Available online at www.sciencedirect.

com

ScienceDirect

Why increases in adolescent depression may be linked to the

technological environment
Jean M Twenge

Between 2011 and 2018, rates of depression, self-harm, and a possible cause of those cohort trends? Second, how is
suicide attempts increased substantially among U.S. technology use related to well-being among individuals?
adolescents. The most probable cause(s) of these trends likely 1)
began or accelerated during these years, 2) affected a large Cohort trends in adolescent depression and
number of people, 3) impacted everyday life, and 4) were psychological well-being
associated with depression. In several large studies, heavy users Between 2011 and 2017–18, rates of adolescent
of technology are twice as likely as light users to be depressed or depression in the U.S. increased at least 60%, with larger
have low well-being. Cohort declines in face-to-face social increases among girls ([4–7]; see Figures 1 and 2).
interaction may also impact even non-users of digital media. Happiness and life satisfaction declined after 2012 [8].
Thus, although technology use is not the cause of most Emergency department admissions for self-harm
depression, increased time spent on technology and the behaviors tripled among 10- to 14-year-old girls between
technological environment may be causes of the sudden 2009 and 2015 [9]. Emergency department admissions
increase in depression since 2011. for suicide attempts or ideation doubled or increased
substantially [10,11,12], as did self-poisonings [13] and
Address
the rate of completed suicides among adolescents, again
Department of Psychology, San Diego State University, 5500 Campanile
Drive, San Diego, CA 92182-4611, United States primarily among girls [14]. However, mental health trends
were weaker or non-existent among U.S. adults ages
Corresponding author: Twenge, Jean M (jtwenge@sdsu.edu) 26 and over, with the rise in depression primarily
appearing in the iGen cohort of Americans born in the
Current Opinion in Psychology 2020, 32:89–94
late 1990s–2000s [7].
This review comes from a themed issue on Socio-ecological
psychology Four steps in exploring causes of the decline
Edited by Ayse K Uskul and Shige Oishi
in adolescent well-being
Determining the causes behind cultural trends is difficult.
Experimental methods are impossible; people cannot be
randomly assigned to be born at different times. However,
https://doi.org/10.1016/j.copsyc.2019.06.036 other types of evidence can be informative. In the case of
2352-250X/ã 2019 Elsevier Ltd. All rights reserved. well-being, four criteria seem most relevant. It is important
to note that this model is not definitive; it is easier to rule
causes out than to definitively rule them in.

The first criterion is the time sequence: does the possible

cause occur during the same years, or just before? This
Just as cultures vary from one region to another, cultures ecological approach, in which possible causes are matched
also vary over time. In recent decades, one of the most by year with outcomes, is a classic way to examine the
substantial cultural changes has been in technology — drivers of cultural change [15]. Given the sudden increase
both the technology available and how people use it. The in depression between 2011 and 2018, the cause must
Internet was widely used beginning in the early 2000s, be something that appeared or strengthened in the early
the first smartphone was introduced in 2007, social media 2010s and continued to grow until the late 2010s. This
was widely used by the early 2010s, and by the end of effectively eliminates the Great Recession, which began in
2012 the majority of Americans owned a smartphone [1]. 2007 and ended around 2009–2010, with the economy then
By 2016, the average U.S. adolescent spent 6 hours a day steadily improving between 2011 and the late 2010s. Other
texting, online, and on social media [2]. This rapid societal factors have been present for decades, such as
adoption of technology is a substantial cultural change, growing income inequality (1980s) and school shootings
but has it had an impact on psychological well-being? (1990s), making them unlikely causes of trends beginning
in the early 2010s. In contrast, smartphones became
In this review, I focus on two primary areas of inquiry. First, common in the early 2010s, and smartphone ownership
is low psychological well-being (such as unhappiness, and time online continued to increase into the late 2010s.
depression, and suicide) higher among U.S. teens born When indicators of the technological environment are
after 1995 (iGen; [3]), and is the technological environment matched by year with adolescent well-being in Granger

www.sciencedirect.com Current Opinion in Psychology 2020, 32:89–94

90 Socio-ecological psychology

Figure 1

29

27

25
% with high depressive symptoms

23 Females

21

19
Males

17

15

13
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010

2012
2013
2014
2015
2016
2017
2011

Year
Current Opinion in Psychology

Percent of U.S. 8th, 10th, and 12th graders high in depressive symptoms, by sex, Monitoring the Future, 1991–2017.

causality analyses, technology precedes declines in well- suggesting it should have a greater influence on personal
being by about a year, whereas economic indicators such as mental health.
unemployment were not correlated [8].
The fourth criterion asks: is the possible cause associated
The second criterion asks: does the possible cause with the outcome among individuals? Here, we would
affect a large number of people in the impacted group? examine whether technology use is associated with
For example, the opioid crisis passes step 1, with opioid low well-being at the individual level, which I do in
addiction and overdoses increasing after 2011. However, the section of this review titled “Technology use and
opioid addiction primarily appeared among adults ages psychological well-being.”
25 and over [16], whereas the increase in depression
appears primarily among those under 25. In addition, Using multiple criteria, rather than solely focusing on the
the opioid epidemic was concentrated among lower time sequence, helps guard against the post hoc ergo
socioeconomic status Whites in the Midwest and South propter hoc fallacy (that if two things occur around the
[17], whereas the rise in depression occurred across same time they must be related). Nicholas Cage films and
class, race, and region [6]. Thus, the opioid crisis is pool drownings rise and fall together [20]. However, any
unlikely to be the cause of the increase in teen supposition that one causes the other would fail the next
depression. three criteria. Thus, the assertion that ‘anything’ that rose
at the same time as teen depression could be a cause is
The third criterion asks: does the possible cause have a false; additional evidence is necessary. That additional
direct impact on everyday life? This argues against a role for evidence points away from many other possible causes
events that directly affect a small number of people but and toward the technological environment.
reach most via news reports. For example, terrorist attacks
had no impact on patterns in Google searches for words Note that these criteria are focused on possible causes for
related to anxiety [18]. In contrast, the shift in teens’ free the increase in depression, not on all of the possible causes
time toward technology and away from face-to-face social of depression, which is an entirely different research
interaction [19] affects most teens’ day-to-day lives, question. Many of the major causes of depression are

Current Opinion in Psychology 2020, 32:89–94 www.sciencedirect.com

 Adolescent depression and technology use Twenge 91

Figure 2

14

13

12

11
% in past 12 months

10

12-17
9
18-25

8 26-49

50+

4
2014
2009

2010

2013

2015
2012

2016

7
2011

201
Year
Current Opinion in Psychology

Major depressive episode in the last 12 months, by age group and sex, National Survey of Drug Use and Health, 2009–2017.

presumably constant over time (for example, genetic Social media and face-to-face interaction provide another
predisposition). Thus, even if technology is the primary example of individual versus group effects. Likely due to
cause of the increase in teen depression since 2011, it is differences in sociability, teens who use social media
only one among many causes of depression overall. more also interact more with their friends face-to-face;
thus, at the individual level, social media time does not
Effects at the level of the group versus the level of the necessarily displace face-to-face time. The picture is
individual different at the group level, however; as technology time
Although the fourth criteria (an association among increased over the years, teens’ face-to-face time
individuals) is worthwhile evidence, it is only part of declined, suggesting that displacement does occur at
the picture. Even if technology use has no effect on the cohort/group level [19].
depression at the individual level, the technological envi-
ronment may have an impact at the level of the group or Technology use and psychological well-being
cohort. Is technology use associated with psychological well-
being among individuals? That question has been the
Imagine a teen in 2017 (let’s call her Abby) who does not source of some controversy. However, recent studies
use social media. At the individual level, social media use based on large samples have provided more definitive
cannot increase Abby’s depression as her use is zero. At answers than previous smaller studies.
the group level, however, Abby is still affected. Without
social media, she will be excluded from some interactions Across several large studies, adolescents and young adults
and may feel unhappy. If she would rather see her friends who spent more time using digital media reported lower
face-to-face than communicate via social media, she may psychological well-being, including lower overall
be unsuccessful (who will she go out with if her friends are psychological well-being [21], lower life satisfaction
all at home on Snapchat?) Thus, the mental health of [22], less happiness [8], more feelings of loneliness and
many teens, not just heavy social media users, may suffer social isolation [23] and more depression [24,25,26].
over time as the culture of teen social life changes. These associations are often considerable; for example,

www.sciencedirect.com Current Opinion in Psychology 2020, 32:89–94

92 Socio-ecological psychology

adolescents who spent five or more hours a day (versus that percent variance explained is not a valid measure of
one hour a day) on electronic devices were 66% more practical importance. For example, aspirin explains only
likely to have at least one risk factor for suicide, and heavy 0.0011% of the variance in heart attacks, but those who
Internet users were twice as likely to be unhappy as light did not take aspirin were twice as likely to have a heart
users [6,8]. Similarly, Kelly et al. [24] found a doubling attack compared to those who took aspirin [30]; this
in depression from no use to heavy use of social media comparison (‘twice as likely’) is a measure of relative
(see Figure 3). risk, an effect size often used in medicine and clinical
contexts. Przybylski and Weinstein [31] concluded that
Choice of effect size device use explained less than 1% of the variance in well-
These findings are in contrast with other studies that being, but in the same data using the same measure and
conclude there are no practically important effects for same control variables, heavy users of smartphones were
technology use and well-being. For example, Orben and twice as likely to be low in well-being as light users ([21];
Przybylski [27] used the same datasets analyzed by Kelly see Figure 3). As Funder and Ozer note, percent variance
et al. [24] and Twenge et al. [6] and concluded that explained “allows writers to disparage certain findings
technology use explained only 0.5% of the variance in that they find incompatible with their own theoretical
well-being, which they deemed not large enough to be predilections”. On both practical and statistical grounds,
practically important. Ferguson [28] makes a similar Funder and Ozer conclude that percent variance
argument, noting a low percent of variance of depression explained is “not merely uninformative; for purposes of
explained by technology use. evaluating effect size, the practice is actively misleading”.

How can we reconcile these opposing conclusions, often In addition, percent variance does not answer the relevant
from the same datasets? Much of the difference lies in the research question. Studies of technology use and well-
statistics used by each set of researchers. Ferguson [28] being are not documenting all of the possible causes of
and Orben and Przybylski [27] relied on percent variance low well-being, which include genetics, trauma, poverty,
explained. However, researchers including Funder and and other factors. Instead, these studies are focused on
Ozer [29] and Rosnow and Rosenthal [30] have shown the association between technology use and well-being.

Figure 3

40

35

Social media,
30 depression,
girls (Kelly)

25 Social media,
depression,
% low well-being

boys (Kelly)

20
Online,
unhappy
(Twenge)
15
Smartphone,
low well-
10 being
(Przybylski)

5
urs

urs

urs

urs

urs

ours
e

our

ur
Non

1 ho

2 ho

3 ho

4 ho

5 ho

6 ho
1/2 h

7+ h

Hours a day of use

Current Opinion in Psychology

Adolescents low in psychological well-being by hours a day of technology use.

NOTE: Social media and depression among girls and boys from the Millennium Cohort Study [24]. Time online and unhappiness from the
Monitoring the Future dataset [8]. Smartphone use and low well-being from the UK sample used by Przybylski and Weinstein [31] and reanalyzed
by Twenge and Campbell [21].

Current Opinion in Psychology 2020, 32:89–94 www.sciencedirect.com

 Adolescent depression and technology use Twenge 93

Comparing well-being across technology use (i.e. with a individual level, several large studies find substantial asso-
relative risk comparison) better answers this question ciations between technology use and low well-being. In
than calculating percent variance explained. considering such questions, the level of analysis (individual
versus group) is an important consideration. Cultural
Orben and Przybylski [27] provided comparisons to other change does not occur in a vacuum, but affects individuals
lifestyle effects (e.g. wearing glasses, eating potatoes, at the level of the cohort as social norms move toward more
believing one is overweight, using marijuana) as a gauge digital interaction and less face-to-face interaction.
of practical importance. However, there is no standard list
of variables to use as comparisons, so these could be Conflict of interest statement
systematically chosen to make a particular argument. J.M.T. has received speaking honoraria and consulting
In this case, for example, marijuana use was included fees for presenting research and is the author of several
while other drugs were not, body-mass index and exercise books, most recently iGen: Why Today’s Super-Connected
were not included although other health-related variables Kids Are Growing Up Less Rebellious, More Tolerant, Less
were, and getting into fights was included but carrying a Happy – and Completely Unprepared for Adulthood.
weapon was not. If they had used different comparisons,
their conclusions about practical importance might have
been different. In contrast, documenting the number of References and recommended reading
Papers of particular interest, published within the period of review,
people affected or using relative risk is more straightfor- have been highlighted as:
ward and less subjective.
 of special interest

Direction of causation 1. Pew Research Center: Mobile Ownership. 2018 https://www.

The studies reviewed thus far are correlational and cannot pewinternet.org/fact-sheet/mobile/.
show causation. At the group level, that may not be 2. Twenge JM, Martin GN, Spitzberg BH: Trends in U.S.
crucial. The increases in adolescent depression after adolescents’ media use, 1976–2016: The rise of digital media,
the decline of TV, and the (near) demise of print. Psychol Pop
smartphones became common after 2011 cannot be Media Cult (in press).
explained by depression causing technology use. If they
3. Twenge JM: iGen: Why Today’s Super-Connected Kids are
were, one would be forced to posit that an unexplained Growing Up Less Rebellious, More Tolerant, Less Happy — And
rise in teen depression caused greater ownership and use Completely Unprepared for Adulthood. New York: Atria Books;
2017.
of smartphones. It seems much more plausible that
smartphones became common, time online increased, 4. Keyes KM, Gary D, O’Malley PM, Hamilton A, Schulenberg J:
Recent increases in depressive symptoms among U.S.
and depression followed. adolescents: trends from 1991 to 2018. Soc Psychiatry Psychiatr
Epidemiol 2019, 54:987-996.
The question of causation is more relevant at the 5. Mojtabai R, Olfson M, Han B: National trends in the prevalence
individual level. Longitudinal studies have generally and treatment of depression in adolescents and young adults.
Pediatrics 2016, 138.
found that more technology use predicts lower well-being
6. Twenge JM, Joiner TE, Rogers ML, Martin GN: Increases in
later (e.g. [32–35]), though some studies find a reciprocal depressive symptoms, suicide-related outcomes, and suicide
relationship [36]. Three experiments also point to causa- rates among U.S. Adolescents after 2010 and links to
increased new media screen time. Clin Psychol Sci 2018, 6:3-17.
tion from technology use to depression. One experiment
randomly assigned Facebook users to either give up their 7. Twenge JM, Cooper AB, Joiner TE, Duffy ME, Binau SG: Age,
period, and cohort trends in mood disorder indicators and
use of the site or continue their normal use; those who suicide-related outcomes in a nationally representative
abstained reported more happiness and less depression dataset, 2005-2017. J Abnorm Psychol 2019, 128:185-199.
[37]. Another experiment asked college students to limit 8. Twenge JM, Martin GN, Campbell WK: Decreases in
their social media use to no more than 30 min a day. psychological well-being among American adolescents after
2012 and links to screen time during the rise of smartphone
Compared to a control group that continued their normal technology. Emotion 2018, 18:765-780.
use, the limiters were less lonely and less depressed after
9. Mercado MC, Holland K, Leemis RW: Trends in emergency
one week and continuing through the next two weeks  department visits for nonfatal self-inflicted injuries among
[38]. Third, college students randomly assigned to spend youth aged 10 to 24 years in the United States, 2001-2015. J Am
Med Assoc 2017, 318:1931-1933.
20 min on their Facebook pages reported less positive Using Centers for Disease Control data, Mercado et al. find that emer-
mood and less meaning in life than those who browsed gency department visits for self-harm behaviors (such as cutting or taking
too many pills) tripled among 10-year to 14-year-old girls between
the internet instead [39]. 2009 and 2015, with substantial increases among 15-year to 19-year-
old girls and young women as well. Trends for 20-year to 24-year-old
young women were less pronounced, as were trends for boys and young
Conclusions men.
In conclusion, adolescent well-being declined significantly
10. Burstein B, Agostino H, Greenfield B: Suicidal attempts and
between 2011 and 2018, with declines in happiness and life  ideation among children and adolescents in US emergency
satisfaction and increases in depression and suicide departments, 2007-2015. JAMA Pediatr 2019, 173:598-600.
Burnstein et al. used a nationally representative sample of emergency
attempts. Using four criteria, the technological environ- room visits collected by the Centers for Disease Control. They found that
ment is a plausible cause of the decline in well-being. At the ED visits for suicidal ideation and attempts nearly doubled between

www.sciencedirect.com Current Opinion in Psychology 2020, 32:89–94

94 Socio-ecological psychology

2007 and 2015 among 5-year to 17-year-olds, with 80% of the increase Using the 2015 administration of the Millennium Cohort Study when
occurring since 2010 and 60% since 2013. participants were 14 years old, Kelly et al. found that girls who are
heavy users of social media are three times more likely to be
11. Kalb LG, Stapp EK, Ballard ED, Holingue C, Keefer A, Riley A: depressed than non-users; among boys, twice as many heavy users
Trends in psychiatric emergency department visits among are depressed. This was true even after controlling for emotional
youth and young adults in the US. Pediatrics 2019, 243: adjustment at age 11. The association between social media use
e20182192. and depression was mediated by online harassment, poor sleep, lower
self-esteem, and body image issues.
12. Plemmons G, Hall M, Doupnik S, Gay J, Brown C, Browning W
et al.: Hospitalization for suicide ideation or attempt: 2008- 25. Lin Ly, Sidani JE, Shensa A, Radovic A, Miller E, Colditz JB et al.:
2015. Pediatrics 2018, 141. Association between social media use and depression among
U.S. young adults. Depress Anxiety 2016, 33:323-331.
13. Spiller HA, Ackerman JP, Spiller NE, Casavant MJ: Sex- and age-
specific increases in suicide attempts by self-poisoning in the 26. Twenge JM, Campbell WK: Associations between screen time
United States among youth and young adults from 2000 to and lower psychological well-being among children and
2018. J Pediatr 2019, 210:201-208. adolescents: evidence from a population-based study. Prev
Med Rep 2018, 12:271-283.
14. Ruch DA, Sheftall AH, Schlagbaum P, Rausch J, Campo J,
Bridge JA: Trends in suicide among youth aged 10 to 19 years 27. Orben A, Przybylski AK: The association between adolescent
in the United States, 1975 to 2016. JAMA Open 2019, 2:e193886. well-being and digital technology use. Nat Hum Behav 2019,
15. Santos HC, Varnum MEW, Grossmann I: Global increases in 3:173-182.
individualism. Psychol Sci 2017, 28:1228-1239. 28. Ferguson CJ: Everything in moderation: moderate use of
16. Hedegaard H, Warner M, Minino AM: Drug Overdose Deaths in screens unassociated with child behavior problems. Psychiatr
the United States, 1999-2016. . NCHS Data Brief No. 294, Q 2017, 88:797-805.
December 2017 2017 https://www.cdc.gov/nchs/products/ 29. Funder DC, Ozer DJ: Evaluating effect size in psychological
databriefs/db294.htm.  research: sense and nonsense.. in press Adv Methods Pract
17. Seth P, Scholl L, Rudd RA: Overdose deaths involving opioids, Psychol Sci 2019.
cocaine, and psychostimulants — United States, 2015-2016. As binary significance testing fades, effect sizes become increasingly
Morb Mortal Wkly Rep 2018 . Centers for Disease Control https:// important. Funder and Ozer review ways of calculating and comparing
www.cdc.gov/mmwr/volumes/67/wr/mm6712a1.htm. effect sizes. They show that percent variance explained is uninformative
and may even be misleading. Instead, they suggest relying on more
18. Stephens-Davidowitz S: Fifty States of Anxiety. New York Times; intuitive and practically meaningful ways of communicating the size of
2016 https://www.nytimes.com/2016/08/07/opinion/sunday/ effects, including number of people affected.
fifty-states-of-anxiety.html.
30. Rosnow RL, Rosenthal R: Effect sizes for experimenting
19. Twenge JM, Spitzberg BH, Campbell WK: Less in-person psychologists. Can J Exp Psychol 2003, 57:221-237.
 social interaction with peers among U.S. adolescents in the
21st century and links to loneliness. J Soc Pers Relat 2019, 31. Przybylski AK, Weinstein N: A large-scale test of the Goldilocks
36:1892-1913. hypothesis: quantifying the relations between digital-screen
Twenge et al. find that iGen teens spend less time interacting with their use and the mental well-being of adolescents. Psychol Sci
friends face-to-face than previous generations did, including hanging out 2017, 28:204-215.
with friends, going to parties, going out on dates, going to movies, and 32. Allen MS, Vella SA: Screen-based sedentary behaviour and
riding in cars. Perhaps as a result, teens’ feelings of loneliness increased psychosocial well-being in childhood: cross-sectional and
sharply after 2012. The declines in face-to-face interaction occurred over longitudinal associations. Ment Health Phys Act 2015, 9:41-47.
the same time period that digital media use increased, and despite
declines or little change in time spent on paid work, homework, and 33. Booker CL, Kelly YJ, Sacker A: Gender differences in the
extracurricular activities. associations between age trends of social media interaction
and well-being among 10-15-year-olds in the UK. BMC Public
20. Vigin T: Spurious Correlations. New York: Hachette Books; 2015. Health 2018, 18:321-333.
21. Twenge JM, Campbell WK: Digital media use is linked to lower 34. Schmiedeberg C, Schroder J: Leisure activities and life
 psychological well-being: evidence from three datasets. satisfaction: an analysis with German panel data. Appl Res
Psychiatr Q 2019, 90:311-331. Qual Life 2017, 12:137-151.
Twenge and Campbell find that heavy users of digital media are often
twice as likely to have poor psychological well-being across different 35. Shakya HB, Christakis NA: Association of Facebook use with
measures of digital media use and different measures of well-being compromised well-being: a longitudinal study. Am J Epidemiol
across three datasets. This includes the large sample of UK adolescents 2017, 185:203-211.
used in Przybylski and Weinstein [31], which the original authors con-
cluded did not show meaningful effects based on using percent variance 36. Kim HH: The impact of online social networking on
explained. However, when the percentage of heavy versus light users low adolescent psychological well-being (WB): a population-level
in well-being are examined, practically important effects are evident. analysis of Korean school-aged children. Int J Adolesc Youth
2017, 22:364-376.
22. Booker CL, Skew AJ, Kelly YJ, Sacker A: Media use, sports
participation, and well-being in adolescence: cross-sectional 37. Tromholt M: The Facebook experiment: quitting Facebook
findings from the UK Household Longitudinal Study. Am J leads to higher levels of well-being. Cyberpsychol Behav Soc
Public Health 2015, 105:173-179. Netw 2016, 19:661-666.
23. Primack BA, Shensa A et al.: Social media use and perceived 38. Hunt MG, Marx R, Lipson C, Young J: No more FOMO: limiting
social isolation among young adults in the U.S. Am J Prev Med social media decreases loneliness and depression. J Soc Clin
2017, 53:1-8. Psychol 2018, 37:751-768.
24. Kelly Y, Zilanawala A, Booker C, Sacker A: Social media use and 39. Yuen EK, Koterba EA, Stasio MJ, Patrick RB, Gangi C, Ash P et al.:
 adolescent mental health: findings from the UK Millennium The effects of Facebook on mood in emerging adults. Psychol
Cohort Study. EClinicalMedicine 2019. Pop Media Cult 2019, 8:198-206.

Current Opinion in Psychology 2020, 32:89–94 www.sciencedirect.com