Content uploaded by Margarida Gaspar de Matos
Author content
All content in this area was uploaded by Margarida Gaspar de Matos on Jun 07, 2020
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Original article
Social Media Use and Cyber-Bullying: A Cross-National Analysis of
Young People in 42 Countries
Wendy Craig, Ph.D.
a
, Meyran Boniel-Nissim, Ph.D.
b
, Nathan King, M.Sc.
c
, Sophie D. Walsh, Ph.D.
d
,
Maartje Boer, M.Sc.
e
, Peter D. Donnelly, M.D.
f
, Yossi Harel-Fisch, Ph.D.
g
,
Marta Malinowska-Cie
slik, Ph.D.
h
, Margarida Gaspar de Matos, Ph.D.
i
, Alina Cosma, Ph.D.
e
,
Regina Van den Eijnden, Ph.D.
e
, Alessio Vieno, Ph.D.
j
, Frank J. Elgar, Ph.D.
k
, Michal Molcho, Ph.D.
l
,
Ylva Bjereld, Ph.D.
m
, and William Pickett, Ph.D.
c
,
*
a
Department of Psychology, Queen's University, Kingston, Canada
b
School of Social Sciences and Humanities, Kinneret Academic College on the Sea of Galilee, Zemach, Israel
c
Department of Public Health Sciences, Queen's University, Kingston, Canada
d
Department of Criminology, Bar-Ilan University, Ramat Gan, Israel
e
Department of Interdisciplinary Social Science, Utrecht University, Utrecht, the Netherlands
f
Dalla Lana School of Public Health, University of Toronto, Toronto, Canada
g
The International Research Program on Adolescent Well-Being & Health, School of Education, Bar Ilan University, Ramat Gan, Israel
h
Department of Environmental Health, Health Sciences Faculty, Jagiellonian University Medical College, Krakow, Poland
i
Health Promotion and Education Centre, FMH/ISAMB, University of Lisbon, Lisbon, Portugal
j
Department of Developmental and Social Psychology, University of Padova, Padova, Italy
k
Institute for Health and Social Policy, McGill University, Montreal, Quebec, Canada
l
Department of Children Studies, National University of Ireland Galway, Galway, Ireland
m
Department of Behavioural Sciences and Learning (IBL), Linköping University, Linköping, Sweden
Article history: Received October 4, 2019; Accepted March 2, 2020
Keywords: Adolescent health; Cyber-bullying; Epidemiology; Social media; Violence
ABSTRACT
Purpose: Social media use (SMU) has become an intrinsic part of adolescent life. Negative con-
sequences of SMU for adolescent health could include exposures to online forms of aggression. We
explored age, gender, and cross-national differences in adolescents' engagement in SMU, then
relationships between SMU and victimization and the perpetration of cyber-bullying.
Methods: We used data on young people aged 11e15 years (weighted n ¼180,919 in 42 countries)
who participated in the 2017e2018 Health Behaviour in School-aged Children study to describe
engagement in the three types of SMU (intense, problematic, and talking with strangers online) by
age and gender and then in the perpetration and victimization of cyber-bullying. Relationships
between SMU and cyber-bullying outcomes were estimated using Poisson regression (weighted
n¼166,647 from 42 countries).
Results: Variations in SMU and cyber-bullying follow developmental and gender-based patterns
across countries. In pooled analyses, engagement in SMU related to cyber-bullying victimization
(adjusted relative risks ¼1.14 [95% confidence interval (CI): 1.10e1.19] to 1.48 [95% CI: 1.42e1.55])
IMPLICATIONS AND
CONTRIBUTION
This 2017e2018 study of
181 thousand adolescents
from 42 countries exam-
ines social media use dur-
ing adolescence. Intense
use, problematic use, and
frequent online contact
with strangers each are
each independently associ-
ated with cyber-bullying.
The universality of such as-
sociations is explored, and
Conflicts of interest: The authors declare they have no conflicts of interest.
Disclosure: This supplement was supported by the World Health Organization
European Office and the University of Glasgow. The articles have been peer-
reviewed and edited by the editorial staff of the Journal of Adolescent Health.
The opinions or views expressed in this supplement are those of the authors and
do not necessarily represent the official position of the funder.
*Address correspondence to: William Pickett, Ph.D., Department of Public
Health Sciences, Queen's University, Kingston K7L3N6, Canada.
E-mail address: will.pickett@queensu.ca (W. Pickett).
www.jahonline.org
1054-139X/Ó2020 Published by Elsevier Inc. on behalf of Society for Adolescent Health and Medicine. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
https://doi.org/10.1016/j.jadohealth.2020.03.006
Journal of Adolescent Health 66 (2020) S100eS108
and perpetration (adjusted relative risk ¼1.31 [95% CI: 1.26e1.36] to 1.84 [95% CI: 1.74e1.95]).
These associations were stronger for cyber-perpetration versus cyber-victimization and for girls
versus boys. Problematic SMU was most strongly and consistently associated with cyber-bullying,
both for victimization and perpetration. Stratified analyses showed that SMU related to cyber-
victimization in 19%e45% of countries and to cyber-perpetration in 38%e86% of countries.
Conclusions: Accessibility to social media and its pervasive use has led to new opportunities for
online aggression. The time adolescents spend on social media, engage in problematic use, and talk
to strangers online each relate to cyber-bullying and merit public health intervention. Problematic
use of social media poses the strongest and most consistent risk.
Ó2020 Published by Elsevier Inc. on behalf of Societyfor Adolescent Health and Medicine. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
high-risk groups for cyber-
perpetration and cyber-
victimizationare identified.
Social media use (SMU) has become a normal part of the lives
of young people [1]. Its benefits include the ability to commu-
nicate with friends, quickly access information and gain new
knowledge, and stay in touch with adult mentors including
parents, family members, and teachers [2]. However, emerging
problems associated with SMU include frequent or intense use,
which may detract from opportunities to participate in other
constructive activities, such as extracurricular and community
events [3]. A recent international study concluded that although
statistically significant, the effects of exposure to digital tech-
nology on adolescent well-being are modest and, in the authors'
opinions, “insufficient to invoke the need for policy change”[4].
In contrast, others have argued that “problematic SMU”(indi-
cated by symptoms of addiction to social media) puts adolescents
at risk for problems because it facilitates potential risky online
interaction with strangers with harmful intentions [5] and con-
tributes to addictive behaviors [6], social withdrawal [7], and
impaired social functioning [6].
Intense and problematic SMU exposes adolescents to online
aggression, including cyber-bullying [6,8e11]. Contemporary
social theories provide frameworks to understand these links.
First, more frequent and intense SMU expose adolescents to
aggressive behavior, including the perpetration of cyber-bullying
[12]. Second, Problem Behavior Theory [13,14] posits that certain
risk behaviors are covaried, organized, and clustered, and specific
profiles of risk underlie vulnerability. Repeated exposure to on-
line aggression can make the behavior seem more acceptable
through role modeling and reinforcement [8e11]. Witnessing
the social rewards of aggression or cyber-bullying, such as
increased social status, also reinforces the behavior [15,16], as
young people become motivated to conform to group norms in
their social environment [15]. In addition, the lack of face-to-face
cues associated with SMU hides the negative consequences of
online aggression or cyber-bullying. Therefore, without this
critical feedback, the aggressive behavior may be more likely to
recur [16,17]. Furthermore, repeated exposure to online aggres-
sion or bullying may result in the “disinhibition effect”[16,17].
That is, these aggressive behaviors may become normalized to
youth over time. Such effects are likely to increase the likelihood
of engaging in cyber-bullying or being cyber-victimized. In
addition, those who use electronic communications excessively
may be differentially vulnerable children, who experience psy-
chosocial problems such as loneliness and social anxiety. These
vulnerable children feel positive about being online but may lack
the social resources and skills to prevent being cyber-victimized
at the moment when it is happening [18]. Thus, frequent and
problematic SMU may increase the likelihood of witnessing and
emulating aggressive online behaviors, both as a perpetrator and
as a victim.
Cross-sectional analyses have shown that both intense and
problematic SMU relate to increased cyber-bullying and cyber-
victimization in adolescents [19e25]. Longitudinal research
supports the implied temporality of these links [26,27]. Although
the opportunity to interact with strangers in online environments
also may play an etiological role [17 ], few studies have examined
this as a potential risk factor for online aggressive behaviors.
In light of this background, through a school-based survey of
adolescents in 42 countries and regions [28], we investigated age,
gender, and cross-national differences in adolescents' engage-
ment in SMU, then relationships between SMU and victimization
and the perpetration of cyber-bullying. We hypothesized that
three SMU variablesdfrequent use, problematic (characterized
by addictive-like behaviors), and involvement with strangersd
would represent a continuum of exposures with varying harms in
terms of online aggression. We examined their unique associa-
tions with cyber-bullying and cyber-victimization while simul-
taneously adjusting for the effects of salient covariates, including
mutual control for each indicator of SMU. Our aim was to provide
foundational information that supports policies that support
adolescent health in a contemporary, digital world.
Methods
Study population and procedures
The 2017e2018 Health Behaviour in School-aged Children
(HBSC) survey was conducted in 47 countries and regions
throughout Europe and Canada in the 2017e2018 academic year.
National research teams surveyed nationally representative
samples of 11-, 13-, and 15-year-old children according to a
common research protocol [28]. Questionnaires were translated
to suit the language of the participating countries following a
standardized protocol that included translation, back-translation
into English, then centralized verification. Sampling procedures
involved the selection of classes within schools with variations in
sampling criteria suited to country-level circumstances. Some
countries oversampled subpopulations (e.g., by geography and
ethnicity), and standardized weights were created to ensure
representativeness.
Our analysis of SMU and cyber-bullying/victimization used
data from 42 countries and regions (illustrated for two indicators
of SMU and cyber-bullying in illust Figures 1 and 2). According to
World Bank Classifications [29], these represented 31 “high-in-
come”and 11 “low- and middle-income”countries. Of the
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108 S101
original 47, three countries did not collect information on SMU,
and two others did not submit data by the time of our analysis.
Each country team obtained approval to conduct the survey from
the ethics review board or equivalent regulatory body associated
with the institution conducting each respective national survey.
Participation was voluntary, and consent (explicit or implicit)
was sought from school administrators, parents, and adolescents
as per national human subject requirements.
We used data on 180,919 adolescents in 42 countries in our
prevalence estimation (Table 1). We tested associations be-
tween variables using a subsample of 166,979 adolescents
(weighted n ¼166,647) from the 42 countries that had com-
plete data on cyber-bullying, SMU, age, gender, and socioeco-
nomic class.
Measures
Cyber-bullying. Using an item modified from the validated
Olweus bullying scale [16], participants indicated how often that
they had been victimized by cyber-bullying in the past couple of
months: that is, sent mean instant messages,email,or text mes-
sages;wall postings;created a website making fun of someone;
posted unflattering or inappropriate pictures online without
permission or shared them with others. Because of the low prev-
alence of cyber-bullying and based on international precedent,
its five ordinal response categories were dichotomized into a
binary outcome (never vs. at least once in the past couple of
months)[30]. A second question asked, using a similar stem and
response categories, how often they had taken part in the
Figure 1. Adjusted relative risks by country for problematic social media use and perpetration of cyber-bullying. Note that *and^indicate statistical significance, based
on a 95% confidence interval, for boys and girls, respectively.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108S102
perpetration of cyber-bullying. Response options were similarly
categorized in a binary fashion [30].
Intense SMU. A4-itemadaptedscalefromtheEUKidsOnline
Survey was used to measure SMU [31]. Respondents were
askedhowoftentheyhaveonlinecontactwiththefollowing
people: close friend(s),friends from a larger friend group,friends
that you got to know through the internet but didn’tknowbefore,
and other people than friends (e.g., parents,brothers/sisters,
classmates,and teachers). For each of these four items, answer
categories ranged from 1 (never/almost never)to5(almost all
the time throughout the day), and a do not know/doesn’tapply
option. Intense SMU was defined as having online contact
almost all the time throughout the day on at least one of the
four items.
Problematic SMU. The Social Media Disorder Scale (
a
¼.89) [32]
measured problematic SMU in 9 dichotomous (yes/no) items
that describe addiction-like symptoms: preoccupation with so-
cial media, dissatisfaction about a lack of time for its use, feeling
bad when not using it, trying but failing to spend less time using
it, neglecting other duties in order to use it, regular arguments
over it, lying to parents or friends about its use, using it to
escape from negative feelings, and having a serious conflict with
family over SMU. Endorsement of 6e9 items indicated prob-
lematic SMU, as recommended by the originators of the scale.
Figure 2. Adjusted relative risks by country for intense social media use and perpetration of cyber-bullying. Note that *and^indicate statistical significance, based on a
95% confidence interval, for boys and girls, respectively.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108 S103
Frequent online contact with strangers. We measured frequent
online contact with strangers using the response of “almost all
the time throughout the day”to an item describing the frequency
of online contact with friends that you got to know through the
internet but did not know before. [31].
Other variables. The HBSC questionnaire also collected data on
self-identified gender group (boy, girl, and in some countries
“neither term describes me”), age group (11, 13, and 15 years),
socioeconomic class (a 6-item measure of material assets in the
home including number of vehicles, bedroom sharing, computer
ownership, bathrooms at home, dishwashers at home, and
family vacations), [33] family support (a 4-item scale describing
the degree of help, emotional support, communication, and
assistance in decision-making perceived to be experienced in
families), and peer support (a 4-item scale describing the degree
of help from friends, ability to count on them, communication of
happy and sad feelings, and communication of problems with
friends) [34].
Statistical analysis
We analyzed the data in SAS 9.4 (SAS Institute, Cary, NC,
2016). Descriptive analyses characterized the international
sample, restricted to participants with complete data for age,
gender, and the cyber-bullying and SMU variables. We then
estimated the prevalence of cyber-bullying (victimization and
perpetration) and then the three types of SMU (intense,prob-
lematic, and online contact with strangers) by age and gender. For
each country, we calculated the prevalence per 100 children and
then summarized these estimates in a pooled analysis using
minimum, median, and maximum values.
In the pooled international sample, we then used Poisson
regression analyses to model cyber-bullying (victimization then
perpetration) as dependent variables with each of the three in-
dicators of SMU as independent variables. We analyzed data on
boys and girls separately and restricted our analyses to records
with complete data on cyber-bullying, SMU, and all covariates
under consideration. Based on a priori consideration of con-
founding, age, family affluence, peer support, and family support
were forced into every adjusted model. Models with victimization
as the dependent variable controlled for perpetration, and vice
versa. Models examining intense SMU were adjusted for prob-
lematic SMU and vice versa, whereas frequent online contact with
strangers was adjusted for problematic SMU. In addition, because
frequent online contact with strangers was one of four items that
also contributed to the intense SMU scale, we performed a
sensitivity analysis that examined the effects of intense SMU, with
and without inclusion of the online contact with strangers item.
All models accounted for the clustered nature of the sampling
scheme via inclusion of school, then country identifiers as
random effects. Beta coefficients and standard errors were used
to generate crude and adjusted estimates of relative risk (RR) and
associated 95% confidence intervals (CIs).
Following this step, we reran each of the adjusted regression
models at the country level. The numbers of countries where we
identified statistically significant (p<.05) effects indicated the
consistency of any observed risks or protections across countries.
We also described these effects across the countries graphically
to illustrate the size of effects and their level of consistency,
irrespective of statistical significance.
Given the large sample sizes involved, the pooled analysis was
90% powered to detect an adjusted RR of 1.06e1.18 in boys and
girls (
a
¼.05, two sided) for each of the relationships under
study. Detectable effects varied but were generally larger (>1.20)
in the country-specific analyses.
Results
Cyber-bullying
The prevalence of reported victimization by cyber-bullying
and perpetration of cyber-bullying varied by country, gender,
and age group (Table 2). Median estimates of both victimization
and perpetration were generally low and, in the pooled analysis,
remained fairly consistent by age group for victimization by
cyber-bullying in boys (p¼.22) but not girls (p¼.02), but
increased with age for perpetration (boys: p¼.01; girls: p¼.02).
The median prevalence of victimization reported by girls was
higher than boys, especially at age 13 years (p¼.02). Conversely,
the median prevalence of perpetration reported by boys was
higher than girls in all age groups (all comparisons, p<.01).
Country-specific estimates for both victimization and perpetra-
tion are provided in Supplementary Table 1.
Social media use
Based on median estimates, intense SMU positively related to
age, especially among girls (p<.01; Table 3). Girls were less likely
than boys to engage in problematic SMU at age 11 years, but
more likely at ages 13 and 15 years. Frequent online contact with
strangers increased with age and was more prevalent among
boys versus girls. In general, among girls, the prevalence of
problematic SMU was higher than the prevalence of frequent
contacts with strangers. Among boys, the prevalence of frequent
contact with strangers was higher than the prevalence of prob-
lematic SMU.
SMU and cyber-bullying
Table 4 presents the regression analysis of cyber-bullying
victimization and perpetration using the pooled sample. For
cyber-victimization, bivariate models showed modest to strong
relationships (adjusted RRs: 1.14 [95% CI: 1.10e1.19] to 1.48 [95%
CI: 1.42e1.55]) between each of the three indicators of SMU and
being victimized by cyber-bullying. Adjusted models showed
Table 1
Description of international study sample, HBSC study, 2018
Descriptor Number
Number of countries reporting, n 42
Total participants, n 180,919
By country, n
Median 3,715
Minimum 1,446 (Albania)
Maximum 11,155 (Wales)
By gender, n (%)
Boys 86,981 (48.1)
Girls 93,938 (51.9)
By age group, n (%)
11 years 56,219 (31.1)
13 years 62,661 (34.6)
15 years 62,039 (34.3)
All values are weighted.
HBSC ¼Health Behaviour in School-aged Children.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108S104
that the observed effects for each of the three indicators were
partially explained by known confounders, as the magnitude of
the RRs decreased after adjustment. However, the relation be-
tween each indicator of SMU and cyber-victimization held up to
these added controls. The consistency of such effects across
countries is shown in the number of countries reporting statis-
tically significant increases in relative risk. There was significant
variation between countries in the significance of the association
between intense use and cyber-victimization (there were eight
countries with a significant association for boys and 25 for girls).
Problematic SMU was most strongly and consistently related to
cyber-victimization. The association was significant in 20 coun-
tries for boys and in 29 countries for girls. Cyber-victimization
related to frequent contact with strangers in 10 countries for
boys and 19 countries for girls.
For perpetration of cyber-bullying bivariate models showed
stronger effects (adjusted RR: 1.31 [95% CI: 1.26e1.36] to 1.84
[95% CI: 1.74e1.95]) between each of the three indicators of SMU
and being a perpetrator when compared with the relative risks
for victimization. Adjustment for confounders attenuated the
overall relative risks, but the effects remained strong and sta-
tistically significant in boys and girls. Again, there was variation
across countries. We observed a statistically significant higher
risk for perpetration when reporting intense (24 and 20 coun-
tries for boys and girls, respectively) and problematic SMU
(22 and 36 countries, boys and girls, respectively) (Figures 1 and
2). We found significant associations with perpetration and
frequent talking to strangers in only 16 countries for boys but 28
countries for girls (Figures 1 and 2).
Finally, the sensitivity analysis used to examine the effects of
intense SMU on the perpetration then victimization by cyber-
bullying outcomes, with and without inclusion of the frequent
online contact with strangers item as part of the intense SMU
measure is presented in Supplementary Table 2. The findings
were very similar for models that did and did not include the
contact with strangers item.
Discussion
The proliferation of SMU among adolescents over the past
decade has led to concerns about its negative consequences for
adolescent health and well-being [1,5e7]. Our study explored
relationships between three types of SMU (intense, problematic,
frequent talking with strangers), and involvement in cyber-
bullying/victimization, and the consistency of these relationships
across gender, age groups and diverse geopolitical contexts. We
note three main findings. First, we observed more consistent
relationships across countries for each of the three types of SMU
with perpetration of cyber-bullying compared with cyber-
victimization. Second, we observed these relationships in more
countries for girls than boys, for both cyber-bullying and cyber-
victimization. Third, for both boys and girls, problematic media
use related to cyber-bullying and cyber-victimization in the most
countries and estimates indicated the presence of modest to
strong effects that merit public health intervention.
Consistent with an “exposure perspective”[35], our findings
suggest that SMU exposes young people to risks for involvement
in cyber-bullying and to more aggressive online behaviors,
Table 2
Reported victimization by and perpetration of cyber-bullying in 42 countries, HBSC study, 2018
Prevalence per 100 children
Within countries by age group and gender
11 years 13 years 15 years
Minimum Median Maximum Minimum Median Maximum Minimum Median Maximum
Boys
Victimization by cyber-bullying 4.0 12.5 27.5 2.2 11.9 24.3 3.2 11.3 28.5
Perpetration of cyber-bullying 1.8 7.6 26.7 3.1 10.3 28.8 3.6 11.8 31.4
Girls
Victimization by cyber-bullying 3.8 12.7 24.5 6.4 13.9 27.8 5.3 12.7 20.9
Perpetration of cyber-bullying .7 6.1 14.0 2.5 7.2 19.1 1.6 7.5 19.4
All values are weighted.
HBSC ¼Health Behaviour in School-aged Children.
Table 3
Reported engagement in sentinel indicators of electronic media communication within countries, HBSC study, 2018
Prevalence per 100 children
Within countries by age group and gender
11 years 13 years 15 years
Minimum Median Maximum Minimum Median Maximum Minimum Median Maximum
Boys
Intense use of social media 14.1 28.9 47.3 17.1 32.4 48.9 18.0 36.8 52.3
Problematic social media use 1.2 5.8 25.3 2.9 6.4 17.8 2.1 6.1 17.5
Frequent social media
contact with strangers
1.2 6.0 14.3 2.4 7.5 12.7 2.7 8.5 13.0
Girls
Intense use of social media 12.5 29.3 48.5 16.9 41.8 60.3 21.1 45.6 64.4
Problematic social media use 1.1 4.7 14.5 3.6 8.4 20.1 3.9 8.8 18.7
Frequent social media contact
with strangers
.6 3.4 7.7 2.7 6.2 11.6 2.2 6.2 13.7
All values are weighted. For Slovenia, only 15-year-olds included for problematic social media user.
HBSC ¼Health Behaviour in School-aged Children.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108 S105
Table 4
Bivariate and adjusted relative risk estimates for victimization byand perpetration of cyber-bullying associated with three indicators of social media use in 42 countries, HBSC study, 2018(weighted n¼166,647 [79,486
boys and 87,161 girls] from 41 countries included in the overall analyses)
Victimization by cyber-bullying Perpetration of cyber-bullying
Overall prevalence Overall relative risk No. countries
reporting
a
Overall prevalence Overall relative risk No. countries
reporting
a
Bivariate Adjusted
b
Adjusted RR
b
Bivariate Adjusted
b
Adjusted RR
b
n (% yes) RR (95% CI) RR (95% CI) >11 <1 n (% yes) RR (95% CI) RR (95% CI) >11 <1
Boys
Intense use
No 54,287 (10.7) 1.00 (ref) 1.00 (ref) 8 34 0 54,287 (9.5) 1.00 (ref) 1.00 (ref) 24 18 0
Yes 25,199 (13.9) 1.29 (1.24e1.35) 1.14 (1.10e1.19) 25,199 (14.2) 1.48 (1.42e1.54) 1.31 (1.26e1.36)
Problematic use
No 74,424 (11.0) 1.00 (ref) 1.00 (ref) 20 22 0 74,424 (10.2) 1.00 (ref) 1.00 (ref) 22 19 1
Yes 5,062 (23.2) 2.12 (2.00e2.24) 1.35 (1.28e1.42) 5,062 (23.2) 2.25 (2.12e2.39) 1.44 (1.37e1.52)
Frequent contact with strangers
No 73,848 (11.2) 1.00 (ref) 1.00 (ref) 10 32 0 73,848 (10.4) 1.00 (ref) 1.00 (ref) 16 26 0
Yes 5,638 (18.6) 1.64 (1.55e1.75) 1.22 (1.16e1.29) 5,638 (19.0) 1.79 (1.69e1.91) 1.34 (1.27e1.41)
Girls
Intense use
No 53,628 (11.9) 1.00 (ref) 1.00 (ref) 25 17 0 53,628 (5.6) 1.00 (ref) 1.00 (ref) 20 22 0
Yes 33,533 (17.2) 1.44 (1.39e1.49) 1.30 (1.25e1.34) 33,533 (9.9) 1.72 (1.63e1.80) 1.39 (1.32e1.45)
Problematic use
No 80,211 (12.6) 1.00 (ref) 1.00 (ref) 29 13 0 80,211 (6.2) 1.00 (ref) 1.00 (ref) 36 6 0
Yes 6,950 (28.9) 2.26 (2.16e2.37) 1.48 (1.42e1.55) 6,950 (19.6) 3.10 (2.92e3.28) 1.84 (1.74e1.95)
Frequent contact with strangers
No 82,323 (13.2) 1.00 (ref) 1.00 (ref) 19 23 0 82,323 (6.8) 1.00 (ref) 1.00 (ref) 28 14 0
Yes 4,839 (26.3) 1.96 (1.85e2.07) 1.39 (1.31e1.47) 4,839 (16.0) 2.30 (2.14e2.48) 1.40 (1.30e1.50)
All analyses are weighted.
Armenia is excluded from the overall pooled analysis because of missing data but included in country-specific analyses adjusting for all covariates except socio-economic class.
CI ¼confidence interval; HBSC ¼Health Behaviour in School-aged Children; ref ¼reference; RR ¼relative risk.
a
Based on CIs that do or do not overlap 1.00.
b
Adjusted for age, county-specific socioeconomic class, family support, and peer support and for clustering at the country and school level. RR estimates for victimization by cyber-bullying are adjusted for
perpetration of cyber-bullying and vice versa. RR estimates for intense user are adjusted for problematic use and vice versa. RRs for frequent contact with strangers are adjusted for problematic use.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108S106
particularly for boys. Time spent online, especially if SMU is
frequent and/or problematic, replaces opportunities to engage in
constructive and protective in-person social activities that pro-
mote socioemotional and moral development [3]. From a theo-
retical perspective, aggressive tendencies may develop in young
people who grow up in environments that “reinforce aggression,
provide aggressive models, frustrate and victimize them, and
teach them that aggression is acceptable and successful”(p. 47)
[36]. Intense and problematic SMU may expose adolescents to
peers and social norms that validate and reinforce different
forms of aggression, including cyber-bullying. Associations be-
tween intense and problematic SMU and cyber-bullying may be
exacerbated by cognitive, emotional, and associated social vul-
nerabilities because SMU provides a safe and anonymous way of
expressing frustrations, which could translate into online
aggression among vulnerable youth [37,38].
Exposure theory also explains gender differences in the con-
sistency of findings across countries. We found more cross-
country consistency for girls than boys in the associations
among problematic use and frequent contact with strangers and
cyber-bullying. Because girls spend more time online and report
more problematic use, they have greater exposure to aggressive
role models, potentially reinforcing the opportunity to engage in
cyber-bullying. Similarly, for girls, intense and problematic use
related to cyber-victimization in most countries but not for boys.
Increased exposure and use of social media among girls may
result in an increased risk of cyber-victimization [39]. Similarly,
the gendered patterns surrounding frequent online contact with
strangers become more pronounced as children age and were
particularly experienced by boys. This may be attributable to
differences in sensation seeking and risk taking, as mediated by
hormonal factors and social norms [40], as well as engagement in
interactive game playing by boys and older adolescents, which by
necessity often involve interaction with strangers [41].
All three of SMU measures independently related to cyber-
bullying and cyber-victimization with modest to strong effect
sizes. Although these associations varied by country, they high-
light the unique contribution of each construct and importance
of broad assessments of SMU that include measures of intense
use, problematic use and talking with strangers. Problematic
SMU related to cyber-bullying and victimization in many coun-
tries. Cross-country variations in associations with cyber-
bullying and cyber-victimization also highlight the importance
of a socioecological approach in understanding these relation-
ships. Cultural, economic, and social factors, such as Internet
access, availability of electronic devices, and social cultural
norms about online behavior, underlie these associations. Future
research should examine the importance of these contextual
differences in explaining cross-national differences found here.
The strengths of our study include large, representative
samples and our use of standardized, validated measures that
differentiated aspects of SMU and cyber-bullying involvement
[31]. The findings are almost certainly generalizable to contem-
porary populations of young people from high-income countries
and may be generalizable beyond such populations, given that
access to the internet is common, even in low- and middle-
income countries [42].
The limitations of the study include the cross-sectional, self-
report nature of the data collection, which limited the potential for
causal inferences. Second, self-reports of sensitive behaviors,
including perpetration and victimization due to cyber-bullying, are
also subject to bias and misclassification because of the social
stigmas associated with their occurrence [43]. Third, HBSC's reli-
ance on a binary indicator of gender does not reflect the experi-
ences of young people whose identity does not match these binary
categories, nor those for whom the sex assigned at birth does not
correspond with that identity. In some circles, the HBSC item is best
considered a measure of “sex at birth”rather than gender.
There are theoretical and methodological implications of the
findings. First, research will benefit from a socioecological approach
to understand variations in association across countries. Exposure
theory may not sufficiently describe cross-country variations in the
pattern of results. Second, research should also assess multiple
aspects of SMU use, given the unique determinants and conse-
quences of its intensity, problematic use, and contact with strangers
online. Third, interventions need to consider a sex-/gender-specific
approach. Frequency of contact with strangers may represent a risk
factor that is particularly important to girls' involvement in cyber-
bullying. The findings also point to the need for further research
that examines the interplay of gender and age in SMU use and
cyber-bullying. On a more practical level, parents, educators, clini-
cians, and others who care for children should be aware of these
correlates of SMU. Problematic and intense SMU as well as talking
with strangers are not innocuous in terms of their potential links to
cyber-bullying and require both awareness and evidence-based
strategies for prevention.
The developing world of electronic social media technology
and its intensive introduction into the daily lives of adolescents
provide new and alternative social settings in which young
people engage in relationships. Although social media environ-
ments replicate much of what is present in traditional face-to-
face activities, these rapidly evolving environments have
changed the meaning and manifestation of social connectedness
among adolescents. Easy access to social media and its pervasive
use have led to new opportunities for cyber-bullying and pre-
sents new challenges and opportunities for health policy and
practice to protect youth from harm.
Acknowledgments
Health Behaviour in School-aged Children is an international
study carried out in collaboration with WHO/EURO. The Inter-
national Coordinator was Jo Inchley (University of Glasgow) for
the 2017/2018 survey. The Data Bank Manager was Professor
Oddrun Samdal (University of Bergen). The 2017/2018 survey
included in this study were conducted by the following principal
investigators in the 42 countries and regions: Albania (Gentiana
Qirjako), Armenia (Sergey G. Sargsyan), Austria (Rosemarie
Felder-Puig), Azerbaijan (Gahraman Hagverdiyev), Flemish
Belgium (Bart De Clercq), French Belgium (Katia Castetbon),
Canada (William Pickett, Wendy Craig, and [the late] John
Freeman), Croatia (Ivana Pavic Simetin), Czech Republic (Michal
Kalman), Denmark (Mette Rasmussen), England (Fiona Broks,
Ellen Klemera), Estonia (Leila Oja, Katrin Aasvee), Finland (Jorma
Tynjälä), France (Emmanuelle Godeau), Georgia (Lela Shengelia),
Germany (Matthias Richter), Greece (Anna Kokkevi), Hungary
(Ágnes Németh), Iceland (Arsaell M. Arnarsson), Ireland (Saoirse
Nic Gabhainn), Italy (Franco Cavallo), Kazakhstan (Shynar
Abdrakhmanova and Valikhan Akhmetov), Lithuania (Kastytis
Smigelskas), Latvia (Iveta Padule), Luxembourg (Helmut Wil-
lems), Malta (Charmaine Gauci), the Netherlands (Gonneke Ste-
vens and Saskia van Dorsselaer), Norway (Oddrun Samdal),
Poland (Joanna Mazur and Agnieszka Ma1kowska-Szkutnik),
Portugal (Margarida Gaspar de Matos), Republic of Moldova
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108 S107
(Galina Lesco), Romania (Adriana Baban), Russian Federation
(Anna Matochkina), Scotland (Jo Inchley), Serbia (Jelena Rakic),
Slovakia (Andrea Madarasova Geckova), Slovenia (Helena Jer-
icek), Spain (Carmen Moreno), Sweden (Petra Lofstedt),
Switzerland (Marina Delgrande-Jordan and Hervé Kuendig),
Ukraine (Olga Balakireva), and Wales (Chris Roberts).
Funding Sources
Grant funding for the researchers involved in this cross-na-
tional manuscript was provided by the (1) Public Health Agency
of Canada; (2) Canadian Institutes of Health Research (operating
grant MOP341188); (3) Ministry of Health, Office of the Director,
Israel; (4) Ministry of Health, Wellbeing, and Sports, the
Netherlands; (5) Institute of Mother and Child, and Warsaw
University, Poland; (6) Public Health Agency of Sweden; (7)
Italian Ministry of Health/Centre for Disease Prevention and
Control; and (8) Department of Health, Ireland.
Supplementary Data
Supplementary data related to this article can be found at
http://doi.org/10.1016/j.jadohealth.2020.03.006.
References
[1] Livingston JD, Cianfrone M, Korf-Uzan K, Coniglio C. Another time point, a
different story: One year effects of a social media intervention on the at-
titudes of young people towards mental health issues. Soc Psychiatry
Psychiatr Epidemiol 2014;49:985e90.
[2] Lenhart A, Smith A, Anderson M, et al. Teens, technology & friendships:
Video games, social media and phones play an integral role in how teens
meet and interact with friends. Pew Research Center. 2015. Available at:
http://www.pewInternet.org/2015/08/06/teens-technology-and-
friendships/. Accessed August 31, 2019.
[3] Jiang X, Peterson RD. Beyond participation: The association between school
extracurricular activities and involvement in violence across generations of
immigration. J Youth Adolesc 2012;41:362e78.
[4] Orben A, Przybylski AK. The association between adolescent well-being
and digital technology use. Nat Hum Behav 2019;3:173e82.
[5] Sasson H, Mesch G. Parental mediation, peer norms and risky online
behavior among adolescents. Comput Hum Behav 2014;33:32e8.
[6] Qiaolei J, Xiuqin H, Ran T. Examining factors influencing internet addiction
and adolescent risk behaviors among excessive internet users. Health
Commun 2018;33:1434e44.
[7] Valkenburg PM, Peter J. Online communication among adolescents: An
integrated model of its attraction, opportunities, and risks. J Adolesc Health
2011;48:121e7.
[8] Boniel-Nissim M, Sasson H. Bullying victimization and poor relationships
with parents as risk factors of problematic internet use in adolescence.
Comput Hum Behav, 88, 176-183.
[9] Nixon CL. Current perspectives: The impact of cyberbullying on adolescent
health. Adolesc Health Med Ther 2014;5:143e58.
[10] Englander E, Donnerstein E, Kowalski R, et al. Defining cyberbullying. Pe-
diatrics 2017;140:S148e51.
[11] Kowalski RM, Giumetti GW, Schroeder AN, Lattanner MR. Bullying in the
digital age: A critical review and meta-analysis of cyberbullying research
among youth. Psychol Bull 2014;140:1073e137.
[12] Gottfredson MR, Hirschi T. A general theory of crime. Redford City, CA:
Stanford University Press; 1990.
[13] Jessor R, Jessor SL. Problem behavior and psychosocial development: A
longitudinal study of youth. New York: Academic Press; 1977.
[14] Jessor R. The origins and development of problem behavior theory. New
York: Springer; 2016.
[15] Blakemore SJ, Mills KL. Is adolescence a sensitive period for sociocultural
processing? Annu Rev Psychol 2014;65:187e207.
[16] Olweus D. Bullying at school: What we know and what can we do. Oxford:
Blackwell publishers; 1993.
[17] Barlett CP, Gentile DA. Attacking others online: The formation of cyber-
bullying in late adolescence. Psychol Popular Media Cult 2012;1:123e35.
[18] Prizant-Passal S, Shechner T, Aderka I. Social anxiety and internet use ea
meta-analysis: What do we know? What are we missing? Comput Hum
Behav 2016;62:221e9.
[19] Erreygers S, Vandebosch H, Vranjes I, et al. The longitudinal association
between poor sleep quality and cyberbullying, mediated by anger. Health
Commun 2019;34:560e6.
[20] Casas JA, Del Rey R, Ortega-Ruiz R. Bullying and cyberbullying:
Convergent and divergent predictor variables. Comput Hum Behav 2013;
29:580e7.
[21] Kircaburun K, Jonason PK, Griffiths MD. The dark tetrad traits and prob-
lematic social media use: The mediating role of cyberbullying and cyber-
stalking. Pers Indiv Differ 2018;135:264e9.
[22] Rice E, Petering R, Rhoades H, et al. Cyberbullying perpetration and
victimization among middle-school students. Am J Public Health 2015;105:
e66e72.
[23] Sampasa-Kanyinga H, Hamilton HA. Use of social networking sites and risk
of cyberbullying victimization: A population-level study of adolescents.
Cyberpsychol Behav Soc Netw 2015;18:704e10.
[24] Machimbarrena J, Calvete E, Fernández-González L, et al. Internet risks: An
overview of victimization in cyberbullying, cyber dating abuse, sexting,
online grooming and problematic internet use. Int J Environ Res Public
Health 2018;15:1e15.
[25] Lee HW, Choi JS, Shin YC, et al. Impulsivity in internet addiction: A com-
parison with pathological gambling. Cyberpsychol Behav Soc Netw 2012;
15:373e7.
[26] Barlett CP, Gentile DA, Chng G, et al. Social media use and cyberbullying
perpetration: A longitudinal analysis. Violence Gend 2018;5:191e7.
[27] Gámez-Guadix M, Borrajo E, Almendros C. Risky online behaviors among
adolescents: Longitudinal relations among problematic internet use,
cyberbullying perpetration, and meeting strangers online. J Behav Addict
2016;5:100e7.
[28] Roberts C, Freeman J, Samdal O, et al. The health behaviour in school-aged
children (HBSC) study: Methodological developments and current ten-
sions. Int J Public Health 2009;54 Suppl 2:140e50.
[29] World Bank. World development report 2020: Trading for development in
the age of global value chains. Washington, DC: World Bank Publications;
2020.
[30] Craig W, Harel-Fisch Y, Fogel-Grinvald H, et al. A cross-national profile of
bullying and victimization among adolescents in 40 countries. Int J Public
Health 2009;54 Suppl 2:216e24.
[31] Mascheroni G, Ólafsson K. Access and use. In: Net children go mobile: Risks
and opportunities. 2nd ed. Milano: Educatt; 2014:11e23.
[32] Van den Eijnden RJJM, Lemmens J, Valkenburg PM. The social media dis-
order scale: Validity and psychometric properties. Comput Hum Behav
2016;61:478e87.
[33] Currie C, Molcho M, Boyce W, et al. Researching health inequalities
in adolescents: The development of the health behaviour in school-
aged children (HBSC) family affluence scale. Soc Sci Med 2008;66:
1429e36.
[34] Zimet GD, Powell SS, Farley GK, et al. Psychometric characteristics of the
multidimensional scale of perceived social support. J Pers Assess 1990;55:
610e7.
[35] Brown JD, Bobkowski PS. Older and newer media: Patterns of use and
effects on adolescents' health and well-being. J Res Adolesc 2011;21:
95e113.
[36] Anderson CA, Carnagey NL. Violent evil and the general aggression model.
In: Miller AG, ed. Social psychology of good and evil. New York, NY:
Guilford; 2004:168e92.
[37] Best P, Manktelow R, Taylor B. Online communication, social media and
adolescent wellbeing: A systematic narrative review. Child Youth Serv Rev
2014;41:27e36.
[38] Aboujaoude E, Savage MW, Starcevic V, Salame WO. Cyberbullying: Review
of an old problem gone viral. J Adolesc Health 2015;57:10e8.
[39] Underwood MK, Ehrenreich SE. The power and the pain of adolescents’
digital communication: Cyber victimization and the perils of lurking. Am
Psychol 2017;72:144e58.
[40] Vermeersch H, Tsjoen G, Kaufman J, Vincke. The role of testosterone in
aggressive and non-aggressive risk taking in adolescent boys. Horm Behav
2008;53:463e71.
[41] Lenhart A, Duggan M, Perrin A, et al. Teens, social media & technology
overview 2015: Smartphones facilitates shifts in communication landscape
for teens. Pew Research Centre. 2015. Available at: https://www.
pewresearch.org/wp-content/uploads/sites/9/2015/10/pi_2015-10-01_
teens-technology-romance_final.pdf. Accessed February 24, 2020.
[42] Organisation for Economic Cooperation and Development. OECD broad-
band portal. Available at: https://www.oecd.org/sti/broadband/broadband-
statistics/. Accessed July 30, 2019.
[43] Juvonen J, Graham S. Bullying in schools: The power of bullies and the
plight of victims. Annu Rev Psychol 2014;65:159e85.
W. Craig et al. / Journal of Adolescent Health 66 (2020) S100eS108S108
