Sources

Miller,C.C. (2023). Everyone Says Social Media Is Bad for Teens. Proving It Is Another
Thing. The New York Times. https://www.nytimes.com/2023/06/17/upshot/social-media-teen-mental-
health.html

Everyone Says Social Media Is

Bad for Teens. Proving It Is
Another Thing.
Parents, scientists and the surgeon general are worried. But there isn’t
even a shared definition of what social media is.

 Share full article



 123

There is little research proving that social media directly causes harmful
outcomes.Credit...Stephen B. Morton for The New York Times
By Claire Cain Miller
June 17, 2023

There have been increasingly loud public warnings that social media is
harming teenagers’ mental health — most recently from the United States
surgeon general — adding to many parents’ fears about what all the time
spent on phones is doing to their children’s brains.

While many scientists share the concern, there is little research to prove
that social media is harmful — or to indicate which sites, apps or features
are problematic. There isn’t even a shared definition of what social media is.
It leaves parents, policymakers and other adults in teenagers’ lives without
clear guidance on what to be worried about.

“We have some evidence to guide us, but this is a scenario where we just
need to know more,” said Jacqueline Nesi, a psychologist at Brown
who studies the topic.
What counts as social media when it comes to teenagers’ health?
The surgeon general, Dr. Vivek Murthy, warned last month that social
media carried a “profound risk of harm,” but he didn’t name any apps or
websites. His report acknowledged that “there isn’t a single, widely
accepted scholarly definition of social media.”
ADVERTISEMENT
SKIP ADVERTISEMENT

Most studies look at platforms with user-generated content, where people

can interact. But that raises a lot of questions. Does it matter if teenagers
see posts from people they know or don’t know? Does it make a difference if
they post or just view? Do multiplayer games count? Dating apps? Group
texts?

YouTube illustrates the challenge. It’s the most popular site among
teenagers by far: 95 percent use it, and almost 20 percent say they do so
“almost constantly,” Pew Research Center found. It has all the features of
social media, yet it hasn’t been included in most studies.
How Often Teens Say They Use Each Platform
YouTubeTikTokSnapchatInstagramFacebook
Almost constantly
19%16%15%10%
Several times a day
41%32%29%27%8%
About once a day
17%9%7%12%8%
Less often
18%9%8%12%14%
Don't use
5%33%41%38%67%

Note: Among teens ages 13 to 17. Those who did not give an answer are not shown.
Figures are rounded.
Source: Pew Research Center survey conducted April 14-May 4, 2022
The New York Times

Some researchers speculated that YouTube may not have as many

detrimental effects, because teenagers often consume it passively, like TV,
and don’t post or comment as often as they do on other apps. Or,
researchers said, it may carry the same risks — it offers endless scrolling
and algorithmic recommendations, similar to TikTok. There is no clear data
either way.

Nesi, J., & Prinstein, M. J. (2015). Using Social Media for Social Comparison and
Feedback-Seeking: Gender and Popularity Moderate Associations with Depressive
Symptoms. Journal of Abnormal Child Psychology, 43(8), 1427–
1438. https://doi.org/10.1007/s10802-015-0020-0

Abstract

This study examined specific technology-based behaviors (social comparison and interpersonal
feedback-seeking) that may interact with offline individual characteristics to predict concurrent
depressive symptoms among adolescents. A total of 619 students (57 % female; mean age 14.6)
completed self-report questionnaires at 2 time points. Adolescents reported on levels of
depressive symptoms at baseline, and 1 year later on depressive symptoms, frequency of
technology use (cell phones, Facebook, and Instagram), excessive reassurance-seeking, and
technology-based social comparison and feedback-seeking. Adolescents also completed
sociometric nominations of popularity. Consistent with hypotheses, technology-based social
comparison and feedback-seeking were associated with depressive symptoms. Popularity and
gender served as moderators of this effect, such that the association was particularly strong
among females and adolescents low in popularity. Associations were found above and beyond
the effects of overall frequency of technology use, offline excessive reassurance-seeking, and
prior depressive symptoms. Findings highlight the utility of examining the psychological
implications of adolescents’ technology use within the framework of existing interpersonal
models of adolescent depression and suggest the importance of more nuanced approaches to the
study of adolescents’ media use.
Keywords: Adolescents, Depressive symptoms, Technology, Social media, Interpersonal
feedback-seeking, Social comparison

Interpersonal models of depression in adolescents emphasize the cyclical

associations among social experiences, depressogenic interpersonal
behaviors, and depressive symptoms (Hames et al. 2013). During the
adolescent transition, changes in the social environment (e.g., more frequent,
complex, and unsupervised peer contact) may complicate this process
(Choukas-Bradley and Prinstein 2014), particularly as social relationships
take on increased importance in shaping self-esteem, well being, and identity
(Harter et al. 1996; Rudolph 2009). For example, as adolescents move toward
developing cohesive self-identities, they typically engage in greater levels of
social comparison and interpersonal feedback-seeking (Harter 2012).
However, depressed or depression-prone individuals may engage in these
social behaviors in maladaptive ways, such that they represent unique
depressogenic-interpersonal behaviors (Hames et al. 2013).

Although much research has investigated the interactions among these social
behaviors, depressive symptoms, and in-person peer interactions in this age
group, relatively little is known regarding social experiences that occur
through technological media, including Social Networking Sites (SNS, e.g.,
Facebook) and text messages. These media have had a revolutionizing impact
on both the frequency and types of peer interaction afforded to today’s
teenagers. In fact, these media may facilitate certain technology-based
behaviors, or behaviors that occur as a result of, or in conjunction with,
technology use, such as technology-based social comparison and feedback-
seeking (SCFS; Manago et al. 2008).

Thus, an investigation of associations between adolescents’ interpersonal

behaviors and depressive symptoms within this online social context remains
critical. The current study will investigate technology-based SCFS as one
important behavior that may be associated with depressive symptoms, while
considering individual characteristics that may influence this association (i.e.
gender and popularity). In examining these associations, it is important to
understand the unique features of adolescents’ online social worlds, mixed
evidence for the association between technology use and psychological
adjustment, the theoretical relevance of SCFS, and the potential influence of
gender and popularity. Each of these factors will be discussed in turn.
Adolescents’ Online Social World

The ubiquity of technology in the interpersonal environments of modern adolescents makes its
influence impossible to ignore, with recent years marking a dramatic increase in technology use
that has transformed the adolescent social world. Over 93 % of American teenagers (ages 12 to
17) are now connected to the Internet, more than any other age group, with an estimated 73 %
belonging to at least one SNS (Jones and Fox 2009; Lenhart et al. 2010). The average young
person now spends approximately 7 hours a day connected to electronic media (Rideout et al.
2010). At least 78 % of adolescents own a cell phone, with at least half of those being a smart
phone (Madden et al. 2013b). Adolescents, in a developmental period during which peer contact
is already increasing, are now afforded almost constant communication with peers, an experience
that may actually increase the importance of the role that peer groups play in development (Uhls
et al. 2011).

In addition to the amount of peer contact, the type of peer interaction afforded by new media is
unique to the current generation of adolescents. SNS, such as Facebook, MySpace, Twitter, and
Instagram, have a number of unique features: a personal profile with photos, links, and text
meant to represent the user; friends, or the collection of other users an individual has allowed
into his or her network; and public commentary on a user’s profiles and photos, visible to others
in the social network (boyd1 2007). These features create an online social world that is
fundamentally different than its offline counterpart.

Additionally, these features allow adolescents to engage with these technologies in unique ways.
The typical adolescent maintains a network of about 300 online friends (Madden et al. 2013a),
meaning that behaviors on social networking sites are performed in the presence of an audience;
every photo, comment, and new online friend provides details about the user to his or her public
network (Manago et al. 2012). Thus, adolescents use the features of SNS to both create and
consume online content, including profiles, photos, and posts. This allows adolescents to receive
constant feedback from their peers and to engage in social comparison processes with them
online (Manago et al. 2008).
Technology Use and Psychological Adjustment

A review of the current literature reveals extremely mixed findings regarding the reciprocal
associations between frequent technology use and psychological adjustment among adolescents.
Some studies have suggested that frequent use of technology may be associated with negative
outcomes. For example, frequent use of social networking sites may be associated with
depressive symptoms (van den Eijnden et al. 2008), short-term declines in subjective well-being
(Kross et al. 2013), romantic jealousy (Muise et al. 2009), and the belief that others are happier
and living better lives than one’s self (Chou and Edge 2012). Other studies have indicated the
opposite: that frequent technology use may be associated with positive adjustment, including
increases in self-disclosure and friendship quality (Valkenburg and Peter 2009). A recent
narrative review reflects these mixed results, indicating associations between online technologies
and positive outcomes such as self-esteem, social support, and self-disclosure, as well as
negative outcomes, such as social isolation, depression, and cyber-bullying (Best et al. 2014).

Importantly, a number of studies have found no association between frequency of technology use
and general well being (Gross 2004), nor between frequency of technology use and depressive
symptoms (Davila et al. 2012; Jelenchick et al. 2013), further highlighting the inconclusive
nature of attempts to characterize overall associations between technology use and psychological
outcomes. In fact, it may not be the quantity of social networking site usage that longitudinally
influences depressive symptoms, but rather the positive or negative quality of peer interactions or
behaviors that occur via these technologies (Davila et al. 2012). As such, researchers have
suggested the need to examine specific technology-based behaviors, as well as individual, offline
characteristics, that may help explain associations between psychological adjustment and
frequent technology use (Bessière et al. 2008; Valkenburg and Peter 2013). An investigation of
these two factors in relation to depressive symptoms is the focus of the current study.

In line with this approach, preliminary findings have implicated reciprocal associations between
various offline characteristics, technology-based behaviors, and negative outcomes. For example,
research suggests that individuals with lower self-esteem and poorer offline friendship quality
are more likely to engage in maladaptive behaviors using technology, such as posting negative or
inappropriate content and using the Internet primarily for content consumption, rather than direct
communication with peers (Mikami et al. 2010; Selfhout et al. 2009). Maladaptive technology-
based behaviors, in turn, may be associated with negative online peer feedback, as well as
increases in loneliness and depressive symptoms (Burke et al. 2010; Forest and Wood
2012; Selfhout et al. 2009), which may influence future technology use in a cyclical fashion.
Technology-Based Social Comparison and Feedback-
Seeking

An important next step in the literature is to draw on existing interpersonal models of

psychopathology to identify specific technology-based behaviors that may act as precursors to
and correlates of depressive symptoms among adolescents. Interpersonal feedback-seeking and
social comparison, which have often been conceptualized as depressogenic interpersonal
behaviors (e.g., Borelli and Prinstein 2006), may actually be facilitated by modern technologies.
These behaviors are important to adolescent identity construction processes, as adolescents seek
to ascertain normative standards of behavior and engage in reflected appraisal (i.e. evaluation of
the self based on perceived opinions of others; Harter et al. 1996). However, high levels of
technology use may be associated with increased levels of these behaviors among some
adolescents.

Modern technologies have transferred many social relations from the private to public sphere
(Subrahmanyam and Greenfield 2008), creating an atmosphere of public commentary and
performance online. As adolescents use selective self-presentation strategies to portray
themselves online in an ideal manner (Chou and Edge 2012; Gonzales and Hancock 2011), self-
focus is heightened, increasing feedback-seeking and social comparison (Manago et al. 2008),
perhaps especially upward comparisons, or comparisons with those believed to be of higher
status than the self. This may serve to intensify the issues of identity development and
interpersonal connectedness, challenging adolescents to confront them with greater constancy
and urgency (Uhls et al. 2011). Furthermore, preliminary findings suggest that negative social
comparison on Facebook may be associated with lower life satisfaction (Krasnova et al. 2013),
as well as increased rumination and depressive symptoms (Feinstein et al. 2013).
Technology Use and Individual Characteristics

In understanding the associations between technology-based SCFS and depressive symptoms, it

is essential to consider preexisting, offline characteristics that may affect these associations. For
example, popularity and gender may play a role. Popular adolescents, who are higher in self-
esteem (Babad 2001), may engage in fewer upward comparisons, experience less negative affect
as a result of these comparisons, and receive more positive feedback from peers (Mikami et al.
2010). In terms of gender, females may be more likely to compare themselves on dimensions of
physical attractiveness based on online photos (Haferkamp and Krämer 2011), perhaps making
such photos more self-relevant and threatening to self-worth (Stefanone et al. 2011).
Furthermore, associations between reassurance-seeking behaviors and depressive symptoms may
be particularly strong among adolescent girls, for whom rates of depression are higher in general
(Starr and Davila 2008), and given known situational continuities between online and offline
contexts (Mikami et al. 2010), this effect may occur online, as well. However, the possibility that
gender and popularity differences may occur within online contexts has not yet been rigorously
examined.
Study Hypotheses

First, it is hypothesized that higher levels of technology-based social comparison and feedback-
seeking behaviors (SCFS) will be concurrently associated with higher levels of depressive
symptoms, controlling for offline ERS, prior depressive symptoms, and overall frequencies of
technology use (i.e., total use of cell phones, Facebook, and Instagram). Second, it is
hypothesized that peer popularity will moderate this association, such that individuals low in
popularity will show the strongest negative association between technology-based SCFS and
depressive symptoms. Third, it is hypothesized that gender will also moderate this association,
such that the effect will be stronger for females.
Methods
Participants

The current study included 619 participants. Students were eighth and ninth grade students in
low to middle socioeconomic status (SES) schools (67 % free or reduced price lunch).
Participants were between the ages of 12 and 16 (mean age 14.6), and 57.3% were female. The
ethnic composition of the sample included 47.9% White/Caucasian, 21.1% African
American/Black, 23.4% Hispanic/Latino, 0.5% Asian, and 5.5% other. This sample closely
matched the demographic makeup of the district from which participants were recruited. All
procedures discussed below were approved by the university human subjects committee.
Procedure

All students in seventh and eighth grade were recruited, except for those in self-contained special
education classrooms, using parental consent and adolescent assent. A total of 868 students
participated in the study’s first wave (baseline). The current study was conducted 1 year later,
when students were in eighth and ninth grades. Of the original sample of 868, 90% of students
participated (n=779). Attrition was due to participants’ moving away from the area (n=14),
moving to a different school (n=20), withdrawal from the school (n=18), withdrawal from the
study (n=20), and absenteeism (n=17).

Out of the 779 students surveyed, 130 students were excluded from the analysis. Of these 130,
53 students were excluded because they indicated that they did not use technology as defined in
the study (use of cell phones, Facebook, or Instagram). The other 77 students did not complete
any of the measures related to technology use, due to the placement of these questions at the end
of the study protocol. Attrition analyses indicated that excluded students were, on average, more
likely to be male, t (775)=3.13, p=0.002, and African American, χ2(3)=24.77, p<0.001.

After preliminary analyses, two outliers were identified in the data, with values more than four
standard deviations below the mean for popularity. These outliers were removed for further
analyses. A separate analysis was conducted using a Winsorising approach, in which outliers’
popularity were set to equal the next closest values (approximately 3.25 standard deviations
below the mean). The pattern of results was consistent in both treatments of outliers; thus, results
for the analysis in which outliers were deleted are reported here. In addition, 14 subjects did not
provide information on baseline depressive symptoms and 14 subjects did not complete measures
of technology-based SCFS. Thus, final model sample was n= 619. Participants were
compensated with $10 gift cards.
Measures

All measures were self-reported and administered to students in classrooms during the school
day using computer-assisted self-interviews (CASI). For all variables except popularity and
socioeconomic status, a mean score of items was computed, with higher scores indicating higher
levels of that variable.
Socioeconomic Status

Participants’ socioeconomic status (SES) was computed by calculating median household

income from US Census tract data (www.census.gov) for each student’s home address. The
mean SES for this sample was identical to the average household income for the town in which
the data were collected, according to census data.
Depressive Symptoms
The Short Mood and Feelings Questionnaire (SMFQ; Angold et al. 1995) was used to assess
depressive symptoms, both at the current time point and 1 year prior (baseline). The SMFQ is a
13-item, unifactorial scale in which subjects endorse statements describing depressive moods and
behaviors over the past 2 weeks on a 3-point scale (0 for not true, 1 for sometimes true, and 2
for true). The SMFQ has good psychometric properties (Sharp et al. 2006) and has been widely
used to assess depressive symptoms in adolescent samples (e.g., Rothon et al. 2009). The current
sample yielded good internal consistency (Cronbach’s alpha 0.94).
Popularity

Sociometric nomination procedures were used to measure peer-reported popularity (Coie et al.
1983). As such, all subjects were presented with a roster of all grademates. Alphabetization of
the roster was reversed for a random half of the participants in order to control for order effects
in participants’ selection of names. Subjects were asked to nominate an unlimited number of
grademates whom they believed to be the most popular and the least popular (Prinstein and
Cillessen 2003). The range of nominations that participants received was between zero and 60.
The vast majority of participants received at least one nomination. Of the full sample, only 19
students did not receive any nominations. For each participant, two sums were calculated: one
for the number of most popular nominations, and one for the number of least popular
nominations. These sums were then standardized within each school grade, and a difference
score was taken between most popular and least popular standardized scores. These differences
scores were then re-standardized to create a measure of popularity, where higher scores indicated
higher levels of popularity (Prinstein and Cillessen 2003). Sociometric nomination procedures
are largely considered the most reliable and valid indices of adolescents’ popularity among peers
(Coie et al. 1983).
Technology-Based Social Comparison and Feedback-
Seeking (Technology-Based SCFS)

The Motivations for Electronic Interaction Scale (MEIS) was designed in order to assess
subjects’ attitudes and behaviors regarding the use of technology, specified as “texting,
Facebook, and other social media.” This measure was developed in three steps. First, a focus
group comprised of recent high school graduates was conducted. Students were asked to generate
examples of technology-based behaviors and attitudes toward technology use that are common
among current high school students (e.g., “I often post a status update if I think it will make
others think I am funny, nice, or cool”). Based on their answers, a pool of 34 items was
generated and administered to a sample of 261 adolescents, living in a nearby school district and
comprised of similar age, gender, and ethnic composition to the current sample. In the second
step, the scale was expanded to include 52 items, with more items added reflecting engagement
in social comparison and feedback-seeking behaviors online. The measure was administered to
158 high school students. Factor analysis revealed a single 10-item factor indicating engagement
in social comparison and feedback-seeking behaviors using technology, as well as other factors
relating to the use of technology for general communication with romantic partners, social
support seeking, and discussions about sexual health topics (Widman et al. 2014).
These items were ultimately included in the final 22-item scale, which was administered to
subjects in the current study. Subjects endorsed the personal relevance of a number of behaviors
on a 5-point scale (1 for Not at all true and 5 for Extremely true). Examples from the 10-item
Social Comparison and Feedback Seeking Subscale (MEIS-SCFS) include, “I use electronic
interaction to see what others think about how I look” and “I use electronic interaction to
compare my life with other people’s lives.” The scale showed good internal consistency
(Cronbach’s alpha 0.92).
Excessive Reassurance-Seeking (ERS)

Joiner and Metalsky (1995) developed the Reassurance-Seeking Scale (RSS) for use with adults
and later adapted it for use with children and adolescents (Joiner 1999) as a subscale of the larger
Depressive Relationships Inventory (DIRI). One critique of the original RSS is that it is very
brief (four items), and that it lacks developmental sensitivity (Starr and Davila 2008). Thus, for
the current study, a Revised ERS scale was created to include 6 additional items, all believed to
be developmentally appropriate to adolescents (e.g. “I often ask people if they think my clothes
look okay”). Another criticism of the original ERS scale is that it lacks detail, simply assessing
how often individuals request assurance that others like and care for them. Thus, the Revised
ERS scale sought to address multiple domains of reassurance-seeking appropriate to adolescents,
including reassurance-seeking about appearance (e.g., “I often ask people if I look attractive”),
gossip (e.g. “I often ask people what other people say about me”), and general liking (e.g., “I
often ask people if other people like me.”).

Similarly to the MEIS, these items were developed through the use of a focus group of recent
high school graduates, followed by pilot testing with a sample of 158 high school students.
Ultimately, the Revised ERS scale was a 10-item measure in which subjects endorsed
reassurance-seeking behaviors on a 5-point scale (1 for Not at all true and 5 for Extremely true).
The original RSS has been shown to have good psychometric properties (Joiner and Metalsky
2001) and has been used to assess depressive symptoms in adolescent samples (e.g., Prinstein et
al. 2005). The Revised ERS scale showed good internal consistency as a unifactorial scale in this
sample (Cronbach’s alpha 0.90).
Frequency of Technology Use

The Electronic Interaction Scale for Time (EIS_T) was developed to determine the average
amount of time subjects spend using specific technologies on “a typical day.” Similar to the
MEIS, this measure was developed over a three-step process. Through the use of multiple focus
groups and pilot testing among a total of 429 high school students, five communication items
were chosen and response options were created to capture the full range of daily electronic
interaction time. For the final EIS_T, subjects indicated the amount of time they spent each day
engaged in in-person communication, voice communication, non-voice cellphone use (i.e., for
“texting, games, or Internet”), Facebook use, and Instagram use. Frequencies were indicated on a
7-point scale (0 for I don’t use this, 1 for Less than 1 h, 6 for 5 or more hours). For the current
study, an average of the final three items (non-voice cellphone use, Facebook use, and Instagram
use) was used to indicate frequency of overall technology use. Self-report measures have been
widely used in previous studies assessing frequency of technology use.
Results
Descriptive Statistics

Descriptive statistics were conducted to examine the means and standard deviations of all study
variables (see Table 2). Independent sample t-tests were used to compare means on study
variables by gender. Interestingly, females reported higher average values of most study
variables, including depressive symptoms, technology-based SCFS, frequencies of technology
use, and excessive reassurance seeking. No gender differences were found in levels of
popularity.
Table 2.
Means (and Standard Deviations) of Study Variables, with Gender and Race/Ethnicity
Comparisons
Full Girls Boys t(df) American Caucasian Latino/Hispanic F(df)
sample African

Frequency of 2.80 3.23 2.21 8.39 3.42 (1.76)a 2.50 2.81 (1.57)c 11.15
technology use (1.57) (1.54) (1.43) (617)** (1.38)b (615)**

Technology-based 1.73 1.82 1.61 3.47 1.87 (0.81)a 1.72 1.56 (0.63)b 5.51
SCFS (0.77) (0.82) (0.68) (617)** (0.75) a (615)**

Depressive symptoms 0.48 0.62 0.28 8.83 0.49 (0.51) 0.44 (0.49) 0.52 (0.51) 1.32 (61
(0.51) (0.56) (0.35) (617)**

Excessive reassurance 1.50 1.60 1.36 4.68 1.55 (0.66)a 1.55 1.35 (0.53)b 3.95
seeking (0.66) (0.73) (0.50) (617)** (0.67)a (615)*

Baseline depressive 0.47 0.58 0.33 6.53 0.52 (0.52) 0.44 (0.48) 0.46 (0.43) 1.58 (61
symptoms (0.49) (0.53) (0.38) (617)**

Popularity 0.07 0.11 0.03 0.992 0.07 (0.85) 0.08 (1.19) 0.005 (0.66) 0.833
(1.00) (0.92) (1.09) (617) (615)

Open in a new tab

Row means that do not share superscripts are significantly different

*

p<0.01
**
p<0.001

Further, participants who reported using technology as defined in the study were compared to
those who indicated that they did not use technology (n=53) on key demographic variables. No
differences emerged between users and non-users of technology in terms of SES or ethnicity.
However, the 53 students who indicated that they did not use technology were more likely to be
male, t (698)=3.83, p<0.001; lower in depressive symptoms, t (698)=−2.86, p=0.004; lower in
ERS, t (698)=−3.51, p<0.001; and lower in peer-reported popularity t (698)=−4.60, p<0.001.

Pearson correlations were conducted to examine bivariate associations among all study variables
(see Table 1). Significant positive associations were found between frequency of technology use,
technology-based SCFS, and offline excessive reassurance-seeking. Popularity was positively
associated with frequency of technology use and technology-based SCFS; however, it was
negatively associated with depressive symptoms. Pearson correlations were also conducted to
examine associations between study variables and socioeconomic status (SES). Individuals lower
in SES reported higher frequencies of technology use and lower levels of popularity (see Table
1). Interestingly, although depressive symptoms were positively correlated with concurrent
technology use frequency, this association was no longer significant after accounting for other
variables in the full regression model, as discussed below.
Table 1.
Bivariate Associations Between Study Variables, Full Sample and by Gender
Full sample By gender

1 2 3 4 5 6 7 1 2

1. Frequency of technology use – – 0

2. Technology-based SCFS 0.28*** – 0.24*** –

3. Depressive symptoms 0.11** 0.34*** – −0.02 0

4. Excessive reassurance seeking 0.15*** 0.50*** 0.47*** – 0.09 0

5. Baseline depressive symptoms 0.06 0.23*** 0.60*** 0.38*** – −0.06 0

6. Popularity 0.12** 0.17*** −0.11** −0.06 −0.11** – 0.14** 0

7. Socio-economic status −0.13*** 0.01 −0.07 −0.02 −0.09* 0.17*** – −0.12* −

Open in a new tab

For associations by gender, males reported above the diagonal in bold, females reported below
the diagonal
*

p<0.05
**

p<0.01
***

p<0.001

Analyses were also conducted to determine whether means and standard deviations of study
variables differed by ethnicity (see Table 2). Results indicated that levels of ERS were
significantly lower among Latino/Hispanic participants versus Caucasian and African American
participants; levels of technology-based SCFS were significantly higher among African
American participants versus Latino/Hispanic participants; and frequency of technology use was
significantly higher among African American participants versus Latino/Hispanic and Caucasian
participants. No differences were found in levels of depression between racial/ethnic groups.
Hypothesis Testing

Hypotheses were tested within a hierarchical multiple linear regression framework using
maximum likelihood estimation in SPSS 22.0. All continuous predictor variables were mean
centered to reduce multicollinearity and to aid in ease of interpretation. Baseline depressive
symptoms, excessive reassurance seeking, and overall frequency of technology use were entered
as covariates in an initial step. The main effects of gender, popularity, and technology-based
SCFS were added in the second step.

In order to test the hypothesis that popularity moderates the relationship between technology-
based SCFS and depression symptoms, an interaction term was created by computing the product
of the centered values of technology-based SCFS and popularity. In order to test the gender
moderation hypothesis, another interaction term was created by computing the product of gender
and the centered value of technology-based SCFS. A third interaction term was created between
gender and popularity. These three two-way interaction terms were added at the third step. A
three-way interaction term was created and entered on a fourth step (see Table 3).
Table 3.
Prediction of Depressive Symptoms by Technology-Based Social Comparison and Feedback-
Seeking (SCFS), Popularity, and Gender (n=619)
 Depressive symptoms

Step statistics Final statistics

Predictors ΔR2 b (se b) β b (se b) β

Step 1, covariates 0.42**

Baseline depressive symptoms 0.51 (0.04) 0.48*** 0.45 (0.04) 0.43***

Excessive reassurance seeking 0.21 (0.03) 0.28*** 0.15 (0.03) 0.19***

Frequency of technology use 0.01 (0.01) 0.04 −0.01 (0.01) −0.04

Step 2, main effects 0.04**

Technology-based SCFS 0.09 (0.02) 0.13*** 0.14 (0.03) 0.21***

Gender (female) −0.19 (0.03) −0.18*** −0.20 (0.03) −0.19***

Popularity −0.04 (0.02) −0.07* −0.04 (0.02) −0.08†

Step 3, two way interactions 0.01*

SCFS × gender −0.09 (0.04) −0.08* −0.09 (0.04) −0.08*

SCFS × popularity −0.05 (0.02) −0.09** −0.07 (0.03) −0.11**

Popularity × gender 0.02 (0.03) 0.02 0.02 (0.03) 0.02

Step 4, three way interaction 0.00

SCFS × gender × popularity 0.03 (0.04) 0.03

Total R2 0.47**

Open in a new tab

Gender was coded as 0 for females, and 1 for males. SCFS=social comparison and feedback-
seeking
†

p<0.06;
*
p<0.05;
**

p<0.01;
***

p<0.001; All variables mean centered with the exception of Gender

The full regression model explained a significant proportion of the variance in depressive
symptoms, R2=0.470, p< 0.001. In support of the first hypothesis, results revealed a significant
main effect of technology-based SCFS on depressive symptoms (B=0.21, p<0.001).
Additionally, in support of the second and third hypotheses, analyses revealed a significant
technology-based SCFS × gender interaction effect (B=−0.08, p<0.05), as well as a significant
technology-based SCFS × popularity effect (B=−0.11, p<0.01). This model was further analyzed
with the addition of ethnicity and SES variables as covariates. The pattern of significant and non-
significant effects remained the same; thus, to present a more parsimonious model, ethnicity and
SES were not included in results.

Interactions were probed following procedures outlined by Aiken and West (1991) and using
interaction utilities provided by Preacher et al. (2006). First, models were reduced by removing
covariates and non-significant interaction terms (Holmbeck 2002). For the gender interaction,
simple intercepts and slopes for the regression of technology-based SCFS on depressive
symptoms were computed for both males and females. Results revealed significant slopes for
both girls, b (se)=0.239 (0.03); p<0.001, and boys, b (se)=0.133 (0.04); p<0.001, indicating that
greater levels of technology-based SCFS were associated with greater levels of depressive
symptoms for both genders. The slope for girls was significantly steeper than for boys, indicating
that the effect of technology-based SCFS on depressive symptoms is stronger among females
(see Fig. 1).
Fig. 1.

Open in a new tab

Plot of simple slopes for technology-based social comparison and feedback-seeking (SCFS) by
gender interaction on depressive symptoms. Note that technology-based SCFS is mean centered

For the popularity interaction, simple intercepts and slopes for the regression of technology-
based SCFS on depressive symptoms were computed at the mean of popularity, as well as at one
standard above and below the mean. Results suggested stronger associations between
technology-based SCFS and depressive symptoms for individuals low in popularity, (−1 SD), b
(se)=0.329 (0.04), p<0.001, than for individuals high in popularity (+SD), b (se)=0.181
(0.03), p<0.001 (see Fig. 2). To further explore this effect, the Johnson-Neyman (J-N) technique
was used to identify the region of significance (Bauer and Curran 2005), that is, the values of
popularity for which technology-based SCFS had a significant effect on depressive symptoms.
Results suggested that the association was significant at all centered values of popularity less
than 2.08 (approximately two standard deviations above the mean). The slope at this boundary
was b (se)=0.10 (0.05), p<0.05. This indicates that technology-based SCFS may be associated
with depressive symptoms for the majority of individuals, but with stronger associations for
those lower in popularity.
Fig. 2.

Open in a new tab

Plot of simple slopes for technology-based social comparison and feedback-seeking (SCFS) by
popularity interaction on depressive symptoms. Note that technology-based SCFS is mean
centered
Discussion

Interpersonal theories of adolescent depression highlight bidirectional associations among

depressive symptoms, social experiences, and interpersonal behaviors that occur within an
increasingly complex social environment during this time period. In recent years, the widespread
adoption of social technologies, including text messaging, cell phone use, and SNS, has
fundamentally transformed the adolescent social landscape; however, little is known about the
role of technology-based social experiences in these depressogenic interpersonal processes.
Researchers have called for a more nuanced understanding of the specific online behaviors and
pre-existing individual characteristics that may influence these associations (Bessière et al.
2008; Davila et al. 2012; Kraut et al. 2002; Valkenburg and Peter 2013). Findings have the
potential to inform interpersonal models of adolescent depression that better account for modern
adolescents’ social environments.
Although results are preliminary, and no strong conclusions about directionality can be made
given the concurrent nature of the data, the current investigation provides a valuable contribution
to the literature. Results identified technology-based SCFS as one online behavior that may have
implications for both online and offline functioning. In support of the first hypothesis,
technology-based SCFS was found to be associated with depressive symptoms, controlling for
overall frequencies of technology use, offline ERS, and prior depressive symptoms. In support of
the second hypothesis, findings indicated that this association might be dependent on the offline
characteristic popularity, with the strongest associations between technology-based SCFS and
depressive symptoms found for adolescents lower in popularity. In support of the third
hypothesis, gender was also found to moderate the association between technology-based SCFS
and depressive symptoms, with the association stronger for females. Each of these findings will
be discussed in turn.
Associations Between Technology-Based SCFS and
Depressive Symptoms

Findings regarding the association between technology-based SCFS and depressive symptoms
may be understood in the context of several theories. First, it may be that the online environment
facilitates higher levels of SCFS. The hyperpersonal model of computer-mediated
communication (CMC; Walther et al. 2011) suggests that certain components of technology-
based interaction serve to intensify the process of identity construction through increased
feedback and decision-making within these environments. Specific components of this model
include selective self-presentation, or the potential for more deliberate portrayals of the self in an
online context, and idealization, or positive assumptions about others for whom limited online
information is available. In other words, the online context fosters idealized self-presentations
and individuals are likely to make positive assumptions about others subsequent to viewing
online content. Given that young people spend the majority of their time on SNS looking at
peers’ profiles and photos, rather than posting or updating their own profiles (Pempek et al.
2009), adolescents may be especially likely to engage in technology-based SCFS and may be
vulnerable to negative comparisons with their peers. Given that identity development is a stage-
salient task characteristic of adolescence, and that social comparison and feedback-seeking are
essential to this process, it is not surprising that a forum that may serve to intensify this process
may be associated with higher levels of these behaviors.

Second, adolescents engaging in technology-based SCFS may form distorted perceptions of their
peers, leading them to engage in harmful upward comparisons, or to doubt the sincerity of
positive feedback that is sought online, and experience decreases in mood or self-esteem. Chou
and Edge (2012) suggest that frequent users of technology employ certain heuristics that lead
them to believe that “life is not fair” and “others are happier and living better lives.” For
example, according to the availability heuristic (Tversky and Kahneman 1973), young people
who frequently engage with technology may more easily recall information encountered online
when forming impressions of others. The tendency for selective self-presentation online may
increase the probability that adolescents encounter, and thus recall, distorted positive perceptions
of their peers’ lives. Furthermore, given the very large size of a typical adolescent’s online social
network, it is likely that many online connections are mere acquaintances offline. Thus, in
forming impressions of individuals that they do not know well offline, correspondence bias
(Gilbert and Malone 1995) may lead adolescents to assume that others’ photos and text reflect
stable personality traits, rather than situational factors.

Third, it may be that depressive symptoms precede and predict technology-based SCFS. Some
research suggests that depressed and dysthymic individuals may be motivated to seek out
negative information about others (Wenzlaff and Beevers 1998). Interpersonal theories of
depression similarly suggest that within offline social worlds, depressed and depression-prone
individuals are more likely to engage in overall higher levels of social comparison (Gibbons and
Buunk 1999) and feedback-seeking behaviors (Hames et al. 2013). Situational continuity
between offline and online contexts (Mikami et al. 2010) suggest that this effect is likely to occur
online, as well. In doing so, depressed individuals may prefer to compare themselves to others
perceived to be equal to or less fortunate than them. This may explain why depression and
technology-based SCFS are significantly associated. Unfortunately, within a computer-mediated,
hyperpersonal environment in which peers selectively portray the most positive aspects of their
lives, however, these downward comparisons may not be possible. In an online world where
users portray themselves in an ideal manner, depressed individuals may be stymied in their
efforts to seek out negative information about others, potentially experiencing a worsening of
symptoms in light of others’ perceived happiness.
Effects of Popularity

Consistent with the second hypothesis, another important finding revealed that popularity
moderated of the association between technology-based SCFS and depressive symptoms.
Notably, findings suggested stronger associations between technology-based SCFS and
depressive symptoms for unpopular individuals. Downward comparisons may be especially
challenging for adolescent low in popularity. Prior research suggests that individuals lower in
popularity and self-esteem receive less positive feedback on their social networking profiles
(Mikami et al. 2010) and post updates that are higher in negativity and lower in positivity (Forest
and Wood 2012). While positive feedback on SNS has been found to enhance adolescents’ self-
esteem, negative feedback has been found to decrease self-esteem (Valkenburg et al. 2006).
Thus, it may be that unpopular adolescents are not only more likely to post or send negative
content, but also to receive negative feedback. In seeking out feedback from peers, unpopular
adolescents may actually be garnering self-relevant information that is harmful to their self-
esteem and related to increases in depressive symptoms. Additionally, given the substantial
overlap between online and offline networks (Reich et al. 2012), it is likely that adolescents who
are unpopular offline have fewer online friends. Manago et al. (2012) posit that larger online
networks and perceived audiences predict life satisfaction and perceived social support. Thus, it
may be that, when seeking feedback online, lower status adolescents perceive smaller audience
sizes for their posted content, resulting in feelings of decreased peer support and overall life
satisfaction.
Effects of Gender
Finally, consistent with the third hypothesis, gender moderated the association between
technology-based SCFS and depressive symptoms. In particular, results suggested that the
association between technology-based SCFS and depressive symptoms was particularly strong
for females compared to males. Prior work has indicated important differences between
adolescent girls and boys that may be relevant to an online context, with girls more likely
experience depressive symptoms as the result of reassurance-seeking behaviors offline (Prinstein
et al. 2005). Furthermore, girls are more likely to prioritize and compare themselves on
dimensions of physical attractiveness online (Haferkamp and Krämer 2011; Jones 2001). Given
the emphasis on photo sharing in today’s popular social networking tools (i.e., Facebook,
Instagram), as well as the increased likelihood that girls will post photos compared to boys, it
may be that online, girls are drawn to comparisons that are more self-relevant, and thus more
threatening to self-worth (Stefanone et al. 2011).

Furthermore, established interpersonal theories of depression show that the link between
interpersonal stressors and depressive symptoms may be particularly strong for girls (Rudolph
2002); thus, insofar as technology-based SCFS present a source of interpersonal stress for girls,
these behaviors are likely to be associated with depressive symptoms. It is important to note that
given limited research in this field, proposed theories on the moderating influences of gender are
speculative; more research is needed to clarify and expand upon this potential moderator.

Findings from this study offer rare data to understand the association between offline
characteristics, online behavior, and adolescent depressive symptoms. Importantly, results did
not identify any overall association between frequency of technology use and depressive
symptoms. Rather, findings suggest the importance of exploring specific technology-based
behaviors and offline, individual characteristics in identifying for whom and under what
conditions associations with depressive symptoms may be present. Although preliminary, these
findings may have important implications for identifying adolescents for whom frequent
technology use may be both a precursor to and an outcome of maladaptive psychological
adjustment. In this emerging field, relevant theories remain speculative and suggest the need for
further research.
Limitations and Conclusions

This study provides a critical initial exploration of the associations among popularity, gender,
online behavior, and depressive symptoms among adolescents and provides a much-need
contribution to the literature on the psychosocial correlates of technology-based behaviors.
However, future research should address these preliminary findings within a prospective
longitudinal framework. Although statistical controls in the model allowed for examination of
effects over and above those of prior depressive symptoms, further work is needed to rigorously
assess temporal relationships between study variables, perhaps testing for the presence of
transactional effects between depressive symptoms and technology-based behaviors (Valkenburg
and Peter 2013).
Another limitation of this study is its reliance on self-report measures, which are subject to recall
and other biases. Adolescents’ reports of technology use frequency in the study are consistent
with nationally representative statistics of over 2000 students, collected by the Kaiser Family
Foundation (Rideout et al. 2010). However, future research should incorporate naturalistic
methods, including observational coding of adolescents’ media output, to determine the accuracy
of reports on technology-based SCFS and other variables. Initial studies using direct observation
of adolescent media content (e.g. Underwood et al. 2012) have shown promising results.
Similarly, given the lack of established measures regarding technology-based behaviors, future
research should aim to develop and validate assessments of adolescents’ engagement with social
technologies. It should also be noted that, as is typical in school-based samples, mean levels of
depressive symptoms were very low in the current analyses. Thus, although providing
preliminary insight into these effects within a community setting, results may not generalize to a
clinical sample. A final limitation of the current study was the inability to fully assess differences
in outcomes by ethnicity or SES. Although the study’s large and diverse sample provided the
opportunity to examine effects across different ethnicities, it is possible that cell sizes were too
small to examine potential interaction effects. Additionally, the use of census tract data provides
only a rough estimate of families’ true SES; more sensitive measures of SES should be used in
future studies.

In summary, the current study provides novel preliminary evidence that technology-based social
comparison and feedback-seeking behaviors may be associated with depressive symptoms
among adolescents, controlling for overall technology use, prior depressive symptoms, and
offline ERS. Furthermore, popularity and gender may play a role in this effect, such that the
association between these behaviors and depressive symptoms is particularly strong among
adolescents low in popularity and among females. Adolescents’ social environments are
increasingly dependent on the existence of social technologies, including cell phones, text
messaging, and SNS. The current findings highlight the importance of understanding how these
modern social environments may intersect with existing interpersonal models of
psychopathology.

Twenge, J. M., Joiner, T. E., Rogers, M. L., & Martin, G. N. (2017). Increases in
depressive Symptoms, Suicide-Related Outcomes, and suicide rates among U.S.
adolescents after 2010 and links to increased new media screen time. Clinical
Psychological Science, 6(1), 3–17. https://doi.org/10.1177/2167702617723376

Abstract
In two nationally representative surveys of U.S. adolescents in grades 8 through 12 (N =
506,820) and national statistics on suicide deaths for those ages 13 to 18, adolescents’
depressive symptoms, suicide-related outcomes, and suicide rates increased between 2010
and 2015, especially among females. Adolescents who spent more time on new media
(including social media and electronic devices such as smartphones) were more likely to report
mental health issues, and adolescents who spent more time on nonscreen activities (in-person
social interaction, sports/exercise, homework, print media, and attending religious services)
were less likely. Since 2010, iGen adolescents have spent more time on new media screen
activities and less time on nonscreen activities, which may account for the increases in
depression and suicide. In contrast, cyclical economic factors such as unemployment and the
Dow Jones Index were not linked to depressive symptoms or suicide rates when matched by
year.

Keles, B., McCrae, N., & Grealish, A. (2020). A systematic review: the influence of social media on
depression, anxiety and psychological distress in adolescents. International journal of adolescence and
youth, 25(1), 79-93. https://www.tandfonline.com/doi/abs/10.1080/02673843.2019.1590851

While becoming inextricable to our daily lives, online social media are blamed for
increasing mental health problems in younger people. This systematic review
synthesized evidence on the influence of social media use on depression,
anxiety and psychological distress in adolescents. A search of PsycINFO,
Medline, Embase, CINAHL and SSCI databases reaped 13 eligible studies, of
which 12 were cross-sectional. Findings were classified into four domains of
social media: time spent, activity, investment and addiction. All domains
correlated with depression, anxiety and psychological distress. However, there
are considerable caveats due to methodological limitations of cross-sectional
design, sampling and measures. Mechanisms of the putative effects of social
media on mental health should be explored further through qualitative enquiry
and longitudinal cohort studies.