Journal of Affective Disorders 275 (2020) 165–174

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Review article

A meta-analysis of the association between adolescent social media use and T

depressive symptoms
Elizabeth J. Ivie, Adam Pettitt, Louis J. Moses, Nicholas B. Allen
⁎

Department of Psychology, University of Oregon, Eugene, OR, USA

ARTICLE INFO ABSTRACT

Keywords: Background: The association of adolescent social media use with mental health symptoms, especially depression,
Meta-analysis has recently attracted a great deal of interest in public media as well as the scientific community. Some studies
Adolesence have cited statistically significant associations between adolescent social media use and depression and have
Depression proposed that parents must regulate their adolescents’ social media use in order to protect their mental health.
Social media
Method: In order to rigorously assess the size of the effect that has been reported in the current scientific lit-
erature, we conducted a meta-analysis of studies that measured the association between social media use spe-
cifically and depressive symptoms amongst early- to mid- adolescents (11-18 years-old). We searched Psychnet,
PubMed, and Web of Science with the following terms: online social networks, social media, internet usage,
facebook, twitter, instagram, myspace, snapchat, and depression.
Results: We found a small but significant positive correlation (k=12 studies, r=.11, p<.01) between adolescent
social media use and depressive symptoms. There was also high heterogeneity (I2=95.22%) indicating sub-
stantial variation among studies.
Conclusions: High heterogeneity along with the small overall effect size observed in the relationship between
self-reported social media use and depressive symptoms suggests that other factors are likely to act as significant
moderators of the relationship. We suggest that future research should be focused on understanding which types
of use may be harmful (or helpful) to mental health, rather than focusing on overall use measures that likely
reflect highly heterogeneous exposures.

Introduction in 2015, 76% of adolescents use social media, and that percentage has
likely increased since this poll was taken (Lenhart et al., 2015). Paral-
Today's youth are growing up with ready access to online social leling these increases in engagement with social media, the prevalence
media via computers, smartphones, and tablets. This novel way of in- of adolescent major depressive episodes (MDEs) has increased from
teracting with others has rapidly become widespread, and considerable 8.7% in 2005 to 11.3% in 2014 (Mojtabai et al., 2016). This increase in
attention has been given to the potential negative effects that adoles- the incidence of depression in adolescence has become a major public
cents’ use these devices may have on youth mental health, particularly health concern, and some have asserted a correlation between the rise
depression (Baker and Algorta, 2016). Recently, it has been argued that in social media use and the rise in depression (e.g., (Pantic et al., 2012,
it may be social media use specifically, rather than digital device use or Twenge et al., 2018)).
screen time more generally, that is most strongly associated with risk On the other hand, some research has found potential benefits of
for depression (Haidt and Allen, 2020). This is especially relevant for social media use in adolescence, such as better self-esteem and per-
adolescents, as the first incidence of a major depressive episode is most ceived social support (Valkenburg et al., 2006). Systematic reviews and
likely to occur during adolescence and young adulthood meta-analyses have attempted to tease these seemingly contradictory
(Avenevoli et al., 2015), and some data suggest that the incidence of ideas apart. One systematic review focused specifically on adolescents
depression has been rising in recent cohorts of adolescent following the and the associations between social media use and well-being, and
widespread adoption of social media (Haidt and Allen, 2020). As such, determined that there appear to be both positive and negative asso-
identifying potential risk factors during this period of life is an espe- ciations, (date range of studies reviewed: January 2003-April 2013)
cially important endeavor. According to a Pew Research poll conducted (Best et al., 2014). Some of the studies reviewed showed benefits to

⁎
Corresponding author.
E-mail address: [email protected] (N.B. Allen).

https://doi.org/10.1016/j.jad.2020.06.014
Received 18 November 2019; Received in revised form 11 June 2020; Accepted 17 June 2020
Available online 24 June 2020
0165-0327/ © 2020 Elsevier B.V. All rights reserved.
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

Fig. 1. PRISMA flowchart of literature search.

well-being, such as increased self-esteem and social support (Best et al., loneliness (Dienlin et al., 2017). They found that increases in online
2014). However, others illustrated negative associations with social communication was related to increases in face-to-face interaction and a
media use, such as social isolation and depression (Best et al., 2014). A small increase in life satisfaction, however, no association was found
recent nationally representative longitudinal cohort study of adoles- between online communication and changes in loneliness.
cents (n=6595) found that adolescents who self-reported the use of The current study takes the investigation of these issues a step
social media platforms more than three hours per day were more likely further by both reviewing the literature on adolescent social media use
to self-report internalizing symptoms at a subsequent follow up as- and depressive symptoms specifically, and by analyzing the reported
sessment approximately one year later (Riehm et al., 2019). A meta- statistical effects of the relationship using the tool of meta-analysis. We
analysis has also investigated the relationship between social net- chose to focus on depressive symptoms because although many studies
working site usage and depression in adults and found a small, yet have investigated the relationship between social media use and well-
significant positive effect (date range of studies reviewed: any time- being during adolescence (e.g., (Best et al., 2014)), relatively few have
February 2018) (Yoon et al., 2019). However, another recent study addressed depressive symptomatology specificially. While well-being
combined three large datasets (n=355,358) and found there to be little and depression are inversely correlated, they are not identical, and
evidence of the link between social media use and well-being depression measures tap into phenomena that are directly relevant to
(Orben and Przybylski, 2019a). A second systematic review in- risk for mental health disorders; including (but not limited to) Major
vestigating the relationship between online social networking and de- Depressive Disorder, which often emerges during adolescence. Al-
pression amongst adults and adolescents determined that, overall, there though there have been previous narrative reviews (e.g., (Baker and
appear to be both benefits and drawbacks to social media usage (date Algorta, 2016)) and one meta-analysis with an adult sample
range of studies reviewed: any time-April 2016) (Baker and (Yoon et al., 2019), this is the first study to use meta-analytic techni-
Algorta, 2016). A study used a two-time point longitudinal design as- ques to examine this association amongst adolescents specifically.
sessed the relationship between different types of online communica- Based on previous findings (Baker and Algorta, 2016, Best et al., 2014),
tion and changes in life satisfaction, face-to-face interactions, and we hypothesized a small positive overall effect for the relationship

166
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

between social media use and depressive symptoms, but a high degree measure(s) of depression 7) type of statistical tests (e.g., correlation) 8)
of heterogeneity of effects across studies. reported statistical value (correlation). Studies were excluded for the
following reasons: 1) no measure of time spent on social media 2) no
2. Methods continuous measure of depressive symptoms 3) not available in English
4) participants older than 18 years-old. Studies that included partici-
Inclusion and exclusion criteria pants who were younger than 11 or older than 18 were excluded.
Studies that did not include a continuous measure of time spent on
We included correlational studies that used both continuous mea- social media or depressive symptoms were excluded in order to increase
sures of depressive symptoms and time spent using social media. the compatibility of methodology and statistical comparison in the
Participants in the studies were adolescents between the ages of 11-18 meta-analysis. We chose to limit our study to continuous measurement
years. Studies must have been available in English. There were no because information about the variance or range in experience is lost
constraints on time period or geographical region on study eligibility. when grouping participants’ responses (e.g., depressed/non-depressed,
problematic social media use/non-problematic social media use)
(Young, 2016).
Search strategy
Statistical methods
We searched the following databases on July 21st, 2017: Psychnet:
online social networks or social media or internet usage; AND depres- Correlations (r) between social media usage and depressive symp-
sion (index terms search), PubMed: "online social networks" OR "social toms were extracted from each study. Correlations (r) were converted
media" Or twitter or myspace or instagram or snapchat or facebook and with the Fisher's Z transformation for analysis. An inverse variance,
depression (all fields), and Web of Science: "online social networks" or restricted maximum likelihood random effects meta-analysis, with the
"social media" or "internet usage" or Facebook or Twitter or Instagram Knapp-Hartung Adjustment to account for small k, was conducted in
or Myspace or Snapchat (all fields). The reference sections of eligible order to account for both within- and between-studies variation
studies were also searched, however, no additional studies were in- (Knapp and Hartung, 2003). The Knapp-Hartung Adjustment employs t
cluded via this method. Studies suggested by Mendeley software that values rather than Z values and is therefore a more conservative esti-
met the inclusion criteria and were published after the search date were mate of the overall effect. Heterogeneity was assessed with Cochran's Q
also reviewed. We completed an updated search on January 17th, 2020 and I2. Publication bias was assessed with visual inspection of funnel
with a restricted date range from January 1st, 2017 to present. We re- plot. In addition, a p-curve analysis was conducted to assess for po-
viewed an additional 1,123 studies. One additional study met the in- tential publication bias based on the distribution of statistically sig-
clusion/exclusion criteria (Thorisdottir et al., 2019). We filtered the nificant p-values. Analyses were performed in R (version 3.3.0) using
results to limit studies conducted with adolescents aged 11-18 years the metafor package (Viechtbauer, 2010).
old. See Figure 1 for flowchart showing the phases of the literature
search and decisions for inclusion and exclusion. Results

Coding procedures Eleven studies, ranging between 2012-2020, with a total of 92,371
participants were included in the meta-analysis (see Table 2 for study
Studies were examined for eligibility by two independent reviewers. characteristics). The random effects meta-analysis with Knapp-Hartung
The reviewers resolved disagreement through discussion after in- Adjustment (k=12) produced a significant small positive effect of the
dependent coding of study characteristics (n=22 out of 1224). A relationship between depressive symptoms and social media usage (r
Cohen's kappa was conducted to assess intercoder reliability (k=.685, =.12, SE=.04, t=3.44, 95% CI= .04-.20, p<.01; Table 2; see Figure 2
p<.001). The quality of studies was assessed with questions from the for forest plot). There was high heterogeneity (Q(df=11) = 362.38,
NHLBI: Quality Assessment Tool for Observational Cohort and Cross- p<.001; I2=97.38%). The statistically significant finding for Q (Q
Sectional Studies that apply to correlational study design (see Table 1 (df=11) = 362.38, p<.001 indicates considerable variation in the true
for list of quality assessment questions; (National Institutes of effect estimates among studies (Table 3). I2 =97.38% is considered
Health, 2014)). Quality in relation to each of nine questions was rated very high and indicates that there are inconsistencies between findings.
as good (2), fair (1), or poor (0) by each independent reviewer. Quality We separately tested mean age and sample size as moderators in the
rating ranged from 0-18, with higher scores indicating better quality. A analysis. Both mean age and sample size accounted for 0% of the var-
standardized data coding system was developed to extract qualitative iance between studies.
and statistical information from each study: Study characteristics: 1) Visual inspection of the funnel plot (Figure 3) illustrated random
authors and citation; 2) sample size and description of the sample (e.g., dispersion of effect sizes among studies, suggesting the absence of small
mean age, ethnicity), 3) main aim of the study, 4) study design (e.g., study bias (and, arguably, little publication bias). We also conducted a
cross-sectional, longitudinal) 5) measure(s) of social media use 6) p-curve analysis to assess publication bias and evidence for a true effect
(citation). Figure 4 illustrates a right-skewed distribution, which in-
Table 1 creased our confidence that the estimate of the overall effect between
Quality assessment. depressive symptoms and social media usage is indicative of a true ef-
Was the research question/objective/hypothesis/aim in this paper clearly stated? fect, albeit a small one.
Was the study population clearly specified and defined?
Were the participants representative of the population of interest? Discussion
Were the characteristics (demographics) of participants provided?
Were all the subjects selected or recruited from the same or similar populations
(including the same time period)? Given that there were relatively few studies that have specifically
Were the measures clearly defined, valid, reliable, and implemented consistently reported findings on the relationship between social media use and
across all study participants? depressive symptoms in adolescence, these findings should be inter-
Are the main findings of the study clearly described? preted cautiously. However, based on the current evidence, there ap-
Were the statistical tests appropriate?
pears to be a small but positive association between social media use
Were key confounding variables measured or addressed on the impact they could
have on outcomes? and depressive symptoms. However, there was high heterogeneity,
which suggests that there are substantial differences between studies.

167
 Table 2
Summary of studies included in meta-analysis on the relationship between social media use and depressive symptoms.
Study ID Authors Publication year Study Correlation (r) Fisher's Z Study design Social media use measure Depressive symptoms Mean age Gender Demographics Methodological
E.J. Ivie, et al.

sample measure (SD); Range ratio (% quality (0-18)

size female)

1 Banjanin et al. 2015 336 0.05 0.05 Cross-sectional Single item frequency: CES-DC 18 (NR); NR 67 Serbian high school 15
"Particpants were also students
asked questions about
their average time spent
on social networking."
M=3.11 hours, SD=2.82
hours
2 Barry et al. 2017 226 0.24 0.2448 Cross-sectional Single item frequency: 82 item DSM-5 checklist 15.27 (1.02); 45.1 81.4% White/Caucasian, 17
times social media was 15-17 6.2% Black/African-
checked per day American, 2.7% Asian,
(“Facebook,” “Twitter,” 2.7% Hispanic, 0.9%
“Snapchat,” “Tumblr,” Native American/
“Instagram,” “Other”), 8- American Indian, 5.3%
item scale: never to more Multi-racial, and 0.9%
than 10 times per day “Other.”; Median
household
income=$60,000
3 Blomfield 2014 1,819 -0.09 -0.09 Cross-sectional Composite of 7 items that 4 item depressed mood 14.6 (1.05); 55 Australian high school 16
et al. asked about frequency of scale (Barber, 2006; 13-17 students: 83.4%
checking and purpose of Barber et al., 2001; Caucasian, 7.2% Asian,
use(0 = never; 1 = less Modecki et al., 2013). 1.5% African, and 1.3%
than once a month; Aboriginal and Torres
2 = every few weeks; Strait Islander; the

168
3 = 1–2 days a week; remaining participants
4 = 3–5 days a week; (6.6%) were drawn from
5 = about once a day; various ethnic groups
6 = several times a day)
4 Brunborg 2017 851 0.32 0.3316 Cross-sectional Composite created with 2 9 item severity measure 15.5* (NR); 53.7 Norwegian adolescents 16
et al. items: (1) How many days for depression (Patient 13-18
per week ranging from Health Questionnaire-9
every day to not at all in the items for Adolescents
last 12 months, (2) hours (Kroenke, Spitzer, &
per day ranging from less Williams, 2001))
than one hour to more than
15 hours
5 Frison et al. 2016 1,621 0.11 0.1104 Cross-sectional 4 item composite score of CES-DC 14.76 48 92% born in Belgium, 5% 18
average weekday/ (1.41);12-18 in another European
weekend/Wednesday/ country, and 3% in a non-
Friday Facebook use. 11 European country
point Likert-scale ranging
from 0 hrs; 0.5 hrs; 0.5–1
hrs; 1–1.5 hrs; 1.5–2 hrs;
2–2.5 hrs; 2.5–3 hrs; 3–4
hrs; 4–5 hrs; more than 5
hrs per day to I'm
constantly ‘online’ and
available for interaction
(continued on next page)
Journal of Affective Disorders 275 (2020) 165–174
 Table 2 (continued)

Study ID Authors Publication year Study Correlation (r) Fisher's Z Study design Social media use measure Depressive symptoms Mean age Gender Demographics Methodological
sample measure (SD); Range ratio (% quality (0-18)
E.J. Ivie, et al.

size female)

6 Morin- 2016 94 -0.097 -0.097 Longitudinal Single item frequency: Child depression 14.5 (1.76); 53.2 Candadian, White, French 17
Major et al. (data How many hours per inventory (CDI) 12-17 speaking adolescents;
incorporated in week spent on Facebook; Middle to high
this meta-analysis 6 point Likert scale (1) socioeconomic strata
were cross- below 1 h, (2)between 2
sectional) and 5 h, (3) between 6 and
10 h, (4) between 11 and15
h, (5) between 16 and 20 h
and (6) above 21 h.
7 Niu et al. 2018 764 0.206 0.209 Cross-sectional Composite of: 8 items CES-D (Chinese version) 14.23 (1.75); 46.9 Chinese adolescents 17
adapted to Qzone use 12-18
from from Facebook
intensity scale, two open
ended questions about
time per day spent on
Qzone and number of
friends on Qzone, and
response to "Qzone is a
part of my everyday
activity" 5 item Likert
scale ranging from strongly
disagree to strongly agree
8 Oberst et al. 2017 1,468 0.158 0.1593 Cross-sectional 5-item Social network Hospital Anxiety and 16.59 (.62); 74.3 Spanish-speaking Latin- 15
intensity scale: e.g., “ Depression Scale (HADS) NR American countries;

169
'Visiting social networking 90.7% reported "high
sites is part of my everyday educational attainment, as
activity,' 7-point-Likert the majority of them
scale from never to always” reported attending
secondary school"
9 Pantic et al. 2012 160 0.15 0.1511 Cross-sectional Single item frequency: BDI-II 18.02 (.29); 68.1 Pozarevac, Central Serbia 15
"Average daily time spent on NR
social networking sites,"
M=1.86 hrs, SD=2.08 hrs
10 Thorisdottir, 2019 10,563 0.202 0.248 Cross-sectional Single item frequency: 10 items from the 15* (NR); 14- 50.3 Icelandic adolescents 15
et al. ‘‘On average, how many depression dimension 16
hours a day do you spend scale of the Original
on social media (e.g., Symptom Checklist
Facebook, Snapchat, Derogatis, Lipman, Covi,
Twitter, and In- 1973)
stagram)?’’ Participants
answered on an 8-point
scale, ranging from almost
no time (= 1) to 6 hours or
more (= 8)."
11 Twenge et al. 2018 75,371 0.05 0.05 Cross-sectional Single item frequency: Six items from the Bentler 15.5**(NR); 50.2 Nationally represtentative 15
How often do you visit Medical and 13-18 U.S. sample
social networking sites; 5 Psychological
point Likert scale (never Functioning Inventory
= 1, a few times a year = depression scale
2, once or twice a month (Newcomb, Huba, &
= 3, at least once a week Bentler, 1981)
= 4, and almost every day
= 5).
Journal of Affective Disorders 275 (2020) 165–174

(continued on next page)

E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

High heterogeneity may reflect differences in study characteristics such

as sample population, study design and measurement tools. High het-
Methodological
quality (0-18)
erogeneity, indicated by the high I2 statistic, may also indicate that
there are salient moderators that are influencing this relationship, such
as individual differences between adolescents, gender differences, de-
16

velopmental differences, temperament, socioeconomic factors, and

sleep, to name a few. However, the heterogeneity estimates should be

Not reported (NR); *mean age calculated from provided age range (14-16); **mean age calculated from provided age range (13-18); *** mean age calculated from provided age range (11-17).
interpreted with caution because of the low power due to the small
Scottish adolescents

number of studies included in this meta-analysis. Some research sug-

gests that Q over-estimates heterogeneity in meta-analyses with small
Demographics

sample sizes and may under-estimate heterogeneity with large sample

sizes (Higgins et al., 2003). There is a large body of research on vul-
nerabilities to depression in adolescence and, given the small effect size
and high heterogeneity between the studies, social media use may be
associated depressive symptoms for some, but not all, adolescents. For
ratio (%
female)
Gender

example, the effect may be stronger for female adolescents, given that
NR

adolescents females are twice as likely to experience depression than

males (Albert, 2015), and females tend to use social media in a different
14*** (NR);
(SD); Range

way, and to a different degree, than do males (Barker, 2009). Un-

Mean age

fortunately, the majority of the studies we reviewed did not assess

11-17

gender differences. Temperament is another factor that is robustly as-

sociated with adolescent onset mental illness and may influence the
relationship between social media use and depression (e.g., (Klein et al.,
Depressive symptoms

Hospital Anxiety and

2015, Lonigan et al., 2003, Oldehinkel et al., 2004)). Temperament is

(HADS; Zigmond &
Depression Scale

easily measured (typically with a questionnaire) and could provide

Snaith, 1983)

some valuable insight into whether specific temperamental traits may

interact with social media use to confer higher risk. Furthermore, there
measure

may be salient mediators of the relationship between social media use

and depression that help explain the association. For example, there has
been a surge in the investigation of the association between sleep and
social media use; the spread

E.g., “How often do you use

Social media use measure

social media? 6-point scale

typical day?” 6-point scale

from “less than 1 hour” to
frequency and duration of

“How many hours do you

number of different social
media sites and devices."
6 item composite score

mental health in adolescence (Blake et al., 2018) and some researchers

created with questions

month” to “daily”) and

throughout the day; the

from "less than once a

that asked about "the

use social media on a

*mean age calculated from provided age range (13-18); ** mean age calculated from provided age range (11-17).

speculate that it may not be the content or use of social media in

of social media use

general that is conferring risk, but the fact that social media use is in-
“6+ hours”).

terfering with sleep (Woods and Scott, 2016). As such, it is possible that
it could be sleep disruption, not social media use itself, that increases
risk of depressive symptoms. Indeed, one of the main findings of one of
the studies included in this meta-analysis was that the effect of night-
time specific social media use significantly worsened adolescents’ sleep
Cross-sectional

compared to general social media use (Woods and Scott, 2016).

Study design

To our surprise, the studies included in this meta-analysis were

conducted in many different countries. This could be considered a
strength in that there has been assessment on the impacts of social
media use across the globe. However, one potential downside is that
Fisher's Z

there is a paucity of studies that attempt to replicate these findings in

0.1104

any given country. Additionally, all but one of the studies in this ana-
lysis used a cross-sectional correlational design, so we cannot infer the
Correlation (r)

temporal relationship between social media usage and depressive

symptoms. One possibility that is difficult to rule out based on corre-
lation design is that the association could be epiphenomenal. For ex-
0.11

ample, it could be that there is a common risk factor (e.g., tempera-

ment, gender) that results in both greater use of social media and higher
sample

levels of depressive symptoms, but that there is no causal relationship

Study

467
size

between social media use and depression per se. Moreover, correlational
studies are subject to interpretations of reverse causality. It could be
Publication year

that increases in social media use lead to greater depressive symptoms,

but another equally valid conclusions from a statistical point of view is
that adolescents experiencing more depressive symptoms are more
2016

likely to use social media to a greater extent. Recently, relevant long-

itudinal studies are starting to emerge (Orben and Przybylski, 2019b),
Woods et al.

so in the future a meta-analysis of longitudinal effects may become

Table 2 (continued)

viable, which will be particularly informative.

Authors

Another notable limitation of the current set of studies is that they

all use self-report questionnaires in their assessment of social media use
Study ID

and depressive symptoms and these measures are subject to self-report

biases or inaccurate reporting of usage. One recent study used time use
12

diary methods to examine the relationship between screen use and

170
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

Fig. 2. Forest plot of the associations between social media use and depressive symptoms by year.

Table 3 manifests quite differently for younger versus older adolescents

Meta-analysis key statistics. (American Psychiatric Association, 2013). For example, young adoles-
Random effects Estimate SE T-value p-value 95% CI cents with depression often display irritability as one of the core fea-
0.231 0.04 3.44 0.006 .04-.20 tures, while reports of sadness (one of the hallmark symptoms of de-
pression) may not be present (American Psychiatric Association, 2013).
Heterogeneity estimates Q-value df p-value I2 (%) Secondly, parents may not allow younger adolescents as frequent access
362.38 11 0.001 97.38
to social media accounts as older adolescents. The age limit for Face-
book, Instagram, and Snapchat users is 13 years old, therefore, the
wellbeing, and consistent with the current findings, found a significant youngest adolescents are still not technically, allowed access. The cur-
but very small association between high screen use and lower well- rent research base thus cannot account for developmental differences
being (Orben and Przybylski, 2019b). The use of self-report metho- both in depressive symptomatology and access. Additionally, the stu-
dology for assessment of time spent on social media is problematic in dies included in this meta-analysis ranged from between 2012-2020.
general. A meta-analysis on the relationship between self-reported Given that the rates of adolescent social media use have increased
media usage and actual use showed that adolescents, in particular, are dramatically (Lenhart et al., 2015) and are likely to continue to in-
poor reporters (Scharkow, 2019). Not only are self-report measures crease, it is possible that earlier studies are not capturing the nature of
poorly correlated with actual use, single-use measures do not appear to adolescent social media use today.
accurately predict behavior (Ellis et al., 2019). For example, Ellis et al. The current findings suggest that there is a small, but significant
(2019) found that even internet addiction scales were not able to pre- relationship between social media use and depression in adolescence.
dict “rapid [device] checking” behaviors that one would expect with Future studies could use experimental designs in order to test for a
addiction. Furthermore, measurements of overall usage do not capture causal relationship between social media use and depression. A study
the ways in which adolescents use social media. Some research has using an experimental design measured the effects of abstaining from
suggested that there are differences in mental health related outcomes Facebook for four weeks in an adult population and found that there
related to nuances of use, such as active versus passive use, (e.g., was a slight decrease (SD=0.09) in depression (Allcott et al., 2020). It
(Escobar-Viera et al., 2018)). Furthermore, self-reported retrospective should be noted, however, that depression was measured with a single
recall of device use is likely to covary with self-reported estimates of question: “How much of the time during the last four weeks… [have
mental health symptoms because of overlapping method and rater bias you felt depressed]” (pg., 642, (Allcott et al., 2020)). Also, it would be
(Podsakoff et al., 2003), thereby potentially inflating the apparent as- important to understand the relationship on a much smaller time scale
sociation between these variables. (e.g., minutes to hour to days) to see whether mood influences social
In addition, the studies in this analysis included adolescents who media engagement or vice versa in adolescents. In a sample of college
ranged in age from 11 to 18 years old, which could be problematic for a students, researchers assessed the temporal relationship between pas-
number of reasons. Firstly, the presentation of depression often sive social media use and depressive symptoms and found that passive

171
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

Fig. 3. Funnel plot.

social media use was not predictive of depression (Aalbers et al., 2019). from adolescents’ devices, depressive symptoms assessed through clin-
In addition, longitudinal study could investigate within-person changes ical interview), as self-report of time spent using digital devices is often
in depressive symptoms and social media use over a much larger scale not accurately reported (e.g., (Boase and Ling, 2013, Kobayashi and
(e.g., years). Currently, with the literature at hand, we can only make Boase, 2012, Vanden Abeele et al., 2013)). The mediators and mod-
between subject assertions (i.e., potentially trait-like differences in so- erators discussed above could be potential variables of interest that
cial media use associated with trait-like depression levels). Studies should be investigated. In addition, studies should assess children in
would also benefit from using data other than self-report to increase the different stages of adolescence in order to account for both develop-
validity of the findings (e.g., collecting social media usage data directly mental differences and access to social media accounts.

Fig. 4. P-curve analysis.

172
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

The small effect size and high variability between studies should 1177/0033294117697090. https://doi.org/.
prompt caution in interpreting the association observed here. The as- Dienlin, T., Masur, P.K., Trepte, S., 2017. Reinforcement or displacement? The reciprocity
of FtF, IM, and SNS communication and their effects on loneliness and life satisfac-
sociation between social media use and depression is considerably tion. J. Comput. Mediated Commun. 22 (2), 71–87.
smaller than effects associated with other known risk factors, including Ellis, D.A., Davidson, B.I., Shaw, H., Geyer, K., 2019. Do smartphone usage scales predict
stress (Kendler et al., 1999), family history of mental disorder behavior? International Journal of Human-Computer Studies 130, 86–92.
Escobar-Viera, C.G., Shensa, A., Bowman, N.D., Sidani, J.E., Knight, J., James, A.E.,
(Weissman et al., 1997), parenting-child and peer relationships Primack, B.A., 2018. Passive and active social media use and depressive symptoms
(Branje et al., 2010) and sleep disruption (Raniti et al., 2017) amongst among United States adults. Cyberpsychol. Behav. Soc. Netw. 21 (7), 437–443.
many others. As such, prevention programs and public policy would be Frison, E, Subrahmanyam, K, Eggermont, S, 2016. The short-term longitudinal and re-
ciprocal relations between peer victimization on Facebook and adolescents’ well-
better served focusing on these well-established risk factors with lager being. J. Youth Adolesc. 45, 1755–1771.
effect sizes than contributing to a moral panic about the effect of Haidt, J., Allen, N., 2020. Scrutinizing the effects of digital technology on mental health.
technology use, especially given the lack of supporting data. Although it Nature 578 (7794), 226–227.
Higgins, J.P., Thompson, S.G., Deeks, J.J., Altman, D.G., 2003. Measuring inconsistency
is likely that social media can have deleterious effects in some cases, we
in meta-analyses. BMJ (Clinical research ed.) 327 (7414), 557–560. https://doi.org/
need to understand much more deeply the mechanisms and moderators 10.1136/bmj.327.7414.557. https://doi.org/.
of these effects before strong recommendations can be drawn. Kendler, K.S., Karkowski, L.M., Prescott, C.A., 1999. Causal relationship between stressful
life events and the onset of major depression. Am. J. Psychiatry 156 (6), 837–841.
Klein, Dyson, Kujawa, Kotov, 2015. Temperament and internalizing disorders. In:
Declaration of Competing Interest Zentner, M., Shiner, R.L. (Eds.), Handbook of Temperament. Guilford Publications,
New York, pp. 541–561.
Knapp, G, Hartung, J, 2003. “Improved tests for a random effects meta-regression with a
The authors report no conflict in interest. single covariate. Statist. Med. 22 (17), 2693–2710.
Kobayashi, T., Boase, J., 2012. No such effect? The implications of measurement error in
Role of Funding Source self-report measures of mobile communication use. Commun. Methods Meas. 6 (2),
126–143.
Lenhart, A., Duggan, M., Perrin, A., Stepler, R., Rainie, H., Parker, K., 2015. Teens, Social
No funding source used for this research. Media & Technology Overview 2015. Pew Research Center [Internet & American Life
Project].
Lonigan, C.J., Phillips, B.M., Hooe, E.S., 2003. Relations of positive and negative affec-
Acknowledgments
tivity to anxiety and depression in children: evidence from a latent variable long-
itudinal study. J. Consult. Clin. Psychol. 71 (3), 465.
None. Modecki, K.L., Barber, B.L., Vernon, L., 2013. Mapping developmental precursors of
cyber-aggression: Trajectories of risk predict perpetration and victimization. Journal
of Youth and Adolescence 42 (5), 651–661. https://doi.org/10.1007/s10964-012-
References 9887-z.
Mojtabai, R., Olfson, M., Han, B., 2016. National trends in the prevalence and treatment
Aalbers, G., McNally, R.J., Heeren, A., De Wit, S., Fried, E.I., 2019. Social media and of depression in adolescents and young adults. Pediatrics, e20161878.
depression symptoms: a network perspective. J. Exp. Psychol. Gen. 148 (8), 1454. Morin-Major, J.K., Marina, M.-F., Durand, N., Wan, N., Juster, R.-P., Lupien, S.J., 2016.
Albert, P.R., 2015. Why is depression more prevalent in women. J. Psychiatry Neurosci. Facebook behaviors associated with diurnal cortisol in adolescents: Is befriending
JPN 40 (4), 219–221. http://doi.org/10.1503/jpn.150205. stressful. Psychoneuroendocrinology 63, 238–246. https://doi.org/10.1016/j.
Allcott, H., Braghieri, L., Eichmeyer, S., Gentzkow, M., 2020. The welfare effects of social psyneuen.2015.10.005.
media. Am. Econ. Rev. 110 (3), 629–676. National Institutes of Health, 2014. Quality assessment tool for observational cohort and
American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental cross-sectional studies. Natl. Heart Lung Blood Inst Avaliable from: www.nhlbi.nih.
Disorders (DSM-5®). American Psychiatric Pub. gov/health-pro/guidelines/in-develop/cardiovascular-risk-reduction/tools/cohort,
Avenevoli, S., Swendsen, J., He, J.P., Burstein, M., Merikangas, K.R., 2015. Major de- Accessed date: 5 November 2015.
pression in the national comorbidity survey–adolescent supplement: prevalence, Niu, G.F., Luo, Y.J., Sun, X.J., Zhou, Z.K., Yu, F., Yang, S.L., Zhao, L., 2018. Qzone use and
correlates, and treatment. J. Am. Acad. Child Adolesc. Psychiatry 54 (1), 37–44. depression among Chinese adolescents: a moderated mediation model. J. Affect.
Baker, D.A., Algorta, G.P., 2016. The relationship between online social networking and Disord. 231, 58–62.
depression: a systematic review of quantitative studies. Cyberpsychol. Behav. Soc. Oberst, U., Wegmann, E., Stodt, B., Brand, M., Chamarro, A., 2017. Negative con-
Netw. 19 (11), 638–648. sequences from heavy social networking in adolescents: The mediating role of fear of
Banjanin, N., Banjanin, N., Dimitrijevic, I., Pantic, I., 2015. Relationship between internet missing out. J. Adolesc. 55, 51–60. https://doi.org/10.1016/j.adolescence.2016.12.
use and depression: focus on physiological mood oscillations, social networking and 008.
online addictive behavior. Comput. Hum. Behav. 43, 308–312. Oldehinkel, A.J., Hartman, C.A., De Winter, A.F., Veenstra, R., Ormel, J., 2004.
Barber, B.L. (2006). To have loved and lost. Adolescent romantic relationships and re- Temperament profiles associated with internalizing and externalizing problems in
jection. In A.C. Crouter & A. Booth (Eds.), Romance and sex in adolescence and emerging preadolescence. Dev. Psychopathol. 16 (2), 421–440.
adulthood: Risks and opportunities (pp. 29–40). Mahwah, NJ: Lawrence Erlbaum Orben, A., Przybylski, A.K., 2019a. The association between adolescent well-being and
Associates. digital technology use. Nat. Hum. Behav. 3 (2), 173.
Barber, B.L., Eccles, J.S., Stone, M.R., 2001. Whatever happened to the jock, the brain, Orben, A., Przybylski, A.K., 2019b. Screens, teens, and psychological well-being: evidence
and the princess? Young adult pathways linked to adolescent activity involvement from three time-use-diary studies. Psychol. Sci. 5 095679761983032–15.
and social identity. J. Adolesc. Research 16, 429–455. https://doi.org/10.1177/ Pantic, I., Damjanovic, A., Todorovic, J., Topalovic, D., Bojovic-Jovic, D., Ristic, S.,
0743558401165002. Pantic, S., 2012. Association between online social networking and depression in
Barker, V., 2009. Older adolescents' motivations for social network site use: the influence high school students: Behavioral physiology viewpoint. Psychiatria Danubina Pantic,
of gender, group identity, and collective self-esteem. Cyberpsychol. Behav. 12 (2), Igor: Institute of Medical Physiology, Faculty of Medicine, University of Belgrade,
209–213. Visegradska 26/11, Belgrade, Serbia and Montenegro, 11000, [email protected]
Barry, C.T., Sidoti, C.L., Briggs, S.M., Reiter, S.R., Lindsey, R.A., 2017. Adolescent social g.ac.rs: Medicinska Naklada.
media use and mental health from adolescent and parent perspectives. J. Adolesc. 61, Podsakoff, P.M., Dennis, W.Scott B.MacKenzie, Lee, Jeong-Yeon, Podsakoff, Nathan P.,
1–11. https://doi.org/10.1016/j.adolescence.2017.08.005. https://doi.org/. 2003. Common Method Biases in Behavioral Research: A Critical Review of the
Best, P., Manktelow, R., Taylor, B., 2014. Online communication, social media and Literature and Recommended Remedies. J. Appl. Psychol. 88 (5), 879–903.
adolescent wellbeing: a systematic narrative review. Children Youth Serv. Rev. 41, Raniti, M.B., Allen, N.B., Schwartz, O., Waloszek, J.M., Byrne, M.L., Woods, M.J., ...,
27–36. Trinder, J., 2017. Sleep duration and sleep quality: associations with depressive
Blake, M.J., Trinder, J.A., Allen, N.B., 2018. Mechanisms underlying the association be- symptoms across adolescence. Behav. Sleep Med. 15 (3), 198–215.
tween insomnia, anxiety, and depression in adolescence_ Implications for behavioral Riehm, K.E., Feder, K.A., Tormohlen, K.N., Crum, R.M., Young, A.S., Green, K.M., ...,
sleep interventions. Clin. Psychol. Rev. 63, 25–40. https://doi.org/10.1016/j.cpr. Mojtabai, R., 2019. Associations between time spent using social media and inter-
2018.05.006. http://doi.org/. nalizing and externalizing problems among US youth. JAMA Psychiatry 1–9.
Blomfield, N., J., C., Barber, B.L., 2014. Social networking site use: Linked to adolescents' Scharkow, M., 2019. The reliability and temporal stability of self-reported media ex-
social self‐concept, self‐esteem, and depressed mood. Aust. J. Psychol. 66 (1), 56–64. posure: a meta-analysis. Commun. Methods Meas. 13 (3), 198–211.
Boase, J., Ling, R., 2013. Measuring mobile phone use: Self-report versus log data. J. Thorisdottir, I.E., Sigurvinsdottir, R., Asgeirsdottir, B.B., Allegrante, J.P., Sigfusdottir,
Comput. Mediat. Commun. 18 (4), 508–519. I.D., 2019. Active and passive social media use and symptoms of anxiety and de-
Branje, S.J., Hale, W.W., Frijns, T., Meeus, W.H., 2010. Longitudinal associations between pressed mood among Icelandic adolescents. Cyberpsychol. Behav. Soc. Netw. 22 (8),
perceived parent-child relationship quality and depressive symptoms in adolescence. 535–542.
J. Abnorm. Child psychol. 38 (6), 751–763. Twenge, J.M., Joiner, T.E., Rogers, M.L., Martin, G.N., 2018. Increases in depressive
Brunborg, G.S., Andreas, J.B., Kvaavik, E., 2017. Social media use and episodic heavy symptoms, suicide-related outcomes, and suicide rates among US adolescents after
drinking among adolescents. Psychol. Rep. 120 (3), 475–490. https://doi.org/10. 2010 and links to increased new media screen time. Clin. Psychol. Sci. 6 (1), 3–17.
Valkenburg, P.M., Peter, J., Schouten, A.P., 2006. Friend networking sites and their

173
E.J. Ivie, et al. Journal of Affective Disorders 275 (2020) 165–174

relationship to adolescents' well-being and social self-esteem. CyberPsychol. Behav. 9 of depressed parents: 10 years later. Arch. Gen. Psychiatry 54 (10), 932–940.
(5), 584–590. Woods, H.C., Scott, H., 2016. #Sleepyteens: Social media use in adolescence is associated
Vanden Abeele, M., Beullens, K., Roe, K., 2013. Measuring mobile phone use: Gender, age with poor sleep quality, anxiety, depression and low self-esteem. J. Adolesc. 51,
and real usage level in relation to the accuracy and validity of self-reported mobile 41–49. https://doi.org/10.1016/j.adolescence.2016.05.008.
phone use. Mobile Media Commun. 1 (2), 213–236. Yoon, S., Kleinman, M., Mertz, J., Brannick, M., 2019. Is social network site usage related
Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. J. Stat. to depression? A meta-analysis of Facebook-depression relations. J. Affect. Disord.
Softw. 36 (3), 1–48. URL. http://www.jstatsoft.org/v36/i03/. Young, M.E., 2016. The problem with categorical thinking by psychologists. Behav.
Weissman, M.M., Warner, V., Wickramaratne, P., Moreau, D., Olfson, M., 1997. Offspring Processes 123, 43–53.

174