Prevalence of School Bullying in Korean Middle School Students | Adolescent Medicine | JAMA Pediatrics | JAMA Network
[Skip to Navigation]
Sign In
August 2004

Prevalence of School Bullying in Korean Middle School Students

Author Affiliations

From the Department of Psychiatry, Hallym University, Anyang, South Korea; Department of Epidemiology, University of California, Berkeley, and Harold E. Jones Child Study Center, Berkeley (Dr Kim); the Department of Child and Family Studies, Yonsei University, Seoul, South Korea (Dr Koh); and the Department of Child and Adolescent Psychiatry and Laboratory of Developmental Neurosciences, Irving B. Harris Center for Developmental Studies, University of Chicago, and Sonia Shankman Orthogenic School, Chicago, Ill (Dr Leventhal).

Arch Pediatr Adolesc Med. 2004;158(8):737-741. doi:10.1001/archpedi.158.8.737

Background  School bullying is the most common type of school violence. Victimization by or perpetration of school bullying has frequently been associated with a broad spectrum of behavioral, emotional, and social problems.

Objective  To investigate the prevalence and demographic characteristics of victims, perpetrators, and victim-perpetrators in a Korean middle school sample.

Methods  We evaluated 1756 middle school students in this cross-sectional study. Students provided demographic information and completed the Korean-Peer Nomination Inventory. Descriptive statistics and the Pearson χ2 test were used.

Results  We found that 40% of all children participated in school bullying. By category, the prevalence of victims, perpetrators, and victim-perpetrators was 14%, 17%, and 9%, respectively. The most common subtypes of victimization were exclusion (23%), verbal abuse (22%), physical abuse (16%), and coercion (20%). Boys were more commonly involved in both school bullying and all 4 types of victimization. The prevalence of bullying was greater in students with either high or low socioeconomic status and in nonintact families.

Conclusions  School bullying is highly prevalent in Korean middle school students. Demographic characteristics can help identify students at greater risk for participation in school bullying.

School bullying, the most common type of school violence, comprises a spectrum of aggressive behaviors that involve both perpetrators and victims. Bullying is usually perpetrated by students who seem to have and try to maintain dominance over others. Bullying is intended to cause mental and/or physical pain to another person.1

Many believe that school bullying is a benign and "normal" part of the childhood and adolescent experience. Furthermore, some people suggest that children learn to manage bullying, resulting in an enhanced character and no adverse consequences.2,3 Scientific reports indicate otherwise. Victimization by or perpetration of school bullying is associated with behavioral, emotional, and social problems. It also appears that bullying can lead to serious mental and physical sequelae.4-6

Two methods have been widely used in school bullying research: self-report and peer nomination. Self-report has most commonly been used and has advantages by providing direct access to the feelings and experiences of children involved in bullying. This is particularly useful because the children are alert to the possibility of peer abuse, have strong emotional reactions to such events, and develop vivid and lasting memories of such experiences.7 However, potential bias may limit its validity. These biases include differences in children's interpretations of abusive interactions, variation in ability to encode or reconstruct such events, and variability in willingness to report painful or embarrassing experiences.8-10 Peer nomination has distinct advantages compared with self-report. It allows for assessment of an individual's behaviors by peers who are most likely to have witnessed or participated in these behaviors. In addition, it permits the aggregation of peer/classmate judgment about individuals' roles in school bullying. Aggregation mitigates against the biases of a particular person while allowing for the richness of individual reports at the level of the child.

This study uses the Korean-Peer Nomination Inventory (K-PNI) to investigate the prevalence of school bullying and victimization subtypes. Demographic characteristics are then used to identify groups involved in school bullying.


Study population

Middle schools were selected in Seoul and Anyang to represent typical Korean public middle school populations. Administrators and parents in the 2 schools agreed to participate in this cross-sectional study in October 2000. The study population was composed of all students in the seventh and eighth grades. The schools differed in size and class composition. The school in Anyang had more classes for each grade and larger class sizes than the school in Seoul (13 classes and 8 to 10 classes per grade, respectively; 41 to 50 and 34 to 41 students per class, respectively). Additionally, the Anyang school had single-sex classrooms, whereas those in Seoul were coeducational.

The Hallym University College of Medicine (Anyang) institutional review board approved the study. Parent and student consent were obtained. Each student completed a peer nomination questionnaire and provided demographic information with the direction of research assistants in each classroom during school hours. The entire survey took 45 to 60 minutes.


Korean-Peer Nomination Inventory

School bullying was identified using the K-PNI. The K-PNI is based on the Peer Nomination Inventory (PNI) developed by Wiggins and Winder11 to determine implicit or explicit judgments about peers as part of the examination of children's social maladjustment. To complete the questionnaire, children were asked to name classmates of the same sex who fit the behavioral type described in each item. The nomination of multiple individuals for each item was allowed. The median score on the PNI represents a natural break in the distribution that identifies children as having been nominated in a given category.

Perry et al9 adapted the original PNI to create a tool for identifying peer victimization and aggression. The modified PNI is composed of 26 items (7 victimization, 7 aggression, and 12 filter items). The reliability and validity of the modified PNI are well established. The K-PNI is an expansion of the modified PNI that not only identifies victims and perpetrators but also allows for subtyping victimization. It has 28 items: 11 for victims, 6 for perpetrators, and 11 filter items. The 11 victimization items are further subdivided into 4 types: exclusion (3 items), verbal abuse (3 items), physical abuse (2 items), and coercion (3 items). Items from the K-PNI questions appear in Table 1. Good to excellent reliability and validity of the K-PNI have previously been reported in Korean children.12

Table 1. 
Victim and Perpetrator Items on the Korean-Peer Nomination Inventory
Victim and Perpetrator Items on the Korean-Peer Nomination Inventory

To aggregate K-PNI data for individual students, victim and perpetrator scales of the K-PNI were expressed in a standardized percentage nomination (SPN) score. The SPN score was calculated by (1) summing the frequencies of nomination in all items of a scale, (2) dividing the summed frequencies by the total number of items in a scale, and (3) dividing this number by the number of same-sex students in a classroom.

An SPN score of 1 meant that an individual had been nominated more than once on either the victim or perpetrator scale. In a skewed distribution, as with the K-PNI, 1 SD does not accurately characterize the population, and a median cutoff is not appropriate when the median score for most children is 0. Thus, an SPN score of greater than 1 was used as the cutoff point for categorizing victims and perpetrators because it was more conservative and identified a more homogeneous group involved in bullying with less misclassification.

For the purpose of analysis, types of school bullying were categorized into 4 separate groups:

  1. None: no involvement in school bullying.

  2. Victim: involvement in school bullying only as a victim. This category includes about 40% of bullying.

  3. Perpetrator: involvement in school bullying only as a perpetrator. This category includes about 40% of bullying.

  4. Victim-perpetrator: involvement in school bullying as both a victim and a perpetrator. This includes about 20% of bullying.

An individual with an SPN score greater than 1 on both the victim and perpetrator scales was categorized as a victim-perpetrator. An SPN score greater than 1 on either scale alone indicated a perpetrator or victim. Victimization was defined as receiving an SPN score greater than 1 on the K-PNI victim scale irrespective of scores on the perpetrator scale, including victims and victim-perpetrators. Finally, an individual with an SPN score of 1 or less on both scales was classified as having no involvement in bullying. Subtyping of victimization was performed with the same cutoff score in each subtype composite.

Demographic Characteristics

Students completed questions about their sex, family members with whom they lived, parents' educational level, and a 5-scale measure of socioeconomic status (SES).

Statistical analysis

Descriptive statistics and the Pearson χ2 test were used to examine the prevalence of school bullying and the relationships between bullying and demographic characteristics.


Study population

Data from 1756 students were used for the final analysis (3 students were excluded owing to lack of validity: 1 form was illegible, and 2 had data entry errors). About one third of the students were from Seoul. Most students came from intact families and were of middle SES. Boys and girls were evenly distributed (Table 2). Seoul and Anyang samples differed in parental educational level and SES. Parents from Seoul were more likely than those from Anyang to have completed college (fathers, 56.4% vs 43.6%; mothers, 73.8% vs 26.2%). Students in Seoul were also more likely to be in the highest SES level (2.5% vs 1.3%) and less likely to have the lowest SES (0.3% vs 1.3%). The significance of geographic differences in this student population cannot be analyzed because these are not epidemiological samples and do not represent either of these regions.

Table 2. 
Study Subjects*
Study Subjects*

Prevalence of school bullying and victimization subtypes

A total of 40% of all students were involved in bullying. Bullying classifications were as follows: victim only, 14%; perpetrator only, 17%; and victim-perpetrator, 9%. Victim and perpetrator SPN scores were compared among those who had no involvement in bullying, victims, perpetrators, and victim-perpetrators. Median victim SPN scores were 0.00, 2.62, 0.21, and 2.88, respectively, suggesting that victims and victim-perpetrators were victimized to the same magnitude. On the other hand, median perpetrator SPN scores were 0.00, 0.00, 2.50, and 4.63, suggesting that victim-perpetrators bullied other students more severely than did perpetrators.

Significantly more boys than girls experienced school bullying. Classification of victimization subtypes ranged from 16% to 23%, with physical abuse the least frequent (16%) and the other 3 subtypes appearing with similar frequencies (exclusion, 23%; verbal abuse, 22%; and coercion, 20%). Although there were sex differences in all victimization subtypes (more bullying among boys than girls), this difference was greatest with physical abuse, for which boys outnumbered girls by more than 2 to 1 (Table 3).

Table 3. 
Frequency of School Bullying in the Study Population*
Frequency of School Bullying in the Study Population*

Demographic characteristics of students involved in school bullying

In terms of demographic characteristics, 2 factors appear to be associated with school bullying: SES and family structure. Bullying was disproportionately represented in the high and low SES groups. Among 29 individuals in the high SES group, 72.4% were involved in school bullying. Similarly, in the low SES group, 81.2% were involved in school bullying. Although the numbers are relatively low in each group, they are in sharp contrast to the 55% to 60% of children in the middle SES categories who had no involvement in school bullying (Figure 1).

Figure 1. 
School bullying prevalence by socioeconomic status (SES). The Pearson χ2 test was used to examine differences in prevalence among 5 levels of SES.

School bullying prevalence by socioeconomic status (SES). The Pearson χ2 test was used to examine differences in prevalence among 5 levels of SES.

To further understand the relationship between SES and school bullying, SPN scores were examined. For perpetrators, those with higher SPN scores were more likely to be in the highest SES group. Median perpetrator SPN scores in perpetrators were 6.03, 5.83, 4.55, 3.79, and 3.92, respectively, in the 5 SES categories from highest to lowest SES. Higher SPN scores for victimization were more likely to be found in both the highest and lowest SES categories. Median victim SPN scores in victims were 4.96, 2.53, 2.69, 1.90, and 3.93, respectively. Therefore, participation in school bullying appears to be a bimodal distribution with a concentration in the highest and lowest SES categories.

Although the numerical differences may be small, family structure also seems to play a role in regard to involvement in school bullying. Children from intact families were significantly less likely to be involved in school bullying (61.8% vs 47.1%). Furthermore, within-group differences were also significant, with children in intact families less likely to be involved in all types of bullying than those from nonintact families (victims, 13.5% vs 16.5%; perpetrators, 16.3% vs 21.5%; victim-perpetrators, 8.4% vs 14.9%) (Figure 2).

Figure 2. 
School bullying prevalence by family structure. The Pearson χ2 test was used to examine differences in prevalence among students from intact and nonintact families.

School bullying prevalence by family structure. The Pearson χ2 test was used to examine differences in prevalence among students from intact and nonintact families.


This study confirms in a Korean population that at a 40% prevalence, school bullying is the most common form of youth violence. Despite years of study by Olweus13 and many others,14 bullying and its many adverse consequences remain a problem for children and adolescents.

Our study adds important information to the literature on school bullying. First and foremost, this study confirms that school bullying is a hazard facing nearly half of our youth, irrespective of the country in which they live. Second, it allows investigators and physicians to examine school bullying through the eyes of the youth. By using the K-PNI, we were able to show that children are well aware of who is a bully and who is a victim. On the basis of children's keen observations, we learn that 40% of children are involved in school bullying with 17% acting as perpetrators, 14% as victims, and 9% as victim-perpetrators.

We have also learned that students of high SES are more likely to be perpetrators and that those of both high and low SES are more likely to be victims. The reasons for this remain unclear. Although one could speculate about the causal effects of SES on school bullying, the size and constitution of this sample make it impossible to arrive at any conclusions.

The lack of an intact family was also associated with an increased risk of participation in bullying. Family structure appears to be a general risk factor for problems in childhood.15-17 Based on this study, it is unclear whether family structure specifically affects school bullying. A different sample with a broader assessment of family structure would be necessary to answer this question.

Our study has some limitations. Even though the sample size is large, it is not epidemiological. This limits some of the conclusions that can be drawn. However, despite this study's limitations, such a large sample from 2 diverse Korean communities provides more than ample evidence that school bullying is ubiquitous. School bullying and its consequences occur in the face of empirical evidence for safe and effective interventions that can reduce this problem. Why are these interventions not being used? Is it because children do not matter? We think not; rather, bias against psychiatric and behavioral disorders leads them to be ignored or receive only cursory research and treatment support. Notwithstanding powerful evidence of significant morbidity associated with school bullying, including depression and suicide, professionals in Europe, the United States, Japan, and now Korea let 40% of their children be placed at risk for the adverse consequences of bullying.

Most articles of this sort end with a call for future studies to provide more clarity about treatment and causality. Although we certainly wish for that, it is not how we choose to end this article. Instead, we hope that these findings will add to the compelling burden of proof that demands a public health intervention to dramatically reduce school bullying before yet another 40% of our youth are damaged. More study is good, but action is better.


What This Study Adds

Despite assertions that bullying is a benign or normal character-building behavior, school bullying is a common form of violence directed at youths. Most research on school bullying, including prevalence studies, has been conducted in Western countries using self-report. Few studies use peer nomination and relate bullying to the demographic characteristics of the students involved.

This study found a 40% prevalence of school bullying in Korean middle school students using peer nomination and provides evidence that school bullying is highly prevalent irrespective of the country in which children live. Students with high SES are more likely to be perpetrators, whereas children in the high and low SES categories are more likely to be victims. Children from nonintact families are also more likely to be involved in bullying. Characteristics that can lead to early identification of and intervention in groups at risk for school bullying deserve further investigation. However, the prevalence is sufficiently high and the consequences severe enough to warrant the initiation of large-scale interventions in Korea and the United States.

Correspondence: Young Shin Kim, MD, MPH, Harold E. Jones Child Study Center, 2425 Atherton St, Berkeley, CA 94720-6070 (

Accepted for publication March 15, 2004.

This study was supported by a Health Promotion Grant 2000 from the Korea Institute for Health and Social Affairs, Seoul. Additional funding came from the Jean Young and Walden W. Shaw Foundation, Chicago, Ill.

Morita  Y Sociological Study on the Structure of Bullying Group.  Osaka, Japan Department of Sociology, Osaka City University1985;
Stutzky  G How do I know if my child is being bullied? Partnership for Learning Web site2004;Available at: February 10, 2004
Bernal  J Bullies seek power.  Texas School Safety Center Newsletter. 2002;Available at: February 10, 2004Google Scholar
Olweus  D Bullying at school: basic facts and effects of a school based intervention program.  J Child Psychol Psychiatry. 1994;351171- 1190PubMedGoogle ScholarCrossref
Salmon  GJames  ASmith  DM Bullying in schools: self reported anxiety, depression, and self esteem in secondary school children.  BMJ. 1998;317924- 925PubMedGoogle ScholarCrossref
Schwartz  DProctor  LJ Community violence exposure and children's social adjustment in the school peer group: the mediating roles of emotion regulation and social cognition.  J Consult Clin Psychol. 2000;68670- 683PubMedGoogle ScholarCrossref
Ladd  GWKochenderfer-Ladd  BJ Identifying victims of peer aggression from early to middle childhood: analysis of cross-informant data for concordance, estimation of relational adjustment, prevalence of victimization, and characteristics of identified victims.  Psychol Assess. 2002;1474- 96PubMedGoogle ScholarCrossref
Graham  SJuvonen  J Self-blame and peer victimization in middle school: an attributional analysis.  Dev Psychol. 1998;34587- 599PubMedGoogle ScholarCrossref
Perry  DGKusel  SJPerry  LC Victims of peer aggression.  Dev Psychol. 1988;24807- 814Google ScholarCrossref
Schwartz  D Self-reports: how the questions shape the answers.  Am Psychol. 1999;5493- 105Google ScholarCrossref
Wiggins  JSWinder  CL The peer nomination inventory: an empirically derived sociometric measure of adjustment in preadolescent boys.  Psychol Rep. 1961;9643- 677Google ScholarCrossref
Kim  YKoh  YNoh  J Development of Korean-Peer Nomination Inventory (K-PNI): an inventory to evaluate school bullying.  J Korean Neuropsychiatry Assoc. 2001;40867- 875Google Scholar
Olweus  D Bullying at school: basic facts and an effective intervention programme.  Promot Educ. 1994;127- 3148PubMedGoogle ScholarCrossref
Juvonen  JGraham  SSchuster  MA Bullying among young adolescents: the strong, the weak, and the troubled.  Pediatrics. 2003;1121231- 1237PubMedGoogle ScholarCrossref
Tousignant  MHabimana  EBiron  CMalo  CSidoli-LeBlanc  EBendris  N The Quebec Adolescent Refugee Project: psychopathology and family variables in a sample from 35 nations.  J Am Acad Child Adolesc Psychiatry. 1999;381426- 1432PubMedGoogle ScholarCrossref
Bergeron  LValla  JPBreton  JJ Pilot study for the Quebec Child Mental Health Survey, II: correlates of DSM-III-R criteria among six to 14 year olds.  Can J Psychiatry. 1992;37381- 386PubMedGoogle Scholar
Gould  MSGreenberg  TVelting  DMShaffer  D Youth suicide risk and preventive interventions: a review of the past 10 years.  J Am Acad Child Adolesc Psychiatry. 2003;42386- 405PubMedGoogle ScholarCrossref