A Large-Scale Evaluation of the KiVa Antibullying Program: Grades 4–6
Antti Ka¨rna¨
University of Turku
Marinus Voeten
Radboud University Nijmegen
Todd D. Little
University of Kansas
Elisa Poskiparta, Anne Kaljonen,
and Christina Salmivalli
University of Turku
This study demonstrates the effectiveness of the KiVa antibullying program using a large sample of 8,237
youth from Grades 4–6 (10–12 years). Altogether, 78 schools were randomly assigned to intervention
(39 schools, 4,207 students) and control conditions (39 schools, 4,030 students). Multilevel regression analyses
revealed that after 9 months of implementation, the intervention had consistent beneficial effects on 7 of the
11 dependent variables, including self- and peer-reported victimization and self-reported bullying. The results
indicate that the KiVa program is effective in reducing school bullying and victimization in Grades 4–6.
Despite some evidence against school-based interventions, the results suggest that well-conceived school-
based programs can reduce victimization.
Bullying is a common problem in schools, affecting
the lives of a large number of students. It is com-
monly characterized as systematic abuse of power
(Smith & Sharp, 1994). More specifically, bullying is
defined as repeated aggressive behavior against a
victim who cannot readily defend himself or herself
(Olweus, 1999). Victims of bullying often experi-
ence insecurity and various forms of psychosocial
maladjustment, such as depression and anxiety;
they sometimes even exhibit self-destructiveness
(for meta-analyses, see Card, 2003; Hawker & Boul-
ton, 2003). For a number of victims, their experi-
ences continue to affect their lives later on in the
forms of depression, low self-esteem, and difficulty
in trusting other people (Isaacs, Hodges, & Salmival-
li, 2008; Olweus, 1994). Not only are victims at risk:
Compared to other children, bullies often become
involved in delinquency and alcohol abuse
(Kaltiala-Heino, Rimpela¨, Rantanen, & Rimpela¨,
2000; Loeber & Dishion, 1983; Nansel et al., 2001;
Nansel et al., 2004; Olweus, 1993a, 1993b). The need
to intervene effectively in bullying is thus clear and
urgent. Accordingly, numerous antibullying pro-
grams have been initiated by researchers, practitio-
ners, and governments. The present study is the first
evaluation of a new antibullying program, designed
for national use in Finnish comprehensive schools.
Antibullying Programs
Several whole-school intervention programs
have been developed to reduce bullying in schools
(for reviews, see Baldry & Farrington, 2007;
Farrington, & Ttofi, 2009; Ferguson, San Miguel,
Kilburn, & Sanchez, 2007; Merrell, Gueldner, Ross,
& Isava, 2008; J. D. Smith, Schneider, Smith, & Ana-
niadou, 2004; P. K. Smith, Ananiadou, & Cowie,
2003; P. K. Smith, Pepler, & Rigby, 2004; Vreeman
& Carroll, 2007). A whole-school approach views
bullying as a systemic problem with multiple
causes at the individual, classroom, and school
levels (J. D. Smith et al., 2004). This layered per-
spective suggests that an intervention must target
the entire school context, rather than just individual
bullies and victims (J. D. Smith et al., 2004). In this
regard, whole-school interventions differ from
This research is part of the KiVa project for developing an
antibullying intervention program for the Finnish comprehensive
schools. The KiVa project is financed by the Finnish Ministry of
Education and Culture. In addition, the present study was sup-
ported by the Academy of Finland Grants 134843 and 135577 to
Christina Salmivalli. We thank the whole KiVa project team, and
especially Marita Kantola and Jonni Nakari, for their contribu-
tion in the data-gathering process.
Correspondence concerning this article should be addressed to
Antti Ka¨rna¨, Department of Psychology, University of Turku,
Assistentinkatu 7, 20014 Turun yliopisto, Finland. Electronic mail
may be sent to ankarna@utu.fi.
Child Development, January/February 2011, Volume 82, Number 1, Pages 311–330
� 2011 The Authors
Child Development� 2011 Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2011/8201-0021
DOI: 10.1111/j.1467-8624.2010.01557.x
more narrowly focused interventions, such as cur-
riculum interventions, social-skills groups, and
counseling (Vreeman & Carroll, 2007).
Most whole-school antibullying programs
were inspired by Dan Olweus’s first Bergen study.
Olweus (1991) utilized a multilayered approach by
targeting individual, class, and school levels with
different intervention components, such as serious
talks with bullies and victims, classroom discus-
sions, and staff meetings. The evaluation was con-
ducted using a cohort-longitudinal design with
time-lagged comparisons. With this design, stu-
dents after the intervention were compared with
students from the same grades in the same schools
before the intervention. For instance, Grade 5 pre-
test data served as a baseline against which the
posttest data from students in Grade 4 were com-
pared after 12 months of intervention. Most com-
parisons showed reductions in victimization and
bullying rates of 50% or more from the baseline fre-
quency. Substantial decreases also emerged for
other antisocial behaviors, such as vandalism, theft,
and truancy, in addition to an increase in general
satisfaction with school life.
Since the first Bergen project, several effective-
ness studies have been conducted in various coun-
tries (e.g., Baldry & Farrington, 2004; Cross, Hall,
Hamilton, Pintabona, & Erceg, 2004; Frey et al.,
2005; O’Moore & Minton, 2004; Pepler, Craig, Zie-
gler, & Charach, 1994; Pitts & Smith, 1995; Roland,
1989; Salmivalli, Kaukiainen, & Voeten, 2005;
Smith & Sharp, 1994; Stevens, De Bourdeaudhuij, &
Van Oost, 2000). Unfortunately, these studies
have shown very inconsistent results, with the
majority of studies nonsignificant, some negative,
and only a few with beneficial outcomes (J. D.
Smith et al., 2004). J. D. Smith et al. (2004) con-
cluded that the amassed evidence is simply too var-
iable to justify adopting such programs to the
exclusion of other procedures.
One potential explanation of the inconsistencies
in the evaluation findings is that Olweus’s remark-
able success is due to the high quality of Scandina-
vian schools, with, for instance, particularly well-
trained teachers (J. D. Smith et al., 2004). In
addition, Baldry and Farrington (2007) proposed
that the inconsistent findings may be associated
with variations in assessment methods and evalua-
tion designs (see also Farrington & Ttofi, 2009;
Vreeman & Carroll, 2007).
The variable and often weak results may be, at
least partly, explained by methodological weak-
nesses of the studies (Baldry & Farrington, 2007).
Several authors have lamented the methodological
problems inherent in the effectiveness studies of
antibullying intervention programs (Baldry & Farr-
ington, 2007; J. D. Smith et al. 2004; Vreeman &
Carroll, 2007). Somewhat surprisingly, all previous
bullying intervention studies lack at least one and
often several methodologically important features,
such as an appropriate control condition, random
assignment, multilevel modeling of hierarchical
data, multimethod and multi-informant outcome
assessment, psychometrically sound measures, sys-
tematic implementation monitoring, proper sample
size, attrition analysis or missing data imputation
(Baldry & Farrington, 2007; Farrington & Ttofi,
2009; J. D. Smith et al., 2004; Vreeman & Carroll,
2007). As a result, the studies clearly fall short of
the standards of evidence required for interventions
to be considered efficacious (see Flay et al., 2005,
for the standards); therefore, only limited empirical
support exists for the effectiveness of school-
based antibullying programs (J. D. Smith et al.,
2004). Numerous reviews of the effectiveness of
the antibullying programs have called for fur-
ther research using higher methodological stan-
dards to rigorously investigate whether such
programs actually are effective or not (Baldry &
Farrington, 2007; J. D. Smith et al., 2004; Vreeman
& Carroll, 2007).
KiVa Antibullying Program
The Finnish Ministry of Education and Culture
funded the development and evaluation of a new,
national antibullying program named KiVa (an
acronym for Kiusaamista Vastaan, ‘‘against bully-
ing’’). The program was developed at the Univer-
sity of Turku, in collaboration between the
Department of Psychology and the Centre for
Learning Research. It was introduced in the inter-
vention schools across Grades 4 through 6 during
the 2007–2008 school year.
Theoretical Background of the KiVa Program
KiVa enjoys a multifaceted theoretical back-
ground (e.g., Salmivalli, Ka¨rna¨, & Poskiparta,
2010a). The program is built on a view of bullying
that is based on two lines of research: (a) studies on
the social standing of aggressive children in general
(e.g., Cillessen & Mayeux, 2004; Rodkin, Farmer,
Pearl, & Van Acker, 2000) and bullies in particular
(Juvonen, Graham, & Schuster, 2003) and (b)
research on participant roles in bullying (Salmivalli,
Lagerspetz, Bjo¨rkqvist, O¨sterman, & Kaukiainen,
1996). Furthermore, social-cognitive theory (Bandura,
312 Ka¨rna¨ et al.
1989) is used as a framework for understanding the
processes of social behavior.
Recent research suggests that bullying behavior
is at least partly motivated by a pursuit of high sta-
tus and a powerful position in the peer group (e.g.,
Juvonen & Galva´n, 2008; Salmivalli & Peets, 2008).
Bullying is also a group phenomenon, in which
bystanders have an effect on the maintenance of
bullying and on the adjustment of the victims
(Salmivalli, 2009; Salmivalli et al., 1996). More spe-
cifically, bystanders can contribute to the mainte-
nance of bullying by assisting and reinforcing the
bully, which provides bullies with the position of
power that they seek. On the other hand, defending
the victim may make bullying an unsuccessful
strategy for attaining and demonstrating high sta-
tus. KiVa is predicated on the idea that a positive
change in the behaviors of classmates can reduce
the rewards gained by bullies and consequently
their motivation to bully in the first place. KiVa
places concerted emphasis on enhancing the empa-
thy, self-efficacy, and antibullying attitudes of
onlookers, who are neither bullies nor victims. This
strategy is based on sound evidence relating these
characteristics to defending and supporting victim-
ized peers (Caravita, DiBlasio, & Salmivalli, 2009;
Po¨yho¨nen, Juvonen, & Salmivalli, 2010; Po¨yho¨nen
& Salmivalli, 2008; Salmivalli & Voeten, 2004). The
aim is to make bystanders show that they are
against bullying and to make them support the vic-
tim, instead of encouraging the bully. As another
equally important component, the KiVa program
includes procedures for handling the acute bullying
cases that come to the attention of the school
personnel.
A prior Finnish bullying intervention study
(Salmivalli et al., 2005) was also based on similar
principles. That program, however, mainly
consisted of teacher education (making the actual
program content rather loose) without concrete
materials that teachers could utilize when
working with students and classrooms. It also
lacked a program manual needed for accurate repli-
cation.
KiVa Program Components
KiVa includes both universal and indicated
actions to prevent the occurrence of bullying as
well as to intervene in individual bullying cases
(e.g., Salmivalli et al., 2010a, 2010b). The program
has three different developmentally appropriate
versions for Grades 1–3, 4–6, and 7–9 (i.e., for 7–9,
10–12, and 13–15 years of age).
Universal actions. The KiVa program for Grades
4–6 includes 20 hr of student lessons (10 double les-
sons) given by classroom teachers during a school
year. The central aims of the lessons are to: (a) raise
awareness of the role that the group plays in main-
taining bullying, (b) increase empathy toward vic-
tims, and (c) promote children’s strategies of
supporting the victim and thus their self-efficacy to
do so. The lessons involve discussion, group work,
role-play exercises, and short films about bullying.
As the lessons proceed, class rules based on the
central themes of the lessons are successively
adopted one at a time.
A unique feature of KiVa is an antibullying com-
puter game included in the primary school versions
of the program. Students play the game during and
between the lessons described earlier. The game
involves five levels, each of them consisting of three
components named: I KNOW, I CAN, and I DO.
Students acquire new information and test their
existing knowledge about bullying (I KNOW), learn
new skills to act in appropriate ways in bullying
situations (I CAN), and are encouraged to make
use of their knowledge and skills in real-life situa-
tions (I DO).
KiVa provides prominent symbols such as
bright vests for the recess supervisors to enhance
their visibility and signal that bullying is taken
seriously in the school and posters to remind stu-
dents and school personnel about the KiVa pro-
gram. Schools get presentation graphics they can
use to introduce the program for the whole per-
sonnel and for parents. Parents also receive a
guide that includes information about bullying
and advice about what parents can do to prevent
and reduce the problem.
Indicated actions. In each school, a team of three
teachers (or other school personnel), along with the
classroom teacher, addresses each case of bullying
that is witnessed or revealed. Cases are handled
through a set of individual and small group discus-
sions with the victims and with the bullies, and sys-
tematic follow-up meetings. In addition, the
classroom teacher meets with two to four prosocial
and high-status classmates, encouraging them to
support the victimized child.
Trainingdays and school networkmeetings. Support
to implement the program is given to teachers and
schools in several ways. In addition to 2 full days
of face-to-face training, networks of school teams
are created, consisting of three school teams each.
The network members meet three times during the
school year with one person from the KiVa project
guiding the network.
Evaluation of KiVa Antibullying Program 313
KiVa naturally shares some features with exist-
ing antibullying programs, such as Olweus’s bully-
ing prevention program (OBPP). These features are
shared principles, or ideas, rather than actual pro-
gram contents. For instance, both OBPP and KiVa
include actions at the level of individual students,
classrooms, and schools, both tackle acute bullying
cases through discussions with the students
involved, and both suggest developing class rules
against bullying. KiVa, however, has at least three
features that, when taken together, differentiate it
from OBPP and other antibullying programs. First,
KiVa includes a broad and encompassing array of
concrete and professionally prepared materials for
students, teachers, and parents. Rather than offer-
ing ‘‘guiding principles’’ or ‘‘philosophies’’ to
school personnel, it provides them with a whole
pack of activities to be carried out with students.
Second, KiVa harnesses the powerful learning
media provided by the Internet and virtual learning
environments. Third, while focusing on the
bystanders, or witnesses of bullying, KiVa goes
beyond ‘‘emphasizing the role of bystanders,’’ men-
tioned in the context of several intervention pro-
grams, by actually providing ways to enhance
empathy, self-efficacy, and efforts to support the
victimized peers. Furthermore, students’ private
attitudes are made salient in order to reduce the
(often false) impression that ‘‘others think that bul-
lying is OK’’ (so-called pluralistic ignorance; see
Juvonen & Galva´n, 2008). Although other programs
share some of these features, none of them has
assembled these features into the coordinated
whole-school, multilayered intervention that is the
hallmark of the KiVa program. With regard to
research, the clear structure, well-defined content
and concrete materials of KiVa make it easy to use
and amenable to replication in further studies.
These features distinguish KiVa from some other
antibullying programs, for which such specific
components are not described in sufficient detail to
enable accurate replication (Vreeman & Carroll,
2007).
The Present Study
The present study expands knowledge about
the effectiveness of antibullying interventions by
examining the effects of the new KiVa antibully-
ing program on bullying, victimization, and other
key outcomes. We focused in this study on
Grades 4–6 because these were included in the
first phase (2007–2008) of program evaluation.
Results involving Grades 1–3 and 7–9 from the
second phase (2008–2009) will be presented in
upcoming reports.
We used several outcome measures to assess the
effectiveness of the KiVa program. The program
effects were examined by comparing intervention-
school students with control school students at two
time points: in the middle and in the end of the
school year (i.e., 4 and 9 months after beginning of
program implementation; 7 and 12 months after
pretest measures). As the main outcomes, we used
self-reported and peer-reported bullying and vic-
timization. We hypothesized that the KiVa program
would yield substantial reductions in these prob-
lem behaviors. We also expected beneficial changes
in other outcomes; specifically, we expected
increases in defending victims and decreases in
assisting and reinforcing bullies. We also hypo-
thesized that the intervention would increase anti-
bullying attitudes, empathy toward victims, and
self-efficacy for defending. Finally, we expected the
program to improve students’ well-being at school.
The evaluation was done using best practice
methodology and stringent standards for effective-
ness (Flay et al., 2005). The present study is respon-
sive to Baldry’s and Farrington’s (2007) call for
high-quality evaluations with theoretically
grounded interventions, randomized designs, and
multiple measures of effectiveness.
Method
Sampling and Design
To recruit schools, letters describing the KiVa
project were sent in the fall of 2006 to all 3,418
schools providing basic education in mainland
Finland. These included both Finnish-language and
Swedish-language schools, because the basic educa-
tion in Finland is given in both official languages.
The letter included information about the goals and
content of KiVa and an enrollment form. In this
first phase of program evaluation (Grades 4–6), the
275 volunteering schools were stratified by prov-
ince and language and 78 of them were randomly
assigned to intervention or control condition (spe-
cial-education-only schools were excluded). We
oversampled Swedish-language schools (15.3% of
the sample schools were Swedish, whereas 9.4% of
all Finnish comprehensive schools are Swedish-
language schools), but adjusted for this in the
analyses. The participating schools were located
throughout the country and resembled other com-
prehensive schools in such characteristics as class
size and proportion of immigrant students. As such,
314 Ka¨rna¨ et al.
they can be considered representative of Finnish
comprehensive schools.
Procedure
The school year in Finland ranges from mid-
August to the end of May. Data collection took
place three times: in May 2007, December 2007 or
January 2008, and May 2008. Students filled out
Internet-based questionnaires in the schools’ com-
puter labs during regular school hours. The process
was administered by the teachers, who were sup-
plied with detailed instructions about 2 weeks prior
to data collection. In addition, teachers were offered
support through phone or e-mail prior to and dur-
ing data collection. Teachers distributed individual
passwords to the students, who used them to log in
to the questionnaire. At the beginning of the ses-
sion, the term bullying was defined for the students
in the way f