Professional Documents
Culture Documents
What Works in Preventing Bullying
What Works in Preventing Bullying
ABSTRACT
This paper summarises the results of a systematic review and meta-analysis of the effectiveness
of anti-bullying programmes in schools. Extensive searches were carried out in 18 databases and
in 35 journals. The number of reports on anti-bullying programmes increased considerably over
time. Nearly 600 reports were found, but only 59 of these (describing evaluations of 30 different
programmes) were eligible for inclusion in our review because they described a high-quality
evaluation. We coded the elements of the intervention in these programmes and key features of
the evaluation and related these to the effects of the intervention. These types of figures have never
been presented in any previous systematic review or meta-analysis of anti-bullying programmes.
Our meta-analysis showed that school-based anti-bullying programmes are effective in reducing
bullying and victimisation (being bullied), which were reduced by about 20Ð23% in experimental
schools, compared with control schools. The most important programme components that were
associated with a decrease in bullying were parent training, improved playground supervision,
disciplinary methods, school conferences, videos, information for parents, work with peers,
classroom rules and classroom management.
KEY WORDS
Systematic review; meta-analysis; anti-bullying programmes in schools; intervention components;
evaluation research.
appraisal of the evidence than traditional narrative (including correlating effect sizes with programme
%-A/-P.N)Z$%).H.3-:'3/2)%-A/-P)'&'(H.-.)=V)H-'%.)#*) components and study features); and d) focusing
/&3-%A-&3/#&)%-.-'%27)B*%#:) ?M!)3#)37-)-&4)#*)+<%/() only on programmes that are specifically designed
=>>MD)'&4)/.)G'.-4)#&)-O3-&./A-)(/3-%'3$%-).-'%27-.N) to reduce bullying and not aggressive behaviour.
Z$%):-3'F'&'(H./.)<%-.-&3.)')[$'&3/3'3/A-).$::'%H) The interested reader should consult our report to
#*)-**-23)./\-.)/&)'&3/FG$((H/&,)<%#,%'::-.)'&4) 37-)QP-4/.7)d'3/#&'()1#$&2/()#&)1%/:-)5%-A-&3/#&)
standardises the evaluation results across studies BJ3#*/0)L'%%/&,3#&)K)C'(4%H0)=>>MD)*#%)'):#%-)4-3'/(-4)
with the aim of making solid inferences about what technical description of our systematic review.
works in preventing bullying, for whom and under
what circumstances.
Criteria for inclusion or exclusion
of studies
Previous research
We aimed to review only the highest quality
]'&H).27##(FG'.-4)/&3-%A-&3/#&)<%#,%'::-.) evaluations. We used the following criteria for
have been devised and implemented in an inclusion of studies in our systematic review.
attempt to reduce school bullying. The most a. The study described an evaluation of a
/&*#%:'3/A-)./&,(-).#$%2-)#*)%-<#%3.)#*)'&3/FG$((H/&,) programme designed specifically to reduce school
<%#,%'::-.)/.)37-)G##S)-4/3-4)GH)5^)Q:/37)'&4) bullying and/or victimisation (being bullied).
2#((-',$-.)B=>>R'D0)P7/27)2#&3'/&.)4-.2%/<3/#&.) b. The study included a clear definition of bullying
#*) !)<%#,%'::-.)/:<(-:-&3-4)/&) )4/**-%-&3) 37'3)P'.)2#&2#%4'&3)P/37)-O/.3/&,)4-*/&/3/#&.)
countries. There are also some reviews containing $.-4)/&)G$((H/&,)%-.-'%27)BL'%%/&,3#&0) ??!T)
.$::'%/-.)#*):'W#%)'&3/FG$((H/&,)<%#,%'::-.)B-,N) Z(P-$.0) ??!DN)C$((H/&,)P'.)4-*/&-4)'.)
6/,GH0)=>>=T)Q:/370)+&'&/'4#$)K)1#P/-0)=>>!T) including physical, verbal, or psychological
C'(4%H)K)L'%%/&,3#&0)=>>_DN)J7-):#.3)%-(-A'&3) attack or intimidation that is intended to cause
-O/.3/&,)%-A/-P.)'%-)GH)"`)Q:/370)Q27&-/4-%0)Q:/37) fear, distress, or harm to the victim, and an
K)+&'&/'4#$)B=>>RD)P7#).$::'%/.-4)-**-23)./\-.) imbalance of power, with a more powerful child
/&) R)P7#(-F.27##()'&3/FG$((H/&,)<%#,%'::-.0) (or children) oppressing less powerful ones.
./O)#*)P7/27)P-%-)$&2#&3%#((-4T)'&4)GH)9%--:'&)
c.) )C$((H/&,)P'.):-'.$%-4)$./&,)/&4/A/4$'().-(*F
K)1'%%#(()B=>>_D0)P7#)%-A/-P-4)=a).27##(FG'.-4)
report questionnaires by students.
<%#,%'::-.0) V)#*)P7/27)2#&2-%&-4)G$((H/&,0)P/37)
evaluations restricted to studies published in the d. The effectiveness of the programme was measured
English language. by comparing students who received it (the
These two prior reviews are of high quality. -O<-%/:-&3'()2#&4/3/#&D)P/37).3$4-&3.)P7#)4/4))
b#P-A-%0)&-/37-%)2'%%/-4)#$3)')*$(():-3'F'&'(H./.) receive it (the control condition). We require that
calculating weighted mean effect sizes and 37-%-):$.3)7'A-)G--&).#:-)2#&3%#()#*)-O3%'&-#$.)
correlations between study features and effect sizes. variables in the evaluation (establishing the
+%-%):-3'F'&'(H3/2)%-A/-P)P'.)<$G(/.7-4)GH) equivalence of conditions) by (i) randomisation, or
L-%,$.#&0)Q'&)]/,$-(0)^/(G$%&)K)Q'&27-\)B=>>_DN) B//D)<%-F3-.3):-'.$%-.)#*)G$((H/&,0)#%)B///D)27##./&,)
However, this included searches in one database some kind of comparable control condition.
#&(H0)*#%)'%3/2(-.)<$G(/.7-4)G-3P--&)37-)H-'%.) ??V) C-2'$.-)#*)(#P)/&3-%&'()A'(/4/3H0)P-)-O2($4-)
'&4)=>>a0)P/37).3$4/-.)37'3)/&2($4-4)G#37)G$((H/&,) uncontrolled studies that only had before and
and aggressive behaviour as outcome measures. We '*3-%):-'.$%-.)#*)G$((H/&,)/&)-O<-%/:-&3'().27##(.)
must emphasise that our research aims to review or classes. However, we include studies that
<%#,%'::-.)37'3)'%-)-O<(/2/3(H)4-./,&-4)3#)%-4$2-) 2#&3%#((-4)*#%)',-)B',-F2#7#%3)4-./,&.DN)
G$((H/&,)'&4)37'3)-O<(/2/3(H):-'.$%-)G$((H/&,N) e. Published and unpublished reports of research
;&)37-)<%-.-&3).H.3-:'3/2)'&4):-3'F'&'(H3/2) 2#&4$23-4)/&)4-A-(#<-4)2#$&3%/-.)G-3P--&) ?M!)
%-A/-P0)P-),#)P'H)G-H#&4)37-)-O/.3/&,)G#4H)#*) and the present are included.
%-.-'%27)GHY)'D)4#/&,):$27):#%-)-O3-&./A-).-'%27-.) f.) );3)P'.)<#../G(-)3#):-'.$%-)37-)-**-23)./\-N)J7-)
*#%)-A'($'3/#&.0).$27)'.)7'&4F.-'%27/&,)'(()A#($:-.) main measure of effect size is the odds ratio.
#*)!V)W#$%&'(.)*%#:) ?M!)$<)3#)37-)-&4)#*)+<%/() g. The minimum initial sample size (total in
=>>MT)GD).-'%27/&,)*#%)/&3-%&'3/#&'()-A'($'3/#&.)/&) -O<-%/:-&3'()'&4)2#&3%#()2#&4/3/#&.D)P'.)=>>N)
M)-(-23%#&/2)4'3'G'.-.)'&4)/&)('&,$',-.)#37-%)37'&)
c&,(/.7T)2D)2'%%H/&,)#$3):#%-)-O3-&./A-):-3'F'&'(H.-.) We set this minimum for the following reasons:
Council on Crime Prevention (Ttofi et al)=>>MDN so we did not code this element as included in
Over all measures of bullying, the weighted 37/.)<%#,%'::-N);&)Figures 1 and 2, studies are
:-'&)Z6)P'.)G-3P--&) NR )'&4) NR!0)/&4/2'3/&,)') organised according to research design.
substantial effect of these programmes on bullying. Element 1)BP7#(-F.27##()'&3/FG$((H/&,)<#(/2HD)
J#),/A-)')2#&2%-3-)-O':<(-0)/*)37-%-)P-%-)=>)G$((/-.) /&A#(A-.)37-)<%-.-&2-)#*)')*#%:'()'&3/FG$((H/&,)
'&4)M>)&#&FG$((/-.)/&)37-)-O<-%/:-&3'()2#&4/3/#&) <#(/2H)/&)37-).27##(N);&):'&H).27##(.0)'.)/&4/2'3-4)
'&4)=a)G$((/-.)'&4)_R)&#&FG$((/-.)/&)37-)2#&3%#() by researchers, such a policy was already in effect.
condition, the OR would be 1.41. Hence, OR = Element 2 (classroom rules) refers to the use of rules
NR )2'&)2#%%-.<#&4)3#)!>e):#%-)G$((/-.)/&)37-) ','/&.3)G$((H/&,)37'3).3$4-&3.)P-%-)-O<-23-4)3#)
2#&3%#()2#&4/3/#&)B#%)2#&A-%.-(H)=!e)*-P-%)G$((/-.) *#((#PN);&):'&H)<%#,%'::-.0)37-.-)%$(-.)P-%-)37-)
/&)37-)-O<-%/:-&3'()2#&4/3/#&DN)) %-.$(3)#*)2#F#<-%'3/A-),%#$<)P#%S)G-3P--&)37-)
With regard to victimisation, over all studies, 3-'27-%.)'&4)37-).3$4-&3.0)$.$'((H)'*3-%).#:-)-O3-&3)
37-)P-/,73-4):-'&)Z6)P'.)G-3P--&) N!!)'&4) #*)-O<#.$%-)#*)37-).3$4-&3.)3#)37-)<7/(#.#<7H)#%)
N!V0)/&4/2'3/&,)./,&/*/2'&3)-**-23.)#*)37-.-) :-..',-.)#*)37-)'&3/FG$((H/&,)<%#,%'::-N);&)
programmes on victimisation (being bullied). To many cases the rules were written on a notice
,/A-)')*$%37-%)/(($.3%'3/A-)-O':<(-0)/*)37-%-)P-%-) that was displayed in a distinctive place in the
=>)A/23/:.)'&4)M>)&#&FA/23/:.)/&)37-)-O<-%/:-&3'() classroom. Element 3 (school conferences) refers
2#&4/3/#&0)'&4)=V)A/23/:.)'&4)_V)&#&FA/23/:.)/&) to the organisation of school assemblies during
37-)2#&3%#()2#&4/3/#&0)37-&)Z6)g) N!!N)b-&2-0)#$%) which children were informed about bullying.
A'($-.)#*)37-)Z6)2#%%-.<#&4)3#)=Ve):#%-)A/23/:.) ;&):'&H)<%#,%'::-.0)37-.-)2#&*-%-&2-.)P-%-)
/&)37-)2#&3%#()2#&4/3/#&)B#%)2#&A-%.-(H0)=>e)*-P-%) #%,'&/.-4)'*3-%)37-)<%-F3-.3)4'3')2#((-23/#&)'&4)
A/23/:.)/&)37-)-O<-%/:-&3'()2#&4/3/#&DN) '/:-4)3#)/&*#%:).3$4-&3.)'G#$3)37-)-O3-&3)#*)
bullying in their school. This was perceived as a
way of sensitising students about bullying and as a
Key elements of the programme means of announcing the formal beginning of the
intervention programme in the school. Element 4
c'27)'&3/FG$((H/&,)<%#,%'::-)/&2($4-4)')A'%/-3H) (curriculum materials) refers to the use of materials
#*)/&3-%A-&3/#&)-(-:-&3.N);&)#%4-%)3#)/&A-.3/,'3-) 'G#$3)G$((H/&,)4$%/&,)2('..%##:)(-..#&.N)Q#:-)
the relationship between intervention elements and programmes involved a new curriculum whereas in
effect size in a comparable way, all elements were #37-%.)3-'27-%.)/&2#%<#%'3-4)'&3/FG$((H/&,):'3-%/'(.)
dichotomised (in order to produce roughly equal into the regular curriculum.
groups of studies, as much as possible). Figure 1 Element 5 (classroom management) refers
shows the elements of the intervention for each to an emphasis on classroom management
.3$4HN);&)2#&.3%$23/&,)37/.)L/,$%-0)P-)2#&.$(3-4) techniques in detecting and dealing with bullying.
the evaluators of the various programmes, and Element 6)B2#F#<-%'3/A-),%#$<)P#%SD)%-*-%.)3#)37-)
sent them our coding of the elements of the 2#F#<-%'3/#&)':#&,)4/**-%-&3)<%#*-../#&'(.)B$.$'((H)
/&3-%A-&3/#&N)CH):/4F"$(H)=>>M0)P-)7'4)%-2-/A-4) among teachers and some other professional
*--4G'2S)#&)=R)#$3)#*)!>)<%#,%'::-.)'&4) groups) in working with bullies and victims of
relevant changes were made to the coding where bullying. Elements 7 and 8 (work with bullies and
appropriate. For instance, even though Bauer, victims) concern individualised work, not offered in
E#\'&#)K)6/A'%')B=>>_D)/&2($4-4)'&)'&3/FG$((H/&,) the classroom, with children involved in bullying
A/4-#0)37/.)'&3/FG$((H/&,):-37#4)P'.)/:<(-:-&3-4) '.)A/23/:.)#%)<-%<-3%'3#%.N);&):#.3)<%#,%'::-.0)
in only two out of seven intervention schools, this service was offered by professionals, such as
X7'3)P#%S.)/&)<%-A-&3/&,)G$((H/&,Y)-**-23/A-)-(-:-&3.)#*)'&3/FG$((H/&,)<%#,%'::-.
De Rosier, 2004 N N N Y N Y Y Y N N N N N N N N N N N N
Fekkes et al, 2006 Y Y Y Y N N N N N Y Y Y Y N N N Y N N Y
Frey et al, 2005 Y Y N Y Y N Y Y N Y Y N N N N N Y N Y N
Hunt, 2007 Y Y Y Y Y N N N N Y Y Y N N N N N N N N
Jenson & Dieterich, 2007 N N Y Y N N N N N Y N N N N N N N N N N
Rosenbluth et al, 2004 Y N Y Y Y Y N Y Y Y Y N N N N N Y EP N N
Salmivalli et al 2009 Y Y N Y Y Y Y Y Y Y Y N Y Y N N Y IN Y Y
Before/after experimental-control comparisons
Andreou et al, 2007 N Y N Y Y N N N N Y N N N N N N Y N N N
Bauer et al, 2007 Y Y Y Y Y N Y Y N Y Y Y N N N N Y N N N
Ciucci & Smorti, 1998 N N Y N N Y N N Y Y N N N N N N Y N N N
Melton et al, 1998 Y Y Y Y Y Y Y Y Y Y Y Y Y N N N Y N Y N
Menard et al, 2008 Y Y N Y Y Y N N N Y Y N N N N N Y CP N N
Bergen 2 [1997–1998] Y Y Y Y Y Y Y Y N Y Y Y Y N N N Y MP Y N
Pepler et al, 2004 Y Y Y Y Y N Y Y Y Y Y Y Y N N N Y IN N N
Rahey & Craig, 2002 Y Y N Y Y Y Y Y Y Y N N N N N N Y IN N N
Rican et al, 1996 N Y N Y Y N N N N Y N N Y N N N N N Y N
Stevens et al, 2000 Y Y Y Y Y Y Y Y N Y Y N N Y Y N Y MP Y N
Whitney et al, 1994 Y Y Y Y Y N Y Y Y Y Y Y N Y N N N N Y N
Other experimental-control comparisons
Evers et al, 2007 N N N N N N N N N Y Y N N N N N N N Y Y
Galloway & Roland, 2004 N N N N N N N N N Y N N N N N N Y N N N
Ortega et al, 2004 Y Y N Y Y Y Y Y Y Y Y N N Y N N Y N N N
Raskauskas, 2007 Y Y N Y Y Y N N N Y Y N N N N N N N Y N
Age-cohort designs
Ertesvag & Vaaland, 2007 Y Y Y N Y N Y Y N Y Y Y N N N N Y IN Y N
Bergen 1 [1983–1985] Y Y Y Y Y Y Y Y N Y Y Y Y N N N Y MP Y N
Oslo 1 [1999–2000] Y Y Y Y Y Y Y Y N Y Y Y Y N N N Y MP Y N
National Norway [2001–2007] Y Y Y Y Y Y Y Y N Y Y Y Y N N N Y MP Y N
Oslo 2 [2001–2006] Y Y Y Y Y Y Y Y N Y Y Y Y N N N Y MP Y N
Salmivalli et al, 2004 Y Y N Y Y N Y Y N Y N N N Y N N Y N N N
_
X7'3)P#%S.)/&)<%-A-&3/&,)G$((H/&,Y)-**-23/A-)-(-:-&3.)#*)'&3/FG$((H/&,)<%#,%'::-.
X7'3)P#%S.)/&)<%-A-&3/&,)G$((H/&,Y)-**-23/A-)-(-:-&3.)#*)'&3/FG$((H/&,)<%#,%'::-.
De Rosier, 2004 A D E G X X M O R S W1 Y 0.87 1.04
Fekkes et al, 2006 A C F H I " M P R S V Y 1.12 1.25
Frey et al, 2005 A D E G " K M O R S W1 Y 1.04 1.09
Hunt, 2007 A D E G X X M O R T W2 Y 1.46 1.26
Jenson & Dieterich, 2007 A D F H X X M O R S W1 Y 1.17 1.63
Rosenbluth et al, 2004 B C E G I L M P R T W1 Y 0.99 0.70
Salmivalli et al 2009 B D F H J L N P R S V Y 1.47 1.66
Andreou et al, 2007 A D E G J L M O R S V Y 1.75 1.48
Bauer et al, 2007 B C E G J K M P R T W1 Y σ 1.01
Ciucci & Smorti, 1998 A D F " I " M O Q S V Y 1.20 1.21
Melton et al, 1998 B C F H " " M P Q T W1 Y 1.52 1.06
Menard et al, 2008 A D F G " L M P R T W1 Y 1.64 1.22
Bergen 2 [1997–1998] B C E H J L N P R T U Y 1.79 1.43
Pepler et al, 2004 B C F " J K M O R S W2 Y 1.69 0.94
Rahey & Craig, 2002 B D E G I " M O Q S W2 Y 1.19 0.79
Rican et al, 1996 A C E " X X M O Q S V Y 2.52 2.43
Stevens et al, 2000 B C E G " L M O Q T V Y σ σ
Whitney et al, 1994 B C F " X X M P Q S V Y 2.12 1.26
Evers et al, 2007 A D " G X X M O R T W1 Z 2.15 2.33
Galloway & Roland, 2004 A D F " J L M O R S U Z 1.20 1.59
Ortega et al, 2004 B D F H J L N O R T V Z 1.63 2.12
Raskauskas, 2007 A D E G X X M P R S W2 Z 1.20 1.35
Ertesvag & Vaaland, 2007 B D F " J L M P R T U Z 1.34 1.18
Bergen 1 [1983–1985] B C F H J L N P Q T U Z 1.69 2.89
Oslo 1 [1999–2000] B C F H J L N P R T U Z 2.14 1.81
National Norway [2001–2007] B C F H J L N P R T U Z 1.78 1.59
Oslo 2 [2001–2006] B C F H J L N P R T U Z 1.75 1.48
Salmivalli et al, 2004 A C F " J L M P R T V Z 1.31 1.30
?
X7'3)P#%S.)/&)<%-A-&3/&,)G$((H/&,Y)-**-23/A-)-(-:-&3.)#*)'&3/FG$((H/&,)<%#,%'::-.
8) :--3/&,.)P/37)<'%-&3.N 37-#%HN)L#%)-O':<(-0)37-)k%--S)'&3/FG$((H/&,)
programme of Andreou, Didaskalou & Vlachou
At the individual level the intervention included: B=>>_D)P'.)/&.</%-4)GH)Q'(:/A'((/I.)B ???D)/4-')
8) )3'(S.)P/37)G$((/-.)'&4)37-/%)<'%-&3.)'&4) 37'3)G$((H/&,)/&A#(A-.).#2/'()%#(-.)'&4)-O<-23'3/#&.)
-&*#%2-:-&3)#*)&#&F7#.3/(-0)&#&F<7H./2'() that are supported by bystanders as well as by
sanctions bullies and victims. Therefore, the programme
8) )3'(S.)P/37)A/23/:.0)<%#A/4/&,).$<<#%3)'&4) targeted all students in raising awareness about
providing assertiveness skills training to help bullying and about the causes and consequences of
them learn how to successfully deal with '4#<3/&,)4/**-%-&3)%#(-.N)b#P-A-%0)&#)'&3/FG$((H/&,)
bullying; also, talks with the parents of victims <%#,%'::-)P'.)G'.-4)#&)P-((F4-A-(#<-4)'&4)3-.3-4)
theories of bullying such as defiance theory or
8) )3'(S.)P/37)27/(4%-&))/&A#(A-4)3#):'S-)37-:)
reintegrative shaming theory (Ttofi & Farrington,
become effective helpers.
=>>M'T)=>>MGDN)6-.-'%27)/.)&--4-4)3#)4-A-(#<)'&4)
This successful programme could be the basis of test better theories of bullying and victimisation as
*$3$%-)'&3/FG$((H/&,)/&/3/'3/A-.N);3)/.)P#%37)/&,) a basis for new intervention programmes.
37'3)37-)37-#%-3/2'()#%/-&3'3/#&)#*)'&3/FG$((H/&,)
programmes (ie. whether they were inspired or Effect size versus study features
based on the OBPP or not) was significantly There have been few other attempts to relate effect
associated with a decrease in bullying. These size to programme elements (see eg. Kaminski et
correlations do not prove a causal effect of these al0)=>>MDN)Table 2 shows the programme elements
components on bullying but they are suggestive. and design features that were significantly (or
Most programmes seem to be based on common nearly significantly in two cases) related to effect
sense ideas about what works in preventing bullying sizes for bullying. Because of insufficient variation,
rather than on specific theories of bullying. */A-)#*)37-)=>)<%#,%'::-)-(-:-&3.)2#$(4))G-)
Q#:-)<%#,%'::-.)P-%-)(##.-(H)G'.-4)#&)') investigated (curriculum materials, information
Design features
Age of children 10- (14) 1.21 11+ (14) 1.57 20.09 .0001
Publication year 04+ (18) 1.31 03- (10) 1.69 18.75 .0001
Outcome measure Other (21) 1.33 2+M (7) 1.74 18.51 .0001
In Norway Rest (21) 1.34 Nor (7) 1.58 7.76 .005
In Europe Rest (12) 1.32 EU (16) 1.53 6.47 .011
Notes: Cat = Category of variable; OR = Weighted mean odds ratio; QB = heterogeneity between groups; Duration in days; Intensity
in hours; Outcome Measure 2+M: two times per month or more (versus other measures)
Design features
Outcome measure Other (22) 1.18 2+M (7) 1.64 49.19 .0001
In Europe Rest (13) 1.13 EU (16) 1.52 40.90 .0001
Design 12 (19) 1.13 34 (10) 1.53 40.73 .0001
In Norway Rest (22) 1.20 Nor (7) 1.55 30.77 .0001
Not in USA US (8) 1.10 Rest (21) 1.45 27.26 .0001
Publication year 04+ (19) 1.23 03- (10) 1.52 21.04 .0001
Notes: Cat = Category of variable; OR = Weighted mean odds ratio; QB = heterogeneity between groups; Design: 12 = randomised
experiments + before/after / experimental-control versus 34 = other experimental-control + age-cohort designs; Duration in days;
Intensity in hours; Outcome Measure 2+M: two times per month or more (versus other measures)