The Effects of City-Wide Implementation of "Second Step" on Elementary School Students' Prosocial and Aggressive Behaviors

Michelle Beaulieu Cooke; Julian Ford; Joan Levine; Cate Bourke; Lisa Newell; Garry Lapidus


J Prim Prev. 2007;28(2):93-115. 

In This Article


All eight elementary schools in Meriden, Connecticut, participated in the city-wide implementation of Second Step. Meriden is a small city with approximately 58,000 residents, 20% of whom are members of racial/minority ethnic groups (Connecticut Department of Economic and Community Development, 1999). Meriden's public school students are ethnoculturally diverse (37% Hispanic and 12% African American), and 13% live in households in which English is not the primary language (Connecticut Department of Education, 2004). Almost half (46%) of the students in Meriden live in low-income households (i.e., are eligible for free or reduced-price meals at school), compared with 29% of students statewide.

Of 986 potential participants, 758 (77%) consented to take part in the evaluation. Seventeen students with parental consent did not complete the study measures, resulting in a final N of 741 (364 third graders and 377 fourth graders). This sample was representative of the student population in the Meriden school district ( Table 1 ): 50.2% female and 49.8% male, 13.5% African American/Black, 47.0% White, 39.0% Hispanic, and 0.7% other races/ethnicities. Racial and ethnic distributions differed between schools, consistent with the varying neighborhood ethnic compositions for each school.

All third and fourth graders from five of the eight elementary schools (due to resource limitations) in Meriden were eligible to participate in the evaluation. The five schools were selected to be representative of the overall elementary school student body and climate. Students in grades 1 and 2 were not eligible to participate because the self-report survey was not believed to be appropriate for this age group, while those in grade 5 were not eligible due to potential confounding by another violence prevention curriculum that was simultaneously being implemented at this grade level.

Of the five schools in the evaluation, two utilized active informed consent and three utilized passive informed consent, following a protocol that was approved by the Institutional Review Board of Connecticut Children's Medical Center and by the Meriden School District Superintendent.

Second Step is a school-based social-emotional curriculum for children in grades pre-kindergarten through middle school. At the elementary school level, lessons are divided into three units teaching skills in Anger and Emotional Management, Empathy, and Impulse Control. The teaching methods employed in the Second Step curriculum include adult and peer modeling, role-playing, and coaching and cueing.

In Meriden, Second Step was implemented for the first time as a city-wide intervention to promote consistent behavioral expectations for children and promote widespread use of modeling and verbal coaching/cueing by teachers, school staff, parents, and community workers who have regular contact with children. Program implementation included all eight elementary schools in the district, as well as its two public middle schools. All training sessions were conducted by certified Second Step trainers, except where indicated. Implementation took place during the 2002-2003 school year and included the following components:

  • Full-day training of all elementary school teachers and staff, including non-classroom teachers (e.g., Music, Physical Education, Special Education) and school administrators using the one-day training module described in the Second Step Trainer's Manual (Committee for Children, 2002);

  • A series of three, 1-h training sessions on each of the three curriculum components (Anger/Emotional Management, Empathy, and Impulse Control) adapted for non-certified school staff (e.g., Custodial staff, Teacher's Aides, Cafeteria Staff) and offered at each school;

  • A series of four, 1-h workshops for parents offered at each school and one secondary location (public library) and conducted by a Parent Education Consultant supervised by a certified Second Step trainer using an adaptation of the Second Step Family Guide;

  • Ongoing technical support provided to each school by certified Second Step trainers including weekly school visits, a Supporting Second Step bulletin board identifying the skills and concepts being taught at each grade level, and periodic Second Step newsletters highlighting teachers using creative approaches to reinforcing Second Step and describing implementation tips for classroom and non-classroom teachers;

  • A series of four flyers released through local newspapers in both English and Spanish to describe Second Step and identify ways in which parents and community members could model and reinforce the specific skills being taught in elementary schools;

  • A 3.5-day Training of Trainers adapted to the needs of those working with children and families in the community attended by members of community agencies and organizations, parents, and select members of the school system.

  • A school and community partnership team formed to aid in the consistent application of Second Step language and principles throughout the school district and the community as a whole.

In addition, during and for the year prior to implementation, program staff met regularly with principals, teachers, and other staff at each of the schools, as well as with parent-teacher organizations. These meetings were designed to build support for Second Step, answer questions and concerns, and assist schools in developing site-based implementation plans based on their individual lesson schedules. Focus groups were conducted with students, parents, and teachers to help tailor the Second Step training programs to meet school-specific needs. Brief evaluation surveys were used to assess the efficacy and acceptability of training sessions and a year-end follow-up teacher survey was performed to examine measures of support for and implementation of the Second Step program.

Assessments were conducted immediately prior to and following implementation of the curriculum. Assessment measures included a student self-report questionnaire that was completed by 3rd and 4th grade students and administered by trained research assistants; a student behavior observation checklist that was conducted by two trained research assistants during two, two-week periods at the start and end of the school year; and a disciplinary referral checklist that was used by a single research assistant to code disciplinary referral records for the 2001-02 and 2002-03 school years. All assessment measures tracked students using an ID number to protect confidentiality and allow for cross-comparisons between assessment measures.

Self-Report Questionnaire. In September 2002 and May 2003, research assistants administered self-report questionnaires to 3rd and 4th grade participants in classroom groups. A total of 639 students (87% of the baseline sample; 67% of all eligible students) completed both the baseline and follow-up questionnaires. Of students who completed the baseline survey, Black students were more likely than White students to not complete the post-test (17% vs. 7%, respectively). The pre-test and post-test samples were similar in all other respects. The 67 item self-report survey was presented to each of the classrooms in a 45-minute session, with a maximum of 25 students per session. The survey items were read orally to the students and administered in a standard order to each group. Students were encouraged to answer each item but were told that they could "pass" (leave blank) any item they did not want to or could not answer. The assessors encouraged students to complete the items but did not in any way influence their specific response choices. Items that were left blank were coded as missing data.

The self-report questionnaire included nine outcome measures taken from four surveys validated for use with school-aged children: Kidcope (La Greca, Silverman, Vernberg, & Prinstein, 1996), the United States Centers for Disease Control and Prevention Youth Risk Behavior Survey (see, the Modified Aggression Scale (Bosworth, Espelage, & Simon, 1999), and the Weinberger Adjustment Inventory (Weinberger & Schwartz, 1990). Psychometrics for each measure based on the present sample are presented in the Results section.

The 12-item Kidcope assesses sub-scales for positive (4 items; problem focused or self-management) and negative (8 items; avoidant/blame/wishful thinking) coping with daily stressors. The positive coping sub-scale has been shown to be internally consistent (Alpha = .77), and both sub-scales have been shown to be temporally reliable and to have convergent validity in relationship to multimodal coping measures (Prinstein, La Greca, Vernberg, & Silverman, 1996).

Three items from the Youth Risk Behavior Survey were summed to provide a score for the frequency of fighting. The items were cumulated in a single score in order to reduce temporal instability (Kappas for individual items over a 10-22 day retest period ranged from .47-.64; Brener et al., 2003) and to provide a measure that could be evaluated for internal consistency. Although no evidence of validity has been reported, the items are the most widely used self-report assessment of youth fighting behavior, having been selected by the Centers for Disease Control and Prevention for its national Youth Risk Behavior Surveillance system.

The Modified Aggression Scale (Bosworth et al., 1999) includes four factor-analytically derived subscales that assess fighting-related beliefs (6 items, Alpha=.71), bullying (4 items, Alpha=.83), anger (4 items; Alpha=.70), and intention to use non-violent behavior (8 items; Alpha=.63). Initial evidence of convergent validity was established with middle school students based on demonstrating predicted inter-correlations of the subscales (Bosworth et al., 1999).

The 30-item Weinberger Adjustment Inventory Self-Restraint Scale includes four factor analytically-derived internally consistent subscales for suppression of aggression (7 items), consideration of others (7 items), impulse control (8 items), and responsibility (8 items). The scale has been shown to be internally consistent (Alpha = .91), temporally reliable (7-month re-test r = .76) and to have convergent validity in relationship to the Aggression and Delinquency sub-scales of the Child Behavior Checklist (r = -.71, -.63, respectively) (Weinberger, 1996). Self-restraint was assessed in order to describe youths' abilities to inhibit impulses/aggression and to interact with others in a prosocial manner consistent with the goals of Second Step.

Behavioral Observations. Behavioral observations were conducted within two weeks of the baseline testing (N = 545; 72% of the baseline sample, 58% of all eligible students) and within two weeks of the follow-up testing (N = 558; 87% of the post-test sample, 60% of all eligible students). Reasons for not observing students included: insufficient time to observe all children in a classroom group within the 2-week time frame, student absent or could not be located on observation day, or student moved.

Observations were conducted in classroom, playground, and cafeteria settings, each focusing for 5 min on one randomly selected student. The only exception to random selection was that students of any ethnicity that represented less than 12% of the class were selected as the first children to be observed from that class, in order to ensure adequate representation of all ethnicities in each class, and to distribute observations equally across classrooms in the time that was allocated to conduct observations. All eligible students to be observed were listed alphabetically by classroom. A random numbers table was then used to generate the order in which students were observed.

Observations were conducted according to the procedures described by Grossman et al. (1997) using categories from the Social Interaction Observation System, 4th edition (SIOS) for child behavior (neutral/positive, pro-social, borderline, negative, aggressive, physical or verbal distress) and teacher behavior (requests, commands, questions). Specific examples of each category were adapted from the SIOS Manual. The neutral category consisted of non-negative behavior and positive behaviors, such as playing catch, pushing a peer on a swing, or neutral verbal conversation. Prosocial behavior covered acts of kindness, such as comforting a peer, offering to share or expressing sympathy or remorse. Borderline behavior consisted of the students joking around, being playful with no negative intent, or horseplay. The negative behavior category included acts of minor violence or hostility, such as a peer being bossy or irritable, tattling, or verbal defiance directed toward the teacher. Aggressive behaviors included harming another student, name calling, teasing, or insulting. Each category was coded as present or absent every 30 s. Data from individual participants were aggregated for analyses and converted into a single present or absent score for each category for each 5-minute session. Two trained observers conducted independent simultaneous observations for 81 5-minute sessions scheduled to sample the entire observation period at baseline and post-test in order to evaluate inter-rater agreement. Inter-rater reliability was unacceptable (<75% for only one category (neutral/positive verbal, no target) and was >85% for the remaining 29 coding categories. The observation category with unacceptable reliability (neutral/positive verbal, no target) was difficult to code due to infrequent occurrence and was dropped from analysis. There was no evidence of rater drift during the course of the study.

Disciplinary Referrals. Research assistants collected disciplinary referrals for the 2002-03 school year and suspension records for 2001-02 and 2002-03 school years in June 2003. School administrative staff provided copies of forms documenting each disciplinary referral. The text of the referral was categorized using a standard coding system (see below) and the student's name was re-coded with an ID number in order to ensure confidentiality.

Discipline referral data were coded for categories derived from an initial inspection by the evaluators (JF, JL) of a random sample of six referrals, including three forms of minor delinquency -- non-violent behavior (e.g., rude, disrespectful, or disruptive behavior), minor violence (e.g., pushing, tripping, snowball throwing), or destroying or throwing objects (e.g., making marks on the floor with shoes, breaking crayons, throwing pencil, throwing down chair) -- and two forms of violence -- threatening violence (e.g., taunting other students, bullying, threatening to hit another student) or violent/physical assault (e.g., fighting, kicking, punching, stabbing). Interrater reliability for the referrals was checked by having two raters independently review 45 referral forms, distributed across the five schools. One school was able to retrieve only two referral forms from the index year and was dropped from the referral analysis. Overall interrater agreement for the disciplinary referral data was Kappa = .87, with consistent agreement across categories. A frequency score was calculated as a count of the number of referrals for minor delinquency and for violence that occurred in the first three months of the school year (September-November) and in the final three months of the school year (March-May).


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: