2012 - Effectiveness of Antibullying School Programmes a Systematic Review by Evidence Levels

8/10/2019 2012 - Effectiveness of Antibullying School Programmes a Systematic Review by Evidence Levels

1/13

Effectiveness of antibullying school programmes: A systematic reviewby evidence levels

Jose Antonio Jimnez Barbero a,, Jose AntonioRuiz Hernndez a,Bartolom Llor Esteban a, Mara Prez Garca b

a University of Murcia, Faculty of Pshychology, Espinardo, CP: 30100 Murcia, Spainb University of Murcia, Faculty of Medicine, Espinardo, CP: 30100 Murcia, Spain

a b s t r a c ta r t i c l e i n f o

Article history:

Received 10 February 2012Received in revised form 14 April 2012Accepted 20 April 2012Available online 27 April 2012

Keywords:

Systematic reviewIntervention programmeSchool violencePreventionAttitudesEffectiveness

Increasingly more educational centres are, therefore, carrying out programmes aimed at preventing or reducingviolence in schools.This study seeks to examine the efciency of such programmes in Primary and Secondary schools. The method-ology used is the systematic search of electronic databases (Medline, Trip Database, Cochrane, Academy SearchPremier, PsycINFO, ERIC and PsycARTICLES) for studies published after January1, 2000, on the assessment of theeffectiveness of schoolinterventionsto preventor reduceviolenceand bullying. The study population comprisesschool-age (616 years) children and adolescents of bothsexes.Initially,299 articles were detected that met theinclusion criteria and that had been independently peer-reviewed. For the nal evaluation, 32 studies wereselected which met the previously established selection and quality criteria, and analysed by level of evidence.The review nds evidence of the efciency of the programmes assessed, although serious limitations are alsodetected, which should be taken into consideration when designing future interventions. The likelihood ofsuccess is enhanced when all the disciplines of a centre are involved, and also the parents. It is also essential toadapt the diverse programmes to the social and cultural characteristics of the school population in which theprogramme is to be carried out. Finally, the ndings indicate the need for continuity in the programmes iftheir long-term efciency is to be guaranteed.

2012 Elsevier Ltd. All rights reserved.

1. Introduction

School violence includes behaviours that cause physical and emo-tional harm, ranging from verbal aggression, humiliation, ostracism,physical harm and destruction of property (Benbenishty & Astor,2005), and including various categories such as classroom disruption,disciplinary problems and maltreatment among classmates (Olweus,1993).

We arefacing a phenomenon that hasprobably always been presentin schools, although it has become the subject of increasing attentionand a social alarm in recent years. Several studies have analysed itsprevalence nding that 20 to 30% of pupils have been involved inviolent episodes, ranging from simple verbal intimidation to physicalor sexual aggression (Currie et al., 2008; Department of Health andHuman Services & Center for Disease Control and Prevention, 2006;Ruiz, Exposito, & Bonache, 2010).

The consequences it may have for children's mental health andfuture behaviour must also be considered (Abada, Hou, & Ram, 2008;stberg, 2003). Some studies show that extended exposure to violenceis linked to the development of: (a) emotional and psychosomatic prob-lems in victims andbullies alike (Bond, Carlin, Thomas, Rubin, & Patton,2001; Gini & Pozzoli, 2009); (b) low self-esteem, depression and suicid-al tendency (Brunstein, Marrocco, Kleinman, Schonfeld, & Gould, 2007;McMahon, Reulbach, Keeley, Perry, & Arensman, 2010); and (c) antiso-cial behaviours, which lead to legal, economic and social problems(Cunningham & Henggeler, 2001).

All of this has led to a heightened awareness of this problem in recentyears and the proliferation of prevention programmes (Farrell, Meyer,Kung, & Sullivan, 2001). Habitually, the main components of these inter-ventions are: (a) globally focused policies emphasizing the democraticparticipation of all school members, which are generally the main partof any long-term interventions; (b) the improvement of the classroomatmosphere, based on pupilpupil and teacherpupil relations; (c)the in-troduction of peer support systems; (d) intervention in the recreationalarea or theschoolsurroundings; (e)pro-social activities in theclassroom,as part of the curriculum; and (f) specic work with bullied students orthose at risk of being bullied (Cowie, 2000; Cunningham et al., 1998).

Despite efforts to establish them in schools, there is a notable lackof evaluation of these programmes, so the effectiveness of which was

Children and Youth Services Review 34 (2012) 16461658

Corresponding author at: Plaza Vistabella, 17, 2L, CP: 30820, Alcantarilla, Murcia,Spain. Tel.: +34 646350366.

E-mail addresses:[email protected](J.A. Jimnez Barbero),[email protected](J.A. Ruiz Hernndez),[email protected](B. Llor Esteban),[email protected](M. Prez Garca).

0190-7409/$ see front matter 2012 Elsevier Ltd. All rights reserved.

doi:10.1016/j.childyouth.2012.04.025

Contents lists available at SciVerse ScienceDirect

Children and Youth Services Review

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c h i l d y o u t h
http://dx.doi.org/10.1016/j.childyouth.2012.04.025http://dx.doi.org/10.1016/j.childyouth.2012.04.025http://dx.doi.org/10.1016/j.childyouth.2012.04.025mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.childyouth.2012.04.025http://www.sciencedirect.com/science/journal/01907409http://www.sciencedirect.com/science/journal/01907409http://dx.doi.org/10.1016/j.childyouth.2012.04.025mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.childyouth.2012.04.025


2/13

really unknown (Surgeon General, 2001). The United States'Agencyfor Healthcare Research and Quality (2010) indicates that meta-analyses of randomized clinical trials (RCTs) are the studies thatoffer the best scientic evidence of therapeutic interventions, fol-lowed by the RCTs themselves. Yet, given their scope and cost, RCTsare beyond the means of many organisations, which are obliged torely on published reviews.

The aim of our study is to examine the effectiveness of the interven-

tion and violence prevention programmes carried out in the last decade.It continues and updates earlier studies (Scheckner, Rollin, Kaiser-Ulrey,& Wagner,2002; Wilson,2000). Studies published since the beginningof2000up to the present were evaluated, selectingthem on thebasis of thequality of their methodological design and on their level of scienticevidence. This allows us to establish the degrees of recommendationof the various programmes (National Institute for Clinical Excellence(NICE), 2008).

2. Methods

2.1. Inclusion criteria

Studies were included in this review if they fullled the followingcriteria: (a) in empirical studies, the aim had to be the evaluation ofthe effectiveness of an intervention programme in reducing violencein the school environment; (b) in review studies,the main aim requiredwas to examine the effect of school violence prevention or reductionprogrammes; (c) the evaluated interventions had to directly targetthe study population (Primary or Secondary school pupils), not theteachers and parents.

2.2. Search strategy

A systematic search was made through the following electronicdatabases: Medline, Trip Database, Cochrane, Academy Search Premier,PsycINFO, ERIC and PsycARTICLES. The keywords and terms used were:bullying, school violence, attitudes toward violence and adoles-cents, intervention or prevention program and self-esteem or empa-

thy, among others. In order to guarantee the currency of the ndings,the search was restricted to works published after January 1, 2000.The titles were examined, as were the abstracts when available, andthose that did not meet the criteria were rejected.

2.3. Selection criteria

The complete texts of the accepted articles were carefully read,and the lists of references were studied to identify possible relevantarticles not detected by our initial search. The selection of studieswas made by two independent reviewers following two stages:

1) Level of evidence: We used the categories proposed by theAgencyfor Healthcare Research and Quality (2010) in order to limit the

review to articles that offered a high level of scientic evidence:(1A)meta-analysis of randomized,controlled clinical trials; (1B)ran-domized controlled clinical trials; (2A) suitably designed, non-randomized controlled studies; and (2B) uncontrolled studies, suchas pre-post studies and cohort studies.

2) Methodological quality: Two independent reviewers assessed themethodological quality of the studies selected with a high inter-rater reliability (Pearson correlation coefcient,r=0.83), using thefollowing evaluation criteria:- For studies of 1A level of evidence (reviews), the criteria used

werethosedescribed byJadad, Moher, and Klassen (1998)forsys-tematic reviews, which assign a quality score from 0 to 8. Studiesthat scored below 4 were rejected. Meta-analyses of randomizedclinical trials were included at this level, as were other meta-

analyses of prospective studies or systematic reviews of special

importance, provided they reached the required score on thisscale.

- For studies of 1B level of evidence (RCTs), studies were includedif they fullled at least one of the following conditions:(a) A score of 6 or above on an ad hoc scale of methodological

quality based on the guide published by the University ofYork in 2001 for the preparation of systematic reviews (NHSCentre for Reviews and Dissemination, 2001). The scale

comprised 10 items: (1) operative de

nition of the constructsand terms used in the study; (2) appropriate method forsample selection; (3) appropriate sample size; (4) a prioridistinction of sub-groups or use of suitable clustering tech-niques; (5) validity of the evaluation (direct collection ofinformation by the researchers); (6) reliability of the evalua-tion (use of a validatedtooland/or a high level of internalcon-sistency to evaluate the intervention); (7) follow-up of theresults; (8) use of outcome measures that match the aim ofthe study; (9) appropriate statistical analysis; and (10) suit-able presentation of the ndings through graphs or similar.

(b) A score of 3 or above on the scale ofJadad et al. (1996),forrandomized clinical tests.

- For the evaluation of the quality of the quasi-experimental studies(2A and 2B levels of evidence) the ad hoc quality scale was used.Articles scoring 6 or above were included in the review.

2.4. Tabulation and analysis of the information

The studies selected were grouped according to their level of scien-tic evidence (1A, 1B, 2A and 2B) and methodological design.

The data from each study were arranged according to the followingcategories: date and country of study; quality of the methodological de-sign; research aim; name of the prevention or intervention programmeevaluated; sample size and age of the study population.

The analysis of the studies selectedwas descriptive, as meta-analysiswas not possible due to the heterogeneity of the results.

3. Results

The electronic search initially returned 9386 publications. Afterreviewing the titles, abstracts and references, 299 potential articleswere identied. Of these, 245 were excluded due to not having a meth-odological design at the levels of evidence 1A, 1B, 2A or 2B. Of theremaining 54 studies, 32 were nally selected that fullled the criteriarequired. Of these, there were 2 meta-analyses of RCTs (1A level of ev-idence), 2 meta-analyses of prospective studies, 1 systematic review, 12RCTs (1B level of evidence), 11 non-randomized controlled studies (2Alevel of evidence), and 4 pre-post uncontrolled studies or studies ofcohorts (2B level of evidence) (seeFig. 1).

3.1. Level of evidence 1A: meta-analysis of RCT and systematic reviews

3.1.1. Description of the studiesWe found 5 studies that fullled the inclusion criteria for this cate-

gory. Of these, two are meta-analyses of RCTs, which constitute themaximum level of evidence (Mytton, DiGuiseppi, Gough, Taylor, &Logan, 2006; Park-Higgerson, Perumean-Chaney, Bartolucci, Grimley,& Singh, 2008). Due to their interest and their relation to the aim ofthe study, we also include two meta-analyses of prospective studies(Merrel, Gueldner, Ross, & Isava, 2008; Tto& Farrington, 2011) andone systematic review of prospective studies (Vreeman & Carroll,2007). Details are shown inTable 1.

The 2 meta-analyses of RCTs (Mytton et al., 2006; Park-Higgersonet al., 2008) included 82 randomized clinical trialsthat assesstheefcien-cy of violence prevention programmes in schools. The meta-analysisby Merrel et al. (2008) and the systematic review by Vreeman and

Carroll (2007)attempt to assess school interventionsaimed at decreasing

1647J.A. Jimnez Barbero et al. / Children and Youth Services Review 34 (2012) 16461658


3/13


4/13

third of the variables measured showed positivechanges following pro-grammes aimed at reducing bullying. They underline that, whilst thesemay be able to inuence awareness, knowledge and self-perception of

competence to cope with bullying, they cannot be expected to have adrastic inuence on the incidence of bullying and victimization.

Lastly, Park-Higgerson et al. (2008) nd no signicant effects forfourof the ve characteristicsexamined in the prevention programmes.

3.1.3. Differences according to age and genderOf the ve studies reviewed, only three interpret the ndings in

terms of age and gender. Of these, two meta-analyses (Mytton et al.,2006; Park-Higgerson et al., 2008) nd no gender differences in theeffect size, although they do highlight age-related differences. Park-Higgerson et al.nd that the effectiveness of theprogrammes is greaterin children of 9 years and above, whereas Mytton et al. conclude thatthe improvements are moresignicantin the follow-up among Second-ary school students.

Vreeman and Carroll (2007) suggest that the age differences ob-served in the effectiveness of the intervention are due to the focus ofthe intervention, as programmes based on social skills obtain morepositive outcomes among younger children, who, notwithstanding,respond worse to interventions based on the cognitive-behaviouralmodel. As regards gender, the same study nds that girls seem torespond better than boys to multidisciplinary interventions.

3.2. Level of evidence 1B: randomized clinical trials (RCTs)

3.2.1. Description of the studiesIn the second level of evidence, we include 12 randomized clinical

trials (RCTs) (seeTable 2). In 8 of the articles, the authors establish astheir main aim the evaluation of the impact of an anti-bullying school

intervention (Baldry & Farrington, 2004; Berry & Hunt, 2009;

DeRosier, 2004; Farrell, Meyer, & White, 2001; Fekkes, Pijpers, &Verloove-Vanhorick, 2006; Frey et al., 2005; Hunt, 2007; Stevens, VanOost, & De Bourdeaudhuij, 2000). A further 3 studies evaluate the

effectiveness of a bullying and school violence prevention programme,(Beets et al., 2009; Jenson & Dieterich, 2007; Teglasi & Rothman,2001). Lastly, one study compares three types of intervention: a pro-gramme designed by the authors (Creating a Peaceful School LearningEnvironment) versus school psychiatric consultancy versus habitualintervention (Fonagy et al., 2009).

The age of the samples ranges from 8 to 16 years: seven interven-tions were evaluated in Primary schools, three in Secondary schoolsand two in Primary and Secondary schools.

The main outcome measures in the evaluation of bullying or intimi-dation at school were: frequencyand type of bullying and victimization,exposure to bullying, frequency of aggressive and violent behaviours,frequency of maladaptive responses to episodes of bullying and vio-lence, bullying-related beliefs(acceptance of bullying, witness's respon-

sibility, adult'sperceivedresponsibility), attitudes towards violence andbullying-related behaviours (direct and indirect aggression and victim-ization). Other measures used were behaviour inside and outside theclassroom, behavioural problems, use of drugs, interaction skills, self-effectiveness/self-perception/self-esteem, anxiety and awareness.

3.2.2. Effectiveness of the interventions evaluatedSignicant differences were found between the intervention group

and the control group in nine of the interventions evaluated (Beetset al., 2009; Berry & Hunt, 2009; DeRosier, 2004; Farrell, Meyer, &White, 2001; Fekkes et al., 2006; Fonagy et al., 2009; Frey et al., 2005;Stevens, Van Oost, & De Bourdeaudhuij, 2000; Teglasi & Rothman,2001). Follow-ups were carried out in eight of these, and the ndingswere maintained in only four cases. Noteworthy is the study by

Fonagy et al.(2009), which comparesa school anti-bullying programme

Table 1

Papers reviewed at level of evidence 1A: meta-analyses and systematic reviews of RCTs.

Authors, year ofpublication

Papers reviewed Age Aims Main results

Mytton et al. (2006) 56 RCTs/meta-analysis. 512 The effects of school violenceprevention programmes for childrenidentied as being aggressive or at riskof being so.

Prevention programmes to reduce aggressivebehaviour are effective (SMD=0.41, CI(95%)=0.56 to 0.26).Interventions to improve social skills are moreeffective than those that teach skills to avoid

responding to provocations.Vreeman and Carroll (2007) 26 papers reviewed without meta-analysis:

9 RCTs11 pre-post with control group2 cohort studies4 pre-post quasi-experimental studies.

613 To evaluate school interventions todecrease bullying.

4 curriculum-based interventions showed a signi-cant reduction of bullying.7 multidisciplinary interventions achieved de-creases in bullying.4 interventions based on increasing social skillsachieved no signicant changes.1 study based on tutoring showed a decrease invictimization.

Park-Higgersonet al. (2008) 26 RCTs/meta-analysis. 616 The characteristics of schoolprogrammes aimed at prevention ofviolence.

No signicant effects were reported for 4 of the 5characteristics of the programmes examined.Only single-focus interventions had any effect onaggressive and violent behaviour (ES=0.15).

Merrell and Gueldner(2008) 16 prospective studies/meta-analysis:2 experimental designs3 mixed design with control group11 quasi-experimental designs.

612 The effectiveness of anti-bullyingschool interventions.

Signicant positive effects for only one third of thevariables.Main variables with positive results: knowledge ofbullying prevention (ES=1.52), global self-esteem

(ES=1.08) and perception of abusive conducts(ES=0.35).Remaining measures showed no signicantchanges.

Tto and Farrington (2011) 53 prospective studies:44 studies/ meta-analysis17 RCTs21 pre-post+ CG4 (other controlled studies)11 cohorts.

616 The effectiveness of the programmecomponents.

The most intensive programmes were moreeffective, as were those that included meetingswith parents (OR =1.57), rm disciplinarymeasures (OR =1.59) and improved supervisionof recreational areas (OR=1.53).

Abbreviations: RCT, randomized clinical trial; CG, control group; SMD, Standardised Mean Difference; ES, Effect Size. pb0.001.

1649J.A. Jimnez Barbero et al. / Children and Youth Services Review 34 (2012) 16461658


5/13

(CAPSLE) with two other interventions, nding that CAPSLE achieved agreater reduction in victimization, aggression and the number of ag-gressive onlookers.

The other three studies evaluated obtained no signicant results(Baldry & Farrington, 2004; Hunt, 2007; Jenson & Dieterich, 2007).

3.2.3. Differences according to age and gender

Frey et al. (2005)report that males are more sensitive to interven-

tion programmes, whereas two other studies (Beets et al., 2009;Farrell, Meyer, & White, 2001) conclude that there are no signicantgender differences in the effectiveness of the intervention, althoughgreater stability was observed in boys during the follow-up. As for age,older children (1315) seem to respond better (Baldry & Farrington,2004; Frey et al., 2005) although one study declares that the changesare most stable in the follow-up of younger children (Stevens, VanOost, & De Bourdeaudhuij, 2000).

3.3. Level of evidence 2A: controlled, non-randomized prospective studies

3.3.1. Description of the studies

Atthis level of evidence,we chose 11studies (see Table 3) with anal-ogous aims, thus allowing inter-comparison.

In these studies, the authors declare their main goals to be: (a) toanalyse the impact of school programmes aimed at reducing violenceand bullying (Ju, Wang, & Zhang, 2009; Sapouna et al., 2010; Stanbury,Bruce, Jain, & Stellern, 2009; Stevens, De Bourdeaudhuij, & VanOost, 2000; Twemlow et al., 2001); (b) to evaluate the effectiveness ofbullying prevention programmes (Bauer, Lozano, & Rivara, 2007;Evers, Prochaska, Van Marter, Johnson, & Prochaska, 2007; Menesini,Codecasa, Benelli, & Cowie, 2003; Rahey & Craig, 2002); (c) to studythe impact of an anti-bullying programme on school achievement(Fonagy, Twemlow, Vernberg, Sacco, & Little, 2005); and (d) to raiseawareness and perception of self-effectiveness when intervening inbullying situations (Andreou, Didaskalou, & Vlachou, 2008). The agerange of the study samples was 6 to 16 years. Five interventions wereperformed in Primary schools (611 years), three in Secondary schools(1116 years) and three in Primaryand Secondary schools(616 years).

The main measures of outcomes in these studies were: bullying be-haviour, the frequency of physical and verbal aggressions, the descrip-tion and characteristics of victimization, the places where this occurs,attitudes towards bullying, empathy, the roles of the participants inthe bullyvictim relationship, peer interaction and perception of self-effectiveness to intervene in bullying incidents.

Other associated variables were also measured, such as the percep-tion of safety, commitment and school performance, disciplinaryproblems, coping strategies, knowledge of bullying, problems of exter-nalizing and internalizing behaviours and the positive reinforcementsemployed by the school staff.

3.3.2. Effectiveness of the interventions evaluatedStatistically signicant results were obtained in most of the mea-

sures in nine of the studies, of which only ve included a follow-up,with the changes persisting in three of these. However, in two studies,the results of the intervention showed no signicant differences(Bauer et al., 2007; Rahey & Craig, 2002).

3.3.3. Differences according to gender and ageRegarding age,Rahey and Craig (2002)andMenesini et al. (2003)

indicatethatvictimization tends to decreasein older children. The latterpaper also reports that episodes of bullying increase in older children,whilst the role of the defender and anti-bullying attitudes prevailmore among younger children. Lastly,Stevens, De Bourdeaudhuij, andVan Oost (2000) nd positive changes in bullying and victimizationonly for Primary schools (i.e., children aged 611 years).

There is a disparity in the results of the diverse authors in terms

of gender. Rahey and Craig (2002) report better results for girls;

Stanbury et al. (2009)onlyobtained signicantchangesin girls; howev-er,Menesini et al. (2003)report that boys were more sensitive to theintervention.

3.4. Level of evidence 2B: non-controlled studies and studies of cohorts

3.4.1. Description of the studies

At this level, we selected three studies of cohorts (Beran, Tutty, &

Steinrath, 2004; Bowllan, 2011; Metzler, Biglan, Rusby, & Spraque,2001) and one pre-post study without a control group (Salmivalli,Kaukiainen, & Voeten, 2005), whose goalwas to determine the efcien-cy of intervention or bullying prevention programmes (seeTable 4).

The age ranges studied vary between 9 and 14 years, with twostudies carried out in Primary and two in Secondary schools.

In order to evaluate the effectiveness of the interventions, the fol-lowing measures wereused: prevalence of bullying, school atmosphere,physical or verbal aggressions, perceived safety, coping strategies, atti-tudes towards victims and bullies, beliefs about self-effectiveness inintervening in situations of aggression, roles those participating inbullying, disciplinary records related to bullying and the degree ofimplementation of the programme.

3.4.2. Effectiveness of the interventions evaluated

Statistically signicant results were obtained for the four studies se-lected. The variables most sensitive to the interventions werefrequencyof observed and experienced bullying, attitudes towards victims andbullies and the number of verbal and physical aggressions. In three ofthe studies, a follow-up was carried outand thendings were reported.

3.4.3. Differences according to gender and ageOnlythe study byBowllan (2011) reports genderandage differences,

with better results obtained by girls of younger ages (1213 years) thanby the remaining subjects of the sample.

4. Discussion

The studies reviewed at the various levels of evidence generally

coincide that school programmes aimed at reducing violence can pro-duce benecial effects in the overall social environment of schools.Thus, the most efcient interventions are those that are developedfrom a multidisciplinary perspective or global focus and that are di-rected at improving social and interpersonal skills and at modifyingattitudes and beliefs.

As in previous studies aheadRigby (1997), the most sensitive vari-ables in the interventions examined seem to be those related to atti-tudes, beliefs and behaviours. Within this concept, students' attitudesand beliefs about bullying, destructive or intimidating behaviour andthe spectator's role are noteworthy. Even in those cases in which theresults are modest or nonsignicant for these measures, some authorsconsider that small changes in attitude within the peer group may con-tribute to general changes in bullying dynamics (Farrell, Meyer, &

White,2001; Frey et al., 2005). Nevertheless, these changes are complexdue tothe difculty of their continuity over time unless accompanied inthe programme by techniques for maintaining behaviour, such as antic-ipating risk situations, positive reinforcement of peers who intervene toput a stop to the bullying, etc. (Stevens, Van Oost, & De Bourdeaudhuij,2000). The authorsreviewedherein alsohighlight the importance of therole of the passive onlooker in bullying (Cowie, 2000; Twemlow et al.,2001). Spectators' silence or complicity may imply a strong supportfor the bullies and indirectly favour a violent culture. Menesini et al.(2003)propose a global approach based on social relationships amongpeers in order to help to break this pact of silence and foster peer re-sponsibility and empathy.

Signicant results in school programmes were also obtained in thefrequency of violent behaviour. In this respect, several authors nd

that the interventions are more effective among adolescents showing

1650 J.A. Jimnez Barbero et al. / Children and Youth Services Review 34 (2012) 16461658


6/13


7/13

Table 2 (continued)

Author/country Aim Sample size Age Programme/intervention Duration/fol

Fekkes et al. (2006)Holland

The effects of a school anti-bullying programme in Primaryschools.

47 PSN=3816IG=3816CG 1= 1552CG 2= 1050

912 Programme based on the OlweusBullying Prevention Programme.

2 academic yNo follow-up

Hunt (2007)Australia

The educational impact of an anti-bullying programme in Secondaryschools.

6 SSN=444IG=152CG=258

1215 Ad h oc prepared interve ntio n. 1 acade mic y1-year follow

Jenson and Dieterich (2007)USA

The effects of a bullying andvictimization preventionprogramme based on Primaryschool children's aptitudes.

28 PSN=1126:IG=670CG=456

810 Youth Matters Programme. 2 academic y12-month fo

Beets et al. (2009)USA(Hawaii)

The effectiveness of a schoolprogramme designed to preventviolent and risk behavioursamong Primary school children.

20 PSN=1714:IG=976CG=738

Pre-test: 1011Post-test:1516

The Positive Action Programme. 4 years.No follow-up

Berry and Hunt (2009)Australia

The effectiveness in anxious,bullied boys of a cognitive-behavioural intervention focusedon factors that increase vulnera-bility to bullying.

7 SSN=46 (boys):IG=22GC=24

1214 Condent Kids Program me. 8 weeks.3-month foll

Fonagy et al. (2009)USA

The effectiveness of the schoolpsychiatric consultation (SPC) vs.a school programme (CAPSLE) vs.treatment at use (TAU) in order toreduce aggression in PS.

9 PSN=1345:CAPSLE=563SPC=422TAU=360

810 SPCCreating a Peaceful SchoolLearning(CAPSLE)TAU

24 months.36-month fo

Abbreviations: N, sample size; IG, Intervention Group; CG, Control Group; PS, Primary Schools; MS, Middle School; SS, Secondary Schools; RR, Relative Risk; OR, Odds Ratio; M, Me pb0.05. pb0.001.


8/13

Table 3Papers reviewed at level of evidence 2A: non randomized controlled prospective studies.

Author/country Aim Sample size Age Programme/intervention Duration/follow-

Stevens, De Bourdeaudhuij,and Van Oost (2000)Blgium

The effectiveness of the anti-bullying approach in emishschools.

9 PS+9 SSN=11041 IG=4352 IG=426CG=243

1016 Programme adapted from theOlweus Bully PreventionProgramme.

7 months.1-year follow-up

Twemlow et al. (2001)USA

The impact of an anti-violenceprogramme based on learningenvironment in a PS with the re-sults from a control school.

2 EPN=192:GI=101GC=91

810 Psychodynamic/social systeminterventions against violence.

3 years.Follow-up: at theacademic year.

Rahey and Craig (2002)Canad

The short-term effects of a schoolanti-bullying programme.

2 Elementary schools (PS and MS )N=491:GI=240GC=251

613 Ad hoc programme based onBully Proong your School.

12 weeks.No followup.

Menesini et al. (2003)Italy

The effectiveness of school anti-bullying intervention.

2 EMN=293:GI=178

GC=115

1114 The Befriending Intervention. 1 school year.No follow-up.

Fonagy et al. (2005)USA

The impact of an anti-bullyingprogramme in educationalachievement in PS.

5 EP and ESN=2206:GI=1106GC=1100

1012 The Creating a Peaceful SchoolLearning Environment (CAPSLE)..

5 years.Follow-up: evaluof each academic

Andreou and Didaskalou (2008)Greece

The effectiveness of anintervention that sought toimprove awareness, self-reection and the capacity tosolve bullying situations.

10 PSN=454CG=206IG=248

912 Ecological approach. Interventionbased on the roles of the bullyingparticipants.

4 weeks.6-month follow-


9/13

Table 3 (continued)

Author/country Aim Sample size Age Programme/intervention Duration/follow-

Bauer et al. (2007)USA

The effectiveness of a bullyingprevention programme.

10 PSN=6518:IG=4959CG=1559

611 Olweus Bullying PreventionProgramme.

2 years.No follow-up

Evers et al. (2007)

USA

The effectiveness of trans-

theoretical model based on schoolinterventions to prevent bullying.

12 MS+12 SS

N=2452CG=7921 IG=8632 IG=797

1116 Build Respect, Stop Bullying. 1 year: 3 session

No follow-up.

Stanbury et al. (2009)USA

A school programme aimed atreducing bullying.

1 SSN=172IG=86CG=86

1314 Programme designed to improveempathy among school children.

7 weeks.No follow-up.

Ju et al. (2009)China

To reduce the incidence ofbullying by establishing arespectful and safe classenvironment.

1 PSN=354IG=233CG=121

810 Programme designed byresearchers based on the actionresearchapproach.

5 weeks.No follow-up.

Sapouna et al. (2010)UKGermany

The effects of virtual teachingaimed at reducing bullying andthe general rates of victimizationamong SS students.

27 PSN=1129:IG.=509CG=560

711 Fun with Empathic Agents toachieve Novel Outcomes inTeaching: FEAR NOT!

3 weeks.5-week follow-u

Abbreviations: N, sample size; IG, Intervention Group; CG, Control Group; PS, Primary Schools; MS, Middle Schools; SS, Secondary Schools; F, FisherSnedecor's F-test; RR, Relative pb0.05. pb0.001.


10/13

Table 4

Papers reviewed at level of evidence 2B: uncontrolled studies and studies of cohorts.

Author/ country Aim Sample size Age Programme/ intervention Duration/ follow-up

Metzler et al. (2001)

USA

The effectiveness of a programme

involving school staff in order toreinforce positive behaviouramong pupils.

3 SS

N=1043:IG=645CG=758

1114 Effect ive Behaviour

Support Intervention.

1 school year.

1-year follow-up

Beran et al. (2004)Canada

To reduce intimidating behavioursand to create a safe schoolenvironment.

4 PSPhase 1EC=25CC=77

Phase 2C1=20C2=70

912 Dare to Care: bully proong yourschool.

Phase 1:Duration: 3 months.No follow-up.Phase 2:

Duration: 1

2 years.Comparison of 2 PS of programme interventi

Salmivalli et al. (2005)UK

The effects of an anti-bullying in-tervention programme.

16 PSN=1220

912 Global intervention based on theroles played by those involved inbullying.

6 months.12-month follow-up.

Bowllan (2011)USA

The prevalence of bullying in MSand the impact of a preventionprogramme.

1 SSN=270EC=112CC=158

1214 Olweus Bullying PreventionProgramme.

1 year.No follow-up.

Abbreviations: N, sample size; EC, Exposed Cohort; CC, Control Cohort; C1, Cohort during rst year of programme; C2, Cohort during second year of programme; PS, Primary Schoocoefcient; F, FisherSnedecor's F-test. pb0.05. pb0.001.


11/13

high levels of violent behaviour (Beets et al., 2009; Farrell, Meyer, &White, 2001). This may be due to the oor effect generated by stu-dents with low levels of aggression. This is consistent with preventionprogrammes that proved more effective among students with highpre-test levels of aggressiveness (Stoolmiller, Eddy, & Reid, 2000).However, in interventions that were not focused on aggression,Fonagy et al. (2009)report signicant reductions in this measure forlow-risk Primary school children.

Signi

cant changes in victimization caused by bullying were alsoobserved in the studies reviewed. The authors underline the importanceof teaching coping strategies and training in social skills in order tobe able to respond to episodes of bullying (Berry & Hunt, 2009; Jenson& Dieterich, 2007), as Olweus reported in previous studies (Olweus,1993, 1996).

The modulating effect of gender and age was not studied in depthbecause most of the studies do not nd (or do not report) signicantdifferences due to these variables. In the studies that indicate genderdifferences, most of the authors report better results amongst males(Beets et al., 2009; Farrell, Meyer, & White, 2001; Frey et al., 2005;Menesini et al., 2003), although some studies contradict this (Jenson &Dieterich, 2007; Rahey & Craig, 2002). The reason for these apparentlycontradictory ndingsdoes not seem to lie in thetheoretical frameworkof the interventions, as most of them stem from a pro-social model,based on the development of anti-bullying beliefs and attitudes, orform mixed models that combine the psychodynamic or globalapproach (Frey et al., 2005; Rahey & Craig, 2002). Nevertheless, webelieve that these ndings may be due to the fact that the tools usedin the evaluation usually measure aggression and direct bullying,which aremore frequent among boys,and they ignore indirect violencebased on more subtle behaviours, such as isolating or rumours, whichare more prevalent among girls (Olweus, 2005). It has been observedthat boys usually reveal higher baselines in scales of aggression, bully-ing and victimization (Menesini et al., 2003). This may explain whyit is easier to obtain more positive results among boys than girls, asthese types of assessment tools are the most common in the studiesexamined. On the other hand, the gender difference observed in thedifferential use of violence may be related to girls' greater predis-

position to develop internalizing behaviours, in contrast to boys, whousually display externalizing ones (Wasserman, McReynolds, Ko, Katz,& Carpenter, 2005; Zahn-Waxler, Shirtcliff, & Marceau, 2008).

Rahey and Craig (2002)report better results among girls and con-sider that this may be due to the intervention being more adapted tothese gender differences. However, an inuential factor may be theuse of indirect assessment tools, such as questionnaires administeredto teachers or parents, who usually give lower scores in victimizationand bullying for girls than for boys. This is consistent with the ndingsof other authors who do not nd signicant gender differences whenthe same variable is measured through self-report questionnaires(Elsea & Smith, 1998). This coincides with the conclusions of Freyet al. (2005), who report that boys generally benet more from theintervention, although girls obtain better results in the acquisition of

social skillsa measure that is assessed through teachers' reports.With regard to age, most of the cases studied that offer ndings on

this variable report better results among older children (Secondaryschools). Indeed, despite the fact that interventions carried out withyounger children (Primary schools) produce signicant improvementin the acquisition of social skills and increased empathy, no changesare observed in the frequency and severity of bullyingand victimizationepisodes or those of physical and verbal aggression. There are several,apparently contradictory, interpretations of these ndings. Someauthors state that children continually exposed to situations of bullying,manage to adapt over time and acquire the social and assertive skillsnecessary to cope with the situation. This would lead to less victimiza-tion when they are older (Jenson & Dieterich, 2007; Smith, Madsen, &Moody, 1999). However,Rahey and Craig (2002)conclude that small

children who are victims of bullying can turn into bullies of younger

children as they get older and stronger, thus perpetrating a cycle ofviolence. This would justify the need to administer intervention pro-grammes simultaneously across the entire school age range. This inter-pretation is consistent with the ndings of other authors who havecarried out programmes at both Primary and Secondary school levels(Stevens, De Bourdeaudhuij, & Van Oost, 2000) and who obtain betterresults for younger aged children.

5. Conclusions

This review allows us to state that there is evidence for the effective-ness of school interventions aimed at reducing or preventing violence.The most effective interventions appear to be those aimed at improvingsocial and interpersonal skills and modifying attitudes and beliefs. Themost sensitive variables used to evaluate the effectiveness of theseinterventions are those related to attitudes, beliefs and behaviours and,to a lesser extent, victimization and frequency of violent behaviours.With regard to themodulating effects of gender and ageon theeffective-ness of the programmes analysed, the ndings are inconclusive, as mostof the studies do not take these variablesinto consideration. The few thatdo so indicate that, in terms of gender, there is a better response amongmales, whereas, in terms of age, older children (Secondaryschools) seemto benet more from the diverse programmes.

6. Recommendations

Our study has revealed certain aspects that need to be borne inmind when designing future interventions:

1) The likelihood of success is greater when all the professional disci-plines of the educational centres and the parents of the childrenare involved.

2) It is essentialto adapt programmes to the social andculturalcharac-teristics of the school population in question and to consider anypossible inuence of age and gender.

3) The long-term effectiveness of interventions is vital. The few studiesthat include follow-ups generally coincide that the effect decreases

when the intervention is not maintained. We suggestthat it is neces-sary to adopt measures that favour this continuity, such as remindersessions or the implementation of prevention or intervention pro-grammes within the academic curriculum.

7. Limitations

This review has a series of limitations that should be taken intoaccount. Firstly, the search strategy restricts the search to articlespublished in English, which may exclude studies of interest in otherlanguages. Unpublished articles and articles providing low levels ofevidence are also excluded. Furthermore, the interventions evaluatedwere generally performed in not very homogeneous populations. Infact, some werecarried out exclusivelyin Primary or Secondary schools,

whereas otherscovered thewhole school-age range,from 6 to 16 years,and therefore include subjects at very different stages of maturity. Thepopulations studied are also of diverse origins (USA, Australia, UK,France, Greece, Canada) and therefore present diverse social andcultural peculiarities. Finally, due to the heterogeneity of the differentprogrammes evaluated, the diverse models, methodologies and toolsused, it is not easy to compare ndings or to apply meta-analysistechniques.

References

Abada, T., Hou, F., & Ram, B. (2008). The effects of harassment and victimization onself-rated health and mental health among Canadian adolescents. Social Science& Medicine,67, 557567.

Agency for Healthcare Research and Quality (AHRQ) (2010). EPC evidence reports.Rockville: AHRQ [Internet]. http://www.ahrq.gov/clinic/epcindex.htm. Accesed

12.12.2010.

1656 J.A. Jimnez Barbero et al. / Children and Youth Services Review 34 (2012) 16461658
http://www.ahrq.gov/clinic/epcindex.htmhttp://www.ahrq.gov/clinic/epcindex.htm


12/13


13/13

2012 - Effectiveness of Antibullying School Programmes a Systematic Review by Evidence Levels

Documents

Transcript of 2012 - Effectiveness of Antibullying School Programmes a Systematic Review by Evidence Levels