Available online at www.sciencedirect.
com
ScienceDirect
Some problems with cyberbullying research
Dan Olweus1 and Susan P Limber2
Research on cyberbullying is plagued by inconsistent findings
and exaggerated claims about prevalence, development over
time, and effects. To build a useful and coherent body of
knowledge, it essential to achieve some degree of consensus
on the definition of the phenomenon as a scientific concept and
that efforts to measure cyberbullying are made in a ‘bullying
context.’ This will help to ensure that findings on cyberbullying
are not confounded with findings on general cyberaggression
or cyberharassment. We tentatively recommend that
cyberbullying should be regarded as a subcategory or specific
form of bullying, in line with other forms such as verbal,
physical, and indirect/relational.
Addresses
1
Psykologkonsult Dan Olweus, Vognstolbakken 16, NO-5096 Bergen,
Norway
2
Department of Youth, Family & Community Studies, 2038 Barre Hall,
Clemson University, Clemson, SC 29634, USA
Corresponding author: Olweus, Dan (olweus@uni.no)
Current Opinion in Psychology 2018, 19:139–143
This review comes from a themed issue on Aggression and violence
Edited by Brad J Bushman
For a complete overview see the Issue and the Editorial
Available online 19th April 2017
http://dx.doi.org/10.1016/j.copsyc.2017.04.012
2352-250X/ã 2017 Published by Elsevier Ltd.
“In the abstract, it need hardly be said that before one proceeds to
explain or to interpret a phenomenon, it is advisable to establish
that the phenomenon actually exists, that it is enough of regu-
larity to require and allow explanation.”
—Robert K. Merton
As suggested in the well-known citation of Merton [1], to
understand and change a phenomenon, it is very impor-
tant that the phenomenon is well identified. A first step in
identifying a new concept or construct is to provide a
preliminary definition of the phenomenon to roughly
indicate the concept’s domain and its boundaries—a kind
of concept mapping. This step concerns the content
validity of the concept [2]. But to make it a useful
scientific concept, its construct validity, including analy-
ses of convergent and discriminant validity, must also be
gradually established [3,4].
www.sciencedirect.com
In the present article, we take a closer look at some
aspects of the concepts of bullying and cyberbullying.
On the basis of a selection of research findings, we focus
in particular on the following issue: Can cyberbullying be
conceptualized as a subcategory or form of traditional
bullying, or should it be best regarded as a distinct
phenomenon with special characteristics that make it
partly different from traditional bullying? [5,6]. In doing
so, bullying defined as a scientific concept – based on but
not identical to the everyday use of the term – will serve
as a starting point for our examination.
In scanning the vast numbers of research publications
about cyberbullying in the past 5–10 years, we have been
struck and concerned by the many disparate and partly
conflicting findings reported. Such heterogeneity sug-
gests that researchers have used different definitions
and operationalizations of the concept. It also indicates
considerable lack of replicability, which will likely create
problems of understanding, intervention, and prevention.
Definitions of bullying and cyberbullying
A common definition of (traditional) bullying is the fol-
lowing: “A student is being bullied or victimized when he
or she is exposed, repeatedly and over time, to negative
actions on the part of one or more other students” [7]. In
order to use the term bullying, there should also be an
imbalance of power, an asymmetric power relationship
[8]. There are three components to this definition: (1) It
concerns purposeful unwanted negative (aggressive)
behavior that (2) typically implies a pattern of behavior
that is repeated, and (3) occurs in an interpersonal rela-
tionship characterized by an imbalance of power or
strength, favoring the perpetrator(s). This definition
makes it clear that bullying often may be considered a
form of peer abuse.
Although the specified criteria of this definition have
sometimes been questioned and even rejected [9,10],
they have been widely used in the research literature
and seem by and large to be well accepted by the research
community [11,12]. This is our basic point of departure
in this article. And when the U.S. Centers for Disease
Control was given the task of developing a ‘universal
definition of bullying’, they landed on a definition that is
basically the same as the one reported above [13].
However, with the advent of cyberbullying, that is, bul-
lying via electronic forms of contact or communication –
such as emails, mobile, chat room, instant messaging,
websites – concerns have been raised about whether
and possibly how both the repetitiveness and the power
Current Opinion in Psychology 2018, 19:139–143
140 Aggression and violence
imbalance criteria in the general definition can be applied
to bullying with electronic means [4]. We generally agree
with the tentative conclusion reached by Smith et al. [12]
that the key criteria defining traditional bullying are
largely applicable to cyberbullying as well. They suggest,
for example, that the imbalance of power can be assessed
“in terms of differences in technological know-how
between perpetrator and victim, relative anonymity,
social status, number of friends, or marginalized group
position” [12, p. 36]. Moreover, the criterion of repeti-
tion may have to be understood in a somewhat different
way with a focus on how many individuals can be reached
with a negative message or image, or the length of time
that a message or image can remain in cyber space, rather
than on the perpetrator’s cyber behavior which is often a
single act [4,12].
Heterogeneity of prevalence estimates
Although there is a considerable degree of consensus, in
principle at least, about how cyberbullying should be
defined, it is obvious that empirical studies published
and often meta-analyzed as cyberbullying studies have
used very different ways of measuring the phenomenon
[14,15]. One consequence is a bewildering array of preva-
lence estimates of cyberbullying, varying between 3–4%
and 40%, with some studies producing estimates at 50%
or even beyond [16,17].
As detailed in a recent paper [18], a good deal of this
heterogeneity is clearly a function of different lengths of
the reference or recall period used. Other differences
Figure 1
Percent
Verbal
22
20 18,1 18,4
18 17,2
15,4
16
14
12
10
8
6 4,1
4
2
0
2007 2008 2009 2010
Time series data for 2007–2010 for verbal bullying (being bullied) and cyber bullying (bullied electronically). Data from all over the USA. Total
n = 447 000.
Current Opinion in Psychology 2018, 19:139–143
come from use of different cut-off points or threshold
values for classifying a respondent as being cyberbullied.
A likely even more important reason for this heterogene-
ity is that cyberbullying has been studied ‘in isolation,’
that is, outside the general context of (traditional) bully-
ing. To put cyberbullying in proper perspective, it is in
our view necessary to study it in the context of (tradi-
tional) bullying more generally. One cannot talk about a
phenomenon as bullying unless a reasonably precise
definition has been provided to the respondents or the
formulation of the questions or other measures used make
it quite clear that the contents conform to what is implied
in (the scientific) concept of bullying [19]. It is, of course,
important not to use cyberbullying/victimization as a
blanket term for any form of negative or aggressive act
[4,20,21,22].
Some empirical prevalence estimates
In a large-scale study of a total of 440 000 U.S. students in
grades 3–12, we compared the prevalence rates of cyber-
bullying measured in the context of traditional bullying
with the prevalence rates for traditional verbal bullying
(the most frequent form of traditional bullying). The
participants belonged to four different cohorts providing
time series data for four different years, from 2007 to 2010,
as shown in Figure 1. The average across-time prevalence
for being verbally bullied ‘2 or 3 times a month or more
often’ was 17.3%, whereas the corresponding figure for
being cyberbullied was 4.5%. A very similar pattern of
results, but at a lower level, was obtained in a study of
Cyber
4,5 5,0
4,3
2007 2008 2009 2010
Current Opinion in Psychology
www.sciencedirect.com
 Some problems with cyberbullying research Olweus and Limber 141
9000 Norwegian students followed over five years from
2010 to 2010 [6,23].
Based on these two large-scale studies in different coun-
tries (and with different designs), results indicate that
cyberbullying measured in the context of traditional
bullying is actually a quite low-prevalence phenomenon,
representing only some 25–35% of the level of traditional
bullying by direct verbal means. There is no doubt that
there are many more children and youth involved in
traditional (verbal) bullying than in cyberbullying. And,
in contrast to a common belief often expressed by both
researchers and the media, basically no systematic
change in prevalence occurred over the time periods
studied, from 2006 to 2010 [6,23]. Norwegian follow-up
data for 2013 and 2014 have continued to indicate no
increase.
Degree of overlap with traditional bullying
Another important issue that has received a good deal of
research attention concerns the degree to which cyber-
bullying overlaps with traditional bullying. In the two
studies mentioned [6,23], the overlap was very high. Of
students who had been exposed to cyberbullying, almost
90% had been bullied in at least one traditional way. This
means that only about 10% of the students involved in
cyberbullying had only been cyberbullied. In other stud-
ies, the degree of overlap has varied from about 50% [24]
to 67% [25], 75% [26, Table 2, p. 380], and even 90%
[27, p. 64]. Although opinions about the degree of over-
lap thus may vary somewhat due to differences with
regard to conceptualization, measurement instrument
and context, authors generally agree that a large propor-
tion of cyberbullied students are also bullied in traditional
ways.
Possible negative effects of cyberbullying?
Although both media and researchers have reported that
there are many serious negative effects of cyberbullying,
it is difficult to know to what extent such effects actually
are a consequence of cyberbullying. This is because a
great majority of cyberbullied children and youth are also
bullied in traditional ways, as noted above. And it is a
well-established fact that there are serious both short-
term and long-term effects of being exposed to traditional
bullying [28,29].
One way to find out more about this important issue
might be to use a regression framework to explore
whether a common outcome variable such as depression
can be predicted from being cyberbullied over and above
being bullied in traditional ways. If being cyberbullied is
found to have a significant incremental effect, this is often
interpreted as evidence that cyberbullying is a distinct
phenomenon with special characteristics that make it
different from traditional forms of bullying. For other
approaches, see Refs. [6,18,23].
www.sciencedirect.com
Three cross-sectional studies with depression as the
outcome variable have reported significant additive
effects of being cyberbullied over and above the effects
of being traditionally bullied [30–32]. At the same time, a
few other cross-sectional studies [33] have produced
mixed or non-significant findings.
More important, a recent longitudinal study [33] with
control for traditional bullying reported a significant
increase in predicted variance 12 months after time
1. The amount of variance added by cybervictimization
was quite small, however, only 1%. And two other lon-
gitudinal studies [34,35] found no significant additional
effects.
Should such a collection of partly discrepant results be
seen as an indication that being cyberbullied is distinct
phenomenon? If one takes as a point of departure the
theoretical position that being cyberbullied is just a form
of being bullied, on a par with other forms of being
bullied, it becomes natural to check if being cyberbullied
is more predictive of a relevant outcome variable than
other forms of being bullied. An empirical check of this
possibility was undertaken, using the outcome variable of
low self-esteem (which correlates strongly with depres-
sion) from a US study [6]. The global variable of being
traditionally bullied predicted 8.3% of the variance in low
self-esteem. Adding the being cyberbullied variable
increased the predicted variance to 11.0%, representing
a 2.7% increase. This result was quite comparable to
adding an item on indirect bullying, such as “being left
out of things on purpose . . . ” [36], to the global variable
which increased the predicted variance to 11.7%, repre-
senting a 3.4% increase. In a similar analysis using the
scale of seven items of different forms of being tradition-
ally bullied (with exclusion of the ‘being left out’ vari-
able), the amount of added variance for the being left out
variable was somewhat smaller but still larger than the
added variance predicted by the being cyberbullied vari-
able. Very likely, most items representing different forms
of being bullied contain some portion of specific variance
that may correlate with a suitable outcome variable over
and above the prediction by a global variable of being
traditionally bullied or a sum scale of the bullied items.
Obviously, finding that being cyberbullied predicts sig-
nificant incremental variance in a suitable outcome vari-
able cannot, without further analyses, be taken as an
indication that cyberbullying should best be regarded
as a distinct phenomenon.
A tentative conclusion
In conclusion, most of the empirical facts and delibera-
tions about cyberbullying reported above are consistent
with the view of cyberbullying as a form of bullying, in
line with other forms such as verbal, physical, and indi-
rect/relational bullying. Because some of the reported
facts are based a limited amount of research, such a
Current Opinion in Psychology 2018, 19:139–143
142 Aggression and violence
conclusion should be regarded as tentative. The emerg-
ing field of cyberbullying is in a relatively early phase, and
there is a clear need for more conceptual, methodological,
and empirical research on many of the issues discussed in
this article [37]. However, to build a useful and coherent
body of knowledge, it essential that future research
efforts measure the phenomenon of cyberbullying in a
‘bullying context.’ This is to ensure that findings on
cyberbullying are not confounded with findings on gen-
eral cyberaggression or cyberharassment where the per-
petrator(s) and the targeted youth do not belong to the
same classroom, school, or other common social unit, and
the youth exposed may have no idea of who the perpe-
trator is.
We want to emphasize that our intention in writing this
article has not been to downplay or trivialize cyberbully-
ing. There is no doubt that some relatively rare forms of
cyberbullying such as having embarrassing pictures or
videos posted very likely will have considerable negative
effects on the targeted youth [26]. Such events must be
taken seriously, as should cyberbullying in general. From
a practical and prevention/intervention perspective,
attention to cyberbullying cases can often represent a
useful new approach to change and also lead to a disclo-
sure of what actually goes on in terms of bullying in the
school context.
Conflict of interest statement
Nothing declared.
References and recommended reading
Papers of particular interest, published within the period of review,
have been highlighted as:
 of special interest
1. Merton RK: Three fragments from a sociologist’s notebooks:
establishing the phenomenon. Specified ignorance, and
strategic research materials. Ann. Rev. Sociol. 1987, 13:1-28.
2. Nunnally JC, Bernstein JH: Psychometric Theory. McGraw Hill;
1994.
3. Campbell DT, Fiske DW: Convergent and discriminant
validation by the Multitrait-Multimethod Matrix. Psychol. Bull.
1959, 56:81-105.
4. Olweus D: School bullying: development and some important
challenges. Rev. Clin. Psychol. 2013, 14:1-30.
5. Menesini E: Cyberbullying: the right value of the phenomenon.
Comment on the paper: cyberbullying: an overrated
phenomenon? Eur. J. Dev. Psychol. 2012, 9:544-552.
6. Olweus D: Invited discussion paper. Cyber bullying: an
 overrated phenomenon. Eur. J. Dev. Psychol. 2012, 9:520-538.
This paper and its follow-up paper [23] argue and document that several
claims about cyberbullying made in the media and by many researchers
are greatly exaggerated. Some measurement issues are also discussed.
7. Olweus D: Bullying at School: What We Know and What We Can
Do. Blackwell Publishers; 1993.
8. Olweus D: Sweden. In The Nature of School Bullying. Edited by
 Smith PK, Morita Y, Junger-Tas-J, Catalano R, Slee P. Routledge;
1999:7-27.
This chapter contains a discussion of the criteria for the definition of
bullying and how bullying is related to the concepts of aggression and
Current Opinion in Psychology 2018, 19:139–143
violence. It also briefly discusses how legislation against bullying, in spite
of original political opposition, became incorporated in Scandinavian
school laws.
9. Bauman M, Underwood MK, Card NA: Definitions: another
perspective and a proposal for beginning with
cyberaggression. In Principles of Cyberbullying Research:
Definition, Measures, and Methods. Edited by Bauman S, Walker J,
Cross D. Routledge; 2013:41-46.
10. Canty J, Stubbert M, Steers D, Collings S: The trouble with
bullying—deconstructing the conventional definition of
bullying for a child-centered investigation into children’s use
of social media. Child. Soc. 2016, 30:48-58.
11. Smith PK, Brain P: Bullying in schools: lessons from two
 decades of research. Aggress. Behav. 2000, 26:1-9.
The history of research and intervention on bullying in schools during the
two decades up to 2000 is summarized. School bullying emerges as an
international issue, and there is an increasing knowledge of its nature and
effects. The paper is an introduction to 10 different articles in this issue
of Aggressive Behavior.
12. Smith PK, del Barrio C, Tokunaga R: Definitions of bullying and
 cyberbullying: how useful are the terms? In Principles of
Cyberbullying Research: Definitions, Measures, and Methods.
Edited by Bauman S, Walker J, Cross D. Routledge; 2013:26-40.
This paper contains a detailed and thoughtful examination of whether and
how the key criteria of traditional bullying can be applied to cyberbullying.
The author’s conclusion is affirmative.
13. Gladden RM, Vivolo-Kantor AM, Hamburger ME, Lumpkin CD:
 Bullying Surveillance Among Youths: Uniform Definitions for Public
Health and Recommended Data Elements, Version 1.0. http://
www.cdc.gov/violenceprenention/pdf/bullyingdefinitions-fina-a.
pdf.
14. Berne S, Frisén A, Schultze-Krumbholz A, Scheithauer H,
Naruskov K, Luik P, Katzer C, Erentaite R, Zukauskiene R:
Cyberbullying assessment instruments: a systematic review.
Aggress. Violent Behav. 2013, 18:320-334.
15. Patchin JW, Hinduja S: Measuring cyberbullying: implications
for research. Aggress. Violent Behav. 2015, 23:69-74.
16. Brochado S, Soares S, Fraga S: A scoping review on studies of
cyberbullying prevalence among adolescents. Trauma Violence
Abuse 2016, 1:1-9.
17. Kowalski RM, Giumetti GW, Schroeder AN, Latanner MR: Bullying
 in the digital age: a critical review and meta-analysis of
cyberbullying research among youth. Psychol. Bull. 2014,
140:1073-1137.
This paper is a careful meta-analytic review of 131 studies of cyberbully-
ing/cybervictimization. However, effect sizes for outcome variables are
not controlled for traditional bullying/victimization.
18. Olweus D: Cyber bullying: a critical overview. In Aggression and
Violence: A Social Psychological Perspective. Edited by Bushman
B. Routledge; 2016:225-240.
19. Schwartz N: Self-reports. How the questions shape the
answers. Am. Psychol. 1999, 54:91-105.
20. Hunter SC, Boyle JME, Warden D: Perceptions and correlates of
 peer-victimization and bullying. Br. J. Educ. Psychol. 2007,
77:797-810.
This is one of the first papers to document empirically that the power
imbalance implicated in the definition of bullying makes bullying and
exposure to general aggression somewhat different phenomena.
21. Olweus D: Understanding and researching bullying: some
critical issues. In Handbook of Bullying in Schools: An
International Perspective. Edited by Jimerson SS, Swearer SM,
Espelage DL. Routledge; 2010:9-33.
22. Solberg ME, Olweus D, Endresen IM: Bullies and victims at
school: are they the same pupils? Br. J. Educ. Psychol. 2007,
77:441-464.
23. Olweus D: Commentary. Comments on cyberbullying article: a
rejoinder. Eur. J. Dev. Psychol. 2012, 9:559-568.
24. Ybarra ML, Mitchell JK: Online aggressor/targets, aggressors,
and targets: a comparison of associated youth
characteristics. J. Child Psychol. Psychiatry 2004, 45:1308-1316.
www.sciencedirect.com