Teacher Ratings of Social Behavior
Frey and colleagues (2005) found that overall there were some significant program effects detected during the first year of the Second Step® program; however many of those effects were not noticed during the second year of the program. During the first year of program implementation, among those students who had a high antisocial baseline rating, the intervention group showed significantly greater declines in antisocial behavior than the control group. Among students with a low antisocial baseline rating, intervention students showed no change in antisocial behavior, whereas control students’ antisocial behavior increased (a significant difference). However, there were no significant group differences in antisocial behavior change during year 2 of the study, regardless of antisocial baseline rating.
During the first year of the program, the intervention group showed significant gains in social competence compared to the control group. There were similar significant gains observed for the intervention group during the second year as well, but they were not as substantial as the first-year gains. Hierarchical linear modeling (HLM) analyses confirmed the group differences in measures of antisocial behavior and social competence in the first year, but the analyses did not confirm the group differences in social competence observed during the second year.
Student Ratings of Hostile Attributions and Intentions
The preliminary analyses of students’ responses to surveys that assessed attributions of hostile goals from hypothetical vignettes of ambiguous provocations found the behavioral intentions had nearly identical patterns of responses to the ambiguous and non-ambiguous provocations. There were no significant differences between groups in hostile attributions and behavioral intentions during both years.
Prisoner’s Dilemma Game
Analyses of the students’ outcomes when playing the Prisoner’s Dilemma game showed some significant differences between the groups. The goals chosen by the intervention group were significantly more likely to be prosocial than those chosen by the control group. Students in the control group expected greater satisfaction for the self-high outcome than intervention group students. However, the groups did not differ in the proportion of cooperative choices made between the pair of students, and there were no group differences in the outcome of the Prisoner’s Dilemma game. The intervention group students were more satisfied with the game outcome and prize division than control students. Although HLM analyses confirmed the group difference in satisfaction with the prize division, the analyses did not confirm the difference in satisfaction with the game outcome.
Overall, Schick and Cierpka (2005) found significant improvements among students in the experimental group in measures of anxiety, but there were few significant differences in other measures. There were also significant differences between experimental and control group girls in some behavioral measures, but not between boys in both groups.
Structured Interviews with Students
At posttest, experimental students reported a significantly reduced fear of loss of control compared to the control group. However, there were no significant differences between the groups in measures of fear of being injured and fear of bad things happening. There were also no significant differences in measures of empathy, self-confidence, and self-esteem. Measures of peer acceptance showed students in the control group felt significantly more accepted by their peers than students in the experimental group. Although aggressive behaviors were significantly reduced for girls and boys in both the experimental and control groups, experimental group girls assessed themselves to be significantly more aggressive than control group girls.
Parent Ratings of Student’s Behavior
The outcome measures showed that parents of students in the experimental group rated their children as exhibiting significantly less anxiety/depression problems compared to ratings from parents of control group students. However, there were no other significant differences between experimental and control group parents in any other measures on the Child Behavior Checklist (including social withdrawal, social problems, attention problems, delinquent behavior, and aggressive behavior). There were also no significant differences between parents’ rating on measures of self-control, assertiveness, and cooperation/social rules.
Teacher Ratings of Student’s Behavior
There were no significant differences between experimental and control group teacher ratings on student’s behaviors measured by the Landau scales of social climate in the classes (LASSO), including measures of rivalry between classmates, aggression against classmates, and extent of clique formation. There was a significant decrease in the extent of discrimination against classmates, but that decrease was observed for both groups.
Holsen, Smith, and Frey (2008) found mixed results when examining the outcomes of the Steg and Steg program, the Norwegian version of Second Step®. The posttest measures of social competence were significantly higher in the sixth-grade intervention group than in their comparison group. There was also a significant increase in social competence scores when examining the seventh-grade intervention group, but only for girls.
For externalizing behaviors, there was a significant decrease in the measures for grade 6, but only for boys. There were no significant effects found for either sex in grade 7. There were also no significant differences between the intervention and comparison groups with regard to the self-reported measure of internalizing problem behaviors for either grade.
Frey and colleagues (2005) examined the impact of the Second Step® program on students’ behavior and social cognitions. The study sample included 15 elementary schools (seven K–fifth grade schools and eight K–sixth grade schools) from three cities across Western Washington. Schools were randomly assigned to the intervention group (two-thirds of the sample) or the control group (one-third of the sample). Schools were reluctant to agree to a waitlist control design; therefore, schools assigned to both intervention and control groups received program materials and teacher training. However, control schools received these benefits for classrooms that were not included in the study sample.
The final sample included 462 students in the intervention group and 436 students in the control group. There were no significant differences between the intervention and control group schools with respect to race/ethnicity. The school populations ranged from 52 to 89 percent European American, while Asian Americans and African Americans populated the next largest ethnic groups. Study participants were 48.2 percent female and between 7 and 11 years old.
Students were followed for 2 school years. The program was implemented in second- and fourth-grade classrooms during year 1 of the study, and in third- and fifth-grade classrooms during year 2. Teachers were administered the School Social Behavior Scale (SSBS) that asks them to report how frequently students engage in each of 32 antisocial and 33 socially competent behaviors. Students were administered surveys that assessed attributions of hostile goals from hypothetical vignettes of ambiguous provocations. After rating the intentions of the vignette characters, students rated how likely they were to respond with physical aggression, verbal aggression, and socially competent behavior on a 5-point scale.
In addition, four same-sex, same-grade children from two different classes were randomly selected to participate in eight trials of the Prisoner’s Dilemma game. Each pair would choose either a cooperative or exclusively self-interested strategy. Students were given the opportunity to earn money for a class party thrown at the end of the school year. The student’s earnings for each pair of answers were determined jointly by his or her selection and that of the other student. Charts displayed overall and for each trial how much students would earn given the four possible combinations (equal-high, self-high, self-low, and equal-low). The students were also audiotaped explaining why they selected their first and second choice and their satisfaction with the outcomes. The decision-making process of each pair was also audiotaped as they jointly selected a cooperative or self-interested strategy on each of eight trials. The outcome measures coded from the observations of the game included the number of prosocial goals; the self-reported expected satisfaction with the outcome and the self-reported actual satisfaction with the outcome of the game; the number of cooperative choices made during joint decision making; the number of positive or negative comments made during negotiations; the need for adult intervention during negotiations; and observed negotiation strategies (cooperative or coercive).
Omnibus multivariate analyses of covariance (MANCOVA) were used to test for significant group differences, followed by confirmatory hierarchical linear modeling (HLM) analyses of individuals nested within classroom.
Schick and Cierpka (2005) evaluated the effectiveness of the Faustlos curriculum, which is the German version of the Second Step® program (“Faustlos” is the German word for “without fists”). The Faustlos curriculum includes 51 lessons divided into three units: empathy, impulse control, and anger management. The lessons start in grade 1 and continue through grade 3.
The study used a pre–post evaluation design with a control group. Pretest data collection occurred in spring 1999 and posttest data was collected in fall 2000, when teachers had delivered 35 of the 51 planned lessons. The study examined changes on the behavioral level of students (improving social competences and reducing aggressive behavior), changes on the emotional level of students (effects on emotions correlated with aggressive behavior), and the gender-specific effects of the program.
The sample included 21 elementary schools from the school districts of Heidelberg and Mannheim, Germany. Fourteen elementary schools (30 classes, 496 students) were randomly assigned to an experimental group, and seven elementary schools (14 classes, 222 students) were randomly assigned to a control group. Parents (n=718) and teachers (n=35) from both groups were administered questionnaires. In addition, two children from each class (one girl and one boy who were randomly selected and whose parents gave informed consent) were also selected for an interview (60 students in the experimental group and 28 students in the control group).
A complete pre–post data set was available for 64 percent of the students, including 161 girls and 174 boys. At the time of the posttest, 17 percent of the students were 5–6 years old, 74 percent were 7–8 years old, and 7 percent were older than 8 years. There were no significant differences between the experimental and control groups, except on gender and the school qualification of mothers. The control group had more boys than girls, whereas the experimental group had more girls than boys, and the mothers’ level of education was higher in the experimental group.
Behavioral problems were assessed by the students’ parents using the German version of the Child Behavior Checklist (CBCL). The social competences of students were measured using several scales from the Heidelberg Competence Inventory (HKI). In addition, a structured interview for the students was developed based on items from different questionnaires. The interview measured students’ empathic competences, self-esteem and competences, aggressive behaviors, and anxieties. Finally, teachers were given six subscales from the Landau scales of social climate in the classes (LASSO 4–13) to assess the effects of the curriculum on the class as a whole.
The data collected from the structured interviews with students, the parents’ questionnaire, and the teachers’ assessments were examined by two-factorial analyses of variance with repeated measurement (group x time). Gender-specific effects based on the students’ structured interviews were examined using three-factorial analyses of variances with repeated measurement (group x time x gender).
Holsen, Smith, and Frey (2008) examined the effectiveness of the Norwegian version of the Second Step® program (called the Steg for Steg program) among students in grades 5 and 6. The study employed an age cohort design. Two sets of complementary analyses were conducted. The first was a repeated measures design. In the second analysis, children of the same age (but from different cohorts) were compared at each measurement time. Since the study schools began implementation simultaneously with fifth-, sixth-, and seventh-grade students, this made it possible to compare age-equivalent groups that had and had not yet participated in the program for a school year. For example, when fifth-grade students reached sixth-grade after one year of the intervention, their posttest data was compared with pretest data from the cohort of sixth-grade children from a time before they had received the intervention.
The primary outcomes of interests included social skills and problem behaviors. Social skills were measured using the 34-item student version of the Social Skills Rating System, which was translated into Norwegian. The scale measures cooperation, assertiveness, empathy, and self-control. Student self-reports of problem behaviors were obtained with the student version of the Problem Behavior Scale. The scale consists of 13 items that measure internalizing problem behavior, externalizing problem behavior, and hyperactivity.
The study included 338 students in grade 5 (age 10 at baseline), 405 students in grade 6 (age 11 at baseline), and 389 students in grade 7 (age 12 at baseline). Baseline data was collected in November 2004 and follow-up data was collected one year later in November 2005. The seventh-grade students were included only as a comparison at baseline. Approximately 670 students participated at the follow-up in 2005. At baseline, sixth- and seventh-grade students differed significantly in social competence scores. Students in different grades did not differ with respect to internalizing and externalizing behaviors. The pattern of gender differences was the same for all grades: girls reported higher levels of social competence and lower levels of externalizing problem behaviors than boys. There were no gender differences for any grade regarding internalizing problem behavior.
The outcome data was analyzed using linear mixed model (LMM) analysis. LMM used correlated residual structures to account for repeated observations within individuals and provide the tools necessary to estimate fixed and random effects in one model. Effect sizes were calculated using the pooled standard deviation of the intervention and comparison groups.
There were some limitations to this type of study design. There was no “true” comparison group available at posttest. The age cohort design controlled for the effects of maturation, but not for the effects of repeated measurements. Unlike the intervention group, the comparison group was measured only at baseline.
Following are CrimeSolutions.gov-rated practices that are related to this program:School-Based Bullying Prevention Programs
Aim to reduce bullying and victimization (being bullied) in school settings. Some interventions aim to increase positive involvement in the bullying situation from bystanders or witnesses. The practice is rated Effective for reducing bullying, bullying victimization, and for increasing the likelihood of a bystander to intervene. The practice is rated No Effects for increasing bystander empathy for victims of bullying.Evidence Ratings for Outcomes:
School-Based Social and Emotional Learning (SEL) Programs
| ||Juvenile Problem & At-Risk Behaviors - Bullying|
| ||Victimization - Being Bullied|
| ||Mental Health & Behavioral Health - Bystander Intervention|
| ||Mental Health & Behavioral Health - Empathy for the Victim|
Designed to foster the development of five interrelated sets of cognitive, affective, and behavioral competencies, in order to provide a foundation for better adjustment and academic performance in students, which can result in more positive social behaviors, fewer conduct problems, and less emotional distress. The practice was rated Effective in reducing students’ conduct problems and emotional stress.Evidence Ratings for Outcomes:
Universal School-Based Prevention and Intervention Programs for Aggressive and Disruptive Behavior
| ||Juvenile Problem & At-Risk Behaviors - Multiple juvenile problem/at-risk behaviors|
| ||Mental Health & Behavioral Health - Internalizing behavior|
Universal school-based prevention and intervention programs for aggressive and disruptive behavior target elementary, middle, and high school students in a universal setting, rather than focusing on only a selective group of students, with the intention of preventing or reducing violent, aggressive, or disruptive behaviors. The practice is rated Effective in reducing violent, aggressive, and/or disruptive behaviors in students.Evidence Ratings for Outcomes:
| ||Juvenile Problem & At-Risk Behaviors|