Content uploaded by Mark D Griffiths
All content in this area was uploaded by Mark D Griffiths on Apr 07, 2019
Content may be subject to copyright.
Problematic Social Media Use: Results from a
Large-Scale Nationally Representative
, Orsolya Kira
, Aniko Maraz
, Zsuzsanna Elekes
, Cecilie Schou Andreassen
, Zsolt Demetrovics
1Institute of Psychology, Eo
´nd University, Budapest, Hungary, 2Doctoral School of Psychology,
´nd University, Budapest, Hungary, 3Institute of Sociology and Social Policy, Corvinus University
of Budapest, Budapest, Hungary, 4International Gaming Research Unit, Nottingham Trent University,
Nottingham, United Kingdom, 5Department of Clinical Psychology, University of Bergen, Bergen, Norway
Despite social media use being one of the most popular activities among adolescents, prev-
alence estimates among teenage samples of social media (problematic) use are lacking in
the field. The present study surveyed a nationally representative Hungarian sample com-
prising 5,961 adolescents as part of the European School Survey Project on Alcohol and
Other Drugs (ESPAD). Using the Bergen Social Media Addiction Scale (BSMAS) and based
on latent profile analysis, 4.5% of the adolescents belonged to the at-risk group, and
reported low self-esteem, high level of depression symptoms, and elevated social media
use. Results also demonstrated that BSMAS has appropriate psychometric properties. It is
concluded that adolescents at-risk of problematic social media use should be targeted by
school-based prevention and intervention programs.
Social media use
Social media use is currently one of the most popular leisure activities among adolescents (e.g.,
[1–3]). Social media (e.g., Facebook,Instagram,Snapchat, etc.) host virtual communities where
users can create individual public and/or private profiles [4–6]. Users can access social media
on different platforms (mobile or computer devices), for different activities (e.g., interacting
with real-life friends, meeting others based on shared interest, chatting, mailing, sharing or
creating pictures / videos, blogging, dating, playing games, gambling; [7–9]).
Facebook is one of the most popular social media among 13–17 years old adolescents in the
USA . According to a recent report, 71% of teenage social media users access more than one
social media and 24% of adolescents are “almost constantly” online due to the widespread use
and popularity of smartphones . Furthermore, there is an increasing interest to explore and
assess the characteristics and prevalence of problematic/excessive use of social media (e.g., [4,
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 1 / 13
´nyai F, Zsila A
´ly O, Maraz A,
Elekes Z, Griffiths MD, et al. (2017) Problematic
Social Media Use: Results from a Large-Scale
Nationally Representative Adolescent Sample.
PLoS ONE 12(1): e0169839. doi:10.1371/journal.
Editor: Susana Jime
Universitari de Bellvitge, SPAIN
Received: November 1, 2016
Accepted: December 23, 2016
Published: January 9, 2017
Copyright: ©2017 Ba
´nyai et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: Data are available at
Funding: This study was supported by the
Hungarian National Research, Development and
Innovation Office (Grant numbers: K111938,
´gnes Zsila was supported by the New
National Excellence Program awarded by the
Ministry of Human Resources. The funding
institutions had no role in the study design or the
collection, analysis and interpretation of the data,
Problematic social media use
To date, there is no consensus among researchers regarding the definition of problematic
social media use due to the conceptual confusion surrounding the classification of problematic
internet use [15,16]. Negative outcomes triggered by the excessive use of social media may
have a detrimental effect on the personal, social, and/or professional lives of the users [8,13,
17–20]. Lee, Cheung, and Thadani  argued that obsessive Facebook users had troubles in
work, academic performance. and interpersonal relationships. For instance, Pantic and Dam-
janovic , Wegmann and Stodt , and Andreassen and Billieux  reported a significant
positive correlation between depression symptoms and social media use, while Malik and
Khan  found negative relationship between self-esteem and high levels of social media use.
Due to the lack of consistency in empirical studies, diagnosis of internet-related disorders
has yet to be established based on the aforementioned theoretical constructs. Internet Use Dis-
order was suggested for consideration in the latest (fifth) edition of the Diagnostic and Statisti-
cal Manual of Mental Disorders (DSM-5; ). However, only one internet-related disorder—
Internet Gaming Disorder—was included in Section 3 of the DSM-5. Another problem is that
various synonyms of problematic social media use exist in the literature with different diagnos-
tic suggestions including (among others) Facebook dependence , Facebook addiction ,
social networking addiction , Twitter addiction , social media addiction , and
Social Media Disorder .
Different theoretical models provide explanations for the development of problematic
social media use (e.g., cognitive-behavioral, social skill, or socio-cognitive models; . These
theoretical models have been developed from a clinical perspective, while the biopsychosocial
model concerns behavioral addictions in general . According to the biopsychosocial model
, problematic social media use can be determined by a range of addiction symptoms
including: mood modification (i.e., excessive social media use leading to specific changes in
mood states), salience (i.e., total preoccupation with social media use), tolerance (i.e., increas-
ing amounts of time using social media), withdrawal symptoms (i.e., negative feelings and psy-
chological symptoms such as irritability, anxiety when social media use is restricted), conflict
(i.e., interpersonal problems as a direct result of social media usage), and relapse (i.e., returning
to excessive social media use after a period of abstinence).
Assessing problematic social media use
To obtain a reliable prevalence rate of problematic social media usage, it is important to use
psychometrically valid measurement tools. Due to the problem of inconsistencies regarding
the definition of problematic, excessive, or addictive social media use, there is also a lack of
reliable and valid psychometric scales to assess the phenomenon of problematic social media
use. More specifically, the existing assessment tools are based on different diagnostic sugges-
tions such as problematic internet use (e.g., Internet Addiction Test; [32–34], Internet Gaming
Disorder ), or other aspects of addictive tendencies (e.g., withdrawal, loss of control,
salience; [35,36]. In addition, some of the measurement tools focus only on specific social
media (e.g., Facebook;  such as the Facebook Addiction Symptoms Scale , the Facebook
Addiction Scale , the Bergen Facebook Addiction Scale , and the Facebook Intrusion
Although the most recent data show that Facebook is the most popular and frequently used
social media among adolescents , empirical research has shown that adolescents use more
than one social media frequently (e.g., ). Therefore, the assessment tools are unable to follow
the ever-changing trends in the area of social media use. Considering the increased usage of
various social media among adolescents [1–3,5] the questionnaires should assess all available
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 2 / 13
writing the manuscript, or the decision to submit
the paper for publication.
Competing Interests: The authors have declared
that no competing interests exist.
social media and the total range of activities on these social media instead of one specific social
media such as Facebook .
Prevalence of problematic social media use
It is difficult to estimate the prevalence of problematic social media use due to the use of vari-
ous assessment tools and the lack of a consensual definition of problematic social media use.
Furthermore, recent research has demonstrated that problematic social media use has a higher
prevalence among female users than males [11,13,40,41]. Unfortunately, in studies that have
assessed different aspects of problematic social media use, the gender distribution was usually
frequently imbalanced in that women were typically over-represented [11,12,15,26,29,37,
42–44], and may be explained by the higher willingness of females to participate in such
Due to different theoretical frameworks and psychometric assessments, the prevalence of
problematic social media use might be underestimated or overestimated. Previous studies
have reported different prevalence rates relating to problematic social media use. For instance,
Olowu and Seri  reported a prevalence rate of 2.8% of addicted social media use among
college students, while Jafarkarimi and Sim  reported a prevalence rate of 47% being
addicted to Facebook among a sample of college students. Explanations for the large difference
in problematic social media use prevalence rates might be the non-representative (self-selected
and typically small participant) samples and different cultural groups examined (e.g., Chinese,
Australian, Nigerian college students, Dutch adolescents [12,15,26,29,37,42–44]. Moreover,
to date, there has only been one nationwide survey assessing problematic (ie., addictive) social
media use  that examined the associations between problematic social media use, narcis-
sism, and self-esteem, and between problematic use of social media, attention-deficit/hyperac-
tivity, obsessive-compulsiveness, anxiety, and depression on cross-sectional convenience
sample of 23,532 Norwegians (although the sample was not nationally representative).
Furthermore, no studies have examined the prevalence of problematic social media use uti-
lizing a representative adolescent sample. Furthermore, only a few studies exist concerning
problematic social media use among adolescents (e.g., ). Previous studies have reported an
increased popularity of social media use among adolescents [1–4,6] and the increased number
of adolescent social media users could explain the higher prevalence of problematic usage in
this group [4,6]. Consequently, the aim of the present study was twofold:
1. To test the psychometric properties of the Bergen Social Media Addiction Scale (BSMAS)
using a nationally representative (Hungarian) adolescent sample.
2. To assess the prevalence of problematic social media use in a nationally representative ado-
Participants and procedure
The data were collected in March 2015 as part of the European School Survey Project on Alco-
hol and Other Drugs (ESPAD; ) that included a nationally representative adolescent sam-
ple. The target population was adolescents aged 16 years. In 2015, Hungary included a short
section to assess internet and social media use in addition to the original questionnaire devel-
oped by the ESPAD Committee.
To obtain a representative sample, two different grades (9
) were included in the Hun-
garian data collection, each containing a proportion of the target population. To reduce
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 3 / 13
sampling error, the grades were divided into non-overlapping, homogeneous subgroups. The
variables to ensure the representativeness of the adolescent sample were as follows: region
(central/western/eastern Hungary), grade (9th, 10th), and type of class (secondary general, sec-
ondary vocational, vocational classes). The data were collected anonymously from the students
in the classrooms of the schools by research assistants.
The refusal rate was 7% on the level of the primary sampling unit (i.e., classes) that led to
skewed nonresponse. To match the composition of the respondents with the sampling frame,
data were weighted by strata with the matrix weighting method recommended by the Educa-
tion Information System 2014/2015 (KIR-STAT; Elekes, 2015). The total sample consisted of
6,664 participants (50.94% male). The youngest participants were 15 years old, while the oldest
were 22 years (mean age 16.62 years; SD = 0.96). The wide age range was due to a very small
number of older students still attending the 9
grades at the age of 19 years or older at
the time of data collection. The questions concerning internet use and social media use were
included for this nationally representative sample of 9
graders in secondary general and
secondary vocational schools. Participant data with severe incompleteness or inconsistencies
were excluded (3.72% of the sample), in addition to those participants who did not use the
internet and/or any social media (an additional 6.83% of the original sample). After removing
these participants, the final sample size was 5,961 (89.45% of the total sample).
This study was approved by the Scientific Ethical Committee of Corvinus University of
Budapest. The study design was based on an international protocol approved by the European
School Survey Project on Alcohol and Other Drugs (ESPAD) Assembly, which was conducted
in full compliance with the principles expressed in the Declaration of Helsinki. Written
informed consent was requested from both the students and their parents (passive on behalf of
Socio-demographics questions. Information regarding gender, age, grade, and residence
Weekly social media use. To assess the adolescents’ weekly time spent on social media on
computer or other devices (e.g., handled devices) two variables were combined: (i) ‘The last 7
days how many days did you use the internet for social networking?’ (Categories were ‘never’,
‘1 day’, ‘2 days’, . . .‘7 days’); (ii) ‘In the last 30 days on an average day how many hours did you
use the internet for social networking?’ (Categories were ‘I don’t use’, ‘less than half an hour’,
‘1 hour’, ‘2–3 hours’, ‘4–5 hours’ and ‘more than 6 hours’).
Bergen Social Media Addiction Scale (BSMAS). To assess problematic social media use,
the Bergen Social Media Addiction Scale (BSMAS; ) was used. The 6-item scale was
adapted from the previously validated Bergen Facebook Addiction Scale (BFAS; ). The
original scale specifically assessed problematic Facebook use during the last year. The scale
incorporated the theoretical framework of the addiction components of the biopsychosocial
model . The BFAS was developed by selecting the items with the highest possible factor
loadings for each component (i.e., salience, mood modification, tolerance, withdrawal symp-
toms, conflict, and relapse) from an item-pool of 18 initial items. In the present study, the Ber-
gen Social Media Addiction Scale (BSMAS) which is based on rephrasing of the BFAS, was to
assess social media use in general over the past 12 months. The scale was translated to Hungar-
ian and then back-translated by independent translators. The back-translation was then com-
pared with the original scale and adjustments were made as necessary. The items are answered
on a 5-point scale (“never” to “always”). The Cronbach’s alpha of the translated BSMAS was
.85 in the present sample.
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 4 / 13
Rosenberg’s Self-Esteem Scale. Self-worth was assessed by the Hungarian version of the
Rosenberg’s Self-Esteem Scale (RSES-HU; , Hungarian version, ). The RSES assessed
global self-esteem (i.e., feeling of self-worth and self-acceptance) with 10 items on a 4-point
scale (“strongly agree” to “strongly disagree”). The score range is between 10–40 and the higher
the score, the higher the self-esteem. Cronbach’s alpha was 0.87 in the present sample.
Center of Epidemiological Studies Depression-Scale. Depressive mood was assessed
with the 6-item short-form of the Center of Epidemiological Studies Depression-Scale
(CES-D; ). The scale assesses the level of depressive symptoms but it was not designed to
diagnose clinical depression. The instrument was translated and then back-translated by Hun-
garian experts in the addiction field. The back-translation was then compared to the original
instrument and adjustments were made where necessary. The items of CES-D were answered
on a 4-point scale (“rarely or never” to “most of the time”). The score range is 4–24, and a
higher score indicates higher level of depressive symptoms. Cronbach’s alpha was 0.84 in the
To test the one-factor model of the BSMAS, confirmatory factor analysis (CFA) was per-
formed with maximum likelihood estimation with robust standard error (MLR) in Mplus 7.3
. To evaluate the model fit, a p-value of Chi-square (χ
) higher than .05 was used for the
test of close fit . Additional fit indices were also included: the comparative fit index (CFI),
the Tukey-Lewis Fit Index (TLI), the root mean square error of approximation (RMSEA) and
its 95% confidence interval (90% CI), and standardized root mean square residual (SRMR).
To indicate a good fit of the model, both CFI and TLI values have to be over than .90 or over
.95 , while the values of RMSEA and SRMR should be less than .05 and .10 respectively
In order to identify the groups of adolescents with high risk of problematic social media
use, a mixture modeling technique called latent profile analysis (performed in Mplus 7.3) was
used. Latent profile analysis is a mixture modeling technique to identify groups of people (cate-
gorical output variable of the analysis) according to their responses to certain continuous vari-
ables (in the present study’s case, the scores given on the six items of the BSMAS). Individuals
with similar responses are classified in the same group . Latent profile analysis was per-
formed with 2 to 4 classes in the full sample (n = 5,961). To determine the number of latent
classes, several indices were used, such as the measures of parsimony of each model (i.e.,
Akaike Information Criteria—AIC, Bayesian Information Criteria—BIC, and the Sample Size
Adjusted Bayesian Information Criteria—SSABIC). The lower values on these indicators, the
more parsimonious the model. The entropy criterion and the interpretability of clusters were
also examined. In the final determination of the number of classes, the likelihood-ratio differ-
ence test (Lo-Mendell-Rubin Adjusted LRT Test) was used that statistically compares the fit of
the estimated model with a model having one less class than the estimated one. The p-value of
less than .05 suggests the tested model fits better than the model with one less class .
To test the construct validity of the BSMAS, the LPA classes were compared along a number
of variables relevant to the phenomenon of SOCIAL MEDIA use (i.e., gender, SOCIAL
MEDIA use hours/ week, level of self-esteem, level of depression). For these comparisons,
Wald’s Chi-square test of mean equality for latent class predictors in mixture modeling was
used because it takes into consideration the probabilistic nature of the LPA classes (for descrip-
tion of the analysis, see www.statmodel.com/download/meantest2.pdf).
To determine the optimal cut-off point for the BSMAS, a sensitivity analysis was performed
and the group with the highest risk of problematic social media use (based on the results of the
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 5 / 13
LPA analysis) was considered as the “gold standard”. The present authors are aware that this
method does not replace the clinical validation process, however, the authors believe that this
is better than using completely ad-hoc cut-off points as several other studies do. The sensitiv-
ity, specificity, positive and negative predictive value (PPV and NPV, respectively), and the
accuracy of each cut-off threshold were calculated and compared to identify the cut-off value
with the best indicators. Sensitivity is the proportion of true positive cases belonging to the at-
risk group based on the LPA (the “gold standard” group in this case). Specificity is the propor-
tion of the true negatives among those who do not belong to the at-risk group based on the
LPA [55,56]. PPV is the proportion of true positives among all participants who scored posi-
tive on the test. NPV was defined as the proportion of true negatives among all participants
with negative test results [56,57]. Finally, accuracy measures the proportions of true negatives
and true positives among all participants [56,57].
All analyses were conducted on the weighted sample. Missing data were treated with Full-
information maximum likelihood (FIML) method . Statistical analyses were carried out
with Mplus 7.3  and IBM SPSS Statistics for Windows, Version 22.0 .
The final sample only comprised those participants who reported using the internet and social
media (n = 5,961, 89.45% of the total sample). Approximately half (49.17%) of the sample was
male (n = 2931). Age ranged between 15 and 22 years (mean age 16.60; SD = 0.94). The mean
number of hours using social media was 23.16 hours per week (SD = 15.57). There was a sig-
nificant difference in weekly social media use between male and female adolescents (mean
= 20.53 hours, SD
= 15.71; mean time
= 25.71 hours, SD
U = 3672101, p<0.001; r = -0.17).
Confirmatory factor analysis
A one-factor model with the six components (salience, tolerance, mood modification, relapse,
withdrawal, conflict) as indicator variables was tested with confirmatory factor analysis. The
analysis provided an acceptable fit to the data (χ
= 5836.190 df = 15 p<0.001; CFI = 0.950;
TLI = 0.917; RMSEA = 0.073 (0.066–0.080) Cfit>0.90; SRMR = 0.034). All factor loadings
were above the recommended threshold (>.50) and ranged from .598 to .814.
Latent profile analysis
The latent profile analysis was performed on the six items of the BSMAS, and according to the
criteria, the three-class solution was selected as the best-fitting model (see Table 1). The AIC,
BIC, and SSABIC values decreased continuously as more classes were added to the analysis.
However, the scale of decrease somewhat diminished after the third latent class was added.
Based on the L-M-R test, the three-class solution was accepted. The entropy of the two-class
solution was the highest, but the entropy of the three-class solution was also adequate.
The features of the three classes are presented in Fig 1 and Table 2. The first class named
‘no-risk’ class represents the majority of social media users (78.3% of social media users; 70.7%
of the total sample) who had the lowest scores on the BSMAS. The second class of social media
users represents ‘low risk’ of problematic use (17.2% and 15.5% respectively), while the third
class represents the population of ‘at-risk’ problematic social media users (4.5% and 4.1%,
respectively). In the ‘at-risk’ group, ‘withdrawal’ and ‘tolerance’ criteria showed elevated levels
compared to the other dimensions. Members of this class (i.e., those at-risk of problematic
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 6 / 13
use) were likely to (a) be female, (b) use the internet and social media for more than 30 hours
per week, (c) have lower self-esteem and higher level of depressive symptoms than social
media users of the other two classes (Table 2).
Suggesting a cutoff score for classification: Sensitivity and specificity
Because of the lack of a clinically diagnosed group of problematic social media users, the third
LPA class (i.e., those at-risk of problematic social media use) was used as the ‘gold standard’ to
determine the optimal cut-off threshold to classify those at-risk of problematic use. Sensitivity,
specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of
the BSMAS at all possible cut-off points were calculated (Table 3). Based on this analysis, a
Table 1. Results of the Latent Profile Analysis.
Fit indices for the Latent Proﬁle Analysis (LPA) of the social media use
Model Log-likelihood Replicated log-likelihood Nr. of free parameters AIC BIC SSABIC Entropy LMR-LRT test p
2 classes -43837 Yes 19 87711 87837 87778 0.96 12838 <0.0001
3 classes -42241 Yes 26 84534 84708 84626 0.94 3140 <0.05
4 classes -41097 Yes 33 82260 82481 82376 0.95 2251 0.69
Note: AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion, SSABIC = sample size adjusted BIC, LMR-LRT = Lo–Mendell–Rubin
Likelihood Ratio Test. Bold data indicate that the three-class solution was selected as a result of the LPA analysis.
Fig 1. The Three Classes Obtained from the Latent Profile Analysis.
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 7 / 13
cut-off score of 19 points was suggested as the ideal threshold at and above which individuals
are classified as at-risk of problematic social media use.
In this case, the specificity is 99% and the sensitivity is 83% (i.e., only 1% of the non-prob-
lematic social media users are identified incorrectly as being at-risk of problematic use by the
scale, while 17% of true cases of problematic social media users are missed). At this value, PPV
is 73% and the NPV is 99%. In other words, 27% of the individuals with a positive test result
are identified incorrectly, while only 1% of individuals with negative test result are identified
incorrectly. This yields an accuracy of 98%. Increasing the cut-off point would result in more
false negative cases, while decreasing it would increase the number of social media users
labeled incorrectly (as being at-risk) by the screening instrument.
To assess the prevalence of problematic social media use in a reliable and a valid way, the psy-
chometric properties of the BSMAS were tested. According to the results, the BSMAS demon-
strated adequate psychometric properties regarding its factor structure, reliability, and validity.
Table 2. Comparison of the Three Latent Classes: Testing Equality for Latent Class Predictors.
No risk class
(n = 4712)
Low risk class
(n = 1035)
(n = 271)
Gender (male %) 50.36
Age (years); Mean (SE) 16.60 (0.02)
Weekly internet use (min 0.5, max 42 hours, mean 23.49, SD
12.73); Mean (SE)
Weekly social media use (min 0.5. max 42 hours, mean 23.13, SD
15.56); Mean (SE)
Self-esteem (min 1, max 4, mean 2.73, SD 0.61); Mean (SE) 2.79 (0.01)
Level of depressive symptoms (min 1, max 4, mean 1.93, SD
0.60); Mean (SE)
Note: Different subscript letters (a, b, c) in the same row reﬂect signiﬁcant (p<0.05) difference between the means while same subscript letters in one row
reﬂect non-signiﬁcant difference between the means according to pair wised Wald χ
test of mean equality for latent class predictors in mixture modeling
Table 3. Cut-off points based on the third class (i.e., those at-risk of problematic social media use) derived from the Latent Profile Analysis.
Cut-off points True positive True negative False positive False negative Sensitivity (%) Speciﬁcity (%) PPV (%) NPV (%) Accuracy (%)
12 243 4304 1224 0 100 78 17 100 79
13 243 4635 895 0 100 84 21 100 84
14 243 4823 701 0 100 87 26 100 88
15 243 4986 539 0 100 90 31 100 91
16 240 5141 386 3 99 93 38 100 93
17 232 5249 278 9 96 95 45 100 95
18 219 5340 188 23 90 97 54 100 96
19 199 5458 74 40 83 99 73 99 98
20 177 5503 29 64 73 99 86 99 98
21 156 5517 17 85 65 100 90 98 98
22 126 5527 8 114 53 100 94 98 98
23 107 5530 4 133 45 100 96 98 98
Note: Bold data indicate that the cut-off score of 19 (and above) was selected as a result of the sensitivity and speciﬁcity analysis. PPV = positive predictive
value; NPV = negative predictive value
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 8 / 13
Using a latent profile analysis on the six items of the Bergen Social Media Addiction Scale
(BSMAS), the adolescent social media users were divided into three different classes, and the
analysis demonstrated that 4.5% of participants could be classified as being at-risk. Previous
studies have shown a wide range of prevalence rates due to various methodological issues such
as convenience sampling, targeting mainly college students, and/ or having small sample sizes
[15,12,26,42,43]. For instance, the prevalence of problematic social media users among Nige-
rian University undergraduates was 1.6% , whereas among Malaysian college students the
reported prevalence was 47% . The results of the present study belong to the more conser-
vative prevalence estimations. Moreover, they correspond to the prevalence rates of the general
problematic and/or addictive Internet use, that range between 1%  and 18.7%  accord-
ing to the recent review .
Regarding validity of the BSMAS, the at-risk group showed the lowest self-esteem and the
highest level of depressive symptoms and the most time spent on internet and social media
use, and was therefore in line with previous research findings [22,24].
In addition, adolescents that were at-risk of social media use were mainly female, and
reported the greatest amount of internet and social media usage. Previous studies have found
similar gender differences in problematic social media use [4,6] and problematic internet use
. For instance, Rehbein and Mo¨ßle  found that among adolescents with problematic
internet use, girls indicated that social media contributed most to their addiction, while boys
also cited online pornography as a primary source of their problems.
Furthermore, the results of the present study showed that within the at-risk group the
withdrawal component had the highest score. Therefore, withdrawal symptoms should be
highlighted when developing prevention and treatment programs in school environments for
adolescents being at-risk of problematic social media use. A cut-off point was calculated using
the third LPA class as the “gold standard” to categorize the risky or problematic social media
users among adolescents in the sample. The suggested cut-off value with the most adequate
sensitivity and specificity values was 19 points. Although, the calculated score cannot replace a
clinically validated cut-off point, it may be more beneficial than using completely ad-hoc
Despite the study’s strengths (most notably the large nationally representative sample using
psychometrically validated instruments), the study is not without its limitations. The study
only included Hungarian adolescents as participants. Therefore, to further test the psychomet-
ric properties of the BSMAS, cross-cultural studies should also be conducted in the future
using different adolescent groups in different countries and cultures. Moreover, the data were
all self-report and self-report measures may lead to different response biases , such as social
desirability bias (e.g., reporting more favorable behavior than the truth), memory recall bias
(difficulty in remembering past events), and response style bias (e.g., scores may show central
tendency or extreme response style).
It is also important to highlight that psychometric screening tools tend to overestimate the
prevalence rates of disorders when the true prevalence rate of the disorder is low. For instance,
an instrument with moderate sensitivity and specificity (i.e., 80.5% and 82.4%, respectively) at
a 2.1% prevalence level of the problematic behavior has a positive predictive value of 8.9%.
This means that out of 100 respondents who score positive on the test only about 9 are true
clinical cases . Consequently, survey-based prevalence rates should be interpreted cau-
tiously to avoid overpathologizing every day behaviors . More generally, the issue of addic-
tion to social networking and social media is a controversial issue and many papers have
questioned whether the activity can be considered an addiction at all [4,27,66–68].
The study had a cross-sectional design, therefore causality cannot be established regarding
the risk factors. Future research should apply longitudinal designs to identify the contributing
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 9 / 13
factors of the problematic behavior among social media users. It should also be noted that
when completing the BSMAS, the participants may have had a different conception of social
media use than intended by the developers of the BSMAS. For instance, on sites such as Face-
book, many different activities can be carried out such as social networking, gaming and gam-
bling. Although the BSMAS is only concerned with social networking, there is always the
possibility that participants’ conception of social media use included some or all of these other
activities and therefore problematic use might be including non-social networking activities.
In conclusion, the results of the present study suggest that the Bergen Social Media Addiction
Scale  is a psychometrically valid scale that is an appropriate tool to identify the signs of
risky social media use among adolescents. This instrument may be especially useful in school
environments to identify those adolescents who are at-risk of problematic social media use
and therefore could be utilized in prevention and intervention programs (i.e., content-control
software, counseling, cognitive-behavioral therapy; ).
This study was supported by the Hungarian National Research, Development and Innovation
Office (Grant numbers: K111938, K111740). A
´gnes Zsila was supported by the New National
Excellence Program awarded by the Ministry of Human Resources.
The funding institutions had no role in the study design or the collection, analysis and
interpretation of the data, writing the manuscript, or the decision to submit the paper for
Conceptualization: ZD MDG OK.
Data curation: ZE FB A
´Z AM OK.
Formal analysis: FB A
´Z OK AM.
Funding acquisition: ZE ZD.
Investigation: ZE ZD AM.
Methodology: ZE ZD AM CSA OK.
Project administration: ZE ZD.
Resources: ZD ZE.
Software: ZE ZD.
Supervision: ZE ZD.
Validation: ZE ZD.
Writing – original draft: FB A
´Z OK AM.
Writing – review & editing: MDG CSA ZD ZE.
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 10 / 13
1. Lenhart A, Purcell K, Smith A, Zickuhr K. Social Media & Mobile Internet Use among Teens and Young
Adults. Washington, DC: Pew Internet & American Life Project; 2010.
2. Lenhart A, Duggan M, Perrin A, Stepler R, Rainie H, Parker K. Teens, social media and technology
overview 2015. Smartphones facilitate shifts in communication landscape for teens. Washington, DC:
Pew Internet & American Life Project; 2015.
3. Ahn J. The effect of social network sites on adolescents’ social and academic development: Current
theories and controversies. Journal of the American Society for Information Science and Technology.
4. Kuss DJ, Griffiths MD. Online social networking and addiction—a review of the psychological literature.
International Journal of Environmental Research and Public Health. 2011; 8(9):3528–52. doi: 10.3390/
ijerph8093528 PMID: 22016701
5. Boyd DM, Ellison NB. Social network sites: Definition, history, and scholarship. Journal of Computer-
Mediated Communication. 2007; 13(1):210–30.
6. Kuss DJ, Griffiths MD. Excessive online social networking: Can adolescents become addicted to Face-
book. Education and Health. 2011; 29(4):68–71.
7. Allen KA, Ryan T, Gray DL, McInerney DM, Waters L. Social media use and social connectedness in
adolescents: The positives and the potential pitfalls. The Australian Educational and Developmental
Psychologist. 2014; 31(01):18–31.
8. Ryan T, Chester A, Reece J, Xenos S. The uses and abuses of Facebook: A review of Facebook addic-
tion. Journal of Behavioral Addictions. 2014; 3(3):133–48. doi: 10.1556/JBA.3.2014.016 PMID:
9. Griffiths MD. Adolescent gambling and gambling-type games on social networking sites: Issues, con-
cerns, and recommendations. Aloma: Revista de Psicologia, Ciències de l’Educacio
´i de l’Esport. 2015;
10. Andreassen CS, Torsheim T, Brunborg GS, Pallesen S. Development of a Facebook addiction scale 1,
2. Psychological Reports. 2012; 110(2):501–17. doi: 10.2466/02.09.18.PR0.110.2.501-517 PMID:
11. Andreassen CS, Pallesen S, Griffiths MD. The relationship between addictive use of social media, nar-
cissism, and self-esteem: Findings from a large national survey. Addictive Behaviors. in press.
12. Wilson K, Fornasier S, White KM. Psychological predictors of young adults’ use of social networking
sites. Cyberpsychology, Behavior, and Social Networking. 2010; 13(2):173–7.
13. Griffiths MD, Kuss DJ, Demetrovics Z. Social networking addiction: An overview of preliminary findings.
Behavioral Addictions Criteria, Evidence, and Treatment. New York: Elsevier; 2014. p. 119–41.
14. Griffiths MD. Social networking addiction: Emerging themes and issues. Journal of Addiction Research
and Therapy. 2013; 4(5):1–2.
15. Wan C. Gratifications & loneliness as predictors of campus-SNS websites addiction & usage pattern
among Chinese college students. Hong Kong: Chinese University of Hong Kong, China. 2009.
16. Wegmann E, Stodt B, Brand M. Addictive use of social networking sites can be explained by the interac-
tion of Internet use expectancies, Internet literacy, and psychopathological symptoms. Journal of
Behavioral Addictions. 2015; 4(3):155–62. doi: 10.1556/2006.4.2015.021 PMID: 26551905
17. Hormes JM. Under the influence of Facebook? Excess use of social networking sites and drinking
motives, consequences, and attitudes in college students. Journal of Behavioral Addictions. 2016; 5
18. Steers M-LN, Wickham RE, Acitelli LK. Seeing everyone else’s highlight reels: How Facebook usage is
linked to depressive symptoms. Journal of Social and Clinical Psychology. 2014; 33(8):701.
19. Muench F, Hayes M, Kuerbis A, Shao S. The independent relationship between trouble controlling
Facebook use, time spent on the site and distress. Journal of Behavioral Addictions. 2015; 4(3):163–9.
doi: 10.1556/2006.4.2015.013 PMID: 26551906
20. Wu AM, Cheung VI, Ku L, Hung EP. Psychological risk factors of addiction to social networking sites
among Chinese smartphone users. Journal of Behavioral Addictions. 2013; 2(3):160–6. doi: 10.1556/
JBA.2.2013.006 PMID: 25215198
21. Lee ZW, Cheung CM, Thadani DR, editors. An Investigation into the Problematic Use of Facebook.
System Science (HICSS), 2012 45th Hawaii International Conference; 2012 4–7 January; Maui, HI:
22. Pantic I, Damjanovic A, Todorovic J, Topalovic D, Bojovic-Jovic D, Ristic S, et al. Association between
online social networking and depression in high school students: behavioral physiology viewpoint. Psy-
chiatria Danubina. 2012; 24(1):90–3.
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 11 / 13
23. Andreassen CS, Billieux J, Griffiths MD, Kuss DJ, Demetrovics Z, Mazzoni E, et al. The relationship
between addictive use of social media and video games and symptoms of psychiatric disorders: A
large-scale cross-sectional study. Psychology of Addictive Behaviors. 2016; 30(2):252–62. doi: 10.
1037/adb0000160 PMID: 26999354
24. Malik S, Khan M. Impact of facebook addiction on narcissistic behavior and self-esteem among stu-
dents. Journal of Pakistan Medical Association. 2015; 65(3):260–3.
25. Association AP. Diagnostic and Statistical Manual of Mental Disorders (DSM-5
). Fifth ed. Arlington,
VA: American Psychiatric Association; 2013.
26. Wolniczak I, Caceres-DelAguila JA, Palma-Ardiles G, Arroyo KJ, Solı
´s-Visscher R, Paredes-Yauri S,
et al. Association between Facebook dependence and poor sleep quality: a study in a sample of under-
graduate students in Peru. PLoS ONE. 2013; 8(3):e59087. doi: 10.1371/journal.pone.0059087 PMID:
27. Griffiths MD. Facebook addiction: concerns, criticism, and recommendations—a response to Andreas-
sen and colleagues. Psychological Reports. 2012; 110(2):518–20. doi: 10.2466/01.07.18.PR0.110.2.
518-520 PMID: 22662405
28. Saaid SA, Al-Rashid NAA, Abdullah Z. The impact of addiction to Twitter among university students.
Future Information Technology. New York: Springer; 2014. p. 231–6.
29. van den Eijnden RJ, Lemmens JS, Valkenburg PM. The Social Media Disorder Scale: Validity and psy-
chometric properties. Computers in Human Behavior. 2016; 61:478–87.
30. Turel O, Serenko A. The benefits and dangers of enjoyment with social networking websites. European
Journal of Information Systems. 2012; 21(5):512–28.
31. Griffiths MD. A ‘components’ model of addiction within a biopsychosocial framework. Journal of Sub-
stance Use. 2005; 10(4):191–7.
32. Young KS. Internet addiction: The emergence of a new clinical disorder. CyberPsychology & Behavior.
33. Armstrong L, Phillips JG, Saling LL. Potential determinants of heavier Internet usage. International Jour-
nal of Human-Computer Studies. 2000; 53(4):537–50.
34. Morahan-Martin J, Schumacher P. Incidence and correlates of pathological Internet use among college
students. Computers in human behavior. 2000; 16(1):13–29.
35. Ehrenberg A, Juckes S, White KM, Walsh SP. Personality and self-esteem as predictors of young peo-
ple’s technology use. CyberPsychology & Behavior. 2008; 11(6):739–41.
36. Walsh SP, White KM, Young RM. Young and connected: Psychological influences of mobile phone use
amongst Australian youth. In: Goggin G, Hjorth L, editors. Proceedings Mobile Media 2007: University
of Sydney; 2007. p. 125–34.
37. Alabi OF. A survey of Facebook addiction level among selected Nigerian University undergraduates.
New Media and Mass Communication. 2013; 10:70–80.
38. C¸am E, Isbulan O. A New Addiction for Teacher Candidates: Social Networks. Turkish Online Journal
of Educational Technology. 2012; 11(3):14–9.
39. Elphinston RA, Noller P. Time to face it! Facebook intrusion and the implications for romantic jealousy
and relationship satisfaction. Cyberpsychology, Behavior, and Social Networking. 2011; 14(11):631–5.
40. Andreassen CS. Online social network site addiction: A comprehensive review. Current Addiction
Reports. 2015; 2(2):175–84.
41. McAndrew FT, Jeong HS. Who does what on Facebook? Age, sex, and relationship status as predictors
of Facebook use. Computers in Human Behavior. 2012; 28(6):2359–65.
42. Zhou SX. Gratifications, loneliness, leisure boredom and self-esteem as predictors of SNS-game addic-
tion and usage pattern among Chinese college students: Hong Kong: The Chinese University of Hong
43. Jafarkarimi H, Sim ATH, Saadatdoost R, Hee JM. Facebook Addiction among Malaysian Students.
International Journal of Information and Education Technology. 2016; 6(6):465–9.
44. Olowu AO, Seri FO. A study of social network addiction among youths in Nigeria. Journal of Social Sci-
ence and Policy Review. 2012; 4:62–71.
´ly O, Zsila A
´, Demetrovics Z. Viselkede
´k. In: Elekes Z, editor. Euro
´l– 2015; Magyarorsza
´nyek. Budapest: Buda-
pesti Corvinus Egyetem; Ta
¨zi Kapcsolatok Kar; Szociolo
´zet; 2016. p. 75–94.
46. Rosenberg M. Society and the Adolescent Self-Image. Princeton, NJ: Princeton University Press;
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 12 / 13
´n R, Szigeti R, Ko
¨nyei G, Demetrovics Z. Global self-esteem and method effects: Competing
factor structures, longitudinal invariance, and response styles in adolescents. Behavior Research Meth-
ods. 2014; 46(2):488–98. doi: 10.3758/s13428-013-0391-5 PMID: 24061931
48. Radloff LS. The CES-D scale a self-report depression scale for research in the general population.
Applied Psychological Measurement. 1977; 1(3):385–401.
´n LK, Muthe
´n BO. Mplus User’s Guide ( 7th edn). Los Angeles, CA: Muthe
´n and Muthe
50. Barrett P. Structural equation modelling: Adjudging model fit. Personality and Individual differences.
51. Lt Hu, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria
versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal. 1999; 6(1):1–55.
52. Steiger JH. Understanding the limitations of global fit assessment in structural equation modeling. Per-
sonality and Individual Differences. 2007; 42(5):893–8.
53. Byrne BM. Structural equation modeling with AMOS: Basic concepts, applications, and programming.
London: Routledge; 2013.
54. Collins LM, Lanza ST. Latent class and latent transition analysis: With applications in the social, behav-
ioral, and health sciences. Chichester: John Wiley & Sons; 2013.
55. Altman DG, Bland JM. Diagnostic tests. 1: Sensitivity and specificity. British Medical Journal. 1994; 308
(6943):1552. PMID: 8019315
56. Glaros AG, Kline RB. Understanding the accuracy of tests with cutting scores: The sensitivity, specific-
ity, and predictive value model. Journal of Clinical Psychology. 1988; 44(6):1013–23. PMID: 3216006
57. Altman DG, Bland JM. Statistics Notes: Diagnostic tests 2: Predictive values. British Medical Journal.
1994; 309(6947):102. PMID: 8038641
58. Released IC. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.; 2013.
59. Rumpf H-J, Vermulst AA, Bischof A, Kastirke N, Guertler D, Bischof G, et al. Occurence of internet
addiction in a general population sample: a latent class analysis. European Addiction Research. 2013;
20(4):159–66. doi: 10.1159/000354321 PMID: 24401314
60. Lin I-H, Ko C-H, Chang Y-P, Liu T-L, Wang P-W, Lin H-C, et al. The association between suicidality and
Internet addiction and activities in Taiwanese adolescents. Comprehensive psychiatry. 2014; 55
(3):504–10. doi: 10.1016/j.comppsych.2013.11.012 PMID: 24457034
61. Pontes HM, Kuss D, Griffiths M. Clinical psychology of Internet addiction: A review of its conceptualiza-
tion, prevalence, neuronal processes, and implications for treatment. Neuroscience and Neuroeco-
nomics. 2015; 4:11–23.
62. Rehbein F, Mo
¨ßle T. Video game and Internet addiction: is there a need for differentiation? Sucht.
´ly O, Nagygyo
¨rgy K, Koronczai B, Griffiths MD, Demetrovics Z. Assessment of problematic internet
use and online video gaming. In: Aboujaoude E, Starcevic V, editors. Mental health in the digital age:
Grave dangers, great promise. New York: Oxford University Press; 2015. p. 46–68.
64. Choi BC, Pak AW. A catalog of biases in questionnaires. Preventing Chronic Disease. 2005; 2(1):A13.
´z A, Kira
´ly O, Demetrovics Z. The diagnostic pitfalls of surveys: if you score positive on a test of
addiction, you still have a good chance not to be addicted. A response to Billieux et al. 2015. Journal of
Behavioral Addictions. 2015; 4(5):151–4.
66. Billieux J, Schimmenti A, Khazaal Y, Maurage P, Heeren A. Are we overpathologizing everyday life? A
tenable blueprint for behavioral addiction research. Journal of Behavioral Addictions. 2015; 4(3):119–
23. doi: 10.1556/2006.4.2015.009 PMID: 26014667
67. Carbonell X, Panova T. A critical consideration of social networking sites’ addiction potential. Addiction
Research and Theory. 2016 Epub ahead of print. http://doi.org/10.1080/16066359.2016.1197915
68. Ryan T, Chester A, Reece J, Xenos S. The uses and abuses of Facebook: A review of Facebook addic-
tion. Journal of Behavioral Addictions. 2014; 3(3): 133–148. doi: 10.1556/JBA.3.2014.016 PMID:
69. Gupta VK, Arora S, Gupta M. Computer-related illnesses and Facebook syndrome: what are they and
how do we tackle them. Medicine Update. 2013; 23:676–9.
Social Media Use in Adolescence
PLOS ONE | DOI:10.1371/journal.pone.0169839 January 9, 2017 13 / 13