Cyberbullying

Abstract

Cyberbullying is aggression carried out through digital media — phones, messaging, social platforms, and the web — in which an individual or group repeatedly and deliberately harms a target who cannot easily defend themselves. MeSH classifies it as a form of bullying, and the field’s central question is whether it is a genuinely new phenomenon or the old one moved online. This article develops the construct as developmental and social psychology treat it: the three criteria inherited from traditional bullying — intent, repetition, and power imbalance — and how each strains in a networked environment; the distinctive features of that environment, including anonymity, a vast audience, and permanence; and the prevalence, role overlap, and psychological correlates the evidence establishes. A worked example derives the shared variance between cyber and traditional bullying, and three interactive demonstrations make the ideas concrete.

Keywords: cyberbullying, bullying, online aggression

Cyberbullying is willful and repeated harm inflicted through computers, phones, and other electronic devices (Hinduja & Patchin, 2010). It is treated as a form of bullying, inheriting that concept's defining criteria — an intent to harm, a repetition of the act, and an imbalance of power between aggressor and target — while relocating them to an environment whose properties are not those of the schoolyard. A message can be sent anonymously, forwarded to hundreds in seconds, and remain visible indefinitely, so the online setting does not merely host the old behavior but reshapes what each criterion means in practice (Smith et al., 2008).

That reshaping is the organizing problem of the field. Early work established that cyberbullying was real, measurable, and associated with the same kinds of distress as traditional bullying (Ybarra & Mitchell, 2004). The debate that followed asked how far the digital medium changes the phenomenon — whether anonymity and audience make it a distinct construct requiring its own theory, or whether it is largely continuous with offline bullying and best understood as one more context in which it occurs (Olweus, 2012; Modecki et al., 2014). The sections below follow that question from the definition, through the features of the online environment, to the prevalence and correlates the evidence now supports.

What Cyberbullying Is

Cyberbullying is the use of electronic communication to bully a person, typically by sending messages of an intimidating, threatening, or demeaning nature (Smith et al., 2008). MeSH classifies it as a form of bullying, and the definition that dominates the field is built by taking the established definition of bullying and applying it to the digital medium: an aggressive act, carried out repeatedly and over time, by an individual or group against a victim who cannot easily defend themselves (Olweus, 2012). The three elements of that older definition — intent, repetition, and a power imbalance — are what make an act bullying rather than an ordinary conflict between equals, and they carry over to the cyber case as the criteria an instance must meet.

The difficulty, and the reason cyberbullying has generated its own literature, is that each criterion behaves differently online (Englander et al., 2017). Repetition need not require the aggressor to act more than once: a single humiliating image, once posted, can be viewed, forwarded, and commented on by an indefinite audience, so the harm repeats even though the perpetrator struck only once. Power imbalance need not rest on physical strength or social standing: anonymity, technical skill, or simply the capacity to reach a target at any hour can confer the asymmetry that defines bullying. Concept analyses of the term have found that authors weight these attributes differently, which is one source of the wide variation in how cyberbullying is defined and measured (Peter & Petermann, 2018). Table 1 sets out the distinctive features of the online environment that drive this reshaping.

Defining Attributes and the Boundary Problem

Whether a given online act counts as cyberbullying depends on which of the three inherited criteria are required and how each is interpreted, and this is not a pedantic matter — it determines what prevalence studies count and whether the construct is coherent (Peter & Petermann, 2018). The least contested criterion is intent: the act must be meant to harm, which distinguishes bullying from an insensitive joke or an accidental slight. The two harder criteria are repetition and power imbalance, and the online environment destabilizes both.

Repetition traditionally means the aggressor acts against the target more than once, establishing a pattern rather than a single incident. Online, a single act can produce repeated harm without any further action by the perpetrator, because others reshare the material and the target re-encounters it (Slonje & Smith, 2008). This has led some authors to argue that repetition should be defined by the victim's repeated exposure rather than the aggressor's repeated action. Power imbalance is similarly transformed: offline it usually rests on physical strength, numbers, or social status, but online it can derive from anonymity — the target cannot retaliate against an unknown aggressor — or from technical proficiency, or from the sheer reach of a post (Englander et al., 2017). These reinterpretations are what Olweus had in mind when he argued that much of what is labeled cyberbullying is either traditional bullying carried into a new medium or isolated incidents that do not meet the repetition-and-imbalance threshold at all, so that the genuinely novel residue is smaller than alarmed reports suggest (Olweus, 2012).

The Distinctive Online Environment

Cyberbullying takes several forms distinguished by the medium and the act: text messages, pictures or video clips, phone calls, email, chat rooms, instant messaging, and postings on websites or social platforms (Smith et al., 2008). Early typologies found that these media differ in their impact — a humiliating picture or video was rated by young people as more harmful than a text message, because the visual record is more concrete and reaches a wider audience (Slonje & Smith, 2008). What unites the forms, and what distinguishes them from a face-to-face taunt, are the features set out in Table 1: anonymity, the breadth and permanence of the audience, and the capacity to reach a target at any hour. Several of these features converge on a single psychological mechanism, the online disinhibition effect: the reduced accountability of anonymity and the absence of the target’s visible reaction loosen the ordinary restraints on conduct, so that people say and do things online they would not say or do face to face (Suler, 2004).

A critical synthesis of the early victimization research argued that these features do not act uniformly (Tokunaga, 2010). Anonymity, for instance, is often invoked as the signature feature of cyberbullying, yet a substantial share of victims know or suspect who is attacking them, so anonymity is better treated as a variable property than as a defining one. The accessibility of digital media — the fact that a phone follows the target home — is more consistently distinctive, because it removes the physical escape that leaving the setting once afforded, and it is part of why cyber victimization is associated with a pervasive sense that there is no safe place. The synthesis also exposed how inconsistently the phenomenon had been measured, with studies varying in the reference period they asked about, the definition they supplied to respondents, and the threshold of frequency they required, which makes prevalence estimates difficult to compare.

Prevalence, Overlap, and Psychological Impact

Because definitions and measurement vary so widely, prevalence estimates for cyberbullying range enormously across studies, which is itself a finding about the construct rather than only about the surveys (Tokunaga, 2010). A meta-analysis that imposed common coding across studies found mean involvement rates of roughly 15% for cyberbullying against roughly 35% for traditional bullying, so the newer form is real and common but markedly less prevalent than the older one (Modecki et al., 2014). The same meta-analysis reported a substantial correlation between involvement in the two — about .47 for perpetration — which is the central quantitative fact of the field: cyber and traditional bullying overlap heavily, so most children involved in one are involved in the other, yet the overlap is far from complete and each retains a distinct component.

The psychological correlates of cyberbullying involvement are well established even where their causal direction is not. A comprehensive meta-analysis linked cyber victimization to depression, anxiety, loneliness, lowered self-esteem, and reduced academic achievement, and cyber perpetration to externalizing problems and substance use, with effect sizes broadly comparable to those for traditional bullying (Kowalski et al., 2014). The most serious correlate is suicidal ideation: youth who experienced cyberbullying, as victims or perpetrators, reported higher levels of suicidal thoughts and attempts than those who had not, consistent with the broader association between peer victimization and self-harm (Hinduja & Patchin, 2010). An important qualification runs through this literature. Because cyber and traditional victimization co-occur, much of the distress associated with cyberbullying may reflect the traditional bullying that accompanies it, and the early comparison of online aggressors and targets already showed that online involvement is embedded in a wider pattern of offline problem behavior and victimization rather than standing apart from it (Ybarra & Mitchell, 2004).

Worked Example

The overlap between cyber and traditional bullying can be quantified from the meta-analytic correlation. Modecki and colleagues reported a mean correlation of r = .47 between involvement in traditional and cyber bullying perpetration (Modecki et al., 2014). The proportion of variance the two share is the coefficient of determination, r², so r² = .47² = .2209, or about 22%. Roughly a fifth of the variance in cyber perpetration is accounted for by its association with traditional perpetration, and the remaining 1 − .2209 = .7791, about 78%, is not.

This single number adjudicates the field's central dispute more cleanly than either slogan does. A correlation of .47 is substantial — it is why the two behaviors so reliably co-occur, and why a child identified as a cyberbully is a good bet to be a traditional bully too — which supports the view that cyberbullying is continuous with the older phenomenon rather than wholly new. But 22% shared variance is far from the near-unity a pure relabeling would predict: 78% of the variation in cyber perpetration lies outside the overlap, which is the quantitative warrant for treating cyberbullying as having a distinct component that merits its own study. The victimization correlation is lower still, about .40, giving r² = .16, so only 16% shared variance — the distinct component is larger on the victim side. The PrevalenceDemo lets the reader vary the assumed correlation and the two prevalence rates and watch the shared variance and the implied overlap of the involved groups update, making concrete how a single coefficient separates the “same thing online” reading from the “genuinely new problem” reading.

Discussion

The study of cyberbullying has, in under two decades, moved from establishing that the phenomenon exists to specifying precisely how it relates to the bullying that preceded it. The foundational contribution was definitional and empirical at once: Smith and colleagues, and Ybarra and Mitchell before them, showed that electronic aggression was measurable, met the criteria of bullying when properly assessed, and carried comparable distress (Ybarra & Mitchell, 2004; Smith et al., 2008). The sceptical correction, that much of the apparent novelty dissolves when the repetition and power-imbalance criteria are applied strictly, forced the field to be precise about what it was counting (Olweus, 2012). The meta-analytic synthesis then supplied the resolution: cyber and traditional bullying are strongly correlated but not identical, so neither the “entirely new problem” nor the “nothing new” position survives the evidence (Kowalski et al., 2014; Modecki et al., 2014).

The open problems concern definition and causation. Concept analyses continue to find that the field has not converged on how to operationalize repetition and power imbalance online, and until it does, prevalence figures will remain hard to compare and the boundaries of the construct will stay contested (Peter & Petermann, 2018). The causal question is harder still: because cyber and traditional victimization co-occur so often, isolating the specific contribution of the online form to outcomes such as depression or suicidal ideation requires designs that can separate the two, and much of the existing evidence cannot. What is not in doubt is that involvement in cyberbullying marks a child at elevated risk, whether or not the online medium is the proximate cause of that risk.

Current Directions

Recent work has shifted from documenting the phenomenon toward understanding its development and what protects against it. A developmental synthesis traced how prevalence and form change with age — rising through early adolescence as online access and social stakes increase — and emphasized protective factors such as parental monitoring, school climate, and social-emotional competence that can be targeted by intervention (Kowalski et al., 2019). This reframes cyberbullying as a developmental problem embedded in the ordinary course of adolescence rather than a free-standing technological threat.

The protective-factors question has itself been subjected to meta-analytic review. A systematic review of meta-analyses identified the factors most consistently associated with reduced bullying and cyberbullying involvement — among them good peer relationships, parental involvement, empathy, and self-control — and noted that the same factors tend to protect against both the traditional and the cyber forms, which is further evidence for their continuity (Zych et al., 2019). A parallel strand of conceptual work continues to press for a defensible definition, arguing that progress on prevention depends on first fixing what the attributes of cyberbullying are and how they should be measured (Peter & Petermann, 2018). Together these lines move the field from describing cyberbullying toward predicting and preventing it.

Common Misconceptions

Cyberbullying is an entirely new problem unrelated to ordinary bullying.

The two overlap strongly: a meta-analytic correlation of about .47 for perpetration means most children involved in one are involved in the other, so cyberbullying is largely continuous with traditional bullying rather than a separate phenomenon (Modecki et al., 2014).

A single hurtful message counts as cyberbullying.

Bullying requires repetition and a power imbalance, not just intent to harm; a one-off insult between equals is a conflict, not bullying — though online a single post can produce repeated harm if others reshare it and the victim re-encounters it (Olweus, 2012).

Cyberbullies are always anonymous strangers.

A substantial share of victims know or suspect who is attacking them; anonymity is a variable feature of cyberbullying that intensifies harm when present, not a defining property of every case (Tokunaga, 2010).

Glossary

Accessibility.

The property of digital media by which a target can be reached at any time and in any place, including the home, removing the respite offline settings once provided.

Aggression.

Behavior directed at another with the intent to cause harm; cyberbullying is a repeated, power-asymmetric form of aggression conducted through digital media.

Anonymity.

The condition in which an aggressor's identity is concealed from the target, which can embolden perpetration and heighten a victim's distress.

Bullying.

Aggressive behavior that is intentional, repeated over time, and marked by an imbalance of power between aggressor and target; cyberbullying is the form conducted online.

Coefficient of determination.

The squared correlation, r², giving the proportion of variance that two measures share; .47² ≈ 22% for cyber and traditional bullying perpetration.

Cyber victimization.

The experience of being the target of cyberbullying, associated with depression, anxiety, lowered self-esteem, and elevated suicidal ideation.

Cyberbullying.

Willful and repeated harm inflicted on a target through electronic devices and digital media, inheriting the intent, repetition, and power-imbalance criteria of bullying.

Intent to harm.

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

Online disinhibition effect.

The tendency to behave with fewer restraints online than face to face, driven by anonymity and the absence of the target’s visible reaction, which helps explain how ordinary people become perpetrators.

Permanence.

The persistence of digital content, which can be retrieved and reshared long after the original act, so that harm repeats without further action by the perpetrator.

Power imbalance.

The asymmetry between aggressor and target that defines bullying; online it may derive from anonymity, technical skill, or audience reach rather than physical strength.

Prevalence.

The proportion of a population involved in cyberbullying; estimates vary widely with definition and measurement, with meta-analytic means near 15% against 35% for traditional bullying.

Protective factor.

A condition associated with reduced involvement in bullying, such as parental monitoring, positive peer relationships, empathy, or self-control.

Repetition.

The criterion that bullying occur more than once; online it may be reinterpreted as repeated exposure of the victim rather than repeated action by the aggressor.

Role overlap.

The co-occurrence of involvement in cyber and traditional bullying, quantified by the correlation between the two and the shared variance it implies.

Suicidal ideation.

Thoughts of or plans for self-harm; found at elevated levels among youth involved in cyberbullying as either victims or perpetrators.

Key Researchers

Sameer Hinduja

(Florida Atlantic University). Co-director of the Cyberbullying Research Center whose large-scale survey work established the association between cyberbullying involvement and suicidal ideation and whose prevention frameworks shape school practice. ORCID - Wikipedia - Wikidata

Robin M. Kowalski

(Clemson University). Social psychologist whose meta-analysis of the cyberbullying literature synthesized its correlates and its overlap with traditional bullying, and whose developmental work identified protective factors against online aggression. ORCID - Google Scholar

Dan Olweus

(1931–2020). Founder of modern bullying research at the University of Bergen, whose repetition-and-power-imbalance definition of bullying is the field's standard and whose sceptical reappraisal questioned whether cyberbullying is a distinct phenomenon. ORCID - Wikipedia - Wikidata

Justin W. Patchin

(University of Wisconsin-Eau Claire). Co-director of the Cyberbullying Research Center whose research on adolescent prevalence and consequences, including the link between online victimization and suicidal ideation, anchors the empirical literature. Google Scholar - University of Wisconsin-Eau Claire

Peter K. Smith

(1943–2025). Developmental psychologist at Goldsmiths, University of London, whose early European studies defined cyberbullying's nature and impact in schools and established the typology distinguishing its forms from traditional bullying. Wikipedia

Michele L. Ybarra

(Center for Innovative Public Health Research). Public-health researcher whose early population-based study of online aggressors and targets established that online involvement is embedded in a wider pattern of offline victimization and problem behavior. Google Scholar - Johns Hopkins

Frequently Asked Questions

What is cyberbullying?

Cyberbullying is willful and repeated harm inflicted on a target through electronic devices and digital media, such as phones, messaging apps, social platforms, and websites, by an individual or group against someone who cannot easily defend themselves (Hinduja & Patchin, 2010).

How is cyberbullying different from traditional bullying?

It shares the same three criteria (intent, repetition, and power imbalance), but the online environment reshapes them: a single post can repeat its harm as others reshare it, power can rest on anonymity or technical skill, and the target can be reached at any hour, including at home (Smith et al., 2008).

Is cyberbullying just bullying moved online?

Partly. Cyber and traditional bullying correlate strongly, about .47 for perpetration, or roughly 22% shared variance, so they overlap heavily, but most of the variation in each is distinct, so cyberbullying is neither wholly new nor a mere relabeling (Modecki et al., 2014).

How common is cyberbullying?

Estimates vary widely with how it is defined and measured, but a meta-analysis imposing common coding found mean involvement near 15%, against about 35% for traditional bullying (Modecki et al., 2014).

Does cyberbullying require repeated acts?

Traditionally yes, but online a single act can cause repeated harm when others reshare the material and the victim re-encounters it, which has led some researchers to define repetition by the victim's repeated exposure rather than the aggressor's repeated action (Slonje & Smith, 2008).

Are cyberbullies usually anonymous?

Not always. A substantial share of victims know or suspect who is attacking them, so anonymity is a variable feature that intensifies harm when present rather than a defining property of every case (Tokunaga, 2010).

What are the psychological effects of cyberbullying?

Cyber victimization is associated with depression, anxiety, loneliness, lowered self-esteem, reduced academic achievement, and elevated suicidal ideation, with effects broadly comparable to those of traditional bullying (Kowalski et al., 2014).

What protects children from cyberbullying?

Meta-analytic review points to good peer relationships, parental involvement, empathy, and self-control, factors that tend to protect against both the traditional and the cyber forms (Zych et al., 2019).

References