2-3 Page paper on case study Social Work Assignment

Positive effects of promoting prosocial behavior in early adolescence: Evidence from a school-based intervention

Gian Vittorio Caprara,1 Bernadette Paula Luengo Kanacri,1

Maria Gerbino,1 Antonio Zuffianò,2 Guido Alessandri,1

Giovanni Vecchio,3 Eva Caprara,1 Concetta Pastorelli,1

and Beatrice Bridglall4

Abstract The purpose of this study was to evaluate the effects of a pilot school-based intervention called CEPIDEA, designed to promote prosocial behavior in early adolescence. The study took place in a middle school located in a small city near Rome. The intervention group included 151 students (52.3% males; Mage ¼ 12.4), and the control group 173 students (50.3% females; Mage ¼ 13.0). Both groups were assessed at three time points, each 6 months apart. A Latent Growth Curve analysis revealed that the intervention group, compared to the control group, showed an increase of helping behavior along with a decrease of physical and verbal aggression across time. Current results also showed that the increase of helping behavior mediated the decline of verbal aggression in adolescents who had attended the intervention. Participants of CEPIDEA also attained higher grades than the control group at the end of middle school. Overall, findings suggest that promoting prosocial behavior may serve to counteract aggressive conduct and enhance academic achievement during adolescence.

Keywords adolescence, aggressive behavior, latent growth curve, school-based intervention, prosocial behavior

*This article was accepted under the editorship of previous Editor-in-Chief, Marcel van Aken.

Prosocial behaviors, including sharing, consoling, and helping, are

voluntary actions undertaken to benefit others (Eisenberg, Fabes, &

Spinrad, 2006). As such, they are considered integral to interven-

tion goals that seek to promote successful youth development.

Indeed, research on adolescent development has evolved, over the

past two decades, into a new vision called Positive Youth Develop-

ment (PYD), which takes a strengths-based approach. That is,

potentialities of adolescents and the plasticity or adaptability of

their developmental trajectories are emphasized (Lerner, Lerner,

Almerigi, & Theokas, 2005). Within the PYD framework, the

promotion of prosocial behavior may be perceived as both a posi-

tive outcome per se and an important avenue for decreasing the

level of aggression among adolescents. In accordance with this

perspective, the scope of the present study is to evaluate the effi-

cacy of a pilot intervention designed to promote prosocial behavior

among adolescents in middle schools.

Beneficial effects of prosocial behavior in adolescence

Recent findings demonstrate that prosocial children, compared with

their less prosocial peers, have better peer relationships (Eisenberg

et al., 2006), are less at risk for externalizing behaviors (e.g. Kokko

& Pulkkinen, 2000), and perform better in school (Caprara, Barbar-

anelli, Pastorelli, Bandura, & Zimbardo, 2000; Wentzel, 1993).

Several studies have identified prosocial behavior as a protective

factor against aggression (e.g. Kokko, Tremblay, Lacourse, Nagin,

& Vitaro, 2006; Pulkkinen & Tremblay, 1992). Although prosocial

and aggressive behaviors are independent behavioral tendencies

stemming from different dispositions (Krueger, Hicks, & McGue,

2001), findings suggest that they are related to each other, espe-

cially in early adolescence (Veenstra et al., 2008). In particular,

Kokko and colleagues (2006) pointed out the compensatory addi-

tive effect of prosocial behavior in reducing the negative conse-

quences of aggression on adolescents’ adjustment.

With the understanding that aggressive behaviors can stem from

scarce behavioral resources and assets, and being aware of the det-

rimental effects of engaging in such behaviors (e.g. Dodge, Coie, &

Lynam, 2006), the importance of promoting prosocial behavior and

related competences in schools cannot be underestimated. Indeed,

researchers have also found that early prosocial behavior strongly

predicted subsequent levels of academic achievement even after

controlling for variation in early academic achievement (Caprara

et al., 2000). A possible explanation may be that adolescents who

1 ‘‘Sapienza’’ University of Rome, Italy 2

University of Toronto, Canada 3

Roma Tre University, Italy 4 Montclair State University, USA

Corresponding author:

Gian Vittorio Caprara, Faculty of Medicine and Psychology, ‘‘Sapienza’’

University of Rome, Via dei Marsi, 78, 00185, Rome, Italy.

Email: [email protected]

International Journal of Behavioral Development

2014, Vol. 38(4) 386–396 ª The Author(s) 2014

Reprints and permissions: sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/0165025414531464 ijbd.sagepub.com

enact prosocial behaviors tend to be more accepted and supported

by their classmates and teachers. Moreover, at a collective level, the

ethos of a prosocial classroom supports active teaching—including

the true give and take between teacher and student— with a focus

on student learning rather than on sanctioning students’ reprehensi-

ble conduct (e.g. Wentzel, 1993). In sum, evidence from these and

other studies suggest that fostering prosocial behavior at school can

contribute to students’ learning and well adjustment while protect-

ing them against the negative consequences of aggression, includ-

ing peer rejection (Bierman, Smoot, & Aumiller, 1993), and

antisocial behavior (Pulkkinen & Tremblay, 1992).

Program background

Within the wider framework of PYD, the specific theoretical ratio-

nale underlying our intervention stems from the integration of var-

ious research traditions related to personality and social

development, which address the personal roots of prosocial beha-

viors. Personality psychologists, for instance, have pointed to a

basic trait like agreeableness (i.e., the tendency to be likeable and

harmonious in relationships with others) as a major determinant

of prosocial behavior (Caprara, Alessandri, & Eisenberg, 2012;

Graziano & Eisenberg, 1997). Developmental psychologists have

stressed the role of empathic and emotion regulation capacities as

critical ingredients for prosocial behavior (Eisenberg et al., 2006).

Indeed, being empathic and able to manage one’s own emotions

is crucial in order to be sensitive to the needs of others while avoid-

ing the contagious effects of others’ negative emotions (Eisenberg

et al., 2006). Values related to self-transcendence, which emphasize

accepting others as equals and displaying concern for their welfare,

also proved to be important motives for prosocial behaviors (e.g.

Schwartz, 2010). Likewise, self-efficacy beliefs attesting individu-

als’ confidence in their abilities to manage themselves across

domains of functioning have proved to contribute significantly to

prosocial behavior by turning traits and values into effective beha-

viors aimed to benefit others (Bandura, 1997; Caprara et al., 2012).

Promoting prosocial behavior at school

Contemporary perspectives regarding both PYD and Positive Psy-

chology (Seligman & Csikszentmihalyi, 2000) suggest the impor-

tance of developing, within the context of schools, the awareness

of the value of prosocial attitudes and behavior and a safe atmo-

sphere in which these skills can be cultivated. Schools are important

socializing venues that can play significant roles in promoting pro-

social behavior (Jennings & Greenberg, 2009). They can nurture

‘‘character strengths’’ such as kindness and compassion, and can set

the stage for exposing students to local, national, and international

concerns for social justice and equality (Yates & Youniss, 1996).

Indeed, preparing adolescents for active civic engagement is a cru-

cial educational goal to meet the challenges of our globally con-

nected world (Council of Europe, 2010).

Currently, the most salient theoretical framework undergirding

effective school-based interventions is Socio and Emotional Learn-

ing (SEL; Durlak, Weissberg, Dymnicki, Taylor, & Schellinger,

2011), which emphasizes the importance of building emotional and

interpersonal capacities, and of developing prosocial behavior.

Among current SEL intervention programs, however, the term pro-

social concerns positive or competent social behavior rather than

specific prosocial actions, such as helping or caring. On the whole,

these interventions have mostly focused on elementary schools (e.g.

Incredible Years, Hutchings et al., 2012; Webster-Stratton, Reid, &

Hammond, 2004) and were directed at children aging from 6 to 12

years (e.g. Second Step; Holsen, Smith, & Frey, 2008). Evidence of

efficacy in increasing prosocial behavior has been found mainly for

elementary children (Webster-Stratton et al., 2004) but not for ado-

lescents, who reported improvement in their prosocial attitudes

(e.g. concern for others, desire to help; Battistich, Schaps, Watson,

Solomon, & Lewis, 2000) rather than their behaviors.

Based on a thorough review of these and other programs and their

underpinnings, we designed a school-based intervention called Pro-

moting Prosocial and Emotional Skills to Counteract Externalizing

Problems in Adolescence (Italian acronym: CEPIDEA). CEPIDEA

is directed to adolescents in middle-school. While many programs

include social skills or prosocial behaviors as one specific compo-

nent of their curriculum, the CEPIDEA program is unique in that it

is entirely designed to promote prosocial behaviors. That is, we con-

nected the development of key skills (including perspective taking

and emotional regulation, for example) anchored specifically on the

promotion of prosocial actions. The enhancement of these skills are

content targets of the CEPIDEA curriculum because of their being

determinants of prosocial behaviors. To this aim, CEPIDEA aligns

with previous findings in viewing self-efficacy in perspective taking

as proxy determinants of prosocial actions (Caprara et al., 2012).

An additional characteristic of the CEPIDEA program is its

focus on early adolescence, namely an age where children are more

sensitive to peer relations (January, Casey, & Paulson, 2011) and

where prosocial behavior may carry manifold beneficial effects.

A recent meta-analysis (January, Casey, & Paulson, 2011) indicates

that early adolescents are more sensitive to interventions focused on

positive peer relations. This susceptibility derives from the value

they place on relationships with their classmates and from their

need to manage new complex social rules. Moreover, since previ-

ous findings suggest a certain stability and even a decline of proso-

cial behavior in middle adolescence (e.g. Luengo Kanacri,

Pastorelli, Eisenberg, Zuffianò, & Caprara, 2013; Nantel-Vivier

et al., 2009), interventions aimed at sustaining behaviors that help

others can result most beneficial.

Program description

The CEPIDEA curriculum included four major components that

reflect the personal determinants of prosocial behavior (Caprara

et al., 2012). These components were implemented in stepwise

fashion over the course of the intervention: (1) sensitization to pro-

social values; (2) development of emotion regulation skills (man-

agement of negative emotions, and expression and reinforcement

of positive emotions); (3) development of perspective-taking skills;

and (4) improvement of interpersonal-communication skills. The

program adopted persuasion, modeling, and mastery experience

(Bandura, 1997) as relevant strategies in developing adolescents’ abil-

ities and self-efficacy beliefs that are conducive to prosocial behaviors.

In addition, the program includes a conclusive section devoted to the

promotion of a collectivistic enactment of prosocial behaviors in the

school context, named ‘‘precursors of civic engagement.’’

Given the support needed by the school’s administration, the

entire school was involved in the CEPIDEA program. Indeed, the

program included two kinds of intervention strategies: (1) prosocial

sessions and (2) prosocial lessons. Prosocial sessions (16 in total)

were delivered once a week during school hours by the CEPIDEA

Caprara et al. 387

research staff and in collaboration with teachers. Sessions included

role playing, modeling, case analysis, interviews and group discus-

sions aimed at setting the conditions for reflecting and experiencing

upon each component of the intervention. In addition to these work-

shops, teachers integrated the concepts of the curriculum through

prosocial lessons (21 in total; 3 per classroom) that were delivered

by teachers in various academic disciplines (e.g. Science, History,

etc.). In these lessons, teachers emphasize the functions of behaving

prosocially and the value this may provide in various contexts of life.

On the whole, while prosocial sessions were crucial to promoting

mastery experiences and encouraging students to apply their new

skills in their daily life at school, prosocial lessons were important

in engaging student reflection regarding how prosocial values can

be applied in their social contexts (e.g. family, neighborhood, etc.)

and the benefits of prosocial behaviors across these and other situa-

tions. Table 1 reports a summary of the sessions of the intervention.

Given the central role teachers play in their student’s lives, an

important strategy of CEPIDEA has targeted teacher training before

the intervention was implemented (Smith, 2011). Thus, teachers

participating in the intervention attended 7 training sessions (2 hours

per session) that were geared toward: (a) providing the theoretical

background and goals of the intervention; (b) learning to recognize

and reinforce students’ prosocial behaviors; (c) acquiring interven-

tion procedures; and (d) promoting a collaborative climate. Subse-

quently, all members of the research staff participated in monthly

school council meetings in order to ensure both program implemen-

tation and fidelity. Research staff also met with teachers individu-

ally and in small groups when necessary to provide support and

address potential concerns. The school principal was involved in

each phase of the program implementation.

The present study

In accordance with the aims of the CEPIDEA program, the present

study focuses on its efficacy as attested by changes in specific pro-

social behaviors. As the tendency to benefit others may find

expression in diverse courses of actions, our primary intention was

to examine the effect of CEPIDEA on two representative dimen-

sions of prosocial behaviours that are consoling and helping. In

focusing on these prosocial behaviors we acknowledged the role

they play in both cognitive and social development. While helping

behavior (i.e., behavior aimed at alleviating an instrumental or

material need) emerge earlier and is crucial to recruit others’ recog-

nition, consoling behavior (i.e., behavior aimed at alleviating emo-

tional needs) emerge later as it rests upon more sophisticated

emphatic abilities (Dunfield, Kuhlmeier, O’Connell, & Kelley,

2011). We expected a positive impact of our intervention on both

helping and consoling, although we did not advance any hypothesis

regarding the degree of change between the two behaviors.

Moreover, in line with our theoretical premises, we assigned a

crucial role to empathic self-efficacy beliefs in fostering prosocial

behaviors. In viewing self-efficacy beliefs as knowledge struc-

tures able to vehicle and sustain helping and consoling, we placed

special emphasis on enabling children to cope efficaciously with

emotion regulation, perspective taking and interpersonal commu-

nication (i.e., the different components of the CEPIDEA pro-

gram). This was achieved through the gradual transmission of

proper knowledge and practices.

Furthermore, we viewed decline in aggression and improve-

ment in academic achievement as additional relevant outcomes

one could expect from promoting prosocial behavior. As stated

above, previous findings suggest that prosocial behaviors may

exert a compensatory function in mitigating the destructive conse-

quences of aggression on adolescents’ development (Kokko et al.,

2006). Thus we reasoned that extending children’s behavioral

repertories, by including models and actions conducive to proso-

cial exchanges, could reduce the probability of engaging in

aggressive behaviors (e.g. Dodge et al., 2006). Similar arguments

as those made for prosocial behavior lead us to distinguish phys-

ical and verbal aggression. A large body of literature agrees with

our reasoning in pointing to physical aggression as more reprehen-

sible and dysfunctional than verbal aggression, in particular in

adolescence and amongst boys (Card, Stucky, Sawalani, & Little,

2008; Tremblay, 2000).

Finally, we expected that promoting prosocial behaviour could

result in an improvement of scholastic achievement in accordance

with previous findings, highlighting the long term effect of proso-

ciality on academic performance (Caprara et al., 2000). We rea-

soned that prosocial behaviour could contribute to academic

achievement, either by setting a climate of mutual collaboration

within the classroom to facilitate learning or by offering children

further opportunities to acknowledge, express and value their

talents and efforts.

Overall, in pursuing the enhancement of prosocial behaviors

within the classroom, CEPIDEA paid close attention to its impact

on peer relationships using peer ratings to assess the changes in

behavior due to the intervention. Time constraints were a matter

of concern, however; which prompted us to choose the most rele-

vant markers of the main behavioral outcomes (i.e., helping and

consoling for prosocial behaviors; and verbal and physical aggres-

sion for aggressive behaviors). This choice enabled us to address

the school’s concerns and maintain participant involvement during

the assessment.

In sum, we anticipated that students who participated in the

intervention, compared to those who did not, would in general score

higher in empathic self-efficacy and prosocial behavior, while scor-

ing lower in aggression and performing better at school. Taking into

Table 1. Summary table of intervention.

Components Prosocial sessions

Sensitization to

prosocial values

� The benefits of prosociality � Monitoring the enactment of prosocial

behaviors within the classroom

� Prosocial values, dilemmas and choices Emotion regulation

skills

� Acknowledging feelings and emotions � Awareness of negative emotions and its

regulation

� Recognizing others’ emotions � Positive emotions and their expression

Perspective-taking

skills

� Recognizing others’ perspectives � Recognizing others’ needs � Helping others in trouble

Interpersonal-

communication

skills

� Dealing with peer group demands � Communication skills and assertiveness � Communication skills and positive

emotions

Precursors of civic

engagement

� From benevolence to universalism � Strategic planning for helping � Goal setting for helping our school/

neighborhood

388 International Journal of Behavioral Development 38(4)

account that trait agreeableness could predispose to prosocial beha-

viors (Caprara et al., 2012), we controlled for the role it might play

in affecting the changes targeted by the intervention. Finally, in

accordance with the PYD approach, we hypothesized that an

increase of prosocial behaviors due to the CEPIDEA program may

influence the reduction of aggressive behaviors and mediate the

effect of the intervention on aggressive conduct.

Method

Participants

The study took place in a public middle school located in Genzano, a

small city near Rome. In agreement with the definition of universal

programs (Institute of Medicine, 1994), the CEPIDEA intervention is

targeted to the general population of early adolescents (i.e., the target

participants of the program have not been identified on the basis of

individual risks). The intervention group included 151 students

(52.3% males; M age ¼ 12.4, SD ¼ .49) in 7 classrooms, and the control group included 173 students (50.3% females; M age ¼ 13.0, SD ¼ .53) in 8 classrooms. Both the intervention and control group primarily comprised of Italian participants (95%) with a small percentage of participants of other ethnicities (5%). The majority of the students in our sample were from intact families (92.4%), with only 7.6% were from single-parent homes (i.e., parents were sepa- rated or divorced). The majority of the parents had a high school

degree (46.4%), whereas 23.7% had a university degree or above. Approximately 20.5% of parents finished middle school and only 9.4% achieved an elementary or less-than-elementary school educa- tion. Approximately 14% of parents were in professional or manage- rial employment; 25% were merchants or operators of other businesses; 31% were skilled workers; 29% were unskilled workers; and 1% were retired. Participants completed baseline measures, which included the assessment of prosocial and aggressive beha-

viors, and all of them scored in ranges typically found in community

samples (Capaldi & Owen, 2001; Caprara et al., 2012).

Design

All participants attended seventh grade in 2008–2009 (control

group) and 2009–2010 (intervention group). As noted in Table 2,

both groups were assessed sequentially at three different time

points (i.e., the pre-test, post-test, and follow-up), each at

6-month intervals. School records of academic achievement at the

end of sixth grade and eighth grade (the first and the last year of

middle school in Italy, respectively) were collected.

Procedures

The entire study was conducted over a 28-month period (from Feb-

ruary 2009 to July 2011) and included assessments, teacher

training, and classroom activities. The first 7 months were primarily

devoted to preparing for the implementation of the intervention,

which was first presented to the School Council and the assembly

of teachers for approval. Parents were informed via letters and an

informational meeting. Informed parental consent was obtained for

almost all participating students (98.5%). Forty teachers had been invited to attend training sessions and almost 95% of them partici- pated. Students from both groups were assessed in three waves

through questionnaires (self-report and peer-report) pertaining to

the variables examined. Questionnaires were administered in each

classroom by two trained and knowledgeable researchers during

school hours. Finally, intervention fidelity was controlled by: (a)

manualization of prosocial sessions; (b) regular communication

with, and ongoing supervision of, teachers; (c) weekly staff meet-

ings; and (d) an ad hoc checklist completed by the staff at the end

of each prosocial session with an aim to evaluate adherence to the

programmed specifications.

Instruments

Agreeableness (AGR). Participants rated their AGR on 6 items (1 ¼ almost never; 5 ¼ almost always) in a condensed version of the Big- Five Questionnaire-Children (BFQ-C, Barbaranelli, Caprara,

Rabasca, & Pastorelli, 2003). The items assessed children’s sensi-

tivity toward others and their needs (e.g. ‘‘I treat my peers with love

and warmth’’). The original BFQ-C has been validated in Italy

(Barbaranelli, Fida, Paciello, Di Giunta, & Caprara, 2008). Cron-

bach’s alphas were .70, .78, and .75, respectively, for pre-test,

post-test, and follow-up.

Empathic Self-Efficacy Beliefs (ESE). Participants rated their per- ceived capability to sense another person’s feelings and to respond

empathetically to another’s distress and misfortune (e.g. ‘‘How well

can you experience how a person in trouble feels?’’) on a 6-item

scale (1 ¼ not well at all; 5 ¼ very well). The psychometric prop- erties of the scale have been validated in Italy and other countries

(Di Giunta et al., 2010). Cronbach’s alphas were .73, .77, and

.80, respectively, for pre-test, post-test, and follow-up.

Prosocial behavior. Participants rated each classmate on two items intended to evaluate the degree of frequency (1 ¼ never/almost never to 5 ¼ often) of helping and consoling behaviors. The two items were derived from the peer nomination of prosocial behavior

(Caprara & Pastorelli, 1993) and were reworded in the third person

(‘‘How many times does ____ help others?’’; ‘‘How many times

does ____ console others?’’). Following standard procedures (e.g.

Visconti & Troop-Gordon, 2010), the consoling and helping beha-

viors of each participant were assessed by averaging the ratings of

at least 20 classmates. Reliabilities for both behaviors were com-

puted through test-retest correlations and they were: .81 and .73

Table 2. Timeline of CEPIDEA assessment.

April 2009 October 2009

November 2009–

March 2010 April 2010 October 2010

Intervention group – Pre-test

(n ¼ 151) CEPIDEA intervention Post-test

(n ¼ 151) Follow-up

(n ¼ 146) Control group Pre-test

(n ¼ 173) Post-test (n ¼ 159) – Follow-up

(n ¼ 140) –

Caprara et al. 389

(pre-test/post-test); .78 and .70 (pre-test/follow-up); and .83 and .76

(post-test/follow-up). The correlations between the two items were

high, and ranged from .71 to .88.

Aggressive behavior. We assessed two different kinds of aggressive behaviors by using a two item peer-report with approximately 20

classmates as evaluators. The two items assessed the degree of fre-

quency (1¼ never/almost never to 5 ¼ often) of physical and verbal aggression of each participant. The items were derived from the self-

report Physical and Verbal Aggression Scale (Caprara & Pastorelli,

1993) and were reworded in the third person (‘‘How many times does

____ kick or hit others?’’; ‘‘How many times does ____ insult

others?’’). Reliabilities for both behaviors (computed through test-

retest correlations) were: .81 and .84 (pre-test/post-test); .77 and

.77 (pre-test/follow-up); and .79 and .83 (post-test/follow-up). The

correlations between the two items ranged from .69 to .87.

Academic achievement. Participants’ achievement scores were col- lected at two time points: (1) at the end of the sixth grade (before the

beginning of any intervention); and (2), at the end of eighth grade

(the end of middle school) from original school records. In the Italian

school system, teachers evaluate their students by using a 10-level

scale for each subject (1 ¼ extremely insufficient to 10 ¼ excellent). We created a composite measure of academic achievement from stu-

dents’ grades in their primary school subjects: Italian, Math, Science,

Foreign Language (English and French), and Social Studies.

Data analytic approach

To examine a potential intervention effect, we employed a Latent

Growth Curve approach (LGC; Bollen & Curran, 2006). Specifi-

cally, LGC enabled us to identify two growth latent factors that rep-

resent the initial status (the intercept) and change over time (the

slope) by using the means of the observed variables at each time-

point. First, we assessed intervention effects by implementing a con-

ditional LGC with treatment condition (TC) as predictor. A significant

effect of TC on slope was the effect of intervention on change over

time in the outcome considered. In order to adjust for initial pre-test

scores, we regressed the slope on the intercept. Moreover, we con-

trolled for possible gender effects and for socio-demographic vari-

ables that were statistically different between the two groups by

inserting them in our LGCs. In addition, we decided to model individ-

ual reactions to TC, controlling for agreeableness by inserting it as a

time-varying covariate. 1

The general LGC model used to examine

intervention effect for each construct is depicted in Figure 1.

Secondly, in light of the univariate results, we tested specific

mediational models by using a parallel process growth curve model

with the slope as the putative mediator (von Soest & Hagtvet,

2011). The mediated effect is indicated by the effect of the interven-

tion on mediator (a) multiplied by the effect of mediator on out- come (b). The distribution of the product coefficients (ab) method is used for calculating the 95% asymmetric confidence intervals (CI) for mediated effects in the PRODCLIN program

(Tofighi & MacKinnon, 2011). Given the non-perfect normality

of the variables, maximum-likelihood with robust standard errors

was chosen as the method for estimating parameters (Muthén &

Muthén, 2006). Due to the sensitiveness of the �2 statistic to sample size, we also considered comparative fit index (CFI) and non-

normative fit index (NNFI) > .90; root mean square of approxima-

tion (RMSEA) < .10; and standardized root mean square residual

(SRMR) < .08 as indicators of acceptable model fit (Kline,

2010). Full information maximum-likelihood (FIML) was used as

a method for estimating missing data. Compared to other traditional

techniques (e.g. single imputation method), FIML produces less

biased estimates of missing values even when the pattern of miss-

ingness is selective and cannot be ignored (Baraldi & Enders,

Figure 1. General conditional latent growth curve model.

Note. Intervention group (n ¼151), Control group (n ¼ 173). Treatment Condition (0 ¼ control group; 1 ¼ intervention group); Gender (0 ¼ male; 1 ¼ female).

390 International Journal of Behavioral Development 38(4)

2010). All the LGCs were analyzed in Mplus 5.1 (Muthén &

Muthén, 2006).

Finally, as we have academic achievement data from two time

points, we assessed intervention effect on academic achievement

at the end of middle school by implementing a hierarchical regres-

sion analysis (HRA) in SPSS 18. In the first step, we modeled aca-

demic achievement at sixth grade; gender; age; and agreeableness

at pre-test. In the second step, we added TC to the model in an effort

to evaluate whether students in the intervention group significantly

increased their academic performance above and beyond the role

played by our control variables.

Results

Baseline comparisons

The comparability of the two groups in terms of gender composi-

tion, participant age, and socioeconomic status (SES) was evaluated

through chi-square and analyses of variance (ANOVA). We found

the intervention and control groups to be equivalent in terms of gen-

der composition � 2 (N ¼ 324, df ¼ 1) ¼ .219, p ¼ .640, and SES by

using father’s and mother’s educational level as a proxy indicator

of SES, respectively, �2(N ¼ 271, df ¼ 5) ¼ 1.223, p ¼ .942, and �

2 (N ¼ 275, df ¼ 5) ¼ 2.760, p ¼ .737). The control group was

6-months older than the intervention group F(1, 309) ¼ 82.026, p ¼ .000. Thus, age was included in LGCs as a covariate. Table 3 reports the descriptive statistics for each variable.

As we worked with latent variables, we assessed differences

between the intervention and control group by correlating the TC

with the initial status (von Soest & Hagtvet, 2011). A statistically

significant correlation between TC and the intercept represented a

significant difference in the initial status between intervention and

control group. As reported in Table 4 (see raw TC-Intercept), the

only difference in the initial level was for helping behavior, which

was higher in the control group. 2

Attrition analyses

All participants in the intervention group were assessed at the

post-test, whereas 14 participants in the control group (4.3%) missed data collection for this assessment. Thirty-eight partici-

pants (11.7%) were missing at the 6-month follow-up (33 subjects in the control group and 5 in the intervention group). Cohen’s d

(1988) was used to offer an indication of the magnitude of effect

size (small d � .20; medium d � .50; large d � .80) between those who were missing at the follow-up and those who were present at

the pre-test.

Participants from the intervention group who dropped out of the

follow-up phase reported higher empathic self-efficacy (respec-

tively, M ¼ 3.80, M ¼ 3.51, d ¼ .48), lower consoling behavior (respectively, M ¼ 2.33, M ¼ 2.84, d ¼ �.79), higher physical (respectively, M ¼ 1.73, M ¼ 1.48, d ¼ .46) and verbal aggression (respectively, M ¼ 2.00, M ¼ 1.72, d ¼ .45), and no meaning- ful differences on helping behavior (respectively, M ¼ 2.71,

Table 3. Descriptive statistics.

Pre-test Post-test Follow-up

Boys Girls Boys Girls Boys Girls

Agreeableness

Intervention 3.35 (0.62) 3.59 (0.59) 3.26 (0.80) 3.52 (0.56) 3.39 (0.73) 3.69 (0.59)

Control 3.16 (0.61) 3.49 (0.65) 3.29 (0.55) 3.49 (0.70) 3.31 (0.63) 3.51 (0.74)

Emphatic self-efficacy

Intervention 3.49 (0.60) 3.54 (0.61) 3.60 (0.67) 3.74 (0.61) 3.57 (0.64) 3.76 (0.65)

Control 3.43 (0.59) 3.71 (0.56) 3.38 (0.56) 3.75 (0.60) 3.44 (0.59) 3.89 (0.64)

Helping

Intervention 2.56 (0.56) 2.95 (0.53) 2.85 (0.58) 3.19 (0.51) 2.85 (0.57) 3.21 (0.51)

Control 2.55 (0.63) 3.06 (0.56) 2.68 (0.57) 3.17 (0.51) 2.81 (0.46) 3.24 (0.42)

Consoling

Intervention 2.57 (0.60) 3.09 (0.60) 2.80 (0.61) 3.25 (0.51) 2.80 (0.59) 3.31 (0.52)

Control 2.51 (0.63) 3.18 (0.54) 2.65 (0.58) 3.35 (0.52) 2.75 (0.54) 3.44 (0.45)

Physical aggression

Intervention 1.73 (0.59) 1.22 (0.30) 1.78 (0.59) 1.22 (0.23) 1.66 (0.50) 1.21 (0.23)

Control 1.73 (0.71) 1.30 (0.36) 1.62 (0.62) 1.23 (0.32) 1.82 (0.57) 1.47 (0.35)

Verbal aggression

Intervention 2.03 (0.65) 1.40 (0.39) 2.15 (0.61) 1.47 (0.36) 2.07 (0.68) 1.50 (0.39)

Control 2.06 (0.74) 1.58 (0.47) 2.02 (0.66) 1.62 (0.44) 2.26 (0.65) 1.88 (0.47)

Sixth grade Eighth grade

Boys Girls Boys Girls

Academic achievement

Intervention 6.90 (0.74) 7.16 (0.87) 6.95 (0.92) 7.55 (1.17)

Control 7.34 (1.31) 7.73 (1.13) 6.95 (1.00) 7.23 (0.91)

Note. Means and standard deviations (in parentheses). Intervention group (n ¼151), Control group (n ¼ 173). Agreeableness, Empathic self-efficacy, Helping, Consoling, Physical aggression, and Verbal aggression are rated on a 5-point Likert scale, ranging from 1 to 5. Possible scores for Academic achievement range from 1 to 10. For all variables, higher scores indicate more of that quality.

Caprara et al. 391

M ¼ 2.74, d ¼ �.06), when compared to their counterparts at the pre-test. Participants who dropped out of the control group

at the follow-up, when compared to their counterparts at the

pre-test, reported higher empathic self-efficacy (respectively,

M ¼ 3.72, M ¼ 3.54, d ¼ .30), lower consoling (respectively, M ¼ 2.58, M ¼ 2.90, d ¼ �.48) and helping behavior (respec- tively, M ¼ 2.64, M ¼ 2.84, d ¼ �.31), and higher physical (respectively, M ¼ 1.90, M ¼ 1.43, d ¼ .82) and verbal aggres- sion (respectively, M ¼ 2.21, M ¼ 1.73, d ¼ .75).

Intervention effects

Intervention effects were estimated by using the latent slope means

derived from our latent variables. As reported in Table 4 (see raw

TC ! Slope), we found positive intervention effects that were small/medium. We computed a similar measure of Cohen’s d mea-

sure of effect size (Cohen, 1988) by using the difference in model-

implied estimated mean for our control and intervention groups

divided by the standard deviation for the outcome at the last time

point. This procedure is more appropriate for models in which

means are a complex function of several covariates (Muthén & Cur-

ran, 1997). We found a small increase in helping behavior (d ¼ .22), a small decrease of verbal aggression (d ¼ �.18), and a small- medium decrease in physical aggression (d ¼�.44) in the interven- tion group relative to the control group.

LGC mediational models

According to the LGCs results, we tested two plausible media-

tion models in which we considered the increase in helping

behavior as responsible for the lower growth rate of verbal

and physical aggression in the intervention group. In Figure 2,

the mediational model with helping as mediator and verbal

aggression as outcome yielded an overall moderate fit

�2(34) ¼ 114.473, Scf ¼ 1.075, p ¼ .000, CFI ¼ .94, NNFI ¼ .91, RMSEA ¼ .08, SRMR ¼ .07. The effect of TC on the helping growth was significant (a ¼ .112, Standard Error (SE) ¼ .050, p ¼ .024) and the effect of the latter on the verbal aggression slope was also significant (b ¼ �.365, SE ¼ .157, p ¼ .020), The estimated mediated effect (ab ¼ �.041, 95% CI ¼ �.102, �.001) was statistically significant as the 95% asymmetric lower and upper CI limits did not include zero.

As the direct effect of TC on verbal aggression was also signif-

icant (�’ ¼ - .079, SE ¼ .036, p ¼ .027), the growth in helping behavior partially mediated the effect of the intervention on

counteracting verbal aggression.

The mediational analysis investigating the role of helping on

counteracting physical aggression, �2(33) ¼ 62.097, Scf ¼ 1.113, p ¼ .002, CFI ¼ .98, NNFI ¼ .96, RMSEA ¼ .05, SRMR ¼ .08, evinced an effect of TC on helping (a ¼ .118, SE ¼ .048, p ¼ .014), but not a statistically significant effect on physical aggression

slope (b ¼ .086, SE ¼ .173, p ¼ .618). The mediated effect was not statistically significant (ab ¼ .010, 95% CI ¼ �.033, .060), whereas the direct effect of TC on physical aggression was signif-

icant (�’ ¼ �.238; SE ¼ .042, p ¼ .000).

Academic achievement

In the first step of the HRA, a large amount of variance has been

explained by our control variables (R 2 ¼ .52). In the last step, TC

significantly predicted (b¼ .315, p ¼ .000) later academic achieve- ment (�R2 ¼ .07; R2 ¼ .59) above and beyond previous academic achievement (b ¼ .764, p ¼ .000), gender (b ¼ .101, p ¼ .010), age (b ¼ .083, p ¼ .056), and AGR pre-test (b ¼ �.018, p ¼ .639), demonstrating that the intervention group, relative to the control

group, significantly improved academically.

Table 4. Conditional latent growth curve models.

Model’s fit � 2 (df) Scf CFI NNFI RMSEA SRMR

Emphatic self-efficacy 20.361 (16) 1.199 0.98 0.97 0.04 0.05

Helping 32.171 (15)** 1.024 0.97 0.96 0.06 0.07

Consoling 27.245 (15) *

0.998 0.98 0.98 0.05 0.05

Physical aggression 23.940 (14) *

1.108 0.99 0.98 0.05 0.06

Verbal aggression 52.164 (16)*** 1.154 0.95 0.94 0.08 0.07

Estimated parameters Emphatic self-efficacy Helping Consoling Physical aggression Verbal aggression

Correlations

Treatment condition – Intercept �.008 �.133* �.092 .083 .053 Treatment condition – Age �.024 �.024 �.024 �.024 �.024 Treatment condition – Gender �.458*** �.458*** �.458*** �.458*** �.458***

Unstandardized beta weights

Treatment Condition ! Slope �.018 .111* �.013 �.231*** �.118*** Gender ! Slope .089* .065 .137** .024 �.001 Age ! Slope �.006 �.006 �.005 .000 .000 Intercept ! Slope �.096 �.203** �.203*** �.078* �.026 Gender ! Intercept .128* .425*** .563*** �.458*** �.539*** Age ! Intercept .002 �.012* �.014** .018** .017** Agreeableness (pre-test) ! y (pre-test) .178*** .068* .083** �.008 �.015 Agreeableness (post-test) ! y (post-test) .178*** .068* .083** �.008 �.015 Agreeableness (follow-up) ! y (follow-up) .178*** .068* .083** �.008 �.015

Note. Intervention group (n ¼ 151), Control group (n ¼ 173). df ¼ Degrees of freedom; Scf ¼ Scaling correction factor; CFI ¼ Comparative fit index; NNFI ¼ Non- normative fit index; RMSEA ¼ Root mean square of approximation; SRMR ¼ Standardized root mean square residual; Treatment Condition (0 ¼ control group; 1 ¼ intervention group); Gender (0 ¼ male; 1 ¼ female). Agreeableness, Empathic self-efficacy, Helping, Consoling, Physical and Verbal aggression are rated on a 5-point Likert scale, ranging from 1 to 5. For all variables, higher scores indicate more of that quality. *p < .05; **p < .01; ***p < .001.

392 International Journal of Behavioral Development 38(4)

Discussion

The current study was conducted to ascertain the extent to which an

intervention entirely designed for the promotion of prosocial beha-

viors in young adolescents has been effective. Our analysis reported

some promising effects of the school-based, pilot program CEPI-

DEA, on fostering helping behaviors, reducing physical and verbal

aggression, and increasing academic achievement. The present

findings indicate that the enactment of prosocial behaviors contrib-

ute to counteracting aggression and suggests that interventions

aimed at promoting positive developmental processes may have the

potential to redirect negative trajectories of functioning (Schwartz

et al., 2011).

In particular, adolescents in the intervention group reported a

small but significant increase in helping behaviors; above and

beyond their own agreeableness tendencies. This result is encoura-

ging, given that participants reported an increase in the most obser-

vable form of prosocial behavior that can lead to immediate awards

and rewards from peers, and to positive chain reactions (Eisenberg

et. al., 2006). Moreover, our hypothesis regarding the beneficial

impact of the CEPIDEA intervention on reducing aggressive beha-

vior has been confirmed, particularly as the intervention group

showed decreases in physical and verbal aggression compared with

the control group.

Alternatively, consoling behavior and empathic self-efficacy

did not increase significantly in the intervention group. This result

may be due to the fact that consoling behavior concerns more inti-

mate and less visible interactions than helping behavior; thus help-

ing might be easily captured by peer-ratings. Furthermore,

consoling behavior, compared to helping behavior, requires sophis-

ticated emotional and interpersonal abilities that are less easy to

detect and develop (Eisenberg et al., 2006).

Regarding the lack of any appreciable increase in empathic self-

efficacy, it cannot be ignored that adolescents may have difficulties

either acknowledging or reporting changes in their emphatic capa-

cities within a relatively short period of time. In particular, one can

guess that as more intervention effects concern complex emotional

phenomena (such as consoling and empathic self-efficacy), more

time is required for changes to be enacted and stabilized.

Overall, although the size of the effects of the CEPIDEA pro-

gram is substantially small (Cohen, 1988), it is noteworthy that the

magnitude of these changes are in line with a recent meta-analysis

of SEL programs in school contexts in which the authors reported a

small effect size (around .20) for positive behaviors and conduct

problems as acceptable (Durlak et al., 2011). However, the effects

of our pilot intervention may be taken as promising points of depar-

tures in view of further improvements.

Among other promising results of the intervention, it is worth

noting that adolescents’ helping behavior mediated the decline of

verbal aggression. This result is in accordance with findings attest-

ing to the compensatory additive effect of prosocial behavior in

counteracting aggressive behaviors (Kokko et al., 2006). However,

the mediational role of helping behavior was not found for

decreases in physical aggression. Probably physical aggression,

as a more strongly sanctioned behavior at school relative to verbal

aggression, garners immediate attention by teachers and the whole

school (Tremblay, 2000). Thus, the effect of the CEPIDEA pro-

gram in reducing physical aggression may be due to other interven-

ing mechanisms involving the classroom context (e.g. teachers’

positive classroom management).

Interestingly, participants targeted by the intervention obtained

better grades at the end of middle school, beyond the high stability

of their previous academic grades. Even if the CEPIDEA program

was not designed directly for the enhancement of adolescents’

learning skills, this result is in line with prior research linking pro-

social actions to scholastic achievement (Caprara et al., 2000;

Wentzel, 1993). In this regard, we cannot ignore that the training

experienced by teachers may have had an effect on their teaching

Figure 2. Latent growth curve mediational model.

Note. Intervention group (n ¼151), Control group (n ¼ 173). For simplicity we omitted the control variables gender, age, and agreeableness. Only statistically significant coefficients (standardized) were reported. Dashed lines represent not statistically significant paths (p > .05). * p <.05; **p < .01.

Caprara et al. 393

styles, which in turn may have affected the classroom climate.

Indeed, recent findings (Jennings & Greenberg, 2009) suggest that

a more positive classroom climate, resulting from improved coop-

eration among students and between students and teachers, can

support a more regulated environment for learning that can con-

tribute to better grades. Nor we cannot exclude that this effect

could be due to other factors such as pre-existing differences

between the trajectories of the two groups (e.g. the control group’s

achievement score trajectory seems to decrease compared to the

intervention group) or the phenomenon of the regression toward

the mean. Although we controlled for previous academic achieve-

ment in our analysis, future implementations of CEPIDEA can

benefit from using randomized designs which can allow for a

more reliable ‘‘causal explanation’’ (Cook & Campbell, 1979)

of results. In addition, future studies should also consider the

specific mechanisms undergirding the effects of the CEPIDEA

program on students’ academic achievement.

Due to the high social desirability of prosocial behavior

(Crothers & Levinson, 2004), one strength of our study was the use

of peer-rating measures, with at least 20 classmates evaluating each

others’ behavior. Nonetheless, we are aware of limitations of our

study. In particular, the lack of randomized design could undermine

the validity of our results. However, as noted by baseline compar-

isons, the intervention and control group results seem comparable,

and we attempted to account for the main differences between the

two groups (e.g. age). In terms of attrition effects, pre-test compar-

isons indicated that both in the intervention and control groups, the

few subjects who dropped-out at the follow-up were initially less

prosocial and more aggressive compared to their counterparts

(i.e., those participants who were assessed at the follow-up).

Although these data might suggest the presence of selective attri-

tion, we are quite comfortable with the quality of our intervention

effects since (i) the attrition rate was relatively small, (ii) we used

FIML estimation that has been found to be an adequate way to deal

with missing data even when the missingness is selective (Baraldi

& Enders, 2010), and (iii) we modeled the change over time of our

outcomes while controlling for initial pre-test scores. Another lim-

itation is the use of single items to evaluate different prosocial and

aggressive behaviors. Although this approach allowed us to have

information about specific behavioral outcomes, it is possible that

we may have lost a more complete and representative dimension

of the constructs assessed.

In addition, we are aware that ‘‘contamination effects’’ (e.g.

teacher behavior, school-climate, etc.) could have occurred in the

control group (in particular at the follow-up) since both intervention

and control groups were present in the same school. In this regard,

future implementations of the CEPIDEA program should be con-

ducted by using groups from different schools in order to avoid dif-

fusion of treatment effects (Cook & Campbell, 1979).

Among other issues to be addressed by future studies is the

necessity to look at CEPIDEA’s effectiveness. Hence, one should

focus on the ecological validity of the program, namely the extent

to which it might be extended across different real world condi-

tions. Further steps should lead to strategies designed to enable

schools to adopt the intervention without such a substantial involve-

ment of researchers.

Despite these limits, the pilot school-based intervention CEPI-

DEA supports and expands existing evidence regarding the

enhancement of prosocial behavior within the school environment

and in the classroom context. Indeed, the present findings may be

helpful for administrators and school counselors as they highlight

the relevance of promoting prosocial behavior in school settings

as a strategy for counteracting aggressive conduct and for enhan-

cing academic achievement during adolescence. Moreover, since

the activities (especially prosocial lessons) were incorporated into

routine educational practices, school psychologists and teachers

interested in promoting positive developmental outcomes in stu-

dents, may find the CEPIDEA program as a useful tool to success-

fully build positive learning environments.

Funding

This study has been funded by the Italian Ministry of Health as part

of a National Strategic Research Program (grant RFPS-2007-5-

641730) on adolescent mental health.

Notes

1. In a previous step, we controlled that agreeableness was not

affected by the treatment condition. We also tested preliminary

models without adding agreeableness as a time-varying covari-

ate, in order to control for possible distortion due to a conceptual

overlap between agreeableness and some outcomes (in particu-

lar, helping and consoling): the intervention effects remained

substantially the same.

2. The full correlation matrix can be requested from the corre-

sponding author.

References

Bandura, A. (1997). Self-efficacy: The exercise of control. New York,

NY: Freeman.

Baraldi, A. N., & Enders, C. K. (2010). An introduction to modern

missing data analyses. Journal of School Psychology, 48, 5–37.

doi:10.1016/j.jsp.2009.10.001

Barbaranelli, C., Caprara, G. V., Rabasca, A., & Pastorelli, C. (2003). A

questionnaire for measuring the Big Five in late childhood. Person-

ality and Individual Differences, 34, 645–664. doi:10.1016/

S0191-8869(02)00051-X

Barbaranelli, C., Fida, R., Paciello, M., Di Giunta, L., & Caprara, G. V.

(2008). Assessing personality in early adolescence through

self-report and other-ratings: A multitrait-multimethod analysis of

the BFQ-C. Personality and Individual differences, 44, 876–886.

doi:10.1016/j.paid.2007.10.014

Battistich, V., Schaps, E., Watson, M., Solomon, D., & Lewis, C.

(2000). Effects of the Child Development Project on students’ drug

use and other problem behaviors. Journal of Primary Prevention,

21, 75–99. doi:10.1023/A:1007057414994

Bierman, K. L., Smoot, D. L., & Aumiller, K. (1993). Characteristics of

aggressive-rejected, aggressive (nonrejected), and rejected (nonag-

gressive) boys. Child Development, 64, 139–151. doi:10.1111/j.

1467-8624.1993.tb02900.x

Bollen, K. A., & Curran, P. J. (2006). Latent curve models: A structural

equation approach. Hoboken, NJ: Wiley. doi:10.1002/0471746096

Capaldi, D., & Owen, L. D. (2001). Physical aggression in a community

sample of at-risk young couples: Gender comparisons for high fre-

quency, injury and fear. Journal of Family Psychology, 15(3),

425–440. doi:10.1037/0893-3200.15.3.425

Caprara, G. V., Alessandri, G., & Eisenberg, N. (2012). Prosociality:

The contribution of traits, values and self-efficacy beliefs. Journal

of Personality and Social Psychology, 102, 1289–1303. doi:10.

1037/a0025626

394 International Journal of Behavioral Development 38(4)

Caprara, G. V., Barbaranelli, C., Pastorelli, C., Bandura, A., & Zimbardo,

P. (2000). Prosocial foundations of children’s academic achievement.

Psychological Science, 11, 302–306. doi:10.1111/1467-9280.00260

Caprara, G. V., & Pastorelli, C. (1993). Early emotional instability,

prosocial behavior, and aggression: Some methodological aspects.

European Journal of Personality, 7, 19–36. doi:10.1002/per.

241007010

Card, N. A., Stucky, B. D., Sawalani, G. M., & Little, T. D. (2008).

Direct and indirect aggression during childhood and adolescence:

A meta-analytic review of gender differences, intercorrelations, and

relations to maladjustment. Child Development, 79(5), 1185–229.

doi:10.1111/j.1467-8624.2008.01184.x

Cohen, J. (1988). Statistical power analysis for the behavioral sciences

(2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.

Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design

and analysis for field settings. Chicago, IL: Rand McNally.

Council of Europe. (2010). Council of Europe Charter on education for

democratic citizenship and human rights education. Strasbourg,

France: Council of Europe Publishing.

Crothers, L. M., & Levinson, E. M. (2004). Assessment of bullying: A

review of methods and instruments. Journal of Counseling and

Development, 82, 496–503. doi:10.1002/j.1556-6678.2004.tb00338.x

Di Giunta, L., Eisenberg, N., Kupfer, A., Steca, P., Tramontano, C., &

Caprara, G. V. (2010). Assessing perceived empathic and social

self-efficacy across countries. European Journal of Psychological

Assessment, 26, 77–86. doi:10.1027/1015-5759/a000012

Dodge, K. A., Coie, J. D., & Lynam, D. (2006). Aggression and antiso-

cial behavior in youth. In: W. Damon & R. M. Lerner (series eds.),

& N. Eisenberg (volume ed.), Handbook of child psychology: Vol. 3.

Social, emotional, and personality development (6th ed., pp.

719–788). New York, NY: Wiley.

Dunfield, K., Kuhlmeier, V. A., O’Connell, L., & Kelley, E. (2011).

Examining the diversity of prosocial behavior: Helping, sharing,

and comforting in infancy. Infancy, 16, 227–224. doi:10.1111/j.

1532-7078.2010.00041.x

Durlak, J. A., Weissberg, R. P., Dymnicki, A. B., Taylor, R. D., &

Schellinger, K. B. (2011). The impact of enhancing students’ social

and emotional learning: A meta-analysis of school-based universal

interventions. Child Development, 82, 405–432. doi:10.1111/j.

1467-8624.2010.01564.x

Eisenberg, N., Fabes, R. A., & Spinrad, T. L. (2006). Prosocial behavior. In

W. Damon & R.M. Lerner (series ed.), & N. Eisenberg (volume ed.),

Handbook of child psychology: Vol. 3. Social, emotional, and person-

ality development (6th ed., pp. 646–718). New York, NY: Wiley.

Graziano, W. G., & Eisenberg, N. (1997). Agreeableness: A dimen-

sion of personality. In R. Hogan, J. Johnson & S. Briggs (Eds.),

Handbook of personality psychology (pp. 795–824). San Diego,

CA: Academic Press.

Holsen, I., Smith, B. H., & Frey, S. K. (2008). Outcomes of the social com-

petence program Second Step in Norwegian elementary schools. School

Psychology International, 29, 71–88. doi:10.1177/0143034307088504

Hutchings, J., Bywater, T., Gridley, N., Whitaker, C., Martin-Forbes, P.,

& Gruffyd, S. (2012). The incredible years therapeutic social and

emotional skills program: A pilot study. School Psychology Interna-

tional, 33, 285–293. doi:10.1177/0143034311415899

Institute of Medicine. (1994). Reducing risks for mental disorders:

Frontiers for preventive intervention research. Washington, DC:

National Academy Press.

January, A. M., Casey, R. J., & Paulson, D. (2011). A meta-analysis of

classroom-wide interventions to build social skills: Do they work?

School Psychology Review, 40, 242–256.

Jennings, P. A., & Greenberg, M. T. (2009). The prosocial classroom:

Teacher social and emotional competence in relation to child and

classroom outcomes. Review of Educational Research, 79,

491–525. doi:10.3102/0034654308325693

Kline, R. B. (2010). Principles and practice of structural equation

modeling (3rd ed.). New York, NY: Guilford.

Kokko, K., & Pulkkinen, L. (2000). Aggression in childhood and

long-term unemployment in adulthood: A cycle of maladaptation

and some protective factors. Developmental Psychology, 36,

463–472. doi:10.1037/0012-1649.36.4.463

Kokko, K., Tremblay, R. E., Lacourse, E., Nagin, D. S., & Vitaro, F.

(2006). Trajectories of prosocial behavior and physical aggression

in middle childhood: Links to adolescent school dropout and phys-

ical violence. Journal of Research on Adolescence, 16, 403–428.

doi:10.1111/j.1532-7795.2006.00500.x

Krueger, R. F., Hicks, B. M., & McGue, M. (2001). Altruism and anti-

social behavior: Independent tendencies, unique personality corre-

lates, distinct etiologies. Psychological Science, 12, 397–402. doi:

10.1111/1467-9280.0037

Lerner, R. M., Lerner, J. V., Almerigi, J., & Theokas, C. (2005). Pos-

itive youth development: A view of the issues. Journal of Early

Adolescence, 25, 10–16. doi:10.1177/0272431604273211

Luengo Kanacri, B. P., Pastorelli, C., Eisenberg, N., Zuffianò, A., &

Caprara, G. V. (2013). The Development of prosociality from ado-

lescence to early adulthood: The role of effortful control. Journal of

Personality, 81(3), 302–312. doi:10.1111/jopy.12001

Muthén, B., & Curran, P. J. (1997). General longitudinal modeling of

individual differences in experimental designs: A latent variable

framework for analysis and power estimation. Psychological

Methods, 2, 371–402. doi:10.1037/1082-989X.2.4.371

Muthén, L. K., & Muthén, B. O. (2006). Mplus user’s guide.

Los Angeles, CA: Muthén and Muthén.

Nantel-Vivier, A., Kokko, K., Caprara, G. V., Pastorelli, C., Gerbino,

M., Paciello, M., . . . Tremblay, R. (2009). Prosocial development

from childhood to adolescence: A multi-informant perspective

with Canadian and Italian longitudinal studies. The Journal of Child

Psychology and Psychiatry, 50, 590–598. doi:10.1111/j.1469-7610.

2008.02039.x

Pulkkinen, L., & Tremblay, R. E. (1992). Patterns of boys’ social

adjustment in two cultures and at different ages: A longitudinal per-

spective. International Journal of Behavioural Development, 15,

527–553. doi:10.1177/016502549201500406

Schwartz, S. H. (2010). Basic values: How they motivate and inhibit pro-

social behavior. In M. Mikulincer & P. R. Shaver (Eds.), Prosocial

motives, emotions, and behavior: The better angels of our nature (pp.

221–241). Washington, DC: American Psychological Association.

Schwartz, S. J., Phelps, E., Lerner, J. V., Huang, S., Brown, H., Lewin-

Bizan, S., . . . Lerner, R. (2011). Promotion as prevention: Positive

youth development as protective against tobacco, alcohol, illicit

drug, and sex initiation. Applied Developmental Science, 14(4),

197–211. doi:10.1080/10888691.2010.516186

Seligman, M. E. P., & Csikszentmihalyi, M. (2000). Positive psychol-

ogy: An introduction. American Psychologist, 55, 5–14. doi:10.

1037/0003-066X.55.1.5

Smith, P. K. (2011). Why interventions to reduce bullying and violence

in schools may (or may not) succeed: Comments on this Special

Section. International Journal of Behavioral Development, 35,

419–423. doi:10.1177/0165025411407459

Tofighi, D., & MacKinnon, D. P. (2011). RMediation: An R package

for mediation analysis confidence intervals. Behavior Research

Methods, 44, 692–700. doi:10.3758/s13428-011-0076-x

Caprara et al. 395

Tremblay, R. E. (2000). The development of aggressive behaviour dur-

ing childhood: What have we learned in the past century? Interna-

tional Journal of Behavioral Development, 24, 129–141. doi:10.

1080/016502500383232

Veenstra, R., Lindenberg, S., Oldehinkel, A. J., De Winter, A. F., Ver-

hulst, F. C., & Ormel, J. (2008). Prosocial and antisocial behavior in

preadolescence: Teachers’ and parents’ perceptions of the behavior

of girls and boys. International Journal of Behavioral Development,

32, 243–251. doi:10.1177/0165025408089274

Visconti, K. J., & Troop-Gordon, W. (2010). Prospective relations

between children’s behavioral responses to peer victimization and

their socioemotional adjustment. Journal of Applied Developmental

Psychology, 31, 261–272. Retrieved from: http://dx.doi.org/10.

1016/j.appdev.2010.05.003

von Soest, T., & Hagtvet, K. A. (2011). Mediation analysis in a latent

growth curve modeling framework. Structural Equation Modeling,

18, 289–314. doi:10.1080/10705511.2011.557344

Webster-Stratton, C., Reid, M. J., & Hammond, M. (2004). Treating

children with early-onset conduct problems: Intervention outcomes

for parent, child, and teacher training. Journal of Clinical Child and

Adolescent Psychology, 33, 105–124. doi:10.1207/S1537442

4JCCP3301_11

Wentzel, K. R. (1993). Does being good make the grade? Social beha-

vior and academic competence in middle school. Journal of Educa-

tional Psychology, 85, 357–364. doi:10.1037/0022-0663.85.2.357

Yates, M., & Youniss, J. (1996). A developmental perspective on com-

munity service in adolescence. Social Development, 5, 85–111. doi:

10.1111/j.1467-9507.1996.tb00073.x

396 International Journal of Behavioral Development 38(4)