Abstract
This study investigated the effectiveness of a life skills program with regard to alcohol consumption, life skills, knowledge, and school bonding for young adolescents. The focus was on the moderating role of gender, based on the assumption that life skills programs may address specific needs of adolescent girls better than those of boys. The universal school-based life skills program IPSY (Information * Psychosocial competence = Protection) was implemented and evaluated over 3 years (longitudinal quasi-experimental design with four measurement points; N = 952, 10 years at T1). Results of (M)ANCOVAs revealed positive general program effects of IPSY on alcohol-related outcomes, life skills, knowledge, and school bonding. As expected, girls profited regarding knowledge of effective communication within groups and self-confidence, but not boys. In contrast, none of the other program effects was moderated by gender. Thus, the program, although lacking gender-specific components, worked well for both boys and girls during early adolescence.
Introduction
Numerous prevention programs have attempted to influence alcohol consumption behaviors positively and to prevent negative developmental pathways in young people. The most often used approach has been to lower the impact of risk factors and increase the influence of protective factors on adolescent substance use (Jessor, Turbin, & Costa, 1998). In addition, prevention programs have aimed at supporting adolescents to resolve developmental tasks appropriate to their age (Weichold, Bühler, & Silbereisen, 2008). This approach is used particularly in universal prevention programs designed to be implemented before the emergence of alcohol use and developed for unselected population groups, such as coeducational school classes (Gordon, 1983).
School-based life skills programs have been shown to exert a positive influence on alcohol consumption patterns in young people (i.e., by delaying onset or by reduction of age-typical increases in frequency and amount), both short- and long-term (Bühler & Kröger, 2006; Tobler et al., 2000). Life skills programs focus on the promotion of general competences in the interpersonal (e.g., assertiveness, positive social relationships) and intrapersonal (e.g., critical thinking, self-esteem) domain. Life skills training, in combination with long-term reinforcement of positive behaviors and the acquisition of substance-specific skills and knowledge, has been shown to enhance young people’s health (World Health Organization [WHO], 1997).
Meta-analytic findings imply that effect sizes of school-based life skills programs with regard to substance use outcomes have been rather small (Tobler et al., 2000). This might be due to moderator variables, such as characteristics of the study participants (e.g., prior consumption experiences, risk factors). For instance, it was speculated that a universal prevention program might not work equally for both genders and that selective program components aimed explicitly at gender-specific risk and protective factors may be more appropriate (Amaro, Blake, Schwartz, & Flinchbaugh, 2001). In addition, few studies have investigated whether the effects of a universal school-based prevention program against alcohol consumption (based on the life skills approach) vary by gender (Cuijpers, 2002; Skara & Sussman, 2003). However, as previous findings show that there are differences in consumption patterns, expectations toward alcohol use, and risk factors for substance use, adolescent boys and girls may not profit equally from such a universal prevention strategy.
First, German adolescent males start drinking earlier, more frequently, and at higher dosages per drinking occasion than females (Bundeszentrale für gesundheitliche Aufklärung [BZgA], 2007). This gender difference in alcohol consumption pattern is rather universal across different cultural backgrounds, but it is more pronounced in German adolescents as compared, for instance, with their age mates from the United States (ages 11 to 13; for weekly alcohol consumption and experiences of drunkenness, see Currie et al., 2004). Because of these marked differences between genders regarding alcohol consumption, universal prevention programs may have limited effectiveness for boys as compared with girls. This may also be true because the content of such programs is often designed to be implemented before the emergence of experimental or regular use, whereas a larger share of boys as compared with girls will already have started consuming alcohol. Thus, boys may need earlier intervention strategies.
Second, it is known that girls and boys differ in their expectations toward and reasons for consuming alcohol. Females in adolescence and adulthood generally expect more negative social reactions and consequences for being drunk (Nolen-Hoeksema & Hilt, 2006), perceive higher peer pressure not to drink than boys, and avoid excessive alcohol consumption, because it carries greater social stigma (Amaro et al., 2001). If adolescent girls drink for other than social reasons, they do so as a means of solving emotional problems, whereas boys see alcohol consumption as acting according to perceived norms and stereotypical behavior shared by their male age-mates (Nolen-Hoeksema & Hilt, 2006). For male adolescents, drinking alcohol is more likely to be an expression of rebelliousness, an attempt to achieve high peer status, a normative initiation ritual expected by others, or a means to approach potential romantic partners (i.e., social and enhancement motives; Kuntsche, Knibbe, Gmel, & Engels, 2006; Zenker, 2009). As male adolescents generally hold positive and more normative expectations toward alcohol consumption, universal prevention strategies may be not as effective for them as for their female counterparts. Thus, boys may need different content and/or a more intense program (e.g., additional program sessions).
Third, girls and boys show gender-typical problems and copings strategies, which emerge and intensify during adolescence (Hill & Lynch, 1983). Such differences are, for example, reflected in a higher vulnerability to stress and in maladaptive coping strategies among girls as compared with boys (e.g., Hampel & Petermann, 2005). In addition, developmental tasks associated with adolescence differ by gender. Among girls, accomplishment of individuation is relational and mastery is defined by range and quality of relationships, and in particular by popularity. For boys, however, mastery relates to physical and intellectual competencies (Miller, 1986). Moreover, girls experience significant declines in academic achievement, motivation, and self-esteem from early to midadolescence (Smaxwill, 2008). Finally, girls and boys have different styles of expressing distress. While girls tend to internalize stress, to attribute failure to personal deficits, and to develop such problems as depression or anxiety (e.g., Chaplin, Gillham, & Seligman, 2009), adolescent boys are more likely to attribute failure to a situation itself and react to stress by acting-out behaviors and externalization (Card, Stucky, Sawalani, & Little, 2008). Both externalizing problems (e.g., low self-control or inappropriate behaviors) and internalizing problems, such as seeing life as meaninglessness and stressful, were found to be gender-specific risk factors for substance use in mid- to late adolescence (e.g., Petraitis, Flay, & Miller, 1995; Zucker, Donovan, Masten, Mattson, & Moss, 2008). In this way, boys and girls seem to need different program content, that is, focusing on internalizing problems for girls and on externalizing behaviors for boys.
Despite the above mentioned gender differences in prevalence rates, expectations, and risk factors for adolescent alcohol use, the vast majority of universal programs developed to prevent adolescent consumption of legal substances do not contain gender-specific or gender-sensitive elements (Blake, Amaro, Schwartz, & Flinchbaugh, 2001) nor do they address possible gender differences in optimal intervention timing. In addition, the typical mode of delivery of life skills programs (i.e., coeducational classroom) may constitute structural constraints to the consideration of gender-typical issues. Due to their modes of facilitation and content, life skills programs, however, may be more appealing for girls than for boys. Life skills programs offer the opportunity to get into close contact and to interact with classmates and teachers, and to receive positive feedback. Moreover, life skills programs contain units designed to promote self-esteem and social skills, which are particularly important features for adolescent girls. The establishment of close bonds to same-aged peers and positive adult role models are vital to promote girls’ self-esteem and self-confidence (Amaro et al., 2001) and to reduce the likelihood of internalizing problems. Finally, the mode of presenting life skills programs (small group work, role play, etc. where teachers are asked to involve all students) should be particularly helpful for girls to promote social investment and to stimulate their integration into same-aged peer groups (O’Donnell, Hawkins, Catalano, Abbott, & Day, 1995). Thereby, life skills programs may work well against the age-typical phenomenon of early adolescent girls lapsing into silence and passivity within the classroom, especially when interacting with boys (Taylor, Gilligan, & Sullivan, 1995). Thus, because both the content and mode of facilitation of life skills programs may better fit girls’ gender-typical needs, higher program effectiveness may be assumed for girls in comparison with boys (at least with regard to intra- and interpersonal competencies).
Several meta-analyses explored the potential gender-specific effectiveness of universal programs combating adolescent alcohol consumption, such as life skills programs (Blake et al., 2001; Bühler & Kröger, 2006; Sussman, 2006). Based on an extensive literature search (covering 1980-2000), Blake and colleagues (2001) identified just 10 studies that investigated the gender-specific effectiveness of substance use prevention programs separately for boys and girls which reported significant gender differences. Of these intervention studies, only two focused on the prevention of alcohol consumption. The first study evaluated a school-based program aiming at the prevention of alcohol use via the training of social competence, coping, and decision making skills among fourth to fifth graders. In this study, after the intervention, only boys but not girls were found to reduce alcohol purchases (Elias, Gara, Schuyler, Branden-Muller, & Sayette, 1991). In the second study, a social development intervention (including components for social skills training, refusal skills training, and improving social bonding to conventional others, Grade 1 to Grade 6) was evaluated. Girls of the intervention group improved in school bonding, perceived more opportunities for social involvement, and also initiated alcohol use later than did controls. In contrast, while boys profited regarding social competencies, academic effort, and commitment to school, no desirable program effects on substance use were found (O’Donnell et al., 1995). In addition, Blake et al. (2001) identified 18 studies investigating treatment by gender interactions effects (without analyzing the subsamples of males and females separately). Only six of them revealed significant moderator effects of gender. Among these, the two studies including life skills programs and assessments of alcohol consumption also revealed mixed findings (Eiser & C. Eiser & R. Eiser, 1987, positive effects for girls; G. J. Botwin, Baker, Filazzola, & E. M. Botwin, 1990, positive effects for boys and girls). Consequently, intervention studies that focus on the gender-specific effectiveness of prevention programs against alcohol consumption are rare and inconclusive, and a large share of these studies revealed no gender differences at all. For most studies that did report gender-specific effects, results indicated slightly greater benefits for girls than boys. In addition, the majority of the studies focused on samples from mid-adolescence onward and few referred to cultural contexts other than the United States, thus underscoring the need for further research.
Against this backdrop, the current study aimed at investigating the general and gender-specific effectiveness of a universal school-based life skills program focusing on alcohol consumption in early adolescence (age 10 at T1) across 3 years in a German context. Thereby, the effects of the program on life skills, knowledge, school bonding (proximal outcomes), and alcohol use (distal outcomes) were examined. We hypothesized (a) that the life skills program promotes intra- and interpersonal life skills, general and substance-specific knowledge, and school bonding and that it has desirable effects on alcohol-related outcomes by lowering the age-typical increase in consumption frequency and positive expectations toward drinking alcohol on a regular basis (general intervention effectiveness). In addition, we hypothesized (b) that the effects vary by gender, such that higher program effectiveness on proximal and distal outcomes is expected for girls as compared with boys (moderation of effects by gender).
Method
Intervention
The life skills program IPSY (Information * Psychosocial competence = Protection; Weichold, 2002) is a comprehensive program for the prevention of the consumption of alcohol and tobacco in early adolescence. This program is based on WHO’s life skills approach (WHO, 1997) as well as on theories and empirical findings concerning the etiology of adolescent substance use and associated risk and protection factors (Weichold et al., 2008). IPSY is a universal prevention strategy that combines the promotion of generic, intra- and interpersonal life skills (e.g., communication skills, coping with anxiety and stress, problem solving) with the training of skills related to substance use (e.g., refusal skills). Moreover, the program transmits knowledge concerning alcohol and tobacco use (e.g., actual prevalence rates, short-term consequences of substance use, advertising strategies) as well as information on the structuring of leisure time. Furthermore, the program includes lessons explicitly focusing on school (e.g., targeting attitudes toward school).
The basic IPSY manual was designed for students in Grade 5 (age 10) and consists of 15 lessons of either 90- or 45-minutes’ duration (10 and 5 units, respectively) with two additional booster sessions, each of seven lessons, for Grades 6 and 7 (3 units of 90-minute and 4 units of 45-minutes’ duration for each year). The program was implemented in coeducational classroom settings by teachers who had participated in a 1-day facilitator training prior to the implementation of the program each year (i.e., in Grades 5, 6, and 7). Facilitation of the program in the classroom was also based on a detailed manual. IPSY applies interactive teaching methods (e.g., role-play, group discussions, etc.) that enable teachers and classmates to get to know each other and to establish close relationships. In addition, teachers that facilitate IPSY are trained explicitly to use positive reinforcement strategies and to apply resource-oriented teaching methods (for more information on the program, see Weichold, 2007).
Evaluation Study
Following a three-year pilot test utilizing a convenience sample, the program IPSY has been implemented and evaluated within a German large-scale longitudinal trial in the Federal State of Thuringia (in the former East Germany) since 2003. For this purpose, a quasi-experimental design (intervention/control group; pretest, posttest, and two follow ups) with school-wise assignment to the respective groups was used (principal investigators: Karina Weichold & Rainer K. Silbereisen). The study in Thuringia was approved by the Ministry of Culture and Education of the Federal State. Participation in the evaluation study and implementation of IPSY was offered to all schools in Thuringia, and 40 accepted the invitation to participate. This represents about 10% of all schools of the federal state. As schools receive similar invitations to take part in special programs on a regular basis, or may have been involved previously in such activities, and as there are no negative consequences for schools that do not agree to participate, the actual response rate is usually rather small. Reasons given for participation in the IPSY study varied and included factors such as a general interest in intervention programs, specific interest in alcohol consumption prevention in schools, and the possible enhancement of school prestige. Out of the 40 schools who agreed to participate, 23 were selected at random to become intervention schools for our evaluation study. The remaining schools (n = 17) were not recruited as control schools (i.e., waiting group) because they wanted the program to be implemented quickly. For this reason, control schools (n = 21) were recruited at random from the remaining schools of the federal state (i.e., those that had not actively indicated their interest in the IPSY study). To control for possible differences between intervention and control group, we investigated the schools’ involvement in earlier intervention efforts. With regard to the level of interest in school-based intervention programs, both intervention and control group schools had been involved in other interventions in the past to a similar degree (i.e., one third of the control and of the intervention schools had taken part in interventions in the past). Hence, they seem not to differ in their general interest in school-based interventions (for more details on the selection of the sample, see Wenzel, Weichold, & Silbereisen, 2009).
Teachers in the intervention schools were trained in IPSY facilitation and students of the intervention group participated in all three program parts (i.e., basic program and two booster sessions) from Grades 5 to 7. Prior to data collection, parents gave their active consent to study participation (only two refused). The data for outcome evaluation (life skills, knowledge, school-related variables, and substance use) were gathered from students via an anonymous questionnaire at each measurement point. This was done in the classroom and administered by project staff or trained teachers. The questionnaires were assigned to students by means of a coding system (i.e., day and month of birth, the last letter of the students’ first name, and first three letters of the mothers’ first name). At T1, project coworkers instructed all students how to build the code, and how to fill out the questionnaire, and were available to answer questions. At the following waves, instructors reminded the students how to build the personal code before they started to fill in the questionnaire, resulting in a low rate of invalid or incomplete codes (about 3% per wave). The personal code enabled us to match questionnaires across waves without breaching anonymity. Nevertheless, between 7% and 20% of the questionnaires could not be matched from one to the next assessment wave (reasons for dropout are described below).
Mean time interval between pretest (T1, beginning of Grade 5) and posttest after the implementation of IPSY basic manual in Grade 5 (T2, end of Grade 5) was 7 months. First follow-up data (T3) were collected 1 year after implementation of the booster session in Grade 6 (end of Grade 6). The second follow-up (T4) was conducted after the booster session in Grade 7 (end of Grade 7). Thus, the data forming the basis for this study span about three years of the adolescents’ development (ages 10 to 13).
Sample
The sample for this study comprised 952 students who took part in all four waves of data collection (longitudinal sample) and represents 56.23% of the original sample at T1 (N = 1,693). Main reasons for dropping of the study were school closure, or school or class amalgamation (two schools from the intervention group and two from the control). Entire classrooms of students dropped out of the study, for instance, if schools (or classes across schools) were merged and principals declined further participation in the study. Furthermore, a small number of students did not participate in each measurement point because of sickness, other reasons for absence, school changes, or because they had to repeat a class. Other grounds for dropping out of the study over time were noncompletion of the questionnaire or an invalid code. By means of attrition analyses, students of the longitudinal sample were compared with those who dropped out directly after pretest or at later measurement points (via t tests and χ2 tests, Bonferroni correction applied) regarding sociodemographic and outcome variables. Significant differences were found concerning age (p < .01; dropouts were older), living circumstances (p < .001; dropouts were more likely to live in rented apartments rather than their own houses), place of residence (p < .001; dropouts typically lived in urban rather than rural areas), and finally number of siblings (p < .01; on average, dropouts had more siblings). No differences were found regarding school type, gender, number of persons per household, marital status of parents, financial background, employment status of mothers or fathers, or assignment to intervention or control condition. Regarding outcome variables, dropouts knew less about adequate behavior within groups (p < .001), otherwise no differences were found. In addition, no differential attrition by treatment condition was found. As attrition was due to dropout of whole units for logistical reasons and not due to the missing of values on single items or scales, and since students who dropped out did not differ substantially from those who remained in the study, we did not use any missing imputation procedures.
From the longitudinal sample, 55.57% were in the intervention group and 44.43% in the control group; 54.31% of the students were female. Students came from college-bound school track (60.99%) or lower track schools (39.01%). At T1, students had a mean age of 10.44 years, and on average one sibling. Of the students, 59.15% reported living with their families in their own house; 56.41% lived in urban areas. The majority came from families where parents were married (71.44%) and reported a good to very good financial status of their families on average. In addition, 39.06% of the mothers and 68.45% of the fathers were employed full-time.
At T1, students of the intervention and control groups of the longitudinal sample were comparable regarding most sociodemographic variables. Differences were found only for place of residence, with students of the intervention group being more likely to come from rural than urban areas (p < .001). There were no differences regarding the other sociodemographic variables (school type, age, sex, number of siblings, financial status, marital status of parents, and employment status of parents) and concerning all outcome variables.
Boys and girls were also compared regarding sociodemographic and outcome variables at T1. No gender differences were found in sociodemographic variables, but boys reported a higher general self-worth (p < .001), less knowledge about adequate group behavior (p < .001), and lower bonding to school in comparison with girls (p < .001). Moreover, girls reported a lower frequency of beer (p < .001) and mixed drinks consumption during the past month (p < .01), and a lower likelihood to drink alcohol on a regular basis during the next year (p < .001). No gender differences were found at T1 concerning other life skills, susceptibility to peer pressure, frequency of consumption during the past 30 days of wine and spirits, and knowledge about alcohol.
Measures
This evaluation study is based on self-reports of adolescents across four measurement points. Besides sociodemographic information, data on intra- and interpersonal life skills, knowledge, reports on school bonding and classroom climate, substance-related knowledge, and alcohol use were gathered. Means and standard deviations for outcomes at T1, scale reliabilities (Cronbach’s alpha) from T1 to T4, example items, and answering format are summarized in Table 1 and Table 2.
Measurement Information on Scales Measuring Life Skills, Susceptibility to Peer Pressure, and School Bonding T1 to T4
Means and Standard Deviations of all Outcome Variables in Intervention and Control Groups From T1 to T4
Note: IPSY = Information * Psychosocial competence = Protection.
General intervention effects.
Effect emerged at T2.
Effect emerged at T3.
Effect emerged at T4.
Intrapersonal life skills
Intrapersonal life skills were assessed with four scales of the Frankfurter Selbstkonzeptskalen (Deusinger, 1986), a multidimensional measure to assess various aspects of self-concept. “Stability against groups” relates to cognitions and emotions of an individual in interaction with others (primarily peers, but also those in authority) and skills to communicate one’s own standpoint and attitudes effectively. “Appreciation through others” represents individual perceptions of appraisal and trust by family, peers, and others. “General self-worth” reflects self-appraisal, i.e., satisfaction with one’s self (also in comparison with peers). Finally, self concept of “problem solving skills” measures an individual’s attitudes toward dealing with problems and challenges of everyday living, tenacity, and optimism toward future life. Furthermore, two scales by Rhodes and Jason (1988) were used to assess psychosocial resources and deficits. Both scales capture various aspects of psychosocial skills as self-perceptions on self-esteem, assertiveness, resistance to peers, and more general communication skills (e.g., self-confidence).
Interpersonal life skills
Interpersonal life skills were measured as knowledge about adequate behavior within groups, knowledge about self-confident behavior and knowledge about effective communication. These measures were developed based on the content transferred to students in the IPSY program (Weichold, 2002). Knowledge about adequate behavior within groups was measured via an open answering format (“To which rules should one stick while discussing in a group? Write down at least five.”). Answers were declared as correct (maximum = 5) according to a list of keywords based on the content of the manual. A sum score for each person was calculated. Knowledge about self-confident behavior was assessed by six items asking students to categorize behavioral styles as aggressive (“not to be willing to make an agreement”), self-confident (“to look someone straight in the eye”), or insecure (“to give in quickly”). Right answers were summed. Knowledge concerning effective communication consisted of four items for rules regarding the listener (e.g., “one should show a friend that one is listening” [nodding, to catch someone’s eye], sum score right answers), and another four items for rules regarding the speaker (e.g., “one should speak loudly and clearly”, sum score right answers). Furthermore, we assessed susceptibility to peer pressure using a scale of Santor, Messervey, and Kusumakar (2000).
School bonding
Bonding to school was measured using a scale of Fend and Schur (1991).
Alcohol-related measures
Information was also collected on the frequency of drinking during the past month prior to data collection. This was done separately for different alcoholic beverages following a standard format used in Germany (BZgA, 2007). Moreover, expectations about regular future consumption were measured (“Will you drink alcohol regularly within the next 12 month?” 0 = definitely not to 3 = definitely). Finally, data were collected about knowledge regarding alcohol use that was conveyed during the intervention (three items, for example, “Alcohol is the most widespread drug”; sum-score of correct answers with 0 = no correct answer to 3 = three correct answers).
Data Analyses
Program effectiveness was investigated by (M)ANCOVA, (multivariate) analyses of covariance with repeated measurements. According to Rausch, Maxwell, and Kelley (2003), this is a more precise procedure to investigate intervention effects within pre-post-follow-up designs as compared with ANOVAs. Factors included in the analyses were time (T1 to T4), group (intervention and control group), and gender. Baseline scores (T1) of all outcome measures were used as covariates. Within this type of analysis, a significant main effect of group indicates differences in outcomes after program participation between the intervention and control group. Furthermore, a significant interaction effect of Group × Time indicates different changes in outcome variables over time between the two groups. Both main effect of group and interaction effect Group × Time can be indicators for program effectiveness (if greater positive changes can be observed in the intervention as compared with the control group).
In order to test for differences in program effectiveness (moderator analyses), we focused on interaction effects with gender within the same (M)ANCOVA. For significant intervention effects found to be moderated by gender, split file analyses were run separately for boys and girls so as to define the existence and direction of the effect within each of the subsamples (Baron & Kenny, 1986; Bortz & Döring, 1995). To analyze general and gender-specific effects within the same analyses, we conducted one MANCOVA including life skills, knowledge, and school bonding as outcomes, and a second with alcohol use variables as dependent variables. Finally, alcohol-related knowledge was analyzed within a separate ANCOVA. Effect sizes are reported for significant intervention effects according to the standards delineated by Derzon, Sale, Springer, and Brounstein (2005).
Table 3 summarizes the results of the univariate tests within MANCOVA including all main and interaction effects. Table 2 gives more detailed information on means and standard deviations of all outcome variables over time in intervention versus control group. Table 4 presents means and standard deviations of outcome variables with significant general intervention effects in intervention and control group over time by gender.
Summary of Results for General and Gender-Moderated Intervention Effects on Life Skills, Knowledge, and School Bonding, and Alcohol-Related Variables
Note: ME = main effect; IA = interaction effect.
General intervention effect.
Gender-moderated intervention effect.
Univariate tests within MANCOVA on life skills, knowledge, and school bonding.
ANCOVA on knowledge on alcohol.
Univariate tests within MANCOVA on alcohol-related variables.
Means and Standard Deviations of Outcome Variables With Significant General Intervention Effects in Intervention and Control Groups by Gender From T1 to T4
Gender-moderated intervention effect.
Results
Program Effectiveness With Regard to Life Skills, Knowledge, and School Bonding (Proximal Outcomes)
General intervention effectiveness
The multivariate analysis of covariance with repeated measurements (MANCOVA) with time as repeated measurement, group membership and gender as factors, T1-variables as covariates, and dependent variables of intra- and interpersonal life skills, knowledge, and school bonding (N = 721) revealed a significant multivariate main effect of time, F(24, 682) = 2.76, p < .001, and group, F(12, 694) = 3.88, p < .001. Time × Group interaction was not significant.
Univariate tests within MANCOVA (see Table 3) showed significant main effects of group for psychosocial resources, F(1, 705) = 7.46, p < .01, deficits in psychosocial skills, F(1, 705) = 6.88, p < .01, and self-concept on appreciation by others, F(1, 705) = 4.64, p < .05, indicating differences in means between the intervention and the control group after controlling for pretest scores. As depicted in Table 2, students of the intervention group reported higher psychosocial resources and lower deficit scores, and felt more appreciated by others than did controls. Regarding knowledge, significant main effects of group were found for knowledge about self-confident behavior, F(1, 704) = 21.53, p < .001 and knowledge about effective communication/listening, F(1, 705) = 7.60, p < .01. Students of the intervention group showed a higher average knowledge regarding self-confident behavior and effective communication than did controls (see Table 2). Finally, we found significant mean differences between groups concerning susceptibility to peer pressure, F(1, 705) = 6.36, p < .05, and school bonding, F(1, 705) = 15.83, p < .001. Students who had taken part in the IPSY program showed lower susceptibility to peer pressure and higher school bonding as compared with those in the control group. A significant Time × Group interaction was found only for knowledge about adequate behavior within groups, F(2, 1410) = 8.50, p < .001, indicating differences in change in knowledge between the intervention and control group over time (i.e., students of the intervention group showed a steeper increase in knowledge than did controls). Single comparisons indicated that significant mean differences between intervention and control group for life skills, knowledge, peer pressure, and school bonding emerged from T2 or T3 onward. Effect sizes of general program effects on these variables ranged from .05 to .27.
Gender-specific intervention effectiveness
We expected program effects on intra- and interpersonal life skills, knowledge, and school bonding to differ by gender, in that we posited more positive program effects for girls than boys. Within the same MANCOVA with life skills, knowledge, and school bonding as outcomes, a significant multivariate main effect of gender, F(12, 694) = 3.44, p < .001, was observed. The multivariate intervention effect of Group × Gender was also significant, F(12, 694) = 2.82, p < .05, but not the Time × Group × Gender interaction. Univariate tests within MANCOVA (see Table 3) revealed significant Group × Gender interactions for knowledge about adequate behavior within groups, F(1, 705) = 3.96, p < .05, knowledge about self-confident behavior, F(1, 705) = 5.71, p < .05, and knowledge about effective communication/listening, F(1, 705) = 5.70, p < .05. In contrast, for the program effects on the remaining skills variables, gender was not found to be a moderator. Subsequent split file analyses revealed program effects on self-confidence in the girls’ sample (ME) group: F(1, 415) = 28.31, p < .001, ES = .20) but not in the boys, whereby girls in the intervention group showed a higher level of self-confidence over time than girls in the control group (see Table 4).
Concerning effective communication, program effects were found for boys (ME group: F(1, 300) = 8.02, p < .01, ES = .02) but not for girls. As depicted in Table 4, boys of the intervention group had more knowledge on effective communication compared with those of the control group, although the effect size was very low, implying low practical relevance of this effect. For knowledge about adequate behavior within groups, separate analyses conducted by gender showed significant Time × Group interactions for boys, F(2, 299) = 4.95, p < .01, ES = .02, and girls, F(2, 414) = 4.67, p < .05, ES = .15. Unexpectedly, boys in the intervention group decreased in their knowledge of adequate behaviors as compared with those in the control group, F(1, 300) = 9.34, p < .01. This finding, however, was only evident at T2 and not found at any subsequent measurement points. In addition, the low effect size suggests limited practical meaning for this difference. Contrary to this finding, a significant main effect of group was found for girls, F(1, 415) = 5.82, p < .05, whereby girls of the intervention group showed higher knowledge on adequate group behavior as compared with girls of the control group (evident at T4).
Program Effectiveness With Regard to Alcohol-Related Variables
General intervention effectiveness
The second MANCOVA, which had time as repeated measurement, group membership and gender as factors, dependent variables measuring alcohol use and expectancies, and T1 outcome variables as covariates (n = 747), revealed significant multivariate main effects of time, F(10, 729) = 15.97, p < .001, and of group, F(5, 734) = 3.63, p < .01. The latter indicates that the groups differ in their mean outcomes after controlling for T1 scores. Time × Group interaction was also significant, F(10, 729) = 2.96, p < .01, indicating differences in mean changes over time in alcohol-related variables between intervention and control group. Univariate tests within MANCOVA (see Table 4) revealed significant main effects of group on the consumption frequency of beer, F(1, 738) = 11.84, p < .01, wine, F(1, 738) = 14.76, p < .001, mixed drinks, F(1, 738) = 12.47, p < .001, and spirits, F(1, 738) = 7.80, p < .01, over the past 30 days, and of alcohol related expectations, F(1, 738) = 11.76, p < .01. In addition, significant Time × Group interactions were found for 30-days’ consumption frequency of beer, F(2, 1476) = 13.34, p < .001, wine, F(2, 1476) = 9.73, p < .001, and mixed drinks, F(2, 1476) = 10.55, p < .001. As depicted in Table 2, students of the intervention group showed a lower age-typical increase of the 30-day consumption frequency of different types of drinks as compared with the control group. They also showed less positive expectations toward alcohol use after program participation. Finally, as the univariate ANCOVA on alcohol-related knowledge also revealed a significant Time × Group interaction, F(2, 741) = 3.91, p < .05, students of the intervention group showed a steeper increase in knowledge over time compared with students of the control group. Additional single comparisons revealed that mean differences between groups on expectancies and knowledge were observed at T3, but those on alcohol use did not emerge before T4. Program effect sizes ranged from 0.13 to 0.41 for alcohol-related variables.
Gender-specific intervention effectiveness
According to our hypotheses, higher program effectiveness with regard to alcohol-related outcomes was expected for girls than boys. Within the multivariate analysis, no significant main effect of gender was found, and the interaction effects Group × Gender and Time × Group × Gender were also not significant (Table 3). Furthermore, from the univariate analyses within the MANCOVA, it was revealed that, for all alcohol-related variables with general intervention effects, gender was not a moderator. The same applied to the ANCOVA with alcohol-related knowledge as outcome.
As the students of our study were nested within schools, observations are not independent and intraclass correlations might bias the results. To test whether general program effects remain significant when taking possible school effects into account, we conducted additional multilevel analyses using the software HLM 5 (Raudenbusch, Bryk, Cheong, & Congdon, 2000). The results of these analyses did not differ from the MANOVA findings presented above (see Wenzel et al., 2009). Thus, the significance of overall effects of the program IPSY was not affected when the nesting of students in schools is taken into account.
Discussion
The present study examined the effectiveness of the universal school-based life skills program IPSY on life skills, knowledge, school bonding, and alcohol-related measures in a sample of German adolescents aged 10 to 13. It was shown that the life skills program positively influenced psychosocial adaptation and prevented the age-typical increase in the frequency of alcohol consumption during early adolescence over a period of 3 years. Thereby, the mean effect sizes of the program were on the upper level of what can be expected of a school-based life skills program (Tobler et al., 2000; i.e., ES = .15).
We were particularly interested in whether the effectiveness of the program IPSY varies by gender. Prior to the main analyses, we compared boys and girls in all outcomes at T1. In line with other studies, we found that boys drank more alcohol than girls (see BZgA, 2007), behaved in a more dominant and assertive manner during social situations, and showed greater rebellion against (school) authority (Zenker, 2009). For girls, lower self-worth and low assertiveness was detected in comparison to boys, possibly reflecting consequences of pubertal development that can contribute to negative self-evaluations and socially insecure behaviors (see Hayward, 2003). Apart from these differences, boys and girls at T1 were very similar concerning a wide range of competencies (i.e., psychosocial deficits and resources), several aspects of self-concept (e.g., related to problem solving), and susceptibility to peer pressure.
Against our expectations, and in spite of the observed gender differences in alcohol consumption and internalizing and externalizing behaviors at T1, gender was not found to moderate the overall program effectiveness on all alcohol-related and the majority of skills variables. Thus, the program IPSY was effective in both boys and girls from age 10 to 13. Other research has also shown that life skills programs may have the same effects on substance use in both genders, particularly when implemented in early adolescence (e.g., Kröger & Reese, 2000). In this life period (as our data also show), initial consumption is still relatively low and the gender-typical behavioral styles (and gender-typical motives to drink; Kuntsche et al., 2006) have only just begun to differentiate. In the same vein, it has been argued that the mechanisms initiating alcohol use in early adolescence may be more similar between both genders than those explaining the progression in drinking patterns and gender-typical consumption cultures in later adolescence (Franzkowiak & Helferich, 1997). Therefore, a life skills program that targets substance use and is taught in a coeducational classroom setting seems to be an appropriate intervention strategy against alcohol consumption in both males and females at the beginning of adolescence. In mid- and late adolescence, however, gender differences may be more pronounced resulting in a greater need for gender-specific program components.
Meaningful gender-specific effectiveness of the program IPSY (in favor of girls) were found only for knowledge on adequate behavior within groups and self-confident behavior. Here, the program IPSY seems to fit better with the interests and age-related needs of girls. Girls define their identity and build their self-worth through social competencies and activities in social contexts and feel a greater need for social support (Kendler, Myers, & Prescott, 2005), while boys shape their identity through more competitive behaviors within adolescence (Amaro et al., 2001). The content and mode of facilitation of the program IPSY may have stimulated girls’ integration and activity within a supportive coeducational classroom setting, therefore reducing the age-typical insecure behaviors of females within mixed gender groups. More specifically, during the program, boys and girls trained self-confident behavior interactively in a naturalistic situation, thereby receiving positive feedback from their peers and teachers. Similarly, knowledge about satisfactory behavior within groups was conveyed by the program, trained within role plays and positively reinforced. In addition, girls seem to be generally more interested than boys in content that demonstrates how to behave optimally during social interactions, resulting in receiving positive feedback (see Leppin, Pieper, Szirmak, Freitag, & Hurrelmann, 1999).
For boys, however, a negative program effect was found for adequate behaviors in groups, pointing to their tendency toward nonconformity, rebelliousness, and testing limits within social interactions (particularly when interactive methods were applied within the classroom, as is typical for life skills programs). In addition, we found that boys benefited from the program with regard to higher knowledge about effective communication when in the role of listener (although the effect size was rather low; ES = .02). This finding implies that boys in particular may have acquired interpersonal skills, such as not interrupting their interaction partner, listening to each other, or asking for information to prevent misunderstandings. As we know from qualitative interviews with the facilitating teachers, these skills seem to have encouraged the girls to speak and interact more openly with the boys, thereby promoting their self-assurance.
In sum, both boys and girls profited from the program alike in most of the skills and competencies and showed a lower than age-typical acceleration in alcohol consumption. The few gender differences observed for program effectiveness regarding knowledge on self-confident and competent communication behaviors suggest higher benefits for girls than boys.
Limitations of the Study and Outlook
Some limitations of the current study and implications for future investigations have to be noted. First, all schools of the intervention group showed an interest in participating in the study and the program (whereas none of the control schools actively indicated an interest in the IPSY program) so that, due to the assignment strategy, there could be a risk of positive selection. Based on the analyses conducted on the general interest of the schools in interventions, and characteristics of teachers and students, however, this seems rather unlikely. One third of both the control and the intervention schools had participated in intervention programs in the past, and teachers assessed during process evaluation did not differ regarding sociodemographic characteristics or school-related self-efficacy at T1. Furthermore, students across both groups did not differ in any outcome variables at T1, indicating no positive selection of the intervention group also in this regard.
Second, the reliability of the self-reported scales indicated by Cronbach’s alpha was relatively low. Note that selective exclusion of single items per scale did not increase Cronbach’s alpha. At T1 some internal consistencies were low, especially for some of the intrapersonal skills measures, but over time they improved (i.e., Cronbach’s alphas ranging from .70 to .88). This trend may be due to the fact that reflexive thinking increases from early to midadolescence and that self-perceptions and associated skills profiles become more coherent, resulting in a more integrative self-concept and thus to higher internal consistencies of the respective scales.
Analyses of gender-specific life skills programs’ effectiveness were rarely conducted (Blake et al., 2001; Bühler & Kröger, 2006) but are necessary to ensure the broad applicability of intervention efforts and to explore whether existing programs should be revised to guarantee higher gender sensitivity. This study shows that in early adolescence a universal prevention program based on the life skills approach can promote positive outcomes and reduce age-typical increases in alcohol use in both boys and girls. Thus, a complex and costly gender-specific adaptation of this program does not seem to be necessary in order to prevent alcohol consumption in early adolescence for both genders. Future studies should, however, focus on universal programs applied to older age groups where gender differences may be more pronounced. Such studies could help to determine more precisely whether, and if so at which age, more gender-related prevention attempts are indeed required.
Footnotes
The authors declared no potential conflicts of interest with respect to the authorship and/or publication of this article.
The authors disclosed that they received the following support for their research and/or authorship of this article: The project IPSY was financially supported by funds of the Federal State of Thuringia (Program: Promotion of Excellence in Scientific Activities at Thuringian Universities awarded to Weichold), the Department of Developmental Psychology, CADS, Center for Applied Developmental Science, and Philip Morris GmbH Germany (procured by the Thuringian Ministry of Education and Culture; PIs: Weichold & Silbereisen).
