Navigation path

Additional tools

  • Print version
  • Decrease text
  • Increase text

Kiva Antibullying Programme

Evidence level:
Best PracticePromising PracticeEmergent Practice
 
Evidence of Effectiveness:
?-0+++
Transferability:
?-0+
Enduring Impact:
?-0+
Review criteria and process

Policy category

Helping Vulnerable Children, Facilitating Positive Transitions to Adulthood

Recommendation Pillars

Improve education systems’ impact on equal opportunities, Put in place mechanisms that promote children’s participation in decision making that affect their lives

Countries that have implemented practice

Finland

Years in Operation

2007  - still operating

Practice Overview

The KiVa programme is a school-wide approach to decreasing the incidence and negative effects of bullying on student well-being at school.  The programme’s impact is measured through self and peer-rated reports of bullying, victimization, defending victims, feeling empathy towards victims, bystanders reinforcing bullying behaviour, anxiety, self-esteem, depression, liking school, and academic motivation and performance, among other factors.  The programme is based on the idea that how peer bystanders behave when witnessing bullying plays a critical role in perpetuating or ending the incident.  As a result, the intervention is designed to modify peer attitudes, perceptions, and understanding of bullying.  The programme specifically encourages students to support victimised peers rather than embolden bullying behaviour and, furthermore, provides teachers and parents with information about how to prevent and address the incidence of bullying.

Practice Overview

The KiVa programme is a school-wide approach to decreasing the incidence and negative effects of bullying on student well-being at school.  The programme’s impact is measured through self and peer-rated reports of bullying, victimization, defending victims, feeling empathy towards victims, bystanders reinforcing bullying behaviour, anxiety, self-esteem, depression, liking school, and academic motivation and performance, among other factors.  The programme is based on the idea that how peer bystanders behave when witnessing bullying plays a critical role in perpetuating or ending the incident.  As a result, the intervention is designed to modify peer attitudes, perceptions, and understanding of bullying.  The programme specifically encourages students to support victimised peers rather than embolden bullying behaviour and, furthermore, provides teachers and parents with information about how to prevent and address the incidence of bullying.

The intervention has been tailored to three age specific groups:  ages 7-9 in grades 1-3, ages 10-12 in grades 4-6, and ages 13 -15 in grades 7-9.  The two main components of the programme are universal (school wide) and targeted (student level) activities.

The universal component of the intervention includes in-class student lessons and virtual learning sessions.  In primary school, teachers conduct 10 90-minute lessons over the 10 month-academic calendar, according to the KiVa teacher’s manual. The manual provides detailed implementation guidelines to facilitate replication elsewhere, including descriptions of course content and systematic discussion techniques to address cases of bullying.  In secondary schools, however, teachers are given the flexibility to instruct the four themes outlined in the manual, either as lessons or as day events, the equivalent of 13-23 lessons. The lessons aim to raise awareness of group influence in maintaining bullying behaviours, to increase empathy towards victims of bullying, to promote victim support strategies and self-efficacy in defending against bullying, and to increase children’s coping skills when victimized.  KiVa also utilizes an antibullying computer game for elementary schools and an internet forum called “KiVa Street” for lower secondary schools to motivate students to apply skills learned through interacting with peers and class material.  The game can be played and the internet forum accessed either during or between lessons.  Finally, KiVa offers a parent guide and promotional awareness material to intervention schools, such as vests for teachers and recess supervisors and posters.

The targeted component of the intervention applies to all age groups and involves adults, namely teachers, directly intervening in specific acute cases of bullying brought to their attention and, to a lesser extent parents, who are offered information about what is currently known about bullying and what can be done at home and at school in order to help bullying victims.  The targeted actions involve private discussions with victims and bullies as well as with “prosocial” classmates, who teachers have identified as being inclined to support the victimized and having a certain amount of social influence or popularity in the classroom.  A team of 3 teachers or relevant school personnel hold discussions with bullies or victims, while the classroom teacher meets separately with 2-4 “prosocial” or high-status classmates who are not directly involved in the incident to encourage them to support the victim.  The indicated actions attempt to enhance empathy towards victimized peers and self-efficacy to support them.

The Finnish Ministry of Education funded the University of Turku in 2006 for 3 years to implement and evaluate the effect of the KiVa antibullying program on the incidence of bullying in mainland comprehensive education schools.

Transferability

Since its inception in Finland as a pilot project in 2006, the KiVa anti-bullying programme has become widely acknowledged for its effectiveness. As a result, Finland nationalised the programme in 2009.  As of 2013, FinnWayLearning reports that 90% of all comprehensive schools in the country are registered KiVa schools (see: http://www.finnwaylearning.fi/en/home).  In addition, KiVa is currently being replicated in other European countries, including the Netherlands, the UK, France and Sweden, as well as in other regions of the world, such as the USA.  In the effort to promote the programme abroad, KiVa materials are being developed in German and Japanese.  Negotiations regarding KiVa implementation in Brazil, Canada, Chile, Estonia, and Korea are equally underway.

The University of Turku provides authorities and schools interested in replicating KiVa with necessary program materials.  The exact content of programme materials and how these are disseminated to implementation partners is not specified on the KiVa website.  However, the evaluation studies do indicate that participating partners accessed most training materials via security protected online accounts.

Evidence of Effectiveness

Evaluation 1

Kärnä, Antti, Voeten, Marinus, Little, Todd D, Poskiparta, Elisa, Kaljonen, Anne, and Salmivalli, Christina, A Large-Scale Evaluation of the KiVa Antibullying Program: Grades 4-6, in Child Development, Vol. 82, No.1, 2011, pp.311-330.

This school-level randomized control trial (also known as a cluster RCT) measures KiVa programme effects on 11 outcome variables gathered from data collected in May 2007 (T1), in December 2007 or January 2008 (T2) and in May 2008 (T3).  A population of 8,237 students ages 10-12 (grades 4-6) from a sample of 78 randomly selected, nationally representative schools participated in the study.

For most outcomes, such as Self-reported bullying and victimisation (measured on scales of 0=not at all to 4=several times a week) and Peer-reported bullying, victimisation, assisting, and reinforcing (measured as proportion scores varying from 0-1), lower mean scores indicate a positive result.  In contrast, higher scores indicate a better outcome for Peer-reported defending victims (measured on a scale of 0-1), Empathy towards victims (on a scale of 0=never to 4=always), Self-efficacy for defending (on a scale of 0=very difficult for me to 3=very easy for me), and Well-being at school (on a scale of 0=I disagree completely to 4=I agree completely).

Summary of Results for Evaluation 1

 

 

Outcome

Treatment Group T2

Control Group T2

Outcomes improved (statistically significant)

Means

Means

Peer-Reported Victimization (range:  0-1)

.059

.070

Peer-Reported Defending (range:  0-1)

.215

.194

Antibullying Attitudes (range: 0-4)

3.186

3.078

Empathy Toward Victims (range: 0-4)

2.003

1.912

Outcomes with no effect

Means

Means

Peer-Reported Bullying (range:  0-1)

.060

.070

Self-Reported Victimization (range: 0-4)

.738

.829

Self-Reported Bullying (range: 0-4)

.355

.432

Peer-Reported Assisting (range:  0-1)

.077

.091

Peer-Reported Reinforcing (range:  0-1)

.116

.127

Self-Efficacy of Defending (range: 0-3)

1.799

1.773

Well-Being at School (range: 0-4)

3.004

2.902

 

 

Outcome

Treatment Group T3

Control Group T3

Outcomes improved (statistically significant)

Means

Means

Self-Reported Victimization (range: 0-4)

.485

.657

Self-Reported Bullying (range: 0-4)

.273

.348

Peer-Reported Victimization (range:  0-1)

.049

.065

Peer-Reported Assisting (range:  0-1)

.071

.086

Peer-Reported Reinforcing (range:  0-1)

.107

.120

Self-Efficacy of Defending (range: 0-3)

1.88

1.809

Well-Being at School (range: 0-4)

2.871

2.748

Outcomes with no effect

Means

Means

Peer-Reported Bullying (range:  0-1)

.054

.070

Peer-Reported Defending (range:  0-1)

.189

.171

Antibullying Attitudes (range: 0-4)

3.134

3.049

Empathy Toward Victims (range: 0-4)

1.673

1.608

Note: Overall, intervention group n = 39 schools, 4,207 students and control group n= 39 schools, 4,030 students.  Table results, however, are based on a sample of intervention n= 4,201 and control n=3,965.


Evaluation 2
Kärnä, Antti, Voeten, Marinus, Little, Todd D., Poskiparta, Elisa, Alanen, Erkki, Salmivalli, Christina, Going to Scale: Nonrandomized Nationwide Trial of the KiVa Antibullying Program for Grades 1-9¸Journal of Consulting and Clinical Psychology, Vol.79, No.6, 2011, pp.796-805.
This study involved a quasi-experimental evaluation of KiVa’s nation-wide dissemination programme implemented over the 2009-2010 academic year in 888 schools covering 150,000 students from grades 1-9.  Data was collected in May 2009 and in May 2010. The evaluation design compares post-test data from students in each grade cohort having undergone the KiVa programme with pre-test data from same-school students from the cohort of the year above, a year earlier who had not yet participated in the KiVa programme.
The outcomes of interest, the percentage of students self-reporting being bullies and the percentage of students self-reporting being bullied (victimisation), were measured using global items from the Olweus Bully/Victim Questionnaire (1996 revision).  Using the Solberg and Olweus (2003) criterion, students who reported being bullied 2-3 times a month, every week, or several times a week during the past couple of months were categorised as victims.  Those reporting they had bullied others at the same frequency were categorised as bullies.  The amount of program implementation per school was recorded as the percentage of fully or partially classes completed, computer games played, and number of bullying cases handled.

Summary of Results for Evaluation 2

Outcome

Treatment Group

Control Group

Outcomes improved (statistically significant)

 

 

Self-Reported Victimisation by Grade

% Self-Reported Victimisation

% Self-Reported Victimisation

1

23.0

25.9

2

20.0

23.5

3

16.1

19.1

4

12.7

16.2

5

10.7

13.5

6

9.8

11.8

1–9

16.3

13.8

Self-Reported Bullying by Grade

% Self-Reported Bullying

% Self-Reported Bullying

1

12.0

13.6

2

9.9

12.5

3

7.7

9.6

4

7.2

9.4

5

7.2

8.7

6

8.5

10.0

1–9

9.6

11.1

Outcomes with no effect

 

 

Self-Reported Victimisation by Grade

% Self-Reported Victimisation

% Self-Reported Victimisation

7

10.6

11.6

8

8.8

10.0

9

8.3

9.3

Self-Reported Bullying by Grade

% Self-Reported Bullying

% Self-Reported Bullying

7

10.8

11.3

8

11.5

11.9

9

12.5

13.2

Note: n=888 schools, including about 150,000 students from grades 1-9, ages 8-16, the equivalent of 28% of the school population

Evaluation 3

 

Kärnä, Antti, Voeten, Marinus, Little, Todd D., Poskiparta, Elisa, Alanen, Erkki, Salmivalli, Christina, Effectiveness of the KiVa Antibullying Program: Grades 1-3 and 7-9, Journal of Educational Psychology, Online First Publication, 2012.

 

This study conducted a cluster RCT of the KiVa programme with students from grades 1-3 and 7-9 that lasted 9 months, from mid-August 2008-May 2009 using data collected in May 2008, in December 2008-February 2009, and in May 2009, from a representative sample of 151 comprehensive basic education on mainland Finland.

 

The study examined 7 outcomes of interest for grades 8-9 and examined two outcomes for grades 2-3.  In terms of self-reported victimisation (i.e. being bullied) and self-reported bullying, the proportion of victims and bullies (on a scale of 0-1) was calculated according to the proportion of students self-reporting bullying or victimisation 2-3 times per month, once per week or several times per week during the past couple of months.  Smaller proportions in the treatment group indicate a positive outcome.  Peer-reported bullying, victimisation, assisting, reinforcing, and defending variables collected peer nominations for each classroom student.  Total nominations for each student were divided by the number of classmate responses, resulting in a score from 0 to 1 for each student on each of the outcome variables.  Smaller proportions in the treatment groups indicate positive change, except for peer-reported defending, where larger mean scores indicate a positive outcome.

 

Summary of Results for Evaluation 3


Outcome

Treatment Group T2

Control Group T2

Outcomes improved (statistically significant)

 

 

1.Self-Reported Victimisation

Mean Proportion

Mean Proportion

Grade 2-3

0.13

.16

Grade 8-9

0.06

0.08

2.Self-Reported Bullying

Mean Proportion

Mean Proportion

Grade 2-3

0.04

0.05

3.Peer-Reported Victimisation

Mean Proportion

Mean Proportion

Grade 8-9

0.06

0.06

7.Peer-Reported Defending

Mean Proportion

Mean Proportion

Grade 8-9

0.08

0.08

Outcomes with no effect

 

 

2.Self-Reported Bullying

Mean Proportion

Mean Proportion

Grade 8-9

0.06

0.06

4.Peer-Reported Bullying

Mean Proportion

Mean Proportion

Grade 8-9

0.05

0.05

5.Peer-Reported Assisting

Mean Proportion

Mean Proportion

Grade 8-9

0.07

0.06

6.Peer-Reported Reinforcing

Mean Proportion

Mean Proportion

Grade 8-9

0.10

0.11

Outcome

Treatment Group T3

Control Group T3

Outcomes improved (statistically significant)

 

 

1.Self-Reported Victimisation

Mean Proportion

Mean Proportion

Grade 2-3

0.13

0.17

2.Self-Reported Bullying

Mean Proportion

Mean Proportion

Grade 2-3

0.04

0.06

3.Peer-Reported Victimisation

Mean Proportion

Mean Proportion

Grade 8-9

0.05

0.05

5.Peer-Reported Assisting

Mean Proportion

Mean Proportion

Grade 8-9

0.05

0.05

7.Peer-Reported Defending

Mean Proportion

Mean Proportion

Grade 8-9

0.06

0.07

Outcomes with no effect

 

 

1.Self-Reported Victimisation

Mean Proportion

Mean Proportion

Grade 8-9

0.07

0.07

2.Self-Reported Bullying

Mean Proportion

Mean Proportion

Grade 8-9

0.05

0.07

4.Peer-Reported Bullying

Mean Proportion

Mean Proportion

Grade 8-9

0.04

0.04

6.Peer-Reported Reinforcing

Mean Proportion

Mean Proportion

Grade 8-9

0.07

0.08

Note: Outcome analysis was only conducted on grades 2-3 (4,704 students ages 8-9 from 36 control and 38 treatment schools) and grades 8-9 (11,070 students ages 14-15 from 35 control and 38 treatment schools) because cohorts from grades 1 and 7 only did not have pre-test measurements.

Evaluation 4

 

Salmivalli, Christina, Kärnä, Antti, and Poskiparta, Elisa, Counteracting bullying in Finland: The KiVa program and its effects on different forms of being bullied, International Journal of Behavioral Development, 35:204, 2011.

 

The cluster RCT for this study was implemented from August 2006-May 2007 on students from grades 4-6, ages 10-12.  Pre-test data was collected in May 2006 and post-test measures in May 2007.  The outcomes of interest for this study were collected via Internet-based student completed questionnaires.  Students were first presented with the global victimisation question from the Revised Olweus’ Bully /Victim questionnaire (Olweus, 1996), followed by 9 items describing specific forms of being bullied: verbal, exclusion, physical, manipulative, material, threat, racist, sexual, and cyber.  Students answered according to a 5 point frequency scale: not at all, once or twice, 2 or 3 times a months, every week, several times a week.  Victims are identified based on the Solberg and Olweus (2003) criteria of being bullied 2-3 times per month.  Negative percentage change in the form of bullying reported amounts to a positive outcome.

 

Summary of Results for Evaluation 4

Outcome

Treatment Group

Control Group

Outcomes improved (statistically significant)

Pre to post-test % change

Pre to post-test % change

Global bullying

-40%

-11%

Outcomes with no effect

Pre to post-test % change

Pre to post-test % change

Verbal bullying

-22%

+9%

Exclusion bullying

-17%

+8%

Physical bullying

-40%

+13%

Manipulative bullying

-25%

+5%

Material bullying

-30%

+33%

Threat bullying

-50%

-29%

Racist bullying

-45%

-18%

Sexual bullying

-17%

+15%

Cyber bullying

-36%

+14%

Note: Table results based on a sample of 5,651 students from 78 schools with complete pre-test and post-test measures.

 

Evaluation 5

 


Salmivalli, Christina, Sainio, Miia, Veenstra, Rene, and Huitsing, Gijs, Same- and Other-Sex Victimization: Are the Risk Factors Similar?, Aggressive Behavior, Vol. 38, 2012, pp.442-455.
This study analyses the results of a cluster RCT of the KiVa programme conducted from 2007-2009 on students from grades 4-6 and grades 8-9 from a sample of 234 schools.  T1 data was collected in May 2007 for grades 3-5 graders and in May 2008 for grades 7-8.  T2 data was collected for grades 4-6 and 8-9 in December 2007-January 2008 and December 2008-January 2009 respectively.  T3 data was collected in May 2008 and May 2009.
Students filled out Internet-based questionnaires administered by trained teachers to harvest data on same-sex, other-sex, and both sex victimisation frequency, self-esteem scores based on questions about how much students agreed with different feelings they could have about themselves and their peers (on a scale of 0=not ture-4=exactly true), friendship frequency per student, peer rejection, and popularity.  Peer rejection, popularity, and friendships nominations were summed and divided by the number of the same and other-sex nominating classmates.  Odds ratios less than one for the intervention group imply a positive outcome for the programme.

 

Summary of Results for Evaluation 5

Outcome

Treatment Group

Outcomes improved (statistically significant)

Odds Ratio

Same-sex victimised between level

0.69

Other-sex victimised between level

0.56

Victimised by both between level

0.53

Outcomes with no effect

Odds Ratio

 

Note: Sample size used for table results: 15,304 students from 145 schools with complete T1, T2, T3 data

Evaluation 6

Williford, Anne, Boulton, Aaron, Noland Brian, Little Todd, Kärnä, Antii and Salmivalli, Christina, Effects of the KiVa Anti-bullying Program on Adolescents’ Depression, Anxiety, and Perception of Peers, Journal of Abnormal Child Psychology, Vol.40, No.2, 2012, pp.289-300.

The study is based on the cluster RCT of the KiVa programme implemented during the 2007-2008 academic year in a sample of 78 schools reviewed in Kärnä et al. (2011a) but focuses on evaluating the KiVa program’s impact on outcomes not previously assessed, namely adolescent depression, anxiety and perception of peers using data collected in May 2007, in December 2007/January 2008, and in May 2008.

For peer-reported victimisation, depression, and anxiety variables of interest, a negative mean difference between the treatment and control groups indicates a positive outcome.  For perception of peers, a positive mean difference between the treatment and control groups indicates a positive outcome.

Summary of Results for Evaluation 6

Outcome

Cohen’s d Effect Sizes for Condition Mean Comparisons

Outcomes improved (statistically significant)

Reported victimisation wave 2

-1.08

Reported victimisation wave 3

-2.19

Peer perception wave 3

.20

Anxiety wave 3

-.13

Outcomes with no effect

 

Reported victimisation wave 1

.13

Peer perception wave 1

.04

Depression wave 1

.02

Depression wave 3

-.09

Anxiety wave 1

-.03


 

 

 

 

 

 

 

 

 

 

 

Note: d<0.2 small effect, 0.2<d<0.5 medium effect, d>0.8 large effect.  Results are based on a sample of 7,741 students (4,056 in the treatment group and 3,685 in the control group) from 78 schools.

 

Outcome

Treatment and Control Mean Differences

Outcomes improved (statistically significant)

Reported victimisation wave 2

-0.00049

Reported victimisation wave 3

-0.00088

Peer perception wave 3

0.000245

Anxiety wave 3

-0.00011

Outcomes with no effect

 

Reported victimisation wave 1

-0.000492125

Peer perception wave 1

-0.000876

Depression wave 1

0.000245

Depression wave 3

-0.000107208

Anxiety wave 1

-0.000492125

Note: Mean differences were calculated from the Cohen’s d effect sizes using the average of the 4 standard errors for the treatment or control condition for each outcome variable presented in Table 3 of the evaluation p. 296.  Results are based on a sample of 7,741 students (4,056 in the treatment group and 3,685 in the control group) from 78 schools.

 

Evaluation 7

Ahtola, Annarilla, Haataja, Anne, Kärnä, Antti, Poskiparta, Elisa, Salmivalli, Christina, For children only? Effects of the KiVa antibullying program on teachers, Teaching and Teacher Education, 28, 2012, pp.851-859.

This study examines the effects of the KiVa programme on grade 1-3 teachers based on the phase of KiVa’s cluster RCT implemented over the course of the 2008-2009 academic year.  Data for 128 teachers belonging to 33 intervention schools and for 110 teachers from 29 control schools was collected between September 17 and October 8, 2008 via web-based questionnaires.

There were three main outcomes of interest.  Teachers’ self-evaluated ability to address bullying problems and teacher confidence in the effectiveness of the KiVa programme were outcomes measured in terms of questions regarding teacher self-perceptions and opinions, which were answered on a 5 point Likert scale, ranging from 0=very little  to 4=very much.  Teacher understanding of bullying as a preventable or treatable phenomenon was measured in terms of agreement with 8 statements, also on a 5 point Likert scale, ranging from 0=disagree completely to 4=agree completely.  On a scale of 0-4, a higher score for the treatment group suggests the programme had a positive outcome for the variable of interest being examined.

 

Summary of Results for Evaluation 7

Outcome

Treatment Group

Control Group

Outcomes improved (statistically significant)

Mean Teacher Perceptions on a Scale of 0-4

Mean Teacher Perceptions on a Scale of 0-4

Ability to address bullying problems (competence to tackle)

2.76

2.60

Outcomes with no effect

 

 

Understanding of bullying as a preventable or treatable phenomenon (malleability of bullying)

3.20

3.18

Confidence in Programme Effectiveness (effectiveness of KiVa)

2.77

2.78


Note: n=128 teachers belonging to 33 intervention schools and 110 teachers from 29 control schools.

Issue to consider

As described above, the effectiveness of the KiVa program varied by age.  In particular, the effects were greatest among children in elementary school.  Additionally, the University of Turku was the program developer and only evaluator. Finally, note that children were only able to participate in the studies discussed here if their parents consented to their participation.

Information on 'Years in Operation':

The KiVa RCT was conducted from 2007-2009.  The quasi-experimental non-randomised evaluation of the nationalised program was conducted during 2009-2011.  As of 2013, 90% of comprehensive schools in mainland Finland were registered KiVa schools implementing the programme.

Information on 'Type of Organization':

National government: the Finish Ministry of Education and Culture hold the copy right

Schools: According to its website 90% of comprehensive schools in the country are registered KiVa schools implementing the programme.

University: the University of Turku owns the right to distribute KiVa licenses to third parties and managed all evaluation studies in Finland

 

Contact Information

Name

Mr. Juha Ollila

Title

Project Manager International Affairs of KiVa

Organization

KiVa Program – FinWay Learning - Turku University

Address

Turku University

Phone

+358 44 0700 044

Email

juha.ollila@utu.fi

Website

Available Resources

http://www.kivaprogram.net/evidence-of-effectiveness

http://www.finnwaylearning.fi/en/home

http://www.ac-paris.fr/portail/jcms/p1_347795/price-programme-experimental-de-prevention-de-la-violence-en-milieu-scolaire?hlText=KiVa

http://www.ac-paris.fr/portail/jcms/p1_359800/des-videos-comme-supports-pour-le-travail-avec-les-eleves?hlText=KiVa

Doll, B., Pfohl, W., and Yoon, J., Editors, Handbook of Youth Prevention Science, Routledge, NY, 2010, pp.238-252.

Evaluation Details

To date, KiVa developers have studied the programme in two separate randomized trials, as well as a quasi-experimental national scale-up study.  According to the published literature reviewed, KiVa studies have found that the programme significantly reduces both self- and peer-reported bullying and victimisation (Kärnä et al. 2011a; Kärnä et al. 2012).  While KiVa was not found to significantly reduce specific forms of bullying (physical, manipulative, cyber, etc.), it was found to reduce bullying in general (Salmivalli et al. 2011).  Evidence further suggests that KiVa reduces same-sex, other-sex, and both-sex victimisation more effectively in elementary schools than in middle schools, where the programme did not have significant effects on other-sex and both-sex victimisation (Sainio et al. 2012).  In addition, the programme was found to reduce anxiety and depression and to increase the perception of a positive peer climate (Williford et al. 2012).  Lastly, 2010 results from the KiVa deployment across the Finnish school system indicated that after a year of implementation both victimisation and bullying in grades 1-9 had reduced significantly (Kärnä et al. 2011b).

Evaluation 1

From 2007-2008, a cluster RCT was conducted involving 8,237 students and 429 classrooms.  In 2006, all 3444 elementary and middle schools from mainland Finland were invited to participate in the KiVa programme.  For the first phase of the program being reviewed, the 275 volunteering schools were stratified by province and language and then 78 were randomly assigned to control and treatment groups of equal sizes and only students from grades 4-6, ages 10-12, were covered. Out of 275 eligible volunteering schools, 78 schools were randomly assigned to treatment and control groups.  In the treatment group, there were 39 schools and 4,207 students.  The control group included 39 schools and 4,030 students.  Students that left school after base line data was collected were dropped from the sample.  The final sample size for the analysis was 8,166 students, 4,201 in the treatment group and 3,965 in the control group.

To be eligible, schools had to provide basic education in mainland Finland in at least one of the official languages, Finnish or Swedish.  To be eligible, students had to come from grades 1-3, ages 7-9, grades 4-6, ages 10-12, and grades 7-9, ages 13-15, and parent consent forms had to be submitted for students to participate.

Intervention effects were calculated by comparing intervention school students with control school students at 7 and 12 months after baseline data was collected.  Outcome estimates accounted for change over time, individual student differences, between classroom differences, and between-school differences.  Covariates controlled for included: gender (boys report higher levels of bullying and victimization), age (previously shown to be an important predictor of bullying and victimization), and language of instruction (Swedish speaking minority students reported lower levels of bullying and victimisation).

The 11 outcomes of interest were measured using different sets of questionnaires.  The Self-reported bullying, Peer-reported bullying, and Self-reported victimization scores were based on the global items scale from the Olweus Bully/Victim Questionnaire (1996 revision).    Peer-reported victimisation, Defending victims, and Assisting and reinforcing bullies scores were based on scales from a 12 item Participant Role Questionnaire (Salmivalli et al. 1996).  Empathy towards victims  used a 7 item empathy scale (Poyhonen, Karna, & Salmivalli, 2008).  Self-efficacy for defending used a 3 item scale derived from the participant role questionnaire items for defending behaviour (Poyhonen et al. 2010).  Finally, well-being at school was measured with items developed by the Finnish National Board of Education such as liking school, academic self-concept, school climate and classroom climate.

Evaluation 2

This study involved a quasi-experimental evaluation of KiVa’s national wide dissemination program from May 2009 to May 2010, an intervention lasting 9 months.  The research design used a cohort-longitudinal design with adjacent cohorts to rule out maturation effects.  In other words, post-test data from students in each grade cohort were compared with pre-test data from students from the same school but from the cohort of the year above a year earlier that had not yet undergone the intervention.  Thus, the data compares different cohorts but only for the same grade level, when the older students were at the same age as the younger cohort.   In other words, first-graders in May 2010 that had been exposed to the intervention for a full academic year were compared with data from students who were in first grade in May 2009 that had not yet undergone the intervention.  Data was collected twice in May 2009 and May 2010.

Grades 1-9, ages 8-16, from comprehensive education schools (offering compulsory schooling in grades 1-9) in mainland Finland were recruited.  The Government offered free materials and training for applications submitted from 2008-2010.  Only schools that participated in web-based pre-test and post-test surveys were included into the final sample.  In total, 888 schools participated in the study, including approximately 150,000 students in 11,200 classrooms (28% of the school population and 30% of the student population).  For victimization, there were samples of 156,634 students for the control group and 141,103 for the treatment group.  For bullying, there were samples of 156,629 students for the control and 141,099 for the intervention.

Self-reported bullying and self-reported victimization were measured using global items from the Olweus Bully/Victim Questionnaire (1996 revision).  Using the Solberg and Olweus (2003) criterion, students who reported being bullied 2-3 times a month, every week, or several times a week during the past couple of months were categorised as victims.  Those reporting they had bullied others at the same frequency were categorised as bullies.  The construct validity of these global measures were tested by examining school level correlations between the global bullying an victimization items, used in the present study on a sample of 25,833 students from grades 1-9 that took part in the RCT of the KiVa program, with the averages for self-reported forms of bullying and victimization as well as peer reports of bullying and victimisation.  Correlation was substantial and statistically significant.  The amount of program implementation per school was also recorded (% of fully or partially classes completed, computer games played, # of bullying cases handled).

Evaluation 3

This study conducted a cluster RCT of the KiVa programme on students from grades 1-3 and 7-9 from a sample of 151 out of 275 comprehensive basic education schools (offering compulsory schooling in grades 1-9) in mainland Finland that had initially volunteered to participate in the programme.  No special education only schools were allowed.  In addition, students who did not obtain parental consent to participate in the study or who left participating schools during the intervention were excluded.

Multilevel modelling was used to establish intervention effects in the presence of nested data structures.  Outcome analysis was only conducted on grades 2-3 (4,704 students ages 8-9 from 36 control and 38 treatment schools) and 8-9 (11,070 students ages 14-15 from 35 control and 38 treatment schools) because cohorts from grade 1 and 7 only had post-test measurement.  At the time of pre-testing, these students were not yet enrolled in the participating schools making data collection for these cohorts more burdensome for schools to collect.  The intervention lasted 9 months, from mid-August-May.  Data was collected in three waves: first in May 2008, then sometime between December 2008-February 2009, and lastly in May 2009.

A total of 78 schools were assigned to the treatment groups: 31 of these schools had previously participated in the control group for the first phase of the study looking at grades 4-6 and the remaining 47 schools were randomly assigned.  The control group consisted in 73 randomly assigned schools.  In total, 40 control and 39 intervention schools were selected for the grades 1-3 sample and 39 control and 39 treatment schools were selected for the grades 7-9 sample.  With one exception, schools participated in either the grades 1-3 sample or the grades 7-9 sample, but not both.

Self and peer-reported bullying and self and peer-reported victimization were measured using global items from the Olweus Bully/Victim Questionnaire (1996 revision).  Using the Solberg and Olweus (2003) criterion, students who reported being bullied 2-3 times a month, every week, or several times a week during the past couple of months were categorised as victims.  Those reporting they had bullied others at the same frequency were categorised as bullies.  Assisting, defending and reinforcing were measured using the Participant Role Questionnaire (Salmivalli & Voeten, 2004).  Implementation fidelity was measured as the average number of lessons and themes conducted per school.

Evaluation 4

The cluster RCT for this study was implemented from August 2006-May 2007 on consenting/authorised students from grades 4-6, ages 10-12, in 78 eligible mainland, basic education schools teaching in either Finnish or Swedish.  Pre-test data was collected in May 2006 and post-test measures in May 2007.  All mainland standard education institutions were contacted and invited to join the study.  The 275 schools that volunteered were then stratified by province and language.  A total of 78 schools were randomly assigned to control and treatment groups of equal sizes.  Parent consent forms were then solicited.  The total sample of 4-6th graders with pre and posttest measures consisted of 5,651 students from grades 4-6.  The 39 intervention schools included 3,347 intervention students and the 39 control schools included 2,304 control students.

For correlational analysis of different forms of bullying a sample of 7,303 pre-test students was used. For pretest-posttest analysis, the sample of 5,651 students with both pre-test and post-test measures was used.  The outcomes of interest for this study were collected via Internet-based student completed questionnaires.  Students were first presented with the global victimisation question from the Revised Olweus’ Bully /Victim questionnaire (Olweus, 1996), followed by 9 items describing specific forms of being bullied: verbal, exclusion, physical, manipulative, material, threat, racist, sexual, and cyber.  Students answered according to a 5 point frequency scale: not at all, once or twice, 2 or 3 times a months, every week, several times a week.  Victims are identified based on the Solberg and Olweus (2003) criteria of being bullied 2-3 times per month.

Evaluation 5

This study analyses the results of a cluster RCT of the KiVa programme conducted from 2007-2009 on students from grades 4-6 and grades 8-9 from 234 basic education elementary and middle schools in Finland.  All elementary and middle, teaching grade 1 up to grade 9 education, were invited to participate.  Stratified random sampling was used to categorize 234 out of 275 volunteering schools into 117 treatment and 117 control schools.  Treatment and control groups were both representative of all 5 provinces in mainland Finland.  Intervention schools were offered free materials and training. Control schools were promised all materials and pre-implementation training after the RCT.  Students had to be from grades 1-9, have guardian consent and have complete explanatory variable data for the three waves of data used for analysis.  Classes with less than 7 students were excluded from the sample.  Classes with less than 60% participation rate at T2 were also excluded.  The final sample with complete data for all three waves included 15,304 students from 145 schools, 78 in the treatment group and 67 in the control group.  T1 data was collected in May 2007 for grades 3-5 graders and in May 2008 for grades 7-8.  T2 data was collected for grades 4-6 and 8-9 in December 2007-January 2008 and December 2008-January 2009 respectively.  T3 data was collected in May 2008 and May 2009.

Students filled out Internet-based questionnaires administered by trained teachers to harvest data on variables of interest, namely: same-sex, other-sex, and both sex victimisation frequency, self-esteem, friendship frequency per student, peer rejection and popularity.

Same–sex and other-sex victimisation definitions were based on the global item victimisation/bullying scale from Olweus (1996).  Students were presented with a roster of classmate names and were asked to mark the names of the students that bullied them.  Students were then categorised into victimised by same-sex or other-sex peers or victimised by both dummy variables.

Self-esteem was measured using a 9-item scale based on the Rosenberg Self-Esteem Scale (Rosenberg, 1965).  Students were asked to rate how much they agreed with different feelings they possibly had about themselves around peers.  For instance, when asked if they felt that they have a number of good qualities, students could respond on a 5 point bipolar scale from not true to exactly true (0-4).  The scores for the ten items were aggregated and averaged.

To measure same and other-sex peer rejection, perceived popularity and friendship status, students were asked to mark names on a class roster that belonged to each definition.  For perceived popularity, they were asked to indicate up to three peers they considered to be the most popular in the class.  For peer rejection and liking, students were asked to nominate who they liked the least and who they liked the most.  Friendship status was measured as a being a reciprocated liking nomination in the first, second or third data collection wave.  Peer rejection, popularity, and friendships nominations were summed and divided by the number of the same and other-sex nominating classmates.

Multinomial logistic regressions were used to predict the likelihood of different outcomes for categorical dependent variables.  Multilevel modelling was conducted to accommodate for clustering at the school and classroom levels in the data and to account for within and between-level variation.  Missing data was addressed using the full information maximum likelihood approach.

Evaluation 6

This study examined the effects of the KiVa programme implemented during the 2007-2008 school year on adolescent depression, anxiety and perception of peers using a sample of 7,741 students from 78 schools.  The study is based on the cluster RCT design from Kärnä et al. (2011a) but focuses on evaluating the KiVa program’s impact on outcome measures not previously assessed.  Participating schools were randomly selected from the pool of 275 schools that had initially volunteered for the programme. The treatment group was composed of 39 schools and 4,056 students.  The control group included 39 school and 3,865 students.  Eligible schools provided basic education in mainland Finland in at least one of the official languages, Finnish or Swedish.  Eligible students were in grades 1-3, ages 7-9, grades 4-6, ages 10-12, (for the 1st phase of the program covered in this publication), or grades 7-9, ages 13-15, and had parental consent to participate in the KiVa programme.

Controlling for gender, age, language of instruction, immigrant status, the four outcomes of interest included: Peer-reported victimisation at waves 1,2,3, Perception of peers at waves 1,3, Depression at waves 1,3, and Anxiety at waves 1,3.  Peer-reported victimisation classified students as victims by the number of peer nominations recorded.  The number of peer nomination for each student was totalised and divide by the number of students in each classroom to estimate the proportion of peer-reported victimisation.  Perception of peers assessed student perception of peer characteristics as supportive, kind and trustworthy or not (Salmivalli et al. 2005) using a 13 item scale.  Depression was measured using a 7-item scale derived from the Beck Depression Inventory (BDI, Beck et al 1996).  BDI items were translated into Finnish and selected based on their computability with children.  Anxiety was measured using a combination of two social anxiety scales, the Fear of Negative Evaluation and the Social Avoidance and Distress (Garcia-Lopez et al. 2001).

Outcomes of interest were collected at three separate times: once in May 2007, once in December 2007/January 2008, and once in May 2008.  At each wave of data collection, teachers administered online surveys during regular class hours

Confirmatory factor analysis (CFA) and chi squared tests were used to compare mean differences in outcome variables between the different study groups.  A multiple group cross lagged panel structural equation model (Little et al. 2007) was estimated to determine when reductions in reported victimisation had positively influenced peer perception, depression and anxiety.

Evaluation 7

This study examines the effects of the KiVa programme on grade 1-3 teachers.  Teachers from grades 1-3 were specifically targeted because the first years of formal schooling affect the direction of children’s entire school career.  The evaluation was based on a cluster RCT design implemented over two academic school years, from 2007-2009.  The data for this study comes from the second phase of the KiVa evaluation program, implemented over the course of the 2008-2009 academic year, and was collected between September 17 and October 8, 2008.  In total, 37 treatment schools and 37 control schools were recruited for the final target sample.  However, complete teacher data was only available for a total of 62 schools.  128 teachers belonged to the 33 intervention schools and 110 teachers belonged to the 29 control schools.

All schools that had been randomly assigned to the control condition in the 1st phase of the evaluation in 2007 and wanted to participate in the 2nd phase were included in the intervention condition.  In addition, new schools were randomly added to the intervention and control groups.  A total of 28 waiting list schools (control schools from the first phase wanting to run the intervention) were included as intervention schools in the second phase.  Overall, the treatment group included 40 schools, 28 waiting list schools and 12 randomly assigned new schools.  The control group was comprised of 39 randomly assigned schools.  All participating schools were selected from the pool of 275 schools that initially volunteered for the evaluation.

The study does not consider the waiting list a serious limitation to the randomisation procedure because all of the schools came from the same pool of volunteers.  This pool was initially randomized into intervention and control groups.  As a result, the study might have ended up with the same samples without the waiting list mechanism.  The waiting list procedure was considered a worthwhile trade off.  It incentivised control schools to stay in the programme without incurring noticeable effects on estimation results.  To avoid overlapping, such as grade 6 control group students counting as grade 7 intervention group students, all of grade 7 observations were dropped from the analysis.

Three outcomes of interest were measured.  Firstly, teachers’ self-evaluated ability to address bullying problems was measured with 2 questions (In your opinion, how much do you know about school bullying? In your opinion, how good are your skills to reduce school bullying?) on a 5 point Likert scale (very little, 0, to very much, 4).  Secondly, teacher confidence in the effectiveness of the KiVa programme was measured with 3 questions (To what extent do you believe that the KiVa programme will decrease bullying incidents? To what extent do your believe that the KiVa program will enhance the well-being of victimised students?) using the same 5 point Likert scale.  Lastly, teacher understanding of bullying as a preventable or treatable phenomenon was measured in terms of agreement with 8 statements (such as Group dynamics are the reason for bullying and school staff can have an influence, or school staff cannot help it, some kids just bully others) on a 5 point Likert scale, ranging from disagree completely, 0, to agree completely, 4.

Bibliography

1.Kärnä, Antti, Voeten, Marinus, Little, Todd D, Poskiparta, Elisa, Kaljonen, Anne, and Salmivalli, Christina, A Large-Scale Evaluation of the KiVa Antibullying Program: Grades 4-6, in Child Development, Vol. 82, No.1, 2011(a), pp.311-330.

2.Kärnä, Antti, Voeten, Marinus, Little, Todd D., Poskiparta, Elisa, Alanen, Erkki, Salmivalli, Christina, Going to Scale: Nonrandomized Nationwide Trial of the KiVa Antibullying Program for Grades 1-9¸Journal of Consulting and Clinical Psychology, Vol.79, No.6, 2011(b), pp.796-805.

3.Kärnä, Antti, Voeten, Marinus, Little, Todd D., Poskiparta, Elisa, Alanen, Erkki, Salmivalli, Christina, Effectiveness of the KiVa Antibullying Program: Grades 1-3 and 7-9, Journal of Educational Psychology, Online First Publication, 2012.

4.Salmivalli, Christina, Kärnä, Antti, and Poskiparta, Elisa, Counteracting bullying in Finland: The KiVa program and its effects on different forms of being bullied, International Journal of Behavioural Development, 35:204, 2011.

5.Sainio, Miia, Veenstra, Rene, Huitsing, Gijs, and Salmivalli, Christina, Same- and Other-Sex Victimization: Are the Risk Factors Similar?, Aggressive Behavior, Vol. 38, 2012, pp.442-455.

6.Williford, Anne, Boulton, Aaron, Noland Brian, Little Todd, Kärnä, Antii and Salmivalli, Christina, Effects of the KiVa Anti-bullying Program on Adolescents’ Depression, Anxiety, and Perception of Peers, Journal of Abnormal Child Psychology, Vol.40, No.2, 2012, pp.289-300.

7.Ahtola, Annarilla, Haataja, Anne, Kärnä, Antti, Poskiparta, Elisa, Salmivalli, Christina, For children only? Effects of the KiVa antibullying program on teachers, Teaching and Teacher Education, 28, 2012, pp.851-859.

Enduring Impact

Although the KiVa was replicated over several waves from 2006-2010, follow-ups have always compared control and treatment cohorts after a one year.  No two year follow-ups have been analysed or studied as of yet.