0.61*      -2.39*     10342*

                            Alcohol

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.46        3.37       1461
T1 Smoking       -0.11       -1.66        888
T1 Drinking       0.76*       2.78*       515*
T1 Drunk         -0.17       -0.44        172
T1 Binge          0.07        0.24       6629
T1 Marijuana      0.53*       3.22*       346*
T1 Inhalants      1.29*       4.98*      1501*
T1 Low SES       -0.05       -0.44        441
TI Prb. Beh.      0.87*       2.74*       700*
LST              -0.33*      -2.30*      2796*
I-LST            -0.23       -1.54       3234

Time 3 (End of 8th Grade)
Intercept         0.52        4.20       1176
T1 Smoking        0.08        0.74        103
T1 Drinking       0.85*       3.62*       205*
T1 Drunk         -0.64       -1.54        120
T1 Binge         -0.11       -0.38        163
T1 Marijuana     -0.12       -0.63        193
T1 Inhalants      1.27*       2.27*        57*
T1 Low SES        0.00       -0.01       3095
T1 Prb. Beh.      0.61        1.81        185
LST              -0.02       -0.15       1503
I-LST            -0.13       -0.94       1086

                          Drunkenness

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.12        1.33       4919
T1 Smoking        0.03        0.73        286
T1 Drinking       0.16        0.78      12630
T1 Drunk          0.32        1.00         88
T1 Binge         -0.15       -0.68       4435
T1 Marijuana      0.68*       2.62*     17641*
T1 Inhalants      1.07*       2.45*     93647*
T1 Low SES       -0.08       -1.08       2123
TI Prb. Beh.      0.45        1.93       1095
LST              -0.14       -1.69       4968
I-LST            -0.08       -0.72       8649

Time 3 (End of 8th Grade)
Intercept         0.16        1.71        903
T1 Smoking        0.07        0.85        106
T1 Drinking       0.66*       3.05*      2172*
T1 Drunk         -0.52       -1.56         51
T1 Binge         -0.35       -1.38        107
T1 Marijuana      0.07        0.42        185
T1 Inhalants     -0.67       -1.80       1991
T1 Low SES        0.02        0.24        509
T1 Prb. Beh.      0.44        1.49        307
LST              -0.02       -0.14       2498
I-LST            -0.07       -0.69       1299

                         Binge Drinking

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.19        1.86       3896
T1 Smoking       -0.04       -0.97        390
T1 Drinking       0.16        0.85        625
T1 Drunk          0.16        0.60         67
T1 Binge         -0.01       -0.07       2162
T1 Marijuana      0.72*       3.08*      4056*
T1 Inhalants      0.57        1.77      29696
T1 Low SES       -0.09       -1.16       3029
TI Prb. Beh.      0.66*       2.22*      2913*
LST              -0.26#      -2.60#      3614#
I-LST            -0.25*      -2.11*     19458*

Time 3 (End of 8th Grade)
Intercept         0.17        1.92        824
T1 Smoking        0.01        0.19        117
T1 Drinking       0.22        1.45        190
T1 Drunk         -0.29       -0.88         50
T1 Binge          0.35        1.34        630
T1 Marijuana      0.03        0.19        225
T1 Inhalants     -0.35       -0.81        143
T1 Low SES        0.07        0.74        455
T1 Prb. Beh.      0.39        1.42        781
LST              -0.09       -0.91       1346
I-LST            -0.09       -1.01        334

                            Marijuana

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.13        1.30       8881
T1 Smoking        0.00       -0.04     119867
T1 Drinking       0.32        1.33       5651
T1 Drunk         -0.37       -1.14       4923
T1 Binge         -0.03       -0.10      49377
T1 Marijuana      0.59*       2.07*     50346*
T1 Inhalants      1.55*       4.58*    209911*
T1 Low SES        0.03        0.39      10945
TI Prb. Beh.      0.56        1.71      12265
LST              -0.24#      -2.41#      4439#
I-LST            -0.26*      -2.18*     10889*

Time 3 (End of 8th Grade)
Intercept         0.12        1.34        783
T1 Smoking        0.06        0.98        175
T1 Drinking       0.18        0.93        631
T1 Drunk         -0.12       -0.43        146
T1 Binge          0.06        0.25       1074
T1 Marijuana      0.14        0.63       1729
T1 Inhalants     -0.23       -0.38        406
T1 Low SES        0.17        1.88       3314
T1 Prb. Beh.      0.40        1.57        219
LST              -0.12       -1.09        690
I-LST            -0.17       -1.71        503

                           Inhalants

                [beta]     t-value       df

Time 2 (End of 7th Grade)
Intercept         0.04        0.62        9388
T1 Smoking        0.04        1.13       77052
T1 Drinking       0.40*       2.04*     303739*
T1 Drunk         -0.65*      -2.41*      67734*
T1 Binge         -0.17       -0.89      280660
T1 Marijuana      0.63*       2.06*    2053222*
T1 Inhalants      1.33*       3.09*    4292196*
T1 Low SES        0.02        0.49       11375
TI Prb. Beh.      0.04        0.22       94582
LST              -0.11*      -2.36*       4290*
I-LST            -0.05       -0.75       10010

Time 3 (End of 8th Grade)
Intercept         0.00       -0.06        5133
T1 Smoking        0.04        0.61       40038
T1 Drinking      -0.03       -0.27       50521
T1 Drunk          0.05        0.47         397
T1 Binge         -0.09       -1.62         197
T1 Marijuana     -0.12       -1.40       13570
T1 Inhalants     -0.20       -1.39        1810
T1 Low SES        0.00        0.03       11444
T1 Prb. Beh.      0.21        1.07       43601
LST               0.04        0.59       14625
I-LST             0.08        1.88        1485

Note: Bolded values are significant at p [less than or equal to] .05
and indicate that the Time 1 variable is a predictor of the outcome
variable at the top of the column for either Time 2 or Time 3. The
degrees of freedom vary due to differences in the rate of missing
information for each of the parameters. Shading denotes a significant
(p [less than or equal to] .05) gender by treatment interaction when
comparing values between Table 1 and Table 2.

Note: Values indicated with * are significant at p [less than or
equal to] .05 and indicate that the Time 1 variable is a predictor
of the outcome variable at the top of the column for either Time 2 or
Time 3.

Values indicated with # denotes a significant (p [less than or
equal to] .05) gender by treatment interaction when comparing values
between Table 1 and Table 2.

References

Amaro, H., Blake, S.M., Schwartz, P.M., & Flinchbaugh, L.J. (2001). Developing theory-based substance abuse prevention programs for young adolescent girls. Journal of Early Adolescence, 21(3), 256-293.

Beane, J.A. (1995), Curriculum integration and the disciplines of knowledge. Phi Delta Kappa, 76, 616-662.

Blake, S.M., Amaro, H., Schwartz, P.M., & Flinchbaugh, L.J. (2001). A review of substance abuse prevention interventions for young adolescent girls. Journal of Early Adolescence, 21 (3), 294-324.

Botvin, G.J., Baker, E., Dusenbury, L., Botvin, E.M., & Diaz, T. (1995). Long-term follow-up results of a randomized drug abuse prevention trial in a White middle-class population. Journal of the American Medical Association, 273, 1106-1112.

Botvin, G.J., Baker, E., Dusenbury, L., Tortu, S., & Botvin, E.M. (1990). Preventing adolescent drug abuse through a multi-model cognitive-behavioral approach: Results of a 3-year study. Journal of Counseling and Clinical Psychology, 58, 437-446.

Botvin, G.J., Baker, E., Filazzola, A.D., & Botvin, E.M. (1990). A cognitive behavioral approach to substance abuse prevention: One-year follow-up. Addictive Behaviors, 15, 47-63.

Botvin, G.J., Batson, H.W., Witts-Vitale, S., Bess, V., Baker, E., & Dusenbury, L. (1989). A psychosocial approach to smoking prevention for urban black youth. Public Health Reports, 104, 573-582.

Botvin, G.J., & Botvin, E.M. (1997). School-based programs. In J. Lowinson, P. Ruiz, R.B. Millman, & J. Langrod (Eds.), Substance abuse: A comprehensive textbook (3rd Ed.). Baltimore: Williams & Wilkins.

Center on Addiction and Substance Abuse (2000). No Place to Hide: Substance Abuse in Mid-size Cities and Rural America. New York, NY: Columbia University.

Cinelli, B., Rose-Colley, M., & Bechtel, L.J. (1995). A strategy for the facilitation of curriculum integration. The Health Educator, 26(2), 27-31.

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Hillsdale, NJ: Lawrence Earlbaum Associates.

Collins, L.M., Schafer, J.L., & Kam, C. (2001). A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological Methods, 6(4), 330-351.

Ellickson, P.L., & Bell, R.M. (1990). Drug prevention in junior high: A multi-site longitudinal test. Science, 247, 1299-1305.

Ennet, S., Ringwalt, C., Thorne, J., Rohrbach, L., Vincus, A., Simons-Rudolph, A. et al. (2002, May). What's the real world implementation of evidence-based substance use prevention programs? Paper presented at the 10th annual meeting of the Society for Prevention Research. Seattle, WA.

Gatewood, T. (1998). Integrated curriculum in today's middle schools? Education Digest, 63, 24-28.

Gorman, D.M. (1998). The irrelevance of evidence in the development of school-based drug prevention policy, 1986-1996. Evaluation Review, 22(1), 118-146.

Hansen, W.B. (1992). School-based substance prevention: A review of the state-of-the-art in curriculum. Health Education Research, 7, 403-430.

Hansen, W.B. (1996). Pilot test results comparing the All Stars Program with seventh grade DARE program integrity and mediating variable analysis. Substance Use & Misuse, 31(10), 1359-1377

Hansen, W.B., Johnson, C.A., Flay, B.R., Graham, J.W., & Sobel, J. (1988). Affective and social influences approaches to the prevention of multiple substance abuse among seventh grade students: Results from project SMART. Preventive Medicine, 17, 1-20.

Henry, K.L., Smith, E.A., & Hopkins, A.M. (2002). The effect of active parental consent on the ability to generalize the results of an alcohol, tobacco, and other drug prevention trial to rural adolescents. Evaluation Review, 26(6), 645-655.

Jacobs, H.H. (1989). Interdisciplinary curriculum: Design and implementation. Alexandria, VA: Association for Supervision and Curriculum Development.

Johnston, L.D., O'Malley, P.M., & Bachman, J.G. (2000). National Survey Results on Drug Use From Monitoring the Future Study. Rockville, MD: National Institute on Drug Abuse.

Johnston, L.D., O'Malley, P.M., & Bachman, J.G. (2001). National Survey Results on Drug Use From Monitoring the Future Study. Rockville, MD: National Institute on Drug Abuse.

Karshin, C., Henry, K., Luppino, C., Swisher, J., Vicary, J., Bechtel, L., et al. (2000). Integrating prevention into education practice. In W.B. Hansen, S.M. Giles, & M.D. Fearnow-Kennedy (Eds.), Increasing Prevention Practice (pp. 281-291). Greensboro, NC: Tanglewood Research.

Maas, C., & Hox, J. (2002). Sample sizes for multilevel modeling. In J. Blasius, J. Hox, E. de Leeuw, & P. Schmidt (Eds.), Social science methodology in the new millennium. Proceedings of the Fifth International Conference on Logic and Methodology. Second expanded edition. Opladen, RG: Leske+Budrich Verlag (CD-ROM).

Muthen, L.K., & Muthen, B.O. (1998). Mplus User's Guide. Los Angeles, CA: Muthen & Muthen.

Palmer, R., Graham, J., White, E., & Hansen, W.B. (1998). The application of multilevel analytic strategies to assess substance use prevention effectiveness among adolescents. Preventive Medicine, 27, 328-336.

Pentz, M.A., Dwyer, H.H., MacKinnon, D.P., Flay, B.R., Hansen, W.B., Wang, E.Y.I. (1989). A multi-community trial for primary prevention of adolescent drug abuse: Effects on drug use prevalence. Journal of the American Medical Association, 261, 3259-3266.

Perry, C.L., Williams, C.L., Veblen-Mortenson, S., Toomey, T.L., Komro, K.A., Anstine, P.S., et al. (1996). Project Northland: Outcomes of a community-wide alcohol use prevention program during early adolescence. American Journal of Pubic Health, 86(7), 956-964.

Rubin, D.B. (1987). Multiple Imputation for Non-response in Surveys. New York: J. Wiley & Sons.

Schafer, J.L. (1997). Analysis of Incomplete Multivariate Data. New York: Chapman and Hall.

Schafer, J.L., & Graham, J.W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7, 147-177.

Snijders, T.A.B., & Bosker, R.J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling (pp. 42-45). London: Sage Publications, Inc.

Swisher, J.D. (2001, June). State Incentive Grant Program: National Cross-Site Evaluation. Presentation at NASADAD/NPN Annual Conference, New Orleans, LA.

Swisher, J.D., Bechtel, LJ., Henry, K.L, Vicary, J.R., & Smith, E.A. (2001). A model substance abuse prevention program. In D.C. Locke, J.E. Myers & E.L. Herr (Eds.), The Handbook of Counseling (pp. 551-559). Thousand Oaks, CA: Sage Publications, Inc.

Swisher, J.D., Shute, R.E., & Bibeau, D. (1984). Assessing drug and alcohol abuse: An instrument for planning and evaluation. Measurement and Evaluation in Counseling and Development, 17(2), 91-97.

Tobler, N.S., & Stratton, H.H. (1997). Effectiveness of school-based drug prevention programs: A meta-analysis of the research. Journal of Primary Prevention, 18, 71-128.

Author Note

Edward A. Smith, John D. Swisher, and Judith IL Vicary, The Pennsylvania Stae University; Loft J. Bechtel, Penn State Altoona; Daphne Minner, Education Development Center, Inc.; Kimberly L. Henry, Colorado State University; Raymond Palmer, University of Texas.

Support for this research was provided by the National Institute on Drug Abuse (Grant DA11254). The authors would like to thank the school districts and students who participated in this research, as well as all members of the ADAPT Project staff.

Correspondence concerning this article should be addressed to Edward A. Smith, Prevention Research Center, 109 South Henderson, The Pennsylvania State University, University Park, Pennsylvania 16802, 814-865-1201, 814-865-2530 (fax), e-mail [e-mail]