3 pages writing

E M P I R I C A L R E S E A R C H

Multi-method Assessments of Sleep over the Transition to College and the Associations with Depression and Anxiety Symptoms

Leah D. Doane • Jenna L. Gress-Smith •

Reagan S. Breitenstein

Received: 30 April 2014 / Accepted: 27 June 2014 / Published online: 18 July 2014

� Springer Science+Business Media New York 2014

Abstract A growing body of research has demonstrated

links between sleep problems and symptoms of depression

and anxiety in community and clinical samples of adoles-

cents and young adults. Scant longitudinal research, how-

ever, has examined reciprocal associations over socio-

contextual shifts such as the transition to college. Using

multiple methods of assessment (e.g., actigraphy, sub-

jective report), the current study assessed whether sleep

quantity, quality or variability changed over the transition

to college and investigated the potential cross-lagged

relationships between adolescents’ sleep and symptoms of

anxiety and depression. The participants (N = 82; 24 %

male) were studied at three time points over approximately

1 year: spring of their senior year of high school (T1), fall

of their first year of college (T2), and spring of their first

year of college (T3). Sleep minutes, sleep efficiency, wake

time variability and anxiety increased over the transition to

college. Subjective reports of sleep problems decreased.

Cross-lagged panel models indicated significant relation-

ships between subjective sleep quality and anxiety symp-

toms over time where subjective sleep problems at T1 were

associated with anxiety at T2, and anxiety at T2 was

associated with subjective sleep problems at T3. In con-

trast, greater depressive symptoms at T1 preceded increa-

ses in subjective sleep problems, sleep latency and sleep

start time variability at T2. Importantly, there were con-

current associations between symptoms of anxiety or

depression at T2 and sleep efficiency, sleep start time

variability, and subjective sleep problems. These findings

suggest that, overall, sleep quantity and quality improved

over the transition to college, although the overall amounts

of sleep were still below developmental recommendations.

However, for some youth, the first semester of college may

be a sensitive period for both sleep problems and symp-

toms of anxiety. In contrast, depressive symptoms were

stable across time but were associated with worsening

sleep problems in the first semester of college. Implications

for future prevention and intervention programs should

include strategies to help youth cope effectively with

adjustment like increased sleep variability and symptoms

of anxiety associated with the transition to college.

Keywords Sleep � Actigraphy � Anxiety symptoms � Depressive symptoms � Adolescence � Transition to college

Introduction

An accumulation of research has documented that sleep

during the adolescent years is shortened, often leading to

daytime dysfunction and health problems such as anxiety,

depressive and somatic symptoms (National Sleep Foun-

dation 2006; Oginska and Pokorski 2006; Roberts et al.

2001). Both biological and psychosocial factors are asso-

ciated with changes in sleep patterns during adolescence

(Carskadon 2011). For example, researchers have identified

changes in adolescents’ circadian rhythm and sleep archi-

tecture, as well as environmental changes such as earlier

school start times, extracurricular activities, and after

school jobs (Carskadon et al. 1998; Dahl and Lewin 2002).

Sleep problems may continue or even become worse for

youth who enter the college environment. Across several

studies, at least 60 % of college students were character-

ized as poor quality sleepers, reported restricted sleep on

weeknights, and frequent ‘‘all-nighters’’ (Lund et al. 2010;

L. D. Doane (&) � J. L. Gress-Smith � R. S. Breitenstein Department of Psychology, Arizona State University,

P.O. Box 871104, Tempe, AZ 85287-1104, USA

e-mail: [email protected]

123

J Youth Adolescence (2015) 44:389–404

DOI 10.1007/s10964-014-0150-7

Pilcher et al. 1997). Social schedules and expectations may

also interact with college students’ biological clocks,

leading to sleep disturbances; this interaction has been

termed ‘‘social jetlag’’ (Wittmann et al. 2006). Further,

first-year college students are more likely to report poor

sleep than second-year students (Suen et al. 2008). Lon-

gitudinal studies indicate that poor sleep quality, quantity

and variability have direct and indirect associations with

college students’ academic performance (Lau et al. 2013;

Orzech et al. 2011), psychosocial functioning (Tavernier

and Willoughby 2014) and alcohol use (Phinney and Haas

2003).

From a developmental perspective, the transition from

high school to college is particularly important to examine

given that most youth experience a socio-contextual shift.

This shift to a new context is often coupled with increased

perceptions of stress, changes in family and peer relation-

ships, greater academic demands and increased responsi-

bilities for their own health and decision-making (Kerr

et al. 2004; Larose and Boivin 1998). Developmental

transitions can also be periods of heightened vulnerability

to stress (Compas et al. 1986). Indeed, first-year college

students report high levels of depression and anxiety (Furr

et al. 2001; Pryor et al. 2010), and poor mental health has

been associated with elevated risk of dropping out of col-

lege (Hartley 2012). Studies of college students have found

significant covariation between affective, stressful and

social experiences and self-reported sleep quality (e.g.,

Galambos et al. 2010). Further, during years of increased

stress, college students demonstrated shorter self-reported

sleep durations, more sleep disturbances, and later waking

times (Galambos et al. 2013). Fewer constraints in college

and variable schedules may lead to significant sleep vari-

ability (e.g., Moore et al. 2009) and reduced sleep quality.

These reductions in quality may be correlated with or

predict a range of developmental outcomes, specifically

those related to emotional health and well-being.

In addition, researchers have identified links (sometimes

reciprocal) between subjective sleep problems and inter-

nalizing problems in community samples of adolescents

and young adults (e.g., Jean Louis et al. 1998; Moo-Estrella

et al. 2005). Longitudinal studies examining associations

between sleep and internalizing problems have shown that

the magnitude of the associations changes over time and

increases from childhood into adolescence (Gregory and

O’Connor 2002; Johnson et al. 2000; Kelly and El-Sheikh

2014). Much prior research has been limited by the reliance

on single subjective or self-report measures of sleep, and

many researchers have solely examined depression or

anxiety, which are highly comorbid in adolescents who

attend college (Eisenberg et al. 2007; Nyer et al. 2013).

Further, little is known about the direction of effects

between sleep and anxiety/depression during adolescence,

particularly across a substantial socio-cultural shift like the

transition to college. In an attempt to address these limi-

tations, the current study examined changes in sleep across

the transition to college and the reciprocal relationships

between adolescent’s sleep and their depressive and anxi-

ety symptoms in a three wave cross-lag design using both

subjective and objective actigraphy-based measures of

sleep quality, quantity and variability.

Sleep, Anxiety, and Depression in Adolescence

and Young Adulthood

Within community and large epidemiological samples of

children and adolescents, research using subjective indi-

cators suggests that there are concurrent associations

between sleep problems, anxiety and depression (e.g.,

Coulombe et al. 2010). However, these associations may

vary by age. Alfano et al. (2009) found that anxiety

symptoms were associated with sleep problems in both

childhood and adolescence, whereas depressive symptoms

were only associated with sleep problems in adolescence.

Evidence from daily diary studies also suggests these

associations interact reciprocally over time. In a diverse

sample of adolescents, Fuligni and Hardway (2006)

showed that stress during the day was associated with

shorter sleep duration, which in turn was associated with

greater anxious and depressive mood the next day.

Studies using subjective reports from college students

typically examine depressive symptoms rather than anxi-

ety; however, the findings are similar to those seen in

adolescent populations. Gress-Smith et al. (2013) reported

high rates of insomnia (i.e., 69.5 % reported at least mild

insomnia severity) and depressive symptoms (i.e., 33.5 %

reported at least mild depressive symptoms) in a large

college student sample. Moo-Estrella et al. (2005) found

that individuals who reported greater sleepiness during the

day reported more depressive symptomatology. Finally, a

recent study demonstrated that female college students who

reported either significant sleepiness during the day or

sleep deprivation (at least 2 h less than recommended

levels) were at greater risk of melancholic symptoms

(Regenstein et al. 2010). In terms of anxiety, one recent

large study of college students found a significant corre-

lation between concurrent self-reported sleep problems and

anxiety symptoms (Orsal et al. 2012). In sum, there is

consistent evidence for concurrent relationships between

self-reported sleep problems, ranging from insomnia to

daytime sleepiness, and symptoms of anxiety and depres-

sion in college students.

Recently, researchers have tested some of the potential

reciprocal relationships between sleep and symptoms of

anxiety and depression (e.g., Kelly and El-Sheikh 2014).

However, most of these studies have been conducted with

390 J Youth Adolescence (2015) 44:389–404

123

children or adults, rather than adolescents. Thus, the

associations among adolescents as they transition to col-

lege remain unclear, and it is possible that these reciprocal

relationships differ for depressive and anxiety symptoms.

Studies using subjective indicators of sleep and sleep

problems are mixed regarding whether poor sleep consis-

tently occurs before anxiety or depressive symptoms, or the

reverse. For example, insomnia preceded impaired psy-

chological functioning during adolescence over the course

of 1 year (Roberts et al. 2002), self-reported sleep prob-

lems at age four predicted depressive and anxiety symp-

toms at age 15 (Gregory and O’Connor 2002), and sleep

problems in childhood predicted depression in pre-adoles-

cence (Gregory et al. 2009). In contrast, a few other studies

have shown the reverse, wherein anxiety or depression

problems predict subsequent insomnia or other sleep

problems (e.g., Johnson et al. 2006). Finally, researchers

have identified bidirectional relationships between sleep

problems, anxiety and depression (e.g., Jansson-Frogmark

and Lindbloom 2008). The most recent study to test the

reciprocal relationships between sleep and anxiety and

depressive symptoms using cross-lag models utilized both

subjective and objective indicators of sleep in a community

sample of children (Kelly and El-Sheikh 2014). The find-

ings suggested differential and complex reciprocal paths

among sleep quality and quantity and anxiety and depres-

sive symptoms from ages 8 to 13. We examined similar

relationships between subjective and objective sleep

quantity, quality and variability later in development as

adolescents navigate the transition to college.

Multi-Method Approach to Measuring Sleep

Recent reviews have suggested that studies of sleep and

associations with adjustment and socioemotional develop-

ment should follow new recommendations to better inform

research and practice (Gregory and Sadeh 2012; Shochat

et al. 2014). The majority of studies have used one type of

assessment (e.g., subjective report, ambulatory assessment

or objective indicators) rather than multiple assessments,

which makes it difficult to understand mechanisms under-

lying associations between sleep and depression and anx-

iety. Researchers have also suggested that subjective sleep

measures may predict different facets of sleep and well-

being than objective sleep measures (Tremaine et al. 2010).

Further, Shochat et al. (2014) argued that developmental

observational studies must be conducted in participants’

naturalistic settings. Following these recommendations, we

examined multiple indicators of sleep in youth’s natural-

istic environments at three measurement periods over

approximately 1 year, using validated subjective reports

and objective indicators of sleep duration, quality and

variability via actigraphy. Thus, we sought to highlight

differential relationships between various indicators of

sleep and anxiety and depressive symptoms over time that

may be important targets for preventive intervention efforts

aimed at improving adjustment for adolescents who tran-

sition to the college environment.

Present Study and Hypotheses

The current study examined cross-lagged relationships

between adolescents’ sleep and symptoms of anxiety and

depression across the transition to college. The first aim of

this study was to identify whether symptoms of anxiety and

depression, sleep quantity, quality or variability changed

over the transition to college. We hypothesized, over the

transition to college (T1–T2), sleep quality (e.g., lower sleep

efficiency, greater sleep latency) would decrease and sleep

variability would increase given less consistent environ-

mental constraints on sleep timing (e.g., school start times,

parental curfew or bedtime). The second aim of this study

was to examine reciprocal relationships between sleep and

symptoms of depression and anxiety separately using cross-

lagged panel models to uncover whether there were differ-

ential associations between sleep and anxiety or depressive

symptoms over a large socio-cultural shift in adolescents’

lives. Based on prior studies (e.g., Gress-Smith et al. 2013;

Kelly and El-Sheikh 2014), we hypothesized that we would

find significant cross-lagged relationships between sub-

jective and objective sleep quality and symptoms of

depression and anxiety. We did not make specific hypotheses

about sleep variability given the lack of prior research on this

construct as it relates to anxiety and depressive symptoms.

Method

Participants

The longitudinal study is comprised of 82 adolescent par-

ticipants from a large Southwestern university. Researchers

recruited participants during the spring of senior year in

high school through college orientation sessions and email.

Inclusion criteria required participants to be accepted to the

university, be seniors in high school, and live within 35

miles of the campus. Assessments occurred at three dif-

ferent time points: Spring of senior year of high school

(T1), fall of their first year of college (T2), and spring of

their first year of college (T3).

Eighty-two adolescents participated in the study at T1

(Mage = 18.05, SD = .41; 24 % male), 76 participated at T2

(93 % retention; 24 % male), and 71 at T3 (87 % retention

from T1; 23 % males). At T2, 17 adolescents (22 %) reported

they were living at home and 59 adolescents (78 %) reported

J Youth Adolescence (2015) 44:389–404 391

123

they were living somewhere other than home (e.g., dormitory,

apartment). Participants were racially and ethnically diverse:

54 % European American, 23 % Latino/Hispanic descent,

13 % multiracial, 5 % African American and 5 % Asian

American/Pacific Islander. Youth also came from varying

socioeconomic backgrounds as measured by their parents’

level of education: 11 % reported that their parents completed

some high school, 26.8 % of parents had a high school

diploma, 22 % had some college, 12.2 % received an asso-

ciate’s degree, 18.3 % received a bachelor’s degree and 9.8 %

received a graduate degree. We examined if significant mean

differences on study and demographic variables (e.g., parental

education, race/ethnicity, living at home) were present as a

result of attrition from T1 to T3 using t tests. Adolescents lost

to attrition were more likely to have parents with more edu-

cation, t(80) = -2.58, p \ .05, M = 1.23, SE = .48, and have greater sleep start time variability at T1, t(78) = 2.45,

p \ .05, M = .77, SE = .31.

Procedure

The university Institutional Review Board approved all pro-

cedures and longitudinal sampling of participants. At T1,

participants completed consent forms in the home, and

researchers collected parental consent for participants who

were under age 18. Participants completed a packet of self-

report questionnaires, including measures of subjective sleep

problems, anxiety and depressive symptoms, and wore the

actigraph watch for four weeknights and three weekdays

during a typical week (most commonly Sunday night through

Thursday morning with some variability of weeknights given

participants’ schedules). Study personnel instructed partici-

pants to press a button on the actigraph watch upon waking in

the morning and when they got into bed at night, which served

as secondary indicators of wake times and bedtimes in addi-

tion to accelerometer-based actigraphy data. Participants

received $40 for protocol completion, as well as feedback

regarding their objective sleep throughout the study (gathered

from the actigraph watches). At T2 and T3, study personnel

delivered all protocol materials to the participants’ dorm

rooms at the university or homes (if not living on campus).

Study protocol did not change from T1. The average time

between T1 and T2 assessments was 5.2 months (SD = .96),

and the average time between T2 and T3 was 4.1 months

(SD = .84). Participants received $50 upon protocol com-

pletion at T2 and $70 at T3.

Measures

Objective Sleep

Participants wore an Actiwatch Score (Phillips Respiron-

ics, Inc.) wrist-watch on their non-dominant wrist for all 4

nights and 3 days of the study. The Actiwatch Score con-

tains an accelerometer, which captures movement

throughout the waking day and during sleep periods. Par-

ticipants pressed a button on the watch upon waking in the

morning and when they got into bed at night. Study staff

cross-checked actigraph sleep periods with self-reported

bedtime and wake time as an additional sleep-period

compliance measure. Researchers scored sleep data using

the Phillips Actiware 6 program, which includes a vali-

dated algorithm to measure sleep (Oakley 1997). Activity

counts within each epoch were calculated based on activity

levels during the adjacent 2 min period. 1

The threshold was

set to 40, with a range of 20–80. Utilizing 1-minute epochs

and based on significant movement after at least 10 min of

inactivity, this algorithm calculates a variety of sleep

parameters. Four sleep variables were emphasized in the

current study: total sleep minutes, sleep onset latency, sleep

efficiency and sleep start time variability. Sleep minutes

was the total amount of sleep measured in minutes

(excluding wake periods during the night). Sleep latency

was the amount of time an individual is in bed before

falling asleep. Sleep efficiency represented the proportion

of time an individual is actually asleep when he or she is in

bed. Time in bed was measured by the actiwatch, and

objective time in bed was cross-checked by staff with diary

self- reports of bedtime. Sleep start time and wake time

variability were within-person standard deviation estimates

across the 4 days of assessment. Actigraphy sleep mea-

surement has been validated against concurrent polysom-

nography (Sadeh et al. 1995).

Objective indicators of sleep were validated with diary

self-reports of bed and wake times to identify significant

outliers and equipment malfunction. Days when there was

equipment malfunction or days in which there was signif-

icant discordance between self-reports and objective mea-

surement were not included in analyses (T1: 9 days; T2:

8 days; T3: 20 days). In total at T1, 95.1 % of adolescents

had valid actigraphy data for all 4 nights, 3.7 % had data

for only 3 nights, and 2.4 % had data for 2 or fewer nights

of sleep. For T2, 93.4 % of participants had actigraphy data

for all 4 nights, 5.2 % had data for only 3 nights of sleep,

and 1.4 % had 2 or fewer nights of sleep. At T3, 85.9 % of

adolescents had actigraphy data for all 4 nights, 11.3 % had

data for only 3 nights, and 5.6 % of participants had data

for 2 or fewer nights of sleep. If data consist of fewer than

3 nights of actigraphy, they provide a poor estimation of

regular sleep (Acebo et al. 1999); thus, such data were

excluded for two participants from T1, one participant from

1 The following algorithm was used where A denotes activity counts

and E denotes epoch: A = E - 2(1/25) ? E - 1(1/5) ? E ? E ?

1(1/5) ? E ? 2(1/25).

392 J Youth Adolescence (2015) 44:389–404

123

T2, and four participants from T3 who had 2 or fewer

nights of valid data.

Subjective Sleep

Participants completed the Pittsburgh Sleep Quality Index

(PSQI) as a measure of subjective sleep problems (Buysse

et al. 1989). The PSQI is a 19-item self-report measure of

sleep quality and sleep disturbance in the past month. The

PSQI includes seven scales: sleep duration, sleep onset

latency, sleep disturbance, daytime dysfunction, sleep

efficiency, sleep quality, and use of sleep medication

(Buysse et al. 1989). Participants answered open-ended

questions such as, ‘‘During the past month, what time have

you usually gone to bed at night?’’ Participants also

answered four-point Likert scale questions such as, ‘‘Dur-

ing the past month, how often have you had trouble

sleeping because you wake up in the middle of the night or

early morning?’’ Responses were coded from 0 (not during

the past month) to 3 (three or more times in a week). Scores

from scales were summed for an overall global sleep dis-

turbance score. Higher total scores indicate poorer sleep

quality and more sleep disturbance. Scores greater than 5

indicate clinically significant sleep problems (Buysse et al.

1989).

Anxiety Symptoms

Participants completed the 14-item Depression, Anxiety

and Stress Scale- Anxiety Subscale (DASS), which asses-

sed anxiety symptoms in the past week (Lovibond and

Lovibond 1995). An example item includes, ‘‘I was wor-

ried about situations in which I might panic and make a

fool of myself’’ (Lovibond and Lovibond 1995). Partici-

pants responded to items using a four-point Likert scale,

from 0 (Did not apply to me at all) to 3 (Applied to me very

much, or most of the time). Item scores were summed to

create an anxiety score for each participant, with higher

scores reflecting more anxious symptomatology. Cron-

bach’s alphas in this sample were acceptable (T1: .75, T2:

.87, T3: .85). Clinical cut-off scores were also determined

for anxiety (DASS Anxiety subscale) at each time point to

show the proportion of the sample falling in the normal

(scores 0–7), mild (scores 8–9), moderate (scores 10–14),

severe (scores 15–19) and extremely severe anxiety ranges

(scores 20?; Lovibond and Lovibond 1995). In total at T1,

68.3 % of adolescents met the criteria for normal, 9.8 %

for mild, 13.4 % for moderate, 4.9 % for severe, and 2.4 %

for extremely severe depressive symptoms. At T2, 51.2 %

of adolescents met the criteria for normal, 12.2 % for mild,

15.9 % for moderate, 8.5 % for severe, and 4.9 % for

extremely severe depressive symptoms. At T3, 50 % of

adolescents met the criteria for normal, 4.9 % for mild,

20.7 % for moderate, 9.8 % for severe, and 3.7 % for

extremely severe anxiety symptoms.

Depressive Symptoms

Participants completed the 20-item Center for Epidemiol-

ogic Studies Depression Scale (CES-D), which assessed

depressive symptoms over the past week (Radloff 1977).

Participants responded to items using a four-point Likert

scale, from 0 (Rarely or none of the time/Less than once a

week to 3 (Most or all of the time/5–7 days). Positively

phrased items (e.g., ‘‘I felt hopeful about the future’’) were

reverse coded. Items were summed to create a total CES-D

score for each participant (0–60), with higher scores

reflecting more depressive symptoms. Cronbach’s alphas in

this sample were acceptable (T1: .86, T2: .86, T3: .91).

Clinical cut-off scores were determined for subjective

depressive symptoms (CES-D) at each time point to show

the proportion of the sample falling in the normal (scores

0–16), mild (scores 16–22) and moderate to severe symp-

tom ranges (scores 23?; Roberts et al. 1990). In total at T1,

58.5 % of adolescents met criteria for normal, 22 % for

mild, and 17.1 % for moderate to severe depressive

symptoms. At T2, 48.8 % of adolescents met the criteria

for normal, 22 % for mild, and 22 % for moderate to

severe depressive symptoms. At T3, 45.1 % of adolescents

met the criteria for normal, 22 % for mild, and 18.3 % for

moderate to severe depressive symptoms.

Covariates

A number of covariates were included given prior research

demonstrating significant relationships with sleep and/or

symptoms of depression or anxiety: sex (Sadeh et al. 2000),

race/ethnicity (Mezick et al. 2008), parental education

(Buckhalt et al. 2007), and whether the participant was

living at home at the time of the assessment.

Data Analytic Plan

Zero-order correlations among all of the objective sleep

parameters, subjective ratings of sleep, depressive symp-

toms, and anxiety symptoms were examined and tested for

normality (e.g., skewness and kurtosis). Next, we examined

whether sleep changed across the transition to college.

Repeated measures analyses of variance were conducted to

test whether or not there statistically significant changes in

the sleep measures and symptoms of anxiety and depres-

sion across the three time points, and to understand whether

the changes were linear or nonlinear by examining Bon-

feronni-adjusted pairwise comparisons and differences in

means. If there were significant differences, effect sizes

were also noted using Cohen’s d (Cohen 1988) with .2, .5,

J Youth Adolescence (2015) 44:389–404 393

123

and .8 used as benchmarks of a small, medium, and large

effect sizes respectively.

Two-tiered autoregressive path models were tested to

examine the cross-lagged or reciprocal effects of anxiety or

depressive symptoms and the domains of sleep (e.g., sleep

latency, subjective sleep) that demonstrated significant

associations with depressive or anxiety symptoms in cor-

relational analyses. Depressive and anxiety symptoms were

run in separate models and as observable scale scores as

opposed to latent variables, due to sample size limitations

(e.g., n \ 100; Loehlin 2004). These models were tested within a structural equation model (SEM) framework to

investigate the directional influence of psychological

symptoms with later sleep disruption, controlling for prior

sleep behavior, or vice versa during the transition to col-

lege. Cross-lagged models allow for construct variability to

be examined over time, as each time point is regressed on

an earlier time point (e.g., depressive symptoms T2 is

regressed on depressive symptoms T1, etc.). Additionally,

significant paths in cross-lagged models also control for

within-construct stability or the autoregressive effect (Cole

and Maxwell 2003).

These models were analyzed in a four step sequence as

outlined by Jonge et al. (2001) to help determine the mech-

anisms of directional influence. The first step in this sequence

used a baseline model that included only stability paths for

comparison for the subsequent steps and more complex

models. The second step involved comparing the baseline

model to models that included depressive or anxiety symp-

toms at T1 analyzed as predictors of sleep variables at T2,

and then sleep variables at T2 predicted affective symptoms

at T3. The third step compared the baseline model to a model

that included paths examining sleep variables at T1 as pre-

dictors of depressive or anxiety symptoms at T2. Depressive

or anxiety symptoms at T2 were then regressed on sleep

parameters at T3. Finally, the baseline model was compared

to a full model including both sets of cross-lagged effects. All

model comparisons were conducted using v2 tests. Final models were determined to be the best fit to the data based on

Chi-squared model comparison tests, and if there was an

improvement in model fit, as determined by v2, standardized root mean square residual (SRMR), root mean square error of

approximation (RMSEA), and comparative fit index (CFI;

Hu and Bentler 1999). The residual variances among the

exogenous variables were allowed to correlate in these

models. Paths to examine the concurrent associations

between depressive or anxiety symptoms and sleep measures

at each time point were also specified (see Fig. 1 as an

example). Models were analyzed in MPlus 6.0 (Muthén and

Muthén 2012) and accounted for missing data with full

information maximum likelihood (FIML), which is consid-

ered superior compared to pairwise deletion, listwise

deletion, and similar response pattern imputation in SEM

(Enders and Bandalos 2001).

Results

Comparison of Means Across the Transition to College

and Descriptive Results

Repeated measures analyses of variance indicated several

significant changes over time for anxiety symptoms, self-

reported sleep problems, sleep efficiency, sleep minutes

and wake time variability (see Table 1). Additional pair-

wise comparisons of means indicated significant differ-

ences in means from T1 (spring of senior year of high

school) to T2 (fall first year of college; and/or T3, spring of

first year of college), but no significant differences in

means from T2 to T3. Specifically, anxiety symptoms

increased from T1 to T2. Wake time variability increased

from T1 to T3. In contrast, many of the sleep indicators

improved from T1 to T2. Subjective sleep problems

decreased from T1 to T2. Sleep efficiency and sleep min-

utes significantly increased from T1 to T2. All effect sizes

were within the small to medium range. Importantly, while

there were significant differences in means across time

points, there were no significant linear trends. Preliminary

analyses also indicated many of the sleep parameters were

positively related to depressive or anxiety symptoms, both

concurrently and prospectively, with the exception of sleep

efficiency (Table 2).

Based on the significant correlations with depressive

and/or anxiety symptoms, subjective sleep problems, sleep

latency, sleep start variability, and sleep efficiency were

selected as exogenous variables in the cross-lag models (r’s

ranging from .27 to .78; all p’s \ .05). Covariates were included on paths that had a significant effect, which were

then included in v2 model comparisons to ensure the inclusion of covariates enhanced model fit. In general, very

few covariates were indicated based on the zero-order

correlations, with the exception of the subjective sleep

problems model. This process resulted in different covar-

iates in each set of models models, accounting for the

varying degrees of freedom reported below.

Cross-Lagged Analyses of Longitudinal Relationships

Sleep Latency

The first model tested the reciprocal effects of sleep latency

and anxiety symptoms across the transition to college (see

Fig. 1). The baseline model with the stability paths of

394 J Youth Adolescence (2015) 44:389–404

123

Table 1 Repeated measures ANOVA and comparison of means across time

Variable Time 1 Time 2 Time 3 F T1–T2 Effect

size (d)

T1–T3 Effect

size (d) Mean SD Mean SD Mean SD

Depressive symptoms 14.53 8.63 15.78 9.12 16.03 10.70 1.12

Anxiety symptoms 5.87a,b 4.66 7.48a 6.86 8.17b 7.69 5.63** .27 .36

Subjective sleep problems 7.85a,b 3.20 5.86a 2.95 5.93b 3.45 20.12*** .65 .58

Sleep start variability .84 .99 .97 1.01 1.00 .93 .68

Sleep latency 11.60 9.68 9.53 7.84 8.71 6.82 1.99

Sleep efficiency 78.95a, b 12.96 83.90a 10.64 83.40b 10.80 11.21*** .42 .37

Sleep minutes (h) 5.95a,b 1.40 6.27a 1.27 6.38b 1.21 4.33* .24 .33

Wake time variability 0.75a 0.61 1.03 1.20 1.14a 0.70 7.35*** .59

T1 time 1, spring of senior year of high school, T2 time 2, fall of first year of college, T3 time 3, spring of first year of college. Means in a row

sharing a subscript are significantly different from each other. There were only significant differences from T1 to T2 and from T1 to T3 so effect

sizes are only presented for those comparisons

* p \ .05; ** p \ .01; *** p B .001

Table 2 Correlations among study variables

Variables 1 2 3 4 5 6 7 8 9 10 11 12

1. Depressive symptoms

T1

–

2. Anxiety symptoms T1 .62** –

3. Sleep problems T1 .35** .11 –

4. Sleep start variability

T1

.16 .07 .27* –

5. Sleep latency T1 -.01 -.01 .06 -.22 –

6. Sleep efficiency T1 -.03 -.13 -.06 .10 -.15 –

7. Sleep duration T1 -.07 .03 -.16 .05 -.01 .57** –

8. Wake time variability

T1

-.01 -.02 -.16 .37** -.14 .01 .02 –

9. Depressive symptoms

T2

.63** .60** .22 -.13 -.03 -.20 -.04 .007 –

10. Anxiety symptoms T2 .56** .63** .27* -.09 .14 -.10 -.05 -.01 .75** –

11. Sleep problems T2 .33** .15 .48* -.03 .03 -.04 -.08 .09 .42** .40** –

12. Sleep start variability

T2

.22 .18 .19 .12 -.10 -.08 -.03 .18 .33** .37** .07 –

13. Sleep latency T2 .21 .02 .46** .03 .30* .08 .07 -.17 .02 .14 .23* -.03

14. Sleep efficiency T2 -.18 -.18 -.06 .04 -.17 .38** .23 .10 -.11 -.16 .20 .05

15. Sleep duration T2 -.05 .09 -.15 -.22 .29* -.04 .26* -.26* .07 -.09 -.03 -.11

16. Wake time variability

T2

.13 -.03 .31** -.02 -.11 -.09 -.16 -.01 .08 .15 -.01 .57**

17. Depressive symptoms

T3

.39** .43** .29* -.17 .17 -.02 .12 -.02 .53** .54** .33** .19

18. Anxiety symptoms T3 .49** .51** .28* -.18 .23 -.02 .10 -.12 .55** .78** .35** .27*

19. Sleep problems T3 .28* .04 .46* .16 -.02 .03 .07 .25* .36** .43** .63** .18

20. Sleep start variability

T3

-.06 -.17 -.03 .05 -.06 -.04 .03 .02 .03 -.13 -.11 -.08

21. Sleep latency T3 .15 .13 .28* .32** .08 .25** .24 .01 -.11 .11 .16 -.01

22. Sleep efficiency T3 -.01 -.03 -.29* -.22 -.18 .41** .33** -.12 -.13 -.27* -.07 .05

23. Sleep duration T3 .09 .09 -.35** -.10 .24 .09 .25* .02 -.10 .002 -.24 -.17

24. Wake time variability

T3

-.01 -.08 -.02 -.16 -.04 .04 .05 -.08 -.003 .12 -.06 .05

J Youth Adolescence (2015) 44:389–404 395

123

anxiety symptoms and sleep latency was the only good

fit to the data v2 (8) = 7.32, p [ .05; CFI = 1.00; RMSEA = .00; SRMR = .06. All other models with

cross-lagged paths did not provide a superior fit to the data.

Within the stability model, only the autoregressive paths

across anxiety and sleep latency were significant (see

Fig. 1).

Upon examining sleep latency and depressive symp-

toms, the depressive symptoms led model was the best fit to

the data v2 (6) = 7.35, p [ .05; CFI = .99; RMSEA = .05; SRMR = .04. The constructs accounted for 14–44 %

of the variance, with depressive symptoms accounting for

more variance than sleep latency at all of the time points.

The direct paths examining depressive symptoms and sleep

latency were significant across time. There was only one

significant prospective effect, with depressive symptoms at

T1 predicting increased sleep latency at T2 (b = .24, p \ .05). The concurrent, and positive, association between

depressive symptoms and latency at T3 was also significant

(b = .26, p \ .05), but no other concurrent correlations were significant.

Sleep Start Variability

The second set of models included anxiety symptoms and

sleep start variability. The anxiety led model was also the

best fit to the data v2 (6) = 3.48, p [ .05; CFI = 1.00; RMSEA = .00; SRMR = .05. The constructs accounted

for 1 to 73 % of the variance, with anxiety accounting for

more than sleep start variability. Anxiety symptoms were

significantly stable and predictive over time; however,

sleep start variability was not. The two cross-over paths

were marginally significant. The path between anxiety at

T1 and start variability at T2 was marginally significant

(b = .22, p = .06). In turn, sleep start variability at T2 and increased anxiety symptoms at T3 was also marginally

Table 2 continued

Variables 13 14 15 16 17 18 19 20 21 22 23

1. Depressive symptoms T1

2. Anxiety symptoms T1

3. Sleep problems T1

4. Sleep start variability T1

5. Sleep latency T1

6. Sleep efficiency T1

7. Sleep duration T1

8. Wake time variability T1

9. Depressive symptoms T2

10. Anxiety symptoms T2

11. Sleep problems T2

12. Sleep start variability T2

13. Sleep latency T2 –

14. Sleep efficiency T2 -.08 –

15. Sleep duration T2 -.01 .29* –

16. Wake time variability T2 -.01 -.15 -.18 –

17. Depressive symptoms T3 .15 -.19 -.02 -.11 –

18. Anxiety symptoms T3 .32** -.22 -.04 -.06 .67** –

19. Sleep problems T3 .22 .01 -.25* -.12 .45** .42** –

20. Sleep start variability T3 -.17 -.05 -.07 .06 -.02 -.07 .16 –

21. Sleep latency T3 .45** .05 -.14 -.01 .14 .15 .26* .05 –

22. Sleep efficiency T3 -.06 .53** .28* -.01 -.15 -.17 -.27* -.18 -.15 –

23. Sleep duration T3 .03 .08 .48** -.11 -.10 -.02 .20 -.40 .04 .16

24. Wake time variability T3 -.003 -.13 -.04 .18 -.07 .17 .09 .03 -.04 -.18 .08

T1 time 1, spring of senior year of high school, T2 time 2, fall first year of college, T3 time 3, spring first year of college

* p \ .05; ** p \ .01

396 J Youth Adolescence (2015) 44:389–404

123

Anxiety Symptoms T2Anxiety Symptoms T1 Anxiety Symptoms T3

Sleep Start Variability T1

Sleep Start Variability T2

Sleep Start Variability T3

.68*** (.06)

.14 + (.08)

.79*** (.05)

.34***(.10).22 + (.12)

Depressive Symptoms T2

Depressive Symptoms T1

Depressive Symptoms T3

Sleep Start Variability T1

Sleep Start Variability T2

Sleep Start Variability T3

.66*** (.07) .57*** (.09)

.28**(.11)

.25*(.11)

Fig. 2 Sleep start variability cross-lag model with depressive symp- toms and anxiety symptoms. Standardized coefficients are provided.

For readability, non-significant paths are dashed grey, marginally

significant paths are dashed black, significant paths are solid black.

T1, spring of senior year of high school; T2, fall of first year of

college; T3, spring of first year of college. ?

p [ .05 and \ .10; *p \ .05; **p B 01; ***p B .001

Anxiety Symptoms T2Anxiety Symptoms T1 Anxiety Symptoms T3

Sleep Onset Latency T1 Sleep Onset Latency T2 Sleep Onset Latency T3

.68*** (.06) .83*** (.04)

.29**(.10) .45**(.11)

Depressive Symptoms T2Depressive Symptoms T1 Depressive Symptoms T3

Sleep Onset Latency T1 Sleep Onset Latency T2 Sleep Onset Latency T3

.66*** (.07) .61*** (.08)

.28**(.11)

.24* (.11)

.45***(.10)

.26 *(.12)

Fig. 1 Sleep onset latency cross-lag model with depressive symp- toms and anxiety symptoms. Standardized coefficients are provided.

For readability, non-significant paths are dashed grey, marginally

significant paths are dashed black, significant paths are solid black.

T1, spring of senior year of high school; T2, fall of first year of

college; T3, spring of first year of college. *p \ .05; **p \ .01; *** p B .001

J Youth Adolescence (2015) 44:389–404 397

123

significant (b = .14, p = .07). There was also a significant concurrent association between anxiety symptoms and

sleep start variability at T2 (b = .34, p = .001), such that increased anxiety was associated with increased sleep start

variability.

In regards to depressive symptoms and sleep start vari-

ability, the depressive symptoms led model was the best

fit to the data v2 (6) = 3.98, p [ .05; CFI = 1.00; RMSEA = .00; SRMR = .04 (see Fig. 2). The amount of

variance accounted for by the constructs ranged from 1 to

44 %, with depressive symptoms accounting for more vari-

ance than sleep start time variability. Only depressive

symptoms showed significant stability across time. One

crossover effect was significant with depressive symptoms at

T1 predicting increased sleep start variability at T2 (b = .25, p \ .05). The concurrent relationship at T2 was significant (b = .28, p = .01), with higher depressive symptoms being significantly associated with increased sleep start variability.

Sleep Efficiency

The third, and final set of models with objective sleep data,

examined the cross-lagged effects between sleep efficiency

and symptoms of anxiety and depression. In both cases, the

stability models including anxiety or depressive symptoms

and sleep efficiency were the best fits of the data v2 (8) = 11.77, p [ .05; CFI = .99; RMSEA = .08; SRMR = .03, and v2 (8) = 11.11, p [ .05; CFI = .99; RMSEA = .07; SRMR = .05, respectively. Thus, no specific cross-lagged

effects were interpreted. The autoregressive effects across all

time points of sleep efficiency, anxiety symptoms, and

depressive symptoms were significant. There was one con-

current, positive, correlation between depressive symptoms

and sleep efficiency at T2 (b = .24, p \ .05).

Subjective Sleep Problems

The final set of models tested the reciprocal effects of

subjective sleep problems, anxiety, and depressive symp-

toms (see Fig. 3). For anxiety symptoms, the full model

with both cross-lagged effects was the best fit to the data v2

(9) = 12.29, p [ .05; CFI = .98; RMSEA = .07; SRMR = .05. The variance accounted for by the constructs

ranged from 39 to 71 % of the variance, with anxiety

accounting for more variance than subjective sleep prob-

lems. Anxiety symptoms and subjective sleep problems

were significant and stable across all three time points.

There were two cross-lagged effects. Subjective sleep

problems at T1 predicted increased anxiety symptoms at

T2 (b = .29, p = .001). There was also a significant path

Anxiety Symptoms T2Anxiety Symptoms T1 Anxiety Symptoms T3

Subjective Sleep Problems T1

Subjective Sleep Problems T2

Subjective Sleep Problems T3

.65*** (.06) .82*** (.05)

.30**(.11)

.29***(.08)

.58*** (.08)

.24*(.29)

Depressive Symptoms T2Depressive Symptoms T1 Depressive Symptoms T3

Subjective Sleep Problems T1

Subjective Sleep Problems T2

Subjective Sleep Problems T3

.66*** (.07) .49*** (.10)

.28*(.11)

.48*** (.10)

.26** (.09)

.67*** (.07)

.38*** (.10)

.32 **(.11)

.57*** (.09)

Fig. 3 Subjective sleep problems cross-lag model with depressive symptoms and anxiety symptoms. Standardized coefficients are

provided. For readability, non-significant paths are dashed grey,

marginally significant paths are dashed black, significant paths are

solid black. T1, spring of senior year of high school; T2, fall of first

year of college; T3, spring of first year of college. *p \ .05; **p \ .01; ***p B .001

398 J Youth Adolescence (2015) 44:389–404

123

from anxiety at T2 predicting increased subjective sleep

problems at T3 (b = .24, p \ .05). Last, there was one significant concurrent relationship between increased anx-

iety and subjective sleep problems at T2 (b = .30, p \ .01).

When examining depressive symptoms and subjective

sleep problems, the depressive symptoms led model was the

best fit to the data and had adequate fit indices v2 (6) = 9.03, p [ .05; CFI = .98; RMSEA = .08; SRMR = .08. The constructs accounted for 35–45 % of the variance, with

depressive symptoms and subjective sleep problems

accounting for similar amounts of variance. Depressive

symptoms and subjective sleep problems were significant

and stable across all three time points. There was one pre-

dictive effect, such that depressive symptoms at T1 predicted

increased subject sleep complaints at T2 (b = .26, p \ .01). There were also positive and significant concurrent associ-

ation between depressive symptoms and subjective sleep

problems at all three time points (all p’s B .01, see Fig. 3).

Discussion

Researchers have accumulated a wealth of evidence sug-

gesting sleep is reduced in adolescence (National Sleep

Foundation 2006; Roberts et al. 2001), and that sleep

problems may continue for youth who enter college (e.g.,

Lund et al. 2010). These sleep problems have been linked

with a variety of negative physical and mental health

outcomes, including obesity, risk for suicide, anxiety, and

depression (Alfano et al. 2009; Gupta et al. 2002; Liu

2004). Specifically, studies have demonstrated longitudinal

associations (sometimes reciprocal) between sleep prob-

lems and symptoms of depression and anxiety in commu-

nity samples of adolescents and young adults (e.g., Johnson

et al. 2006; Roberts et al. 2002). Most prior research,

though, has been limited by subjective self-report mea-

surements of sleep and little is known about the direction of

effects between sleep and anxiety/depression during ado-

lescence, particularly across a substantial socio-cultural

shift like the transition to college. We utilized a multi-

method approach for assessing sleep quantity and quality in

order to establish a better understanding of which indica-

tors of sleep quantity or quality might change over this

developmental transition, and whether those changes were

associated concurrently or longitudinally with changes in

anxiety and depressive symptoms. We also used rigorous

statistical modeling to best understand if reciprocal rela-

tionships existed over time. Several objective and sub-

jective indicators of sleep improved over the transition to

college, while symptoms of anxiety increased. Further, we

found differential associations between sleep and symp-

toms of depression and anxiety over time. Specifically, we

identified longitudinal relationships between subjective

sleep quality and anxiety symptoms over time. In contrast,

higher depressive symptoms before the transition to college

(T1) preceded increases in subjective sleep problems, sleep

latency and start time variability in the first semester of

college (T2). These findings build on research in child and

adolescent populations (e.g., Kelly and El-Sheikh 2014),

and suggest some important targets in the prevention of

stress-related symptoms and disorders over the transition to

college.

Our findings revealed significant improvements in sev-

eral sleep patterns across the transition to college. Across

the whole sample, we found increases in sleep efficiency,

sleep minutes, and subjective sleep quality from pre- to

post-transition but no changes after the transition (from T2

to T3). Although these results were contrary to our

hypotheses, they reflect recent findings from a national

dataset indicating that sleep duration improves from ado-

lescence into young adulthood (Maslowsky and Ozer

2014). It is important to note that the sleep minutes expe-

rienced by our sample at all three measurement occasions

were significantly lower than recommended levels of 9 h a

night for adolescents and 7–8 h a night for adults (National

Sleep Foundation 2011). Further, our study findings

regarding increases in sleep efficiency and decreases in

sleep latency may reflect compensatory sleep mechanisms:

when either sleep duration or sleep quality decreases,

individuals typically adapt by experiencing improvements

in other components of sleep (Sadeh et al. 2003). In the

current sample, adolescents perceived their sleep to be of

higher quality as they transitioned into college, and even

though sleep duration improved across the transition, val-

ues were still below recommended levels for this age

group.

Wake time variability increased from pre- to post-tran-

sition, likely due to varying external demands like class

times or work schedules or increased autonomy and deci-

sion-making about sleep timing. Our findings partially

support recent studies by researchers who demonstrated

that particular subgroups of youth had high sleep vari-

ability during the first semester of college (Ari and Shul-

man 2012). Although we did not collect objective

indicators of weekend sleep, it is possible that many youth

in our study were experiencing ‘‘catch up’’ sleep on

weekends. This could have resulted in an overall percep-

tion that their sleep was more positive, even when their

weekday durations were shorter than recommended levels.

The variability in timing and overall low level of duration

indicates that many youth were experiencing what has been

termed ‘‘social jetlag’’ (Wittmann et al. 2006), which has

been linked with poor academic performance (Haraszti

et al. 2014) and poor physical and psychological well-being

in morning chronotype college students (Lau et al. 2013).

J Youth Adolescence (2015) 44:389–404 399

123

Importantly, our study was able to examine sleep patterns

both pre- and post-transition to identify whether poor

sleeping habits were established prior to college entry and

whether they continued to improve over time or only

improved in the first semester of college. Our descriptive

findings about sleep specifically indicate that efforts aimed

at improving length of sleep duration or quality are likely

best aimed at adolescents before they enter the college

environment, whereas wake time variability may be a

problem established during the first year of college.

Studies examining reciprocal relationships among

depression and sleep have typically been conducted using

subjective indicators of sleep and have found support for

sleep problems preceding depression in childhood and ado-

lescence (e.g., Gregory et al. 2009). Although some studies

have found support for bidirectional relationships (Cousins

et al. 2011; Kelly and El-Sheikh 2014), none to our knowl-

edge have been conducted over a substantial socio-cultural

shift such as the transition to college. Interestingly, we

consistently found that pre-transition depressive symptoms

were associated with subsequent post-transition subjective

and objective sleep problems (but not the reverse), and post-

transition depressive symptoms were concurrently associ-

ated with greater sleep problems (increased variability and

latency, decreased subjective quality). Our findings com-

plement available research on the existence of concurrent

self-reported sleep problems and depressive symptoms in

adolescents and college students (e.g., Gress-Smith et al.

2013, Orzech et al. 2011; Regenstein et al. 2010). It is also

notable that there were not significant increases in depressive

symptoms across the transition to college. Indeed, results

indicated that depressive symptoms before the transition

may be a risk factor for poor sleep quality in college. Their

co-occurrence may be a result of worsening symptoms of

depression over a stressful transition.

Given other research suggesting the bidirectional inter-

play between daily experiences of stress, anxiety symp-

toms, depressive symptoms and sleep (e.g., Doane and

Thurston 2014, Fuligni and Hardway 2006; Galambos et al.

2010), it is possible that students who were depressed were

less likely to utilize adaptive coping strategies and thus

may have experienced reduced sleep quality (e.g., greater

latency) or greater sleep start time variability. Coping or

emotional regulation strategies were not examined in the

present study; however, other studies have shown the

mediating role of emotion regulation between social ties

and sleep problems in college students (Tavernier and

Willoughby 2014), suggesting that these skills may be

integral when adapting to stressful environments. Although

these hypotheses should be tested in larger samples, it

seems that both depressive symptoms and poor sleep

hygiene are likely responsive targets to prevention efforts

for college student adjustment.

Our results also illustrated that anxiety symptoms

increased over the transition to college and were concur-

rently related to both subjective and objective post-transition

sleep problems (e.g., subjective sleep problems, sleep start

time variability). Although some prior research in children

and adolescents has found that sleep problems typically

precede anxiety symptoms and disorders (e.g., Gregory et al.

2005), we did not find this to be the case over the transition to

college. However, poor subjective sleep quality in high

school (T1) was associated with subsequent anxiety in the

first semester of college (T2). Further, post-transition anxiety

symptoms (T2) were associated with concurrent and pro-

spective subjective sleep problems. The nature of these paths

is similar to those identified in children (Kelly and El-Sheikh

2014) and implicate mechanisms that are common across

sleep disturbance and anxiety symptoms.

We did not identify associations among sleep minutes

and symptoms of depression and anxiety, which was

inconsistent with our hypotheses and some previous

research (e.g., Perlman et al. 2006). Some theorists have

argued that sleep quality is a better predictor of adjustment

than quantity (Krystal and Edinger 2008), but most studies

have only examined these differential associations using

subjective reports. A recent study in a nationally repre-

sentative sample found that sleep duration increased from

adolescence into young adulthood (Maslowsky and Ozer

2014), which was hypothesized to reflect fewer early

morning environmental demands (i.e., early high school

start times). Similarly in this sample, youth likely have

more freedom to set their bed and wake times in college

such that sleep duration most likely reflects choices or

preferences (e.g., eveningness) rather than impairments due

to increases in symptoms.

Researchers have hypothesized potential mechanisms

underlying associations among poor sleep and symptoms of

depression and anxiety. Theories involving emotional reg-

ulation (Dahl 1996) and evidence from neuroimaging studies

suggest that the prefrontal cortex (PFC) and amygdala may

play critical roles. For example, greater activation of the PFC

occurred in response to negative emotion and affect fol-

lowing poor sleep (Ochsner et al. 2004; Urry et al. 2006),

suggesting links with emotion regulation (Wang and Sau-

dino 2011). Sleep deprivation studies have illustrated overall

declines in PFC function after insufficient sleep (Drummond

and Brown 2001; Thomas et al. 2000). Further, researchers

have found that amygdala reactivity to fearful stimuli was

associated with greater depressive symptoms in poor sleep-

ers, but not good sleepers, suggesting that sleep modulates

associations between amygdala reactivity and emotion reg-

ulation (Prather et al. 2013). Finally, ruminating thought

processes and emotional distress are correlates of poor sleep

(for review, see Kahn et al. 2013) and symptoms of anxiety

and depression (McLaughlin and Nolen-Hoeksema 2011),

400 J Youth Adolescence (2015) 44:389–404

123

suggesting that they may serve as mediating factors under-

lying longitudinal relationships over the transition to college.

Despite contributions, several limitations should be

considered. These measurements occurred in a small

sample of youth who were planning to attend a 4-year

university close to home and who were disproportionately

female. Further, our sample was unusual in that higher SES

youth were more likely to be lost to attrition. Given the

characteristics of our sample, our findings may not gener-

alize to adolescents who are transitioning to work or

community college. In addition, given the size of our

sample, we were unable to test the associations between

sleep and anxiety symptoms and sleep and depressive

symptoms simultaneously, and were likely underpowered

to detect significant paths with smaller effect sizes. For

example, we identified two marginally significant cross-

lagged paths between anxiety and sleep start variability.

With a larger sample size we may have had the power to

detect paths with small-medium effect sizes like these at a

statistically significant level. Future work should examine

pathways among sleep parameters (e.g., sleep variability),

depressive symptoms and anxiety symptoms concurrently

over a greater number of days in a larger sample of youth.

We examined a specific transition period that is common

for many adolescents. It is possible that other patterns

might emerge over other periods of development, a longer

study period, or during a time of more stability. Finally, we

did not examine other adverse sleep hygiene behaviors

(e.g., alcohol use) that might contribute to sleep problems

and adjustment. Future research should also include indi-

cators of stress-related coping behaviors to understand

potential mechanisms underlying associations between

sleep and symptoms of depression and anxiety.

Conclusion

Collectively, our findings contribute to the literature

examining developmental trajectories of sleep and symp-

toms of depression and anxiety. We highlighted the lon-

gitudinal and reciprocal relationships among these

indicators of adjustment over a period of time typically

characterized by change and uncertainty for many adoles-

cents. Further, our results underscore the importance of

utilizing multiple methods of assessment (e.g., subjective,

actigraphic) to capture dimensions of sleep duration,

quality and variability within participants’ naturalistic

environments. Our findings have direct relevance for high

school and university administrators seeking to promote

well-being in their student populations and by identifying

potential targets to help promote successful adjustment

during the transition to college. Our research may have

implications for future prevention and intervention

programs in both high school and college that include

strategies to help youth cope effectively with adjustment

problems like sleep variability and symptoms of depression

and anxiety.

Acknowledgments The authors thank Michael R. Sladek and Scott Van Lenten for comments on earlier drafts of this manuscript. This

research was conducted with the support of the Institute for Social

Science Research at Arizona State University (LD, Principal

Investigator).

Author contributions LD conceived of the study, participated in its design, led coordination and data collection and drafted the manu-

script; JGS participated in the design of the study, helped draft the

manuscript and performed the statistical analysis; RS participated in

the design and helped draft the manuscript. All authors read and

approved the final manuscript.

References

Acebo, C., Sadeh, A., Seifer, R., Tzischinsky, O., Wolfson, A. R.,

Hafer, A., et al. (1999). Estimating sleep patterns with activity

monitoring in children and adolescents: How many nights are

necessary for reliable measures? Sleep, 22(1), 95–103.

Alfano, C. A., Zakem, A. H., Costa, N. M., Taylor, L. K., & Weems,

C. F. (2009). Sleep problems and their relation to cognitive

factors, anxiety, and depressive symptoms in children and

adolescents. Depression and Anxiety, 26, 503–512.

Ari, L. L., & Shulman, S. (2012). Pathways of sleep, affect, and stress

constellations during the first year of college: Transition

difficulties of emerging adults. Journal of Youth Studies, 15(3),

273–292.

Buckhalt, J. A., El-Sheikh, M., & Keller, P. (2007). Children’s sleep

and cognitive functioning: Race and socioeconomic status as

moderators of effects. Child Development, 78(1), 213–231.

Buysse, D. J., Reynolds, C. F, I. I. I., Monk, T. H., Berman, S. R., &

Kupfer, D. J. (1989). The Pittsburgh Sleep Quality Index: A new

instrument for psychiatric practice and research. Psychiatry

Research, 28, 193–213.

Carskadon, M. A. (2011). Sleep in adolescents: The perfect storm.

Pediatric Clinics of North America, 58(3), 637–647.

Carskadon, M. A., Wolfson, A. R., Acebo, C., Tzischinsky, O., &

Seifer, R. (1998). Adolescent sleep patterns, circadian timing,

and sleepiness at a transition to early school days. Sleep, 21(8),

871–881.

Cohen, J. (1988). Statistical power analysis for the behavioral

sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Cole, D. A., & Maxwell, S. E. (2003). Testing mediational models

with longitudinal data: Questions and tips in the use of structural

equation modeling. Journal of Abnormal Psychology, 112(4),

558–577.

Compas, E. B., Wagner, B. M., Slavin, L. A., & Vannatta, K. (1986).

A prospective study of life events, social supports, and

psychological symptomatology during the transition from high

school to college. American Journal of Community Psychology,

14(3), 241–257.

Coulombe, J. A., Reid, G. J., Boyle, M. H., & Racine, Y. (2010).

Concurrent associations among sleep problems, indicators of

inadequate sleep, psychopathology, and shared risk factors in a

population-based sample of healthy Ontario children. Journal of

Pediatric Psychology, 35(7), 790–799.

Cousins, J. C., Whalen, D. J., Dahl, R. E., Forbes, E. E., Olino, T. M.,

Ryan, N. D., et al. (2011). The bidirectional association between

J Youth Adolescence (2015) 44:389–404 401

123

daytime affect and nighttime sleep in youth with anxiety and

depression. Journal of Pediatric Psychology, 36(9), 969–979.

Dahl, R. E. (1996). The regulation of sleep and arousal: Development and

psychopathology. Development and Psychopathology, 8(1), 3–27.

Dahl, R. E., & Lewin, D. S. (2002). Pathways to adolescent health:

Sleep regulation and behavior. Journal of Adolescent Health,

31(65), 175–184.

Doane, L. D., & Thurston, E. C. (2014). Associations among sleep,

daily experiences, and loneliness in adolescence: Evidence of

moderating and bidirectional pathways. Journal of Adolescence,

37, 145–154.

Drummond, S. P. A., & Brown, G. G. (2001). The effects of total

sleep deprivation on cerebral responses to cognitive perfor-

mance. Neuropsychopharmacology, 23(S5), S68–S73.

Eisenberg, D., Gollust, S. E., Golberstein, E., & Hefner, J. L. (2007).

Prevalence and correlates of depression, anxiety, and suicidality

among university students. American Journal of Orthopsychia-

try, 77(4), 534–542.

Enders, C. K., & Bandalos, D. L. (2001). The relative performance of

full information maximum likelihood estimation for missing data

in structural equation models. Structural Equation Modeling, 8,

430–457.

Fuligni, A. J., & Hardway, C. (2006). Daily variation in adolescents’

sleep, activities, and psychological well-being. Journal of

Research on Adolescence, 16(3), 353–378.

Furr, S. R., Westefeld, J. S., McConnell, G. N., & Jenkins, J. M.

(2001). Suicide and depression among college students: A

decade later. Professional Psychology: Research and Practice,

32(1), 97.

Galambos, N. L., Howard, A. L., & Maggs, J. L. (2010). Rise and fall

of sleep quantity and quality with student experiences across the

first year of university. Journal of Research on Adolescence,

21(2), 324–349.

Galambos, N. L., Vargas Lascano, D. I., Howard, A. L., & Maggs, J.

L. (2013). Who sleeps best? Longitudinal patterns and covariates

of change in sleep quantity, quality, and timing across four

university years. Behavioral Sleep Medicine, 11(1), 8–22.

Gregory, A. M., Caspi, A., Eley, T., Moffitt, T. E., O’Connor, T. G.,

& Poulton, R. (2005). Prospective longitudinal associations

between persistent sleep problems in childhood and anxiety and

depression disorders in adulthood. Journal of Abnormal Child

Psychology, 33(2), 157–163.

Gregory, A. M., & O’Connor, T. G. (2002). Sleep problems in

childhood: A longitudinal study of developmental change and

association with behavioral problems. Journal of the American

Academy of Child and Adolescent Psychiatry, 41(8), 964–971.

Gregory, A. M., Rijsdijk, F. V., Lau, J. Y., Dahl, R. E., & O’Connor,

T. G. (2009). The direction of longitudinal associations between

sleep problems and depressive symptoms: A study of twins aged

8 to 10 years. Sleep, 32(2), 189–199.

Gregory, A. M., & Sadeh, A. (2012). Sleep, emotional and behavioral

difficulties in children and adolescents. Sleep Medicine Reviews,

16, 129–136.

Gress-Smith, J. L., Roubinov, D. S., Adreotti, C., Compas, B. E., &

Lueken, L. J. (2013). Prevalence, severity and risk factors for

depressive symptoms and insomnia in college undergraduates.

Stress Health. doi:10.1002/smi.2509.

Gupta, N. K., Mueller, W. H., Chan, W., & Meininger, J. C. (2002). Is

obesity associated with poor sleep quality in adolescents?

American Journal of Human Biology, 14(6), 762–768.

Haraszti, R. A., Ella, K., Gyongyosi, N., Roenneberg, T., & Kaldi, K.

(2014). Social jetlag negatively correlates with academic

performance in undergraduates. Chronobiology International,

31, 1–10.

Hartley, M. T. (2012). Assessing and promoting resilience: An

additional tool to address the increasing number of college

students with psychological problems. Journal of College

Counseling, 15, 37–51.

Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in

covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary

Journal, 6(1), 1–55.

Jansson-Frogmark, M., & Lindbloom, K. (2008). A bidirectional

relationship between anxiety and depression, and insomnia? A

prospective study in the general population. Journal of Psycho-

somatic Research, 64, 443–449.

Jean Louis, G., von Gizycki, H., Zizi, F., & Nunes, J. (1998). Mood

states and sleepiness in college students: Influences of age, sex,

habitual sleep, and substance use. Perceptual and Motor Skills,

87, 507–512.

Johnson, E. O., Chilcoat, H. D., & Breslau, N. (2000). Trouble

sleeping and anxiety/depression in childhood. Psychiatry

Research, 94, 93–102.

Johnson, E. O., Roth, T., & Breslau, N. (2006). The association of

insomnia with anxiety disorders and depression: Exploration of

the direction of risk. Journal of Psychiatric Research, 40,

700–708.

Jonge, J., Dormann, C., Janssen, P. P., Dollard, M. F., Landeweerd, J.

A., & Nijhuis, F. J. (2001). Testing reciprocal relationships

between job characteristics and psychological well-being: A

cross-lagged structural equation model. Journal of Occupational

and Organizational Psychology, 74(1), 29–46.

Kahn, M., Sheppes, G., & Sadeh, A. (2013). Sleep and emotions:

Bidirectional links and underlying mechanisms. International

Journal of Psychophysiology, 89, 218–228.

Kelly, R. J., & El-Sheikh, M. (2014). Reciprocal relations between

children’s sleep and their adjustment over time. Developmental

Psychology, 50(4), 1137–1147.

Kerr, S., Johnson, V. K., Gans, S. E., & Krumrine, J. (2004).

Predicting adjustment during the transition to college: Alexithy-

mia, perceived stress, and psychological symptoms. Journal of

College Student Development, 45(6), 593–611.

Krystal, A. D., & Edinger, J. D. (2008). Measuring sleep quality.

Sleep Medicine, 9(1), 10–17.

Larose, S., & Boivin, M. (1998). Attachment to parents, social

support expectations, and socioemotional adjustment during the

high school-college transition. Journal of Research on Adoles-

cents, 8(1), 1–27.

Lau, E. Y. Y., Wong, M. L., Ng, E. C. W., Hui, C. H., Cheung, S. F., &

Mok, D. S. Y. (2013). ‘‘Social jetlag’’ in morning-type college

students living on campus: Implications for physical and psycho-

logical well-being. Chronobiology International, 30(7), 910–918.

Liu, X. (2004). Sleep and adolescent suicidal behavior. Sleep-New

York Then Westchester, 27, 1351–1358.

Loehlin, J. C. (2004). Latent variable models: An introduction to

factor, path, and structural equation analysis (4th ed.). Mahwah,

NJ: Erlbaum.

Lovibond, P. F., & Lovibond, S. H. (1995). The structure of negative

emotional states: Comparison of the Depression Anxiety Stress

Scales (DASS) with the Beck Depression and Anxiety Invento-

ries. Behavioral Research Therapy, 33(3), 335–343.

Lund, H. G., Reider, B. D., Whiting, A. B., & Prichard, J. R. (2010).

Sleep patterns and predictors of disturbed sleep in a large

population of college students. Journal of Adolescent Health, 46,

124–132.

Maslowsky, J., & Ozer, E. J. (2014). Developmental trends in sleep

duration in adolescence and young adulthood: Evidence from a

national United States sample. Journal of Adolescent Health,

54(6), 691–697.

McLaughlin, K. A., & Nolen-Hoeksema, S. (2011). Rumination as a

transdiagnostic factor in depression and anxiety. Behaviour

Research and Therapy, 49, 186–193.

402 J Youth Adolescence (2015) 44:389–404

123

Mezick, E. J., Matthews, K. A., Hall, M., Strollo, P. J., Buysse, D. J.,

Kamarck, T. W., et al. (2008). Influence of race and socioeco-

nomic status on sleep: Pittsburgh SleepSCORE Project. Psycho-

somatic Medicine, 70(4), 410–416.

Moo-Estrella, J., Perez-Benitez, H., Solis-Rodriguez, F., & Aran-

kowsky-Sandoval, G. (2005). Evaluation of depressive symp-

toms and sleep alterations in college students. Archives of

Medical Research, 36, 393–398.

Moore, M., Kirchner, H. L., Drotar, D., Johnson, N., Rosen, C.,

Ancoli-Israel, S., et al. (2009). Relationships among sleepiness,

sleep time, and psychological functioning in adolescents.

Journal of Pediatric Psychology, 34(10), 1175–1183.

Muthén, L. K., & Muthén, B. (2012). Mplus short courses:

Exploratory factor analysis, confirmatory factor analysis, and

structural equation modeling for continuous outcomes. Recuper-

ado em novembro.

National Sleep Foundation. (2006). Summary of findings: 2006 Sleep

in America poll. Retrieved from http://sleepfoundation.org/sites/

default/files/2006_summary_of_findings.pdf.

National Sleep Foundation. (2011). How much sleep do we really

need? Retrieved from http://www.sleepfoundation.org/article/

how-sleep-works/howmuch-sleep-do-we-really-need.

Nyer, M., Farabaugh, A., Fehling, K., Soskin, D., Holt, D.,

Papakostas, G. I., et al. (2013). Relationship between sleep

disturbance and depression, anxiety, and functioning in college

students. Depression and Anxiety, 30, 873–880.

Oakley, N. R. (1997). Validation with polysomnography of the

Sleepwatch sleep/wake scoring algorithm used by the Actiwatch

activity monitoring system. Bend: Mini Mitter, Cambridge

Neurotechnology.

Ochsner, K. N., Ray, R. D., Cooper, J. C., Robertson, E. R., Chopra,

S., Gabrieli, J. D., et al. (2004). For better or for worse: Neural

systems supporting the cognitive down- and up-regulation of

negative emotion. Neuroimage, 23, 483–499.

Oginska, H., & Pokorski, J. (2006). Fatigue and mood correlates of

sleep length in three age-social groups: School children,

students, and employees. Chronobiology International, 23(6),

1317–1328.

Orsal, O., Orsal, O., Alparslan, G. B., & Unsal, A. (2012). Evaluation

of the relation between quality of sleep and anxiety among

university students. HealthMed, 6(7), 2244–2255.

Orzech, K. M., Salafsky, D. B., & Hamilton, L. A. (2011). The state

of sleep among college students at a large public university.

Journal of American College Health, 59(7), 612–619.

Perlman, C. A., Johnson, S. L., & Mellman, T. A. (2006). The

prospective impact of sleep duration on depression and mania.

Bipolar Disorders, 8, 271–274.

Phinney, J. S., & Haas, K. (2003). The process of coping among

ethnic minority first-generation college freshman: A narrative

approach. The Journal of Social Psychology, 143(6), 707–726.

Pilcher, J. J., Ginter, D. R., & Sadowsky, B. (1997). Sleep quality

versus sleep quantity: Relationships between sleep and measures

of health, well-being and sleepiness in college students. Journal

of Psychosomatic Research, 42(6), 583–596.

Prather, A. A., Bogdan, R., & Hariri, A. R. (2013). Impact of sleep

quality on amygdala reactivity, negative affect, and perceived

stress. Psychosomatic Medicine, 75(4), 350–358.

Pryor, J. H., Hurtado, S., DeAngelo, L., Blake, L. P., & Tran, S.

(2010). The American freshman: National norms fall 2010.

Higher Education Research Institute, University of California.

Radloff, L. S. (1977). The CES-D Scale: A self-report depression

scale for research in the general population. Applied Psycholog-

ical Measurement, 1(3), 385–401.

Regenstein, Q., Nataranjan, V., Pavlova, M., Kawaski, S., Gleason,

R., & Koff, E. (2010). Sleep debt and depression in female

college students. Psychiatry Research, 176, 34–39.

Roberts, R. E., Andrews, J. A., Lewinsohn, P. M., & Hops, H. (1990).

Assessment of depression in adolescents using the Center for

Epidemiologic Studies Depression Scale. Psychological Assess-

ment: A Journal of Consulting and Clinical Psychology, 2(2), 122–128.

Roberts, R. E., Roberts, C. R., & Chen, I. G. (2001). Functioning of

adolescents with symptoms of disturbed sleep. Journal of Youth

and Adolescence, 30(1), 1–18.

Roberts, R. E., Roberts, C. R., & Chen, I. G. (2002). Impact of

insomnia on future functioning of adolescents. Journal of

Psychosomatic Research, 53, 561–569.

Sadeh, A., Gruber, R., & Raviv, A. (2003). The effects of

sleep restriction and extension on school-age children: What

a difference an hour makes. Child Development, 74(2),

444–455.

Sadeh, A., Hauri, P. J., Kripke, D. F., & Lavie, P. (1995). The role of

actigraphy in the evaluation of sleep disorders. Sleep, 18(4),

288–302.

Sadeh, A., Raviv, A., & Gruber, R. (2000). Sleep patterns and sleep

disruption in school-age children. Developmental Psychology,

36(3), 291–301.

Shochat, T., Cohen-Zion, M., & Tzischinsky, O. (2014). Functional

consequences of inadequate sleep in adolescents: A systematic

review. Sleep Medicine Reviews, 18(1), 75–87.

Suen, L. K. P., Hon, L. K. E., & Tam, W. W. S. (2008). Association

between sleep behavior and sleep-related factors among univer-

sity students in Hong Kong. Chronobiology International, 25(5),

760–775.

Tavernier, R., & Willoughby, T. (2014). A longitudinal examination

of the bidirectional association between sleep problems and

social ties at university: The mediating role of emotion

regulation. Journal of Youth and Adolescence, 1–14. doi:10.

1007/s10964-014-0107-x.

Thomas, M., Sing, H., Belenky, G., Holcomb, H., Mayberg, H.,

Dannals, F., et al. (2000). Neural basis of alertness and cognitive

performance impairments during sleepiness. I. Effects of 24 h of

sleep deprivation on waking human regional brain activity.

Journal of Sleep Research, 9, 335–352.

Tremaine, R. B., Dorrian, J., & Blunden, S. (2010). Subjective and

objective sleep in children and adolescents: Measurement, age,

and gender differences. Sleep and Biological Rhythms, 8,

229–238.

Urry, H. L., van Reekum, C. M., Johnstone, T., Kalin, N. H., Thurow,

M. E., Schaefer, H. S., et al. (2006). Amygdala and ventromedial

prefrontal cortex are inversely coupled during regulation of

negative affect and predict the diurnal pattern of cortisol

secretion among older adults. Journal of Neuroscience, 26,

4415–4425.

Wang, M., & Saudino, K. J. (2011). Emotion regulation and stress.

Journal of Adult Development, 18(2), 95–103.

Wittmann, M., Dinich, J., Merrow, M., & Roenneberg, T. (2006).

Social jetlag: Misalignment of biological and social time.

Chronobiology International, 23(1–2), 497–509.

Leah D. Doane Ph.D is an Assistant Professor of Psychology at Arizona State University. She received her doctorate in Human

Development and Social Policy from Northwestern University’s

School of Education and Social Policy. Her major research interests

include adolescent and emerging adult development, particularly with

regard to identifying and understanding psychophysiological mech-

anisms underlying adolescent and young adult everyday stress

experiences in naturalistic settings.

Jenna L. Gress-Smith Ph.D received her doctorate in Clinical Psychology (Health Psychology emphasis) and Master of Arts in

J Youth Adolescence (2015) 44:389–404 403

123

Psychology at Arizona State University. Her major research interests

include the physiological processes associated with mental health

outcomes during major life transitions, such as pregnancy, postpartum

adjustment, and emerging adulthood. She is also interested in

resilience theory and how physiological and biological processes,

such as cortisol and sleep, are associated with stress, affect, and well-

being in populations facing health disparities.

Reagan S. Breitenstein is a candidate for Master of Arts in Develop- mental Psychology at Arizona State University. Her major research

interests include understanding relations between adolescent sleep and

mental and physical health outcomes. She is also focused on identifying

genetic and environmental contributions to associations between inter-

personal relationships and middle childhood sleep and physiology.

404 J Youth Adolescence (2015) 44:389–404

123

Copyright of Journal of Youth & Adolescence is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.