Bullying

Abstract

Bullying is a type of aggression: intentional harm directed at a target, carried out repeatedly and marked by an imbalance of power that leaves the victim unable to defend themselves. Those three criteria — intent, repetition, and power imbalance — separate bullying from ordinary conflict between equals, and they organize a literature reaching from the schoolyard to adult psychiatric outcomes decades later. This article develops the construct as developmental and social psychology treat it: the defining criteria and their boundary problems; the recognition that bullying is a group process of participant roles rather than a two-child transaction; its prevalence, correlates, and long-term consequences; and what controlled evaluations show about preventing it. A worked example derives how many children a program must reach to avert one case, and three interactive demonstrations make the ideas concrete.

Keywords: bullying, aggression, peer victimization

Bullying is a subtype of aggressive behavior in which a person is exposed, repeatedly and over time, to negative actions on the part of one or more others, and in which there is an imbalance of power between the parties (Olweus, 1994). The three elements of that definition — an intent to harm, a repetition of the act over time, and an asymmetry of power between aggressor and target — are what make an act bullying rather than a one-off quarrel, and they have remained the field’s standard for three decades. A single shove between equally matched children is a fight; the same shove delivered week after week by a stronger child to one who cannot retaliate is bullying.

What began as the study of a dyad — a bully and a victim — has become the study of a group. Bullying overwhelmingly happens in front of peers, and the responses of those bystanders, whether they reinforce the aggressor, defend the target, or merely watch, help determine whether it persists (Salmivalli, 2010). The sections below follow the construct from this definition, through the participant roles and forms it takes, to the prevalence, consequences, and prevention the evidence now supports.

What Bullying Is

Bullying is aggressive behavior that is intentional, repeated over time, and carried out in a relationship where there is an imbalance of power (Olweus, 1994). The intent criterion requires that the act be meant to harm, which excludes an accidental injury or an insensitive remark that was not aimed at the target. The repetition criterion requires a pattern rather than a single episode, establishing that the victim faces an ongoing threat rather than a one-time event. The power-imbalance criterion requires that the target be unable to defend themselves effectively, whether because the aggressor is physically stronger, more numerous, higher in social status, or in possession of information that can be used against the victim. An act must satisfy all three to count; this is why researchers distinguish bullying sharply from the broader category of peer conflict, in which two children of roughly equal standing disagree (Juvonen & Graham, 2014).

Each criterion carries a boundary problem that measurement must confront (Cook et al., 2010). Repetition is a matter of degree: how many incidents, over what interval, constitute a pattern is a judgment that different surveys resolve differently, and a child who has been targeted once in a way that leaves them in continuing fear may meet the spirit of the criterion without its letter. Power imbalance is often invisible to an outside observer, resting on a social or informational asymmetry rather than a physical one. Table 1 sets out the three defining criteria and the way each demarcates bullying from adjacent behaviors.

Types of Bullying

In the Medical Subject Headings (MeSH) vocabulary, bullying is indexed under aggression and carries one narrower descriptor, giving the single formal subtype below. MeSH is an indexing classification built to retrieve the biomedical literature, not a psychological taxonomy, so its hierarchy marks how the field files its studies rather than a claim that this is the only way to partition the behavior. Researchers more often cut bullying along orthogonal dimensions that cross this one — by mode (physical, verbal, relational), by medium (in-person versus electronic), and by the role a child occupies (bully, victim, or bully-victim) — and a single episode can be classified on all of these at once.

- Cyberbullying — Bullying carried out through digital media, in which the three criteria are relocated to a networked environment whose anonymity, audience, and permanence reshape what each criterion means in practice.

The other distinctions the literature draws are dimensional rather than hierarchical. By mode, bullying is commonly divided into physical (hitting, taking belongings), verbal (name-calling, threats), and relational or indirect forms (exclusion, rumor-spreading) that damage a target’s social standing rather than their body (Juvonen & Graham, 2014). By role, children are classified as bullies, victims, or the especially high-risk bully-victims who both perpetrate and suffer it (Copeland et al., 2013).

Participant Roles and Forms

The most consequential shift in how bullying is understood was the recognition that it is a group phenomenon (Salmivalli, 2010). Bullying episodes are usually witnessed by other children, and those witnesses are not neutral: their behavior feeds back into the episode and helps determine whether it is rewarded or stopped. The participant-role framework assigns children, beyond the bully and the victim, to roles that capture this structure — assistants who join in, reinforcers who provide the audience and encouragement that bullies seek, defenders who support or console the victim, and outsiders who withdraw and do nothing. Because much bullying is motivated by the pursuit of status and dominance within the peer group, the reinforcing audience is part of the reward that maintains it, and the presence or absence of defenders is one of the clearest correlates of whether a child who is targeted continues to be.

This reframing changed what prevention aims at. If bullying were a property of a few aberrant individuals, the remedy would be to identify and treat them; because it is sustained by the responses of the surrounding group, effective programs instead try to shift the behavior of the bystanders — to convert reinforcers and outsiders into defenders and to remove the social rewards of aggression (Menesini & Salmivalli, 2017). The demonstration below lets the reader apply the three definitional criteria to a series of episodes and see which ones qualify as bullying.

The forms bullying takes vary with age and sex. Direct physical bullying is more common among younger children and boys; relational and indirect forms, which exploit the social group rather than the body, become relatively more prominent with age and are reported more among girls (Cook et al., 2010). The participant-role demonstration below shows how the composition of the bystander group shapes the reward a bully receives.

Prevalence and Psychological Impact

Bullying is common. A nationally representative survey of US students in grades 6 through 10 found that roughly 30% reported moderate or frequent involvement — as a bully, a victim, or both — and that involvement in any role was associated with poorer psychosocial adjustment (Nansel et al., 2001). A later meta-analysis that imposed common coding across contexts estimated mean involvement near 35% for traditional bullying, against about 15% for its online form (Modecki et al., 2014). Prevalence figures vary with the definition supplied, the reference period, and the frequency threshold required, so these numbers mark a band rather than a point, but the order of magnitude — somewhere around a third of children touched by bullying in some role — is robust.

The psychological correlates of victimization are among the best-established findings in the field. A meta-analysis of two decades of cross-sectional research found that victimized children were more depressed, more anxious, more lonely, and lower in self-esteem than their peers, with the association to depression the strongest (Hawker & Boulton, 2000). The harm is not only psychological: a meta-analysis found that bullied children carried roughly twice the risk of psychosomatic problems such as headaches, stomachaches, and sleep difficulties (Gini & Pozzoli, 2013). A systematic review and meta-analysis of the consequences of victimization confirmed robust links to internalizing problems, self-harm, and suicidal behavior (Moore et al., 2017). Most striking is the persistence of these effects: a prospective study that followed children into adulthood found that those who had been bullied, and especially the bully-victims who both bullied and were bullied, showed elevated rates of anxiety disorders, depression, and suicidality years after the bullying had stopped (Copeland et al., 2013).

Prevention and Intervention

The first evidence that bullying could be reduced came from Olweus’s large-scale Norwegian intervention, a whole-school program combining school-wide rules, classroom discussion, and consistent adult response, which reported substantial reductions in self-reported bullying (Olweus, 1994). The hopeful early figures prompted a generation of programs and, eventually, the meta-analyses needed to estimate their average effect across many controlled evaluations. Those syntheses tell a consistent and sobering story: school-based anti-bullying programs work, but modestly. An early systematic review found that programs reduced bullying perpetration by about 20% to 23% and victimization by about 17% to 20% on average (Ttofi & Farrington, 2011). An updated meta-analytic review a decade later, drawing on a larger body of rigorous evaluations, found similar average reductions — on the order of 19% to 20% for perpetration and 15% to 16% for victimization (Gaffney et al., 2019).

Two lessons follow. First, prevention is real: across diverse programs and settings the direction of effect is reliably favorable, and the best-evaluated programs, such as the group-focused KiVa program built on the participant-role model, achieve larger reductions than the average (Menesini & Salmivalli, 2017). Second, the average effect is small enough that many children must be reached to prevent one case, which the worked example below quantifies. Alongside program delivery, research has turned to the protective factors that buffer children independently of any formal intervention — among them supportive peer relationships, parental involvement, empathy, and self-control — which a systematic review of meta-analyses found protect against both traditional and cyber forms (Zych et al., 2019).

Worked Example

How many children must a prevention program reach to avert one case of bullying? The answer follows from two verified quantities: a baseline prevalence and a relative reduction. Take the meta-analytic mean prevalence of traditional bullying perpetration, about 35%, as the baseline rate in an untreated school (Modecki et al., 2014), and the lower bound of the meta-analytic effect of anti-bullying programs, a relative reduction of about 20% in perpetration (Ttofi & Farrington, 2011).

A relative reduction of 20% applied to a baseline of 35% gives a post-program prevalence of 0.35 × (1 − 0.20) = 0.35 × 0.80 = 0.28, or 28%. The absolute risk reduction is therefore 0.35 − 0.28 = 0.07, seven percentage points. The number needed to treat — the number of children who must experience the program for one additional child to be spared involvement in bullying — is the reciprocal of the absolute reduction: 1 / 0.07 ≈ 14.3, which rounds up to 15 children per case prevented.

This single figure reconciles the two things the evaluation literature says at once. The program genuinely works — a relative reduction of a fifth is not nothing, and at the scale of a school system it spares many children — yet the number needed to treat of roughly 15 shows why bullying is so stubborn: because the baseline rate is high and the relative effect is modest, the absolute gain per child is small, and a program must be delivered broadly to show a visible dent in prevalence. The InterventionDemo lets the reader vary the baseline rate and the relative reduction and watch the absolute risk reduction and the number needed to treat update, making concrete how a modest percentage effect translates into the reach a program requires.

Discussion

The study of bullying has matured from a definition into a developmental science. Olweus’s contribution was to fix the construct — intent, repetition, power imbalance — precisely enough that it could be measured and that interventions could be tested against it (Olweus, 1994). The decisive conceptual advance that followed was to relocate the phenomenon from the individual to the group: once bullying was seen as a pursuit of status enacted before, and sustained by, a peer audience, both its persistence and the leverage points for changing it came into focus (Salmivalli, 2010). The epidemiological and longitudinal work then established the stakes — that involvement is common, that its correlates are serious, and that they persist into adulthood (Copeland et al., 2013; Moore et al., 2017).

The open problems are causal and practical. Much of the evidence linking victimization to poor outcomes is correlational, and disentangling the harm bullying causes from the pre-existing vulnerabilities that raise the risk of being bullied requires longitudinal and genetically informed designs that the field is only beginning to deploy at scale. On the practical side, the gap between the whole-school programs that work in careful trials and the diluted versions implemented at scale remains wide, and the modest average effect sizes mean that improving program fidelity and targeting is where the next gains lie. What is not in doubt is that bullying is a tractable public-health problem rather than an immutable feature of childhood: it can be measured, its consequences are grave, and it can be reduced.

Current Directions

Recent work has concentrated on sharpening both the estimate of what prevention achieves and the understanding of why some children are protected. The updated meta-analytic review of prevention programs not only confirmed the modest average effect but began to identify the program components associated with larger reductions — whole-school approaches, parent involvement, and sustained delivery — which points implementation toward what actually carries the effect (Gaffney et al., 2019). This moves the field from asking whether programs work to asking which of their parts do.

In parallel, the protective-factors literature has been synthesized to ask what reduces involvement independently of any formal program. A systematic review of meta-analyses identified the factors most consistently associated with lower bullying and victimization — good peer relationships, parental involvement, empathy, and self-control — and found that the same factors protect against both the traditional and the online forms, evidence that the two share underlying mechanisms (Zych et al., 2019). Together with the consolidation of the consequences literature into firm meta-analytic estimates (Moore et al., 2017), these lines are shifting the field from documenting that bullying harms toward specifying what prevents it and for whom.

Common Misconceptions

Any aggressive act between children is bullying.

Bullying requires all three criteria together — intent, repetition, and a power imbalance. A single fight between equally matched children is a conflict, not bullying, because it lacks both repetition and the asymmetry of power (Olweus, 1994).

Bullying is a problem between two children, the bully and the victim.

It is a group process. Episodes are typically witnessed by peers whose responses — reinforcing the bully or defending the target — help determine whether the bullying continues, which is why effective prevention targets the bystanders, not only the bully (Salmivalli, 2010).

Children grow out of bullying with no lasting harm.

Prospective research finds that childhood victimization, and especially the bully-victim pattern, predicts elevated anxiety, depression, and suicidality that persist well into adulthood (Copeland et al., 2013).

Glossary

Absolute risk reduction.

The difference in the rate of an outcome between an untreated and a treated group; for a 35% baseline cut by a fifth, 0.35 − 0.28 = 0.07, or seven percentage points.

Aggression.

Behavior directed at another with the intent to cause harm; bullying is the repeated, power-asymmetric subtype of aggression.

Bully-victim.

A child who both perpetrates bullying and is victimized by it; the role carries the highest risk of adverse adult psychiatric outcomes.

Bullying.

Aggressive behavior that is intentional, repeated over time, and marked by an imbalance of power between aggressor and target.

Cyberbullying.

Bullying carried out through electronic devices and digital media, the one formal MeSH subtype of bullying.

Defender.

A participant role in which a bystander supports or consoles the victim or intervenes against the bully; the presence of defenders predicts whether victimization persists.

Intent to harm.

The criterion that an act be meant to injure the target, distinguishing bullying from an accidental slight or an insensitive joke.

Number needed to treat.

The reciprocal of the absolute risk reduction; the number of children who must receive a program for one additional case to be prevented — about 15 for the worked example.

Participant role.

The position a child occupies in a bullying episode — bully, victim, assistant, reinforcer, defender, or outsider — in the group-process account of bullying.

Peer victimization.

The experience of being the repeated target of peers’ aggression, associated with depression, anxiety, loneliness, and psychosomatic complaints.

Power imbalance.

The asymmetry between aggressor and target that defines bullying; it may rest on physical strength, numbers, social status, or information.

Reinforcer.

A participant role in which a bystander provides the attention, laughter, or encouragement that rewards the bully and helps sustain the episode.

Relational bullying.

Indirect bullying that harms a target’s social standing through exclusion or rumor-spreading rather than physical or verbal attack.

Repetition.

The criterion that bullying recur over time, establishing an ongoing pattern rather than an isolated incident.

Whole-school program.

A prevention approach that combines school-wide rules, classroom work, and consistent adult response rather than targeting individual bullies; the design of the original Olweus program.

Key Researchers

David P. Farrington

(1944–2024). Criminologist at the University of Cambridge whose systematic and meta-analytic evaluations established the average effect of school-based prevention programs and situated bullying within the developmental trajectory of antisocial behavior. Google Scholar - University of Cambridge - Wikipedia - Wikidata

Sandra Graham

(UCLA). Educational and developmental psychologist whose attributional and ethnic-context research reframed who is victimized and why, showing how classroom diversity shapes the experience and attributions of victims. UCLA - Wikidata

Jaana Juvonen

(UCLA). Developmental psychologist whose work on the social cognition of victimization and the ethnic-diversity hypothesis clarified who is victimized and why, and the social and academic costs borne by victims. ORCID - Google Scholar - UCLA

Dan Olweus

(1931–2020). Founder of modern bullying research at the University of Bergen, whose intent-repetition-power-imbalance definition became the field’s standard and whose large-scale Norwegian intervention first demonstrated that whole-school programs can reduce bullying. Wikipedia - Wikidata

Christina Salmivalli

(University of Turku). Developmental psychologist whose participant-role theory recast bullying as a group process of assistants, reinforcers, defenders, and outsiders, and who developed and evaluated the widely adopted KiVa anti-bullying program. University of Turku - Wikipedia - Wikidata

Dieter Wolke

(University of Warwick). Developmental psychopathologist whose longitudinal studies traced the long-term mental-health and psychosomatic consequences of childhood bullying into adulthood and established the elevated risk carried by bully-victims. University of Warwick - Google Scholar - Wikipedia - Wikidata

Frequently Asked Questions

What is bullying?

Bullying is aggressive behavior that is intentional, repeated over time, and carried out in a relationship marked by an imbalance of power, so that the target cannot easily defend themselves (Olweus, 1994).

How is bullying different from an ordinary fight or conflict?

A conflict is a disagreement between children of roughly equal power, and a single fight is a one-time event. Bullying requires all three of intent, repetition, and a power imbalance, so a fair fight between equals is not bullying (Juvonen & Graham, 2014).

What are the main types of bullying?

Bullying is commonly divided by mode into physical, verbal, and relational (indirect) forms, and by medium into in-person bullying and cyberbullying, its one formal MeSH subtype, conducted through digital media (Modecki et al., 2014).

Why is bullying described as a group process?

Episodes are usually witnessed by peers who take on roles: assistants and reinforcers who reward the bully, defenders who support the victim, and outsiders who withdraw. These bystander responses help determine whether the bullying continues (Salmivalli, 2010).

How common is bullying?

Roughly a third of school-age children are involved in some role; a US national survey found about 30% reporting moderate or frequent involvement, and a cross-context meta-analysis estimated mean involvement near 35% for traditional bullying (Nansel et al., 2001; Modecki et al., 2014).

What are the psychological effects of being bullied?

Victimization is associated with depression, anxiety, loneliness, lowered self-esteem, and roughly double the risk of psychosomatic complaints, and these effects can persist into adulthood as elevated anxiety, depression, and suicidality (Hawker & Boulton, 2000; Gini & Pozzoli, 2013; Copeland et al., 2013).

Can bullying be prevented?

Yes, but modestly: meta-analyses of controlled evaluations find that school-based programs reduce bullying perpetration by about a fifth and victimization somewhat less, with whole-school and group-focused programs achieving the larger reductions (Ttofi & Farrington, 2011; Gaffney et al., 2019).

What protects children from bullying?

A systematic review of meta-analyses points to supportive peer relationships, parental involvement, empathy, and self-control, factors that protect against both the traditional and the cyber forms (Zych et al., 2019).

References