Journal of Adolescent Health 58 (2016) 426e432

www.jahonline.org

Original article

What’s Keeping Teenagers Up? Prebedtime Behaviors

and Actigraphy-Assessed Sleep Over School and Vacation
Emily Harbard a, Nicholas B. Allen, Ph.D. b, c, John Trinder, Ph.D. b, and Bei Bei, D.Psych., Ph.D. a, d, *
a
Institute of Cognitive and Clinical Neurosciences, School of Psychological Sciences, Faculty of Biomedical and Psychological Sciences, Monash University, Victoria, Australia
b
Melbourne School of Psychological Sciences, University of Melbourne, Victoria, Australia
c
Department of Psychology, University of Oregon, Eugene, Oregon
d
Department of Psychiatry, Centre for Women’s Mental Health, Royal Women’s Hospital, University of Melbourne, Victoria, Australia

Article history: Received June 10, 2015; Accepted December 8, 2015

Keywords: Prebedtime behaviors; Adolescents; Social media; Vacation; School-term; Bedtime; Chronotype; Presleep arousal; Family;
Sleep duration

A B S T R A C T
IMPLICATIONS AND
CONTRIBUTION
Purpose: Technology-related behaviors (e.g., computer use) before bedtime (BT) have been asso-
ciated with poorer and shorter sleep in adolescents; however, less is known about other behaviors
This study assessed a wide
in relation to sleep. This study characterized a range of behaviors in the hour before bed (i.e., pre- range of prebedtime behav-
BT behaviors [PBBs]) and examined their relationship with sleep parameters during school and iors in adolescents during
vacation periods (i.e., restricted and extended sleep opportunities, respectively). Mechanistic roles both school and vacation
of chronotype and cognitive presleep arousal (PSAcog) were also examined. periods and examined their
Methods: During the last week of a school term and throughout a 2-week vacation, 146 adoles- associations with objective
cents (47.26% male, age M
standard deviation ¼ 16.2
1.0 years) from the general community sleep parameters. Findings
completed daily sleep measure using actigraphy, self-report measures on PBBs and PSAcog (Pre- identified behaviors detri-
sleep Arousal Scale) for both school and vacation periods, and chronotype (MorningnesseEven- mental to, and protective of,
ingness Questionnaire). adolescents’ sleep. Potential
Results: Adolescents engaged in a variety of behaviors before bed. Notably, playing video games mechanisms described have
was associated with significantly later school and vacation BT and shorter school sleep duration practical implications for
(controlling for chronotype). During vacation, online social media was associated with significantly future interventions.
longer sleep onset latency, and this relationship was mediated by higher PSAcog. In contrast, on
school nights, spending time with family was associated with significantly earlier BT and longer
sleep duration.
Conclusions: Technology-related PBBs video games and online social media were risk factors for
shorter and poorer sleep, whereas time with family was protective of sleep duration. In addressing
sleep problems in adolescents, therapeutic procedures that target the potentially addictive nature
of technology use and reduce PSAcog were implicated.
Ó 2016 Society for Adolescent Health and Medicine. All rights reserved.

Insufficient and poor sleep in adolescents, especially during

Conflicts of Interest: The authors have indicated no financial conflicts of
interest.
* Address correspondence to: Bei Bei, D.Psych.(Clinical), Ph.D., School of Psy-
chological Sciences, Faculty of Medicine, Nursing and Health Sciences, Monash
University, 18 Innovation Walk (Bld 17) Clayton Campus, Victoria 3800, Australia.
E-mail address: bei.bei@monash.edu (B. Bei).
school days, is widely reported [1] and is associated with
numerous consequences, such as impaired academic perfor-
mance [2], increased risk of motor vehicle accidents [3], and
emotional difficulties [4]. The causes of sleep problems in ado-
lescents are multi-factorial. Biological (e.g., delays in circadian

1054-139X/Ó 2016 Society for Adolescent Health and Medicine. All rights reserved.
http://dx.doi.org/10.1016/j.jadohealth.2015.12.011

Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432
phase and decreases in homeostatic drive) and psychosocial
factors (e.g., decreased parental control [5], increased academic
and social demands [6]) are permissive of later bedtime (BT),
while school start times contribute to early risetime (RT), pro-
moting sleep restriction [7]. Pre-bedtime behaviors (PBBs) are
important to understand, as they are modifiable factors that may
be relevant in addressing sleep problems in adolescents.
Technology-related prebedtime behaviors
Technology-related PBBs are consistently associated with
shorter sleep duration and lower sleep quality. More time spent
watching television, using the Internet, and computer gaming
are associated with later BT and shorter time in bed on weekdays
and later RT on weekends [8]. Relevant to sleep are the location
and timing of technology use. Compared with adolescents
without bedroom access to this technology, those with access in
bedrooms use these devices more and have later BT and shorter
total sleep time (TST) on weekdays [9]. Within the hour before
bed, frequent video gaming, computer use, listening to Mp3, and
use of mobile phones are associated with later self-reported BT
on both school and weekend nights [10]. Weekday sleep duration
is also significantly shorter for adolescents who usually/always
engaged in watching television, playing video games, using
computers, using mobile phones for calling/texting or listening
to music before bed, than those who sometimes/never engaged
[11]. More frequent engagement in technology-related PBBs,
such as using Mp3 players, computers, and mobile phones before
bed, is also associated with longer sleep onset latency (SOL; time
taken to initiate sleep) [10e12].
Three potential mechanisms have been proposed to explain
the links between technology-related behaviors and adolescents’
sleep [13]: (1) technology-related behaviors may displace sleep
time by delaying BT and shortening sleep duration; (2) evening
media use may contribute to greater cognitive arousal (i.e.,
mental activation such as thoughts and worries), which is asso-
ciated with later BT and longer SOL in both adults [14] and ad-
olescents [15]; and (3) light from screen-based devices may
supress melatonin and delay sleep onset [16].
Nontechnology-related prebedtime behaviors
Few studies have examined the relationship between
nontechnology-related PBBs and sleep. Reading at BT was shown
to promote earlier school BT and thus is protective of sleep
duration [17]. While not directly examined in relation to sleep,
spending more time with family is associated with greater
parental control over BT and media usage [18], and greater
parental control is associated with earlier BT [10]. Conversely,
homework, spending time with friends, and physical exercise can
be hurdles to adequate sleep in adolescents [19,20].
Gaps in the literature and the present study
There are several important gaps in the literature: (1)
nontechnology-related PBBs are poorly understood; (2) with one
exception [11], studies have not controlled for chronotype, an
important confounder of the association between PBBs and sleep
timing, making it unclear whether PBBs lead to later BT beyond
circadian preference or whether evening preferences drive
engagement in PBBs; (3) most studies have relied on retrospec-
tive self-reports of sleep, which is open to recall and subjective
Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
427
biases [21]; (4) the relationship between PBBs and sleep during
relatively unconstrained sleep opportunity (i.e., vacation) is un-
explored [22]; and (5) despite plausible theories [13], mecha-
nisms underlying the relationships between PBBs and sleep have
not been well examined. Therefore, the present study aimed to:
(1) Characterize patterns of both technology and
nontechnology-related PBBs during school (School) and
vacation (Vacation) periods using a longitudinal design.
(2) Examine how PBBs relate to objectively measured sleep
timing (BT and RT) and duration (TST) controlling for chro-
notype and SOL (chronotype was not controlled for in SOL
analyses as it shared negligible correlation with SOL (r ¼ .02
and .04 for School and Vacation, respectively). Based on the
existing literature, these sleep variables represent key as-
pects of sleep that might be directly affected by PBBs [13].
(3) Examine whether cognitive presleep arousal (PSAcog) is a
mechanism mediating significant associations between
technology-related PBBs and longer SOL.
Methods
Participants
This study aimed to recruit a community sample of adoles-
cents attending Years 10, 11, and 12 in schools in Melbourne,
Australia. As described in the procedures, no exclusion criteria
were applied at the time of recruitment.
Materials
Actigraphy. Actigraphy is widely used to study sleep/wake pat-
terns in adolescents [23], providing objective estimates of sleep
duration and quality close to PSG [24]. This study used compa-
rable models of Actiwatch-2 and Actiwatch-64 (Mini Mitter,
Bend, OR) [25]. Data were collected with 1-minute epochs and
analyzed based on “medium” threshold for sleep/wake detection
in Actiware 5.5. Both actigraph models contain an “event
marker” button for registering BT and RT. The following variables
were generated using Actiware: BT, RT, TST, and SOL (see [22] for
actigraphy data processing). For the School variables, data from
Monday to Thursday nights (i.e., 24-hour cycles that started and
ended on a school day) were averaged; for Vacation, data during
a 2-week vacation were averaged.
Prebedtime behaviors. The Prebedtime Behavior Questionnaire
(PBBQ) is a self-report inventory that assesses frequencies of 25
evening behaviors adolescents commonly engaged in, such as
reading a book, web browsing (Table 1). No existing scale mea-
sures PBBs in adolescents; therefore, items in the PBBQ were
selected through a focused discussion among seven researchers
(including two licensed Clinical Psychologists) specialized in
adolescent sleep and mental health. An “other” item was
included for capturing any PBBs not listed on the PBBQ. Partici-
pants were asked to rate how often they engaged in each
behavior within the hour before intending to sleep over the past
week on a four-point scale: 0 (never), 1 (once or twice a week), 2
(three to four times a week), and 3 (most nights).
Chronotype. The MorningnesseEveningness Questionnaire
(MEQ) is a 19-item scale on preferred timing of physical and
mental activities (i.e., chronotype) [26] with higher scores
428 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432

Table 1 actigraph by mail after the first week of the following school
Items in the pre-BT behavior questionnaire term. Questionnaires were conducted through online surveys,
1. Reading a book 14. Pray or meditate with demographic information and MEQ administered during
2. School work (e.g., assignments, 15. Watching television the last 2 weeks of the school term, and the PBBQ and Presleep
studying for examinations)
Arousal Scale in the last 2 weeks of the school term and again in
3. Sports and exercise 16. Watching DVDs
4. Snacking 17. Listening to music the second week of the vacation.
5. Drinking caffeinated beverages 18. Listening to talking audios such as
(e.g., tea, coffee, caffeinated radio, audiobooks, or podcasts
sodas) Statistical analyses
6. Having drinks that contain 19. Playing video games (including
alcohol games on video-game consoles, Preliminary analyses. Responses to the PBBQ were screened and
computer, iPhone/iPod, mobile
edited. First, all responses to the other item (e.g., “watching TV
phone, or other portable gaming
devices) shows”) fell into an existing behavior (e.g., watching television)
7. Household obligations 20. E-mail and were merged with the existing item by adopting the
8. Personal hygiene 21. Online chat/discussions higher frequency of the two. Second, >70% of PBBs at both
(chatroom, instant messengers, time points had low (<20%) endorsement for 3 (most nights);
including audio and video chats)
9. Part-time job 22. Online social networking (e.g.,
therefore, this response was merged with three to four times a
Facebook, Twitter, Myspace) week and recoded as 2 (three or more times a week) for all
10. Spending time with family 23. Web browsing for blogs, news, behaviors. This resulted in a relatively equal number of re-
and information sponses for the three frequency categories never (0), once or
11. Spending time with friends 24. Streaming/downloading online
twice a week (1), and three or more times a week (2), which
media (music, photos, videos)
12. Phone and text messaging 25. Other (please specify) were used for all subsequent analyses. Finally, items with more
13. Writing in a diary or journal _______________ than a 75% endorsement of never were excluded from further
analysis due to small sample sizes (see Supplementary Table 1
Participants were asked to rate how often they engaged in each behavior within
the hour before intending to sleep over the past week on a four-point scale: for details).
0 (never), 1 (once or twice a week), 2 (three to four times a week), and 3 (most As a necessary assumption for linear regression analyses, only
nights). PBBs that demonstrated linear relationships with sleep outcomes
(assessed using comparative boxplots [31]) were used. All vari-
ables were screened for distributions, outliers, and other condi-
indicating a stronger morning preference. The MEQ is well vali- tions required by statistical assumptions.
dated in adults, [27] and used successfully in adolescents [28,29].
In this study, the MEQ showed good reliability (a ¼ .80). Primary analyses. Analyses were conducted in SPSS (version
20; IBM Corp., Armonk, NY). Descriptive statistics were used
Presleep arousal. The Presleep Arousal Scale [30] is a well- to describe the pattern of PBBs during School and Vacation
validated 16-item self-report measure of presleep arousal (PSA) (Aim 1). Regression analyses were used to examine the re-
over the past week, with eight items assessing PSAcog (e.g., being lationships between PBBs and sleep variables (Aim 2). For BT
mentally alert, active) and eight assessing physiological PSA. This and RT, hierarchal regressions were conducted with MEQ in
study utilized the PSAcog subscale, which showed good reli- the first and individual PBBs entered in the second step.
ability in this sample (a ¼ .87 and a ¼ .84 for School and Vaca- Assuming a PBB uniquely explains 5.5% of the variance in the
tion, respectively). sleep variable, a sample size of 140 provides at least 80% po-
wer with or without chronotype as a covariate.
Mediation analysis was conducted to examine whether PSA-
Procedures
cog mediated significant relationships between PBBs and SOL
(Aim 3). Statistical significance was based on two-tailed p < .05
This study was approved by the Human Research Ethics
and bootstrapped confidence intervals from 20,000 resamples.
Committee of the University of Melbourne and the Monash
Missing data (<5.59% in all variables) were addressed using
University Human Research Ethics Committee. Participants
pairwise (regression) and listwise (mediation) deletion.
were recruited via flyers in community centres and school
e-newsletters with two movie vouchers as incentives. Informed
consent was obtained from participants and their guardians. Results
The Australian school year consists of four terms with a typical
2-week vacation in between (except a 5-week summer vaca- Sample characteristics
tion). Data were collected around five school-vacation transi-
tions in 2011 and 2012; none involved the summer vacation. In A total of 146 participants (47.3% male, age M
SD ¼ 16.2

each round, participants were mailed an actigraph, which they 1.0 years) were recruited (65.1% Caucasian, 26.7% Asian). A small
wore continuously on the nondominant hand during the last proportion of participants self-reported a depressive, anxiety, or
school term week (we have previously shown that sleep pat- attention deficit hyperactivity disorder (6.16%) or sleep condi-
terns at the start and end of a school term were similar [22]; tions such as insomnia and bruxism (11.64%; see further details
School; Monday to Thursday night) and the following 2-week in [22]). These participants were not excluded to maintain the
vacation (Vacation; Friday night until the last Saturday night representativeness of a community sample. Three females were
before the next school term; more repeated measures during excluded from analysis, one due to the presence of multiple sleep
Vacation were necessary given its elevated intraindividual disorders with daily naps, one due to a lost actigraph, whereas
variability compared with School [22]). They returned the the third dropped out due to illness.

Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432 429

Table 2 Sleep timing (bedtime and risetime). Evening chronotype was

Descriptive statistics of continuous variables associated with later sleep timing during School and Vacation (all
School Vacation p < .001), explaining 21.4% (School) and 24.9% (Vacation) of BT
M SD n M SD n variance, and 18% (School) and 16.7% (Vacation) of RT variance.
During School, time with family was associated with earlier BT
PSAcog 19.77 6.44 138 17.62 6.04 141
MEQ 48.25 8.51 142 e e e (p ¼ .010) and accounted for an additional 3.8% of BT variance
BT 23:14 61.96 141 00:38 79.95 143 beyond MEQ. Conversely, snacking predicted later BT (p ¼ .024)
RT 07:19 42.50 141 09:38 80.39 143 and uniquely accounted for 3% of variance. During School and
TST (min) 398.01 46.07 141 430.49 47.28 143 Vacation, playing video games predicted later BT controlling for
SOL (min) 15.64 11.29 141 20.52 12.59 143
chronotype (both p < .05), explaining 2.6% and 2.1% of additional
All repeatedly measured values are significantly different between School and variance, respectively; similarly, more frequent online chat pre-
Vacation at p < .001, N ¼ 143. dicted a later RT during Vacation (p ¼ .011, DR2 ¼ .039). No sig-
BT ¼ bedtime; M ¼ mean; MEQ ¼ MorningnesseEveningness Questionnaire;
PSAcog ¼ cognitive presleep arousal; RT ¼ risetime; SD ¼ standard deviation;
nificant associations were found between School RT and any
SOL ¼ sleep onset latency; TST ¼ total sleep time. PBBs.

Sleep duration (total sleep time). MEQ was a significant predictor

Patterns of prebedtime behaviors over school and vacation (Aim 1)
Descriptive statistics for continuous variables are in Table 2,
and responses to PBBs over School and Vacation are shown in
Figure 1. Compared with School, Vacation pre-BT engagement in
school work, listening to music, personal hygiene, Web browsing,
and social mediaerelated behaviors were significantly lower,
although watching television, watching DVDs, and spending time
with friends were significantly higher based on Wilcoxon tests
(all p < .05).
Prebedtime behaviors and sleep (Aim 2)
Table 3 summarizes the findings below; PBBs that shared a
linear but nonsignificant relationship with sleep parameters are
in Supplementary Table 2.
of TST (p ¼ .011, R2 ¼ .047). Controlling for chronotype, during
School, more frequent time with family and video games predicted
longer (p ¼ .002, DR2 ¼ .063) and shorter (p ¼ .001, DR2 ¼ .070)
TST, respectively. No PBBs in this study significantly predicted
TST during Vacation.
Sleep onset latency. During School, no PBBs in this study signif-
icantly predicted SOL. During Vacation, however, more frequent
online chat, social networking, and listening to music all predicted
longer SOL (all p < .05), accounting for 4.2%, 4.0%, and 6.5% of the
variance, respectively.
The mediating role of cognitive presleep arousal (Aim 3)
Social networking and online chat both showed significant
relationships with Vacation SOL. Given their conceptual overlap

Figure 1. Comparison of average engagement in PBBs over School and Vacation periods ranging between 0 (never), 1 (once or twice a week), and 2 (three or more
times a week). *p < .05 based on Wilcoxon-Signed Rank Statistics.

Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
430 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432

Table 3
Significant relationships between pre-BT behaviors and sleep variables for school and vacation

School Vacation

PBB B (CI) PBB B (CI)

BTa Time with family 15.46* (27.14 to 3.77) Video games 13.61* (.09e27.13)
Snacking 14.29* (1.94e26.64)
Video games 11.77* (.88e22.66)
RTa None None Online chat 19.79* (4.70e34.88)
TSTa Time with family 14.81** (5.33e24.29) None None
Video games 14.39** (23.12 to 5.66)
SOL None None Online chat 3.23* (.66e5.81)
Social networking 3.11* (.56e5.66)
Listening to music 3.82** (1.40e6.25)

B ¼ unstandardized regression coefficients; BT ¼ bedtime (minutes past midnight); CI ¼ 95% confidence interval; PBB ¼ pre-BT behavior; RT ¼ risetime (minutes after 6
A.M.); SOL ¼ sleep onset latency; TST ¼ total sleep time.
a
Signifies that when BT and RT were examined as outcomes, coefficients presented here were estimated with chronotype (scores on the MorningnesseEveningness
Questionnaire) controlled for.
*p < .05; **p < .01.

(i.e., both involve social interaction over the Internet), they were
summed as online social media, which demonstrated a linear
relationship with SOL and PSAcog during Vacation. Mediation
analysis (summarized in Figure 2) showed that both PSAcog and
online social media had significant direct effects on Vacation SOL,
with more frequent online social media and higher PSAcog
associated with significantly longer SOL. The indirect effect from
online social media to SOL via PSAcog was also statistically sig-
nificant (b ¼ .072, standard error ¼ .32, confidence interval ¼
.23e1.54), suggesting that PSAcog mediated the relationship
between online social media and SOL during Vacation. Although
listening to music significantly predicted longer SOL, it was not
linearly associated with PSAcog, and therefore mediation anal-
ysis was not conducted.
Discussion
Adolescents engaged in a variety of technology and
nontechnology-related PBBs during school and vacation periods.
During School, video games were associated with later BT and
shorter TST, whereas time with family was associated with earlier BT
and longer TST. During Vacation, video games were associated with
later BT, whereas online social media was associated with longer SOL,
and this relationship was partially mediated by higher PSAcog.
Overall patterns of prebedtime behavior
Consistent with the high rates of technology use in adoles-
cents [32], engagement in technology-related behaviors was
high over both School and Vacation. Pre-BT social media
Figure 2. Unstandardised regression coefficients for the relationship between
online social media and SOL as mediated by PSAcog. The unstandardized
regression coefficients between online social media and SOL controlling for
PSAcog (direct effect) is in parentheses. N ¼ 141. *p < .05; **p < .01; ***p < .001.
engagement was higher during School, whereas passive media
(e.g., television and DVD) was higher during Vacation. During
Vacation, there may be more daytime opportunity for social
media and more in-person contact with friends at night time.
Unsurprisingly, pre-BT school work engagement was higher
during School than that during Vacation (44% vs. 12% reported
3 times/week). The frequency of spending time with family was
comparable over both periods.
School
Consistent with existing literature, chronotype strongly
predicted sleep timing. Importantly, this study showed that the
associations between some PBBs and sleep timing were inde-
pendent of chronotype. During School, controlling for chro-
notype, every unit increase in video games frequency was
associated with 12 minutes later BT, which might have contrib-
uted to the associated 14 minutes shorter TST on school nights
given RT is constrained by school schedules. This is consistent
with Arora et al. [11] and suggested that adolescents may
persevere in playing games and go to bed later beyond the effect
of chronotype. This supports the displacement hypothesis in
which video games might shorten school night sleep by replacing
sleep time [13].
In contrast, every unit increase in time with family frequency
was associated with 15 minutes earlier BT controlling for chro-
notype and 15 minutes longer TST on school nights. Although
speculative, it may be that those who more often spend time
with family were subjected to higher parental control over BTs
and media use, which are associated with earlier BT [10]. It is also
possible that those who reported more frequent pre-BT family
engagement engaged in technology-related activities within a
family space, which was associated with earlier BT and longer
TST compared with bedroom usage [9].
Previously reported associations between later BT, shorter
TST, and usage of technology on school nights [10,11] were not
confirmed. A number of factors could have contributed to this.
First, unlike previous research, this study controlled for chro-
notype, a probable confounder that has been associated with
video gaming and surfing the Web before sleep in young adults
[33]. Second, previous studies typically compared differences in
sleep between groups of adolescents with different levels of
engagement (e.g., never, sometimes, always [11]), whereas this

Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432
study examined linear dose effects of PBBs on sleep variables.
Third, unlike studies using self-report BT and TST [9,11,12], sleep
was measured objectively in this study.
No PBBs showed significant relationships with school-term
RT, which is largely influenced by school schedules [6]. Further-
more, no PBBs showed a significant relationship with SOL during
School. School-related sleep restriction [22] might have
contributed to high homeostatic drive, which might have had a
strong effect in shortening SOL, such that the effects of PBBs were
masked.
Vacation
Apart from a marginally significant association of video games
with later BT, no PBBs were significantly associated with sleep
timing during Vacation, controlling for chronotype. With more
flexible schedules during Vacation, adolescents were more likely
to choose BT and RT based on circadian preference, and therefore
chronotype may exert stronger influences on sleep timing than
PBBs. Flexible sleep schedules may also explain why no PBBs
were related to TST during vacation, as adolescents were afforded
relatively unconstrained sleep opportunity.
Consistent with previous studies, interactive and stimulating
behaviors such as online chat and listening to music were asso-
ciated with a longer SOL [10,11], but only during Vacation with
relatively unconstrained sleep opportunity. PSAcog mediated the
relationship between more frequent social media and longer SOL.
Social media can stimulate cognitive activation, as it is a dynamic
platform that encourages users to make social exchanges and
judgments, forming reciprocal relationships between social
needs and gratifications [34]. It is possible that this cognitive
activation could lengthen sleep onset. Listening to music was
associated with longer SOL but did not share a linear relationship
with PSAcog, suggesting a more complex relationship between
presleep music use and SOL. Although listening to music may
increase difficulty falling asleep [11], adolescents with sleep
onset difficulties might also use music as a sleep aid [17]. It is
worth noting that although statistically significant, the effects of
listening to music and online social media on SOL were small.
Nevertheless, our findings provide further support for the hy-
pothesis that presleep arousal plays a mechanistic role between
PBBs and SOL [13].
Limitations
There are several limitations. (1) Like previous studies [8,10],
individual behaviors were measured using single items.
Although single items made it feasible to assess a large number
of PBBs at once and can be valid and reliable, they are not stable
or precise compared with multi-item measures [35]. (2) Light
emitted from screen-based devices was not measured, and its
potential effects on alertness [13] were not assessed. (3) The role
of puberty was not assessed as most participants (95.9%) were
15 years and likely to be of Tanner Stage V [36]. This focus on
older adolescents could limit generalizability of findings to
younger adolescents. (4) Diagnoses of sleep disorders were not
objectively screened. To maintain the representativeness of a
community sample, we did not exclude (with one exception) the
small percentage of participants who self-reported current sleep
disorders and acknowledge that the presence of sleep disorders
may be a confounding factor. (5) Only the last week of a school
term was assessed, and this week may not be representative of
Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
431
the whole school term. (6) Although associations between PBBs
and sleep variables were found, causality cannot be inferred. (7)
As with all self-report data, answers to the PBBQ are subject to
recall and self-report bias. (8) Owing to sample size issues, it was
not possible to examine the relationship between PBBs with low
endorsement (e.g., drinking alcoholic drinks) and sleep. (9) Only
the frequency of behavioral engagement was measured. The
quality of behavioral engagement (e.g., positive/negative expe-
rience, intensity) may also contribute to associations with sleep.
(10) Rapid advances in technology change how adolescents
engage in technology-related behaviors. Findings in this study
were applicable to 2011e2012 when data were collected but
might not generalize to settings where new technologies are
available (e.g., tablets and eReaders).
Implications
To our knowledge, this is the first study to characterize a wide
range of adolescent presleep behaviors during school and vaca-
tion periods and examine their relationships with objectively
measured sleep. Controlling for chronotype, only a small number
of PBBs shared significant relationships with sleep timing, sug-
gesting that biological processes underlying later circadian
timing in adolescents are strong determinants of sleep timing
and are necessary confounders to control for in future studies.
Similarly, school-related sleep restriction likely contributes to
different associations between PBBs and sleep, with implications
for timing of measures in future studies.
During school nights, some pre-BT technologyerelated be-
haviors such as playing video games and online social media are a
risk factor for shorter and poorer sleep, whereas time with family
was protective of sleep duration. The detrimental effects of video
games on sleep timing and duration suggest that therapeutic
components (e.g., motivational interviewing [37]) in addition to
sleep education for addressing the potentially addictive nature of
gaming [38] could help encourage earlier BT and longer sleep
duration. The mediating role of PSAcog in the association be-
tween pre-BT social media and longer SOL may have clinical
implication for adolescents who are vulnerable to, or who
already experience, poor sleep and/or sleep-related mood
problems [39]. Therapeutic components that aim to reduce
cognitive arousal, such as relaxation and mindfulness may be
particularly relevant [40]. Finally, findings in this study support a
strong protective role of family involvement in adolescents’ sleep
duration on school nights, and this has practical implications for
future policies and interventions that aim to address inadequate
sleep in adolescents.
Acknowledgments
There is no off-label or investigational use in this study.
Funding Sources
This work was supported by the Melbourne School of Psy-
chological Sciences at the University of Melbourne, Australia.
Supplementary Data
Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.jadohealth.2015.12.011.
432 E. Harbard et al. / Journal of Adolescent Health 58 (2016) 426e432
References
[1] Matricciani L, Olds T, Petkov J. In search of lost sleep: Secular trends in the
sleep time of school-aged children and adolescents. Sleep Med Rev 2012;
16:203e11.
[2] Wolfson AR, Carskadon MA. Understanding adolescent’s sleep patterns and
school performance: A critical appraisal. Sleep Med Rev 2003;7:491e506.
[3] Danner F, Phillips B. Adolescent sleep, school start times, and teen motor
vehicle crashes. J Clin Sleep Med 2008;4:533e5. Available at: http://www.
aasmnet.org/jcsm/.
[4] Roane BM, Taylor DJ. Adolescent insomnia as a risk factor for early adult
depression and substance abuse. Sleep 2008;31:1351e6. Available at:
www.journalsleep.org.
[5] Russo PM, Bruni O, Lucidi F, et al. Sleep habits and circadian preference in
Italian children and adolescents. J Sleep Res 2007;16:163e9.
[6] Carskadon MA. Patterns of sleep and sleepiness in adolescents. Pediatrician
1990;17:5e12. Available at: pediatrics.aappublications.org.
[7] Colrain I, Baker F. Changes in sleep as a function of adolescent develop-
ment. Neuropsychol Rev 2011;21:5e21.
[8] Van Den Bulck J. Television viewing, computer game playing, and internet
use and self-reported time to bed and time out of bed in secondary-school
children. Sleep 2004;27:101e4. Available at: http://www.journalsleep.org/.
[9] Schochat T, Flint-Bretler O, Tzischinsky O. Sleep patterns, electronic media
exposure and daytime sleep-related behaviours among Israeli adolescents.
Acta Paediatr 2010;99:1396e400.
[10] Pieters D, De Valck E, Vandekerckhove M, et al. Effects of pre-sleep media use
on sleep/wake patterns and daytime functioning among adolescents: The
moderating role of parental control. Behav Sleep Med 2012;12:427e43.
[11] Arora T, Broglia E, Thomas GN, et al. Associations between specific tech-
nologies and adolescent sleep quantity, sleep quality, and parasomnias.
Sleep Med 2014;15:240e7.
[12] Hysing M, Pallesen S, Stormark KM, et al. Sleep and use of electronic de-
vices in adolescence: Results from a large population-based study. BMJ
Open 2015;5:e006748.
[13] Cain N, Gradisar M. Electronic media use and sleep in school-aged children
and adolescents: A review. Sleep Med 2010;11:735e42.
[14] Takano K, Sakamoto S, Tanno Y. Repetitive thought impairs sleep quality:
An experience sampling study. Behav Ther 2014;45:67e82.
[15] Hiller RM, Lovato N, Gradisar M, et al. Trying to fall asleep while cata-
strophising: What sleep-disordered adolescents think and feel. Sleep Med
2014;15:96e103.
[16] Cajochen C, Frey S, Anders D, et al. Evening exposure to a light-emitting
diodes (LED)-backlit computer screen affects circadian physiology and
cognitive performance. J Appl Phys 2011;110:1432e8.
[17] Eggermont S, Van den Bulck J. Nodding off or switching off? The use of
popular media as a sleep aid in secondary-school children. J Paediatr Child
Health 2006;42:428e33.
[18] Musick K, Meier A. Assessing causality and persistence in associations
between family dinners and adolescent well-being. J Marriage Fam 2012;
74:476e93.
[19] Noland H, Price JH, Dake J, et al. Adolescents’ sleep behaviors and per-
ceptions of sleep. J Sch Health 2009;79:224e30.
[20] Adam EK, Snell EK, Pendry P. Sleep timing and quantity in ecological and
family context: A nationally representative time-diary study. J Fam Psychol
2007;21:4e19.
Downloaded for Anonymous User (n/a) at University of Chile Catholic ALERTA from ClinicalKey.com by Elsevier on April 08, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
[21] Lauderdale DS, Knutson KL, Yan LL, et al. Self-reported and measured
sleep duration: How similar are they? Epidemiology 2008;19:838e45.
[22] Bei B, Allen NB, Nicholas CL, et al. Actigraphy-assessed sleep during school
and vacation periods: A naturalistic study of restricted and extended sleep
opportunities in adolescents. J Sleep Res 2014;23:107e17.
[23] Sadeh A. The role and validity of actigraphy in sleep medicine: An update.
Sleep Med Rev 2011;15:259e67.
[24] Sadeh A, Sharkey KM, Carskadon MA. Activity-based sleep-wake identifi-
cation: An empirical test of methodological issues. Sleep 1994;17:201e7.
Available at: http://www.journalsleep.org/.
[25] Philips Respironics. Equivalence of activity recordings and derived sleep
statistics; 2008. Available at: https://www.cpapaustralia.com.au/media_
files/actiwatch-performance-new.pdf. Accessed April 20, 2014.
[26] Horne JA, Ostberg O. A self-assessment questionnaire to determine
morningness-eveningness in human circadian rhythms. Int J Chronobiol
1976;4:97e110. Available at:http://informahealthcare.com/cbi.
[27] Taillard J, Philip P, Chastang J-F, et al. Validation of Horne and Ostberg
morningness-eveningness questionnaire in a middle-aged population of
French workers. J Biol Rhythms 2004;19:76e86.
[28] Megdal SP, Schernhammer ES. Correlates for poor sleepers in a Los Angeles
high school. Sleep Med 2007;9:60e3.
[29] Giannotti F, Cortesi F, Sebastiani T, et al. Circadian preference, sleep and
daytime behaviour in adolescence. J Sleep Res 2002;11:191e9.
[30] Nicassio PM, Mendlowitz DR, Fussell JJ, et al. The phenomenology of the
pre-sleep state: The development of the pre-sleep arousal scale. Behav Res
Ther 1985;23:263e71.
[31] Sparks R, Carter C, Donnelly JB, et al. Remote access methods for explor-
atory data analysis and statistical modelling: Privacy-preserving analy-
ticsÒ. Comput Methods Programs Biomed 2008;91:208e22.
[32] Gradisar M, Wolfson AR, Harvey AG, et al. The sleep and technology use of
Americans: Findings from the National Sleep Foundation’s 2011 sleep in
America poll. J Clin Sleep Med 2013;9:1291e9.
[33] Fossum IN, Nordnes LT, Storemark SS, et al. The association between use of
electronic media in bed before going to sleep and insomnia symptoms,
daytime sleepiness, morningness, and chronotype. Behav Sleep Med 2013;
12:343e57.
[34] Wang Z, Tchernev JM, Solloway T. A dynamic longitudinal examination of
social media use, needs, and gratifications among college students. Comput
Human Behav 2012;28:1829e39.
[35] Bowling A. Just one question: If one question works, why ask several?
J Epidemiol Community Health 2005;59:342e5.
[36] Tanner J, Whitehouse R. Clinical longitudinal standards for height, weight,
height velocity, weight velocity, and stages of puberty. Arch Dis Child
1976;51:170e9.
[37] Cain N, Gradisar M, Moseley L. A motivational school-based intervention
for adolescent sleep problems. Sleep Med 2011;12:246e51.
[38] Weinstein AM. Computer and video game addictiondA comparison be-
tween game users and non-game users. Am J Drug Alcohol Abuse 2010;36:
268e76.
[39] Bei B, Wiley JF, Allen NB, et al. A cognitive vulnerability model on sleep and
mood in adolescents under naturalistically restricted and extended sleep
opportunities. Sleep 2015;38:453e61.
[40] Bei B, Byrne ML, Ivens C, et al. Pilot study of a mindfulness-based, multi-
component, in-school group sleep intervention in adolescent girls. Early
Interv Psychiatry 2013;7:213e20.