Skip to main content
NCBI home page
Psychological Science logoLink to Psychological Science
. 2020 Jun 19;31(7):822–834. doi: 10.1177/0956797620919432

Effortful Control Moderates the Relation Between Electronic-Media Use and Objective Sleep Indicators in Childhood

Sierra Clifford 1,✉, Leah D Doane 1, Reagan Breitenstein 2, Kevin J Grimm 1, Kathryn Lemery-Chalfant 1
PMCID: PMC7492726  PMID: 32558622

Abstract

Electronic-media use is associated with sleep disruptions in childhood and adolescence, although research relies primarily on subjective sleep. Effortful control, a dimension of self-regulation, may mitigate this association by helping children disengage from and regulate responses to media. We examined associations between media use and multiple actigraph-measured sleep parameters at mean and day levels and tested children’s effortful control as a moderator of mean-level relations. We collected actigraph data and parents’ diary reports of children’s prebedtime television, video-game, laptop, desktop, cell-phone, and tablet use in 547 twin children (7–9 years old; 51.74% female). Mean-level media use was associated with bedtime and sleep duration. For the proportion of nights on which twins used media, but not the average number of media types, effortful control attenuated associations between media use and reduced sleep duration and efficiency. Day-level media use was related only to bedtime. Findings replicate and extend existing research and highlight self-regulation as a potential protective factor.

Keywords: sleep, electronic media, temperament, childhood development

Electronic-media use is associated with poor sleep in children and adolescents, including reduced duration, later onset, and disordered sleep (Hale & Guan, 2015). This is concerning, given associations of sleep with physical and mental health and academic performance (Astill, Van der Heijden, Van IJzendoorn, & Van Someren, 2012; Matthews & Pantesco, 2016) as well as widespread media use among children and adolescents (Chassiakos, Radesky, Christakis, Moreno, & Cross, 2016). However, with recent exceptions, research has relied on parent- or self-reported sleep rather than objective indicators (Hale & Guan, 2015). The few objective studies of children and adolescents differ in sleep indicators studied (e.g., efficiency, duration, bedtime), measurement of media (e.g., screens in the bedroom, daily screen time), and actigraph methodology (placement, nights collected), and they often found no association between media use and sleep duration (e.g., Harrex et al., 2018). With the exception of one study in adolescents (Tavernier, Heissel, Sladek, Grant, & Adam, 2017), objective research has also focused on person-level analyses averaged across the study week. Finally, there has been little consideration of moderating factors such as temperament that may attenuate associations between media use and sleep. Thus, the goal of this study was first to replicate person-level associations between media use and objectively measured sleep in a large community sample in middle childhood. Second, we extended these associations to a multilevel day-to-day framework predicting nightly sleep from prebedtime media use. Finally, we examined children’s effortful control as a moderating factor that may influence associations between media use and sleep.

Sleep and Media Use

Electronic-media use is theorized to affect sleep through three main complementary mechanisms: displacement of sleep by media use, heightened cognitive and physiological arousal near bedtime, and circadian rhythm disruptions due to short-wavelength light exposure (LeBourgeois et al., 2017). Considerable research, including research with large nationally representative samples, has found associations between children and adolescents’ media use and parent- or self-reported sleep quantity, quality, and timing (Carter, Rees, Hale, Bhattacharjee, & Paradkar, 2016; Hale & Guan, 2015). Television and total screen time are most commonly studied, but findings have been extended to computer, video-game, cell-phone, and tablet use. Across all media, findings are most consistent for sleep duration and bedtime delay, whereas sleep efficiency and latency are more rarely studied and often unassociated with media use (Hale & Guan, 2015).

Objective studies of sleep and media are rare. Experimental studies of young adults have supported an effect of prebedtime exposure to light-emitting devices on melatonin suppression, later sleep onset, and subjective evening alertness but are limited by small samples (< 20 individuals; e.g., Chang, Aesbach, Duffy, & Czeisler, 2015; Chinoy, Duffy, & Czeisler, 2018). Of the four studies considering media and actigraph-measured sleep in children, three are in large samples but subject to methodological limitations, including reliance on 1 night of actigraphy (Nixon et al., 2008), use of waist-worn actigraphs (Chaput et al., 2014), and dichotomization of screen time (Harrex et al., 2018). Several studies found no associations between weekly or daily screen time and sleep (e.g., Harrex et al., 2018; Nixon et al., 2008), whereas two other studies found associations between daily screen time and efficiency (but not duration; Chaput et al., 2014) and later bedtimes and shorter sleep duration (Muller, Signal, Elder, & Gander, 2017). Different measurement and outcomes complicate interpretation, but the finding in two studies (Chaput et al., 2014; Muller et al., 2017) that screens in children’s bedrooms were related to objective sleep when average daily media use was not suggests that media use is a better predictor near bedtime than across the day, consistent with associations between subjective sleep and prebedtime media use (Hale et al., 2018; Owens et al., 1999).

In adolescents, nighttime cell-phone use and disruption (Tashjian, Mullins, & Galaván, 2019) and playing video games before bed (Harbard, Allen, Trinder, & Bei, 2016) are associated with lower sleep duration, although several other media types (e.g., Web browsing) showed no unique relation to objective sleep (Fobian, Avis, & Schwebel, 2016; Harbard et al., 2016). Finally, one study of 71 adolescents (Tavernier et al., 2017) examined person- and day-level associations between multiple media types and actigraph-assessed sleep, albeit over only 3 nights. Neither video games nor television were associated with any sleep parameter, but working on the computer was associated with lower mean sleep duration, day-level sleep duration, and day-level sleep efficiency. This examination across levels of analysis is valuable because it suggests that perturbations to daily routine (e.g., more computer time than usual) matter for sleep duration, even accounting for consistent individual differences in media use.

Sleep and Effortful Control

Associations between media use and sleep may differ across individuals because of characteristics of both the environment and the person (e.g., risk taking; Smith, Gradisar, King, & Short, 2017). Mitigating factors likely include those that help children down-regulate physiological and cognitive arousal and maintain a consistent sleep schedule despite the presence of appealing activities. One such factor is effortful control, a component of temperament indexing multiple dimensions of self-regulation, including inhibitory, activational, and attentional control (Rothbart & Bates, 2006). Regulation of attention and goal-driven behavior are closely linked to sleep (Dahl, 1996), and children low in effortful control may have greater difficulty regulating emotional and physiological responses to media as well as disengaging with media at bedtime, making effortful control a plausible moderator of the relation between media use and sleep.

The Current Study

We aimed to replicate and extend research on children’s media use and sleep by examining multiple indicators of actigraph-assessed sleep (duration, efficiency, latency, bedtime) at both person and day-to-day levels and by testing children’s effortful control as a moderator of person-level relations between media use and sleep. We expected that media use in the hour before bed would be associated with poorer sleep across all indicators, at both person and day levels, but also that media use would interact with effortful control such that the relation between sleep and media use would be strongest for children low in effortful control.

Method

Participants

Participants were drawn from the Arizona Twin Project, an ongoing longitudinal study of twin children’s health and development (Lemery-Chalfant, Clifford, McDonald, & O’Brien, 2013). The current sample included 547 twins (273 pairs; 28.57% monozygotic, 39.56% same-sex dizygotic, 31.14% opposite-sex dizygotic, 0.73% unknown zygosity) who participated in the objective sleep and daily-diary portion of the study at 8 years of age. The sample was recruited with the aim of examining genetic and environmental influences on child development in a representative community sample. Twins in the current sample were 51.74% female and ranged from 6.97 to 9.91 years of age (M = 8.45 years, SD = 0.61). They were primarily non-Latino European American (58.53%) and Hispanic/Latino (23.48%), with the remaining twins falling into the categories of Asian American (3.30%), African American (3.70%), Native American (2.75%), Native Hawaiian or Pacific Islander (1.10%), and other or multiethnic (7.14%). Annual family income ranged from less than $20,000 to more than $150,000 (median = $80,001–$90,000). A majority of primary caregivers had a graduate or professional degree (23.73%), at least 2 years of graduate education (2.64%), or a 4-year college degree (37.48%), with the remainder having either some college education (25.99%), a high school degree (9.41%), or less than a high school degree (0.75%). This pattern was similar for secondary caregivers: 19.22% had a graduate or professional degree, 3.02% had at least 2 years of graduate education, 35.42% had a 4-year college degree, 25.49% had some college education, 15.55% had a high school degree, and 1.30% had less than a high school degree.

For families who participated only at 12 months and those who participated in the sleep and diary assessment at 8 years, there were no significant differences in ethnicity, t(278) = 1.50, p = .13, or sex, t(279) = 0.53, p = .59, but families who participated only at 12 months were lower on socioeconomic status, t(256) = −3.21, p = .001.

Thirteen individuals were excluded from the final sample of 547 twins prior to all analyses because of physical or cognitive disabilities (e.g., Down syndrome, autism). No other exclusions were made on the basis of extreme scores or missing data.

Procedure

All participating families were offered the opportunity to participate in an intensive study of sleep and health, including two home visits approximately 1 week apart, a study week between the two home visits, and online or paper interviews with primary caregivers, secondary caregivers, and teachers. Approval was obtained from the Arizona State University Institutional review board, and families were compensated for all components of the study.

During the first home visit, two trained project staff obtained informed consent from parents and verbal assent from twins, and they trained primary caregivers on study-week procedures. During the study week, the twins wore actigraph watches on their nondominant wrists for 7 consecutive days (removed only for bathing and swimming), and primary caregivers completed an assessment table recording twins’ bedtimes and wake times to validate actigraph data. In addition, primary caregivers completed daily diaries each evening (55.66% online, 40.63% paper, 3.71% both) asking about the twins’ sleep, activities, mood, and nutrition (85.19% of the sample had at least seven diaries, 8.59% had six diaries, 2.19% had five diaries, and 4.03% had fewer than five diaries). Project staff contacted primary caregivers daily to answer questions and ensure that study protocol was being followed. At the second home visit, study materials were collected, and primary caregivers completed an in-home paper survey asking about twins’ environment, behavior, and temperament, including effortful control. Finally, primary caregivers provided demographic information in an online or paper survey.

Measures

Media use

On each evening of the study week, primary caregivers completed online or paper diaries, which included questions on whether each of their twins participated in a list of 20 activities in the hour before bed on the previous night (0 = no, 1 = yes). This list included four items tapping electronic-media use: “Used a desktop/laptop computer,” “Watched TV,” “Played video games,” and “Used a phone or tablet for games/internet.” At the day level, media use was defined in two ways: first, a binary variable indicating whether or not twins used any of the four types of media (0 = no, 1 = yes) on the previous night, and second, a sum of all media types used on the previous night (possible range = 0–4). At the person level, two variables were computed from these daily scores: first, the proportion of study nights when twins used any of the four types of media (possible range = 0–1), and second, the mean number of media types used per night across the study week (possible range = 0–4).

Objective sleep

Objective sleep data were collected via actigraphy using the Motionlogger Micro Watch (Ambulatory Monitoring, Ardsley, NY), a wrist-based accelerometer worn on the nondominant wrist for 7 consecutive days. Activity was measured in 1-min epochs using a zero-crossing mode, and periods of sleep and waking were detected using the Sadeh algorithm in Action W-2 (Version 2.7.1; Ambulatory Monitoring), in conjunction with cross-validation by primary-caregiver report on the sleep-assessment table. Parents recorded the time their children went to bed, and this parent report of bedtime was used in conjunction with actigraph-detected ambient light in the room and physical activity to determine the time when children first attempted sleep and first fell asleep. Actigraph data collected over at least 5 nights and validated by sleep diaries offer a nonintrusive, valid, and reliable objective measure of sleep within children’s naturalistic environments (Acebo et al., 1999).

In light of recent recommendations to include multiple objective sleep parameters in a study (Gregory & Sadeh, 2012), we used the following parameters: duration, efficiency, latency, and bedtime. Bedtime (measured on a 24-hr scale) is defined as the time when the child first attempted to sleep, established through a combination of parent report of bedtime, ambient light, and reduction in actigraph-measured physical activity. Latency is the time between bedtime and first sleep onset. Sleep duration (in hours) is defined as the total time spent asleep between the period of first sleep onset to sleep offset, excluding all bouts of waking and latency prior to first onset. Sleep efficiency is the ratio of time spent asleep to total time in bed, with total time in bed including true sleep, bouts of waking, and latency prior to first sleep onset.

Day-level sleep parameters were winsorized to 3 standard deviations outside the mean prior to analysis. Person-level sleep parameters (mean duration, efficiency, latency, and bedtime across all study nights) were first computed from nonwinsorized day-level data and then winsorized to 3 standard deviations outside the mean. At both the day level and the person level, the natural log of latency (taken after winsorizing) was used rather than the raw score because of expected deviations from normality (day level: skewness = 5.06, kurtosis = 41.75; person level: skewness = 2.79, kurtosis = 10.01).

There were full missing data for 30 twins because of mechanical failure, a lost or submerged watch, or refusal to wear the watch but continuing to participate in other study procedures. Of the 481 twins with at least partial actigraph data, 418 (86.90%) had at least 7 nights of data, 46 (9.56%) had 6 nights, 5 (1.04%) had 5 nights, 7 had 4 nights (1.46%), and 5 (1.04%) had 3 nights. Excluding twins with fewer than 5 nights of actigraph data from the analyses did not change the results.

Effortful control

Effortful control was measured using primary-caregiver report on the Temperament in Middle Childhood Questionnaire (TMCQ; Simonds, Kieras, Rueda, & Rothbart, 2007; Simonds & Rothbart, 2004), intended for children 7 to 10 years of age. The full TMCQ consists of 157 items assessing children’s temperament across 17 scales, measured on a Likert-type scale ranging from 1 (almost always untrue of your child) to 5 (almost always true of your child), within the past 6 months. The TMCQ contains 3 subscales intended to tap facets of children’s effortful control: the 15-item Activational Control scale (e.g., “Can make him/herself do homework, even when s/he wants to play”; Cronbach’s α = .75), the 7-item Attentional Focusing scale (e.g., “Is easily distracted when listening to a story”; Cronbach’s α = .90), and the 8-item Inhibitory Control scale (e.g., “Has an easy time waiting to open a present”; Cronbach’s α = .68). All 3 scales were moderately correlated (r = .40–.56), and a mean effortful-control composite was formed (Cronbach’s α = .87).

Covariates

All person-level and day-level analyses included all of the following covariates: age of twins (in years), sex (1 = female), race (1 = non-Latino European American), socioeconomic status (a standardized mean composite of primary-caregiver education, secondary-caregiver education, and family income-to-needs ratio based on 2016 standards; for more detail, see Doane et al., 2019), and vacation status (0 = study week took place during the school year, 1 = study week took place during holiday or summer vacation). All day-level, but not person-level, analyses included weekend versus school night (0 = Sunday night through Thursday night, 1 = Friday or Saturday night) as a covariate.

Data-analysis plan

Two- and three-level modeling in Mplus (Version 7.0; Muthén & Muthén, 2012) was used to examine relations among media use, effortful control, and sleep parameters, with separate models tested for each sleep parameter. In a sample with daily assessments nested within twins nested within families, three possible sources of variance exist: differences between days within individual (Level 1), differences between individuals within family (Level 2), and differences between families (Level 3). Although “pure” Level 1 variance can be explained only by Level 1 variables (i.e., you cannot predict differences between days from a variable that is constant across days), variables measured at Level 1 may contain a mixture of Level 1, Level 2, and Level 3 variance. For instance, a twin’s sleep on any given day is a combination of daily fluctuations, an individual’s tendency to sleep more or less on average, and a family’s tendency for both twins to get more or less sleep on average than children in other families (e.g., because of differences in bedtime or ambient noise levels). Multilevel modeling in Mplus accounts for the nested structure of twin and daily-diary data by adjusting standard errors and the chi-square test of model fit to correct for nonindependence of data within cluster and by modeling variances, covariances, and means at each level separately. Missing data were handled with maximum likelihood estimation using the maximum likelihood robust (MLR) estimator for two-level models and the maximum likelihood with standard order approximation using the first-order derivative (MLF) estimator for three-level models. Effect size is reported as a pseudo R2, which quantifies the reduction in unexplained variance at each level for the intercept (unexplained between-family and between-twin differences in sleep) and, when applicable, random slopes (e.g., unexplained differences in the strength of the relation between effortful control and sleep). Effect sizes for main effects are not reported when an interaction is significant, because they cannot be calculated without also dropping the interaction from the model.

Prior to conducting the multilevel analyses, we fitted unconditional models to examine the proportion of variance across days (Level 1), individuals (Level 2), and families (Level 3). We next examined whether the effect of predictors on sleep parameters differed across twin and family levels (i.e., examined contextual effects) by testing whether the regression coefficients predicting sleep could be equated across group-mean-centered individual scores and family-level cluster means without significant loss of fit. Finally, we tested random slopes examining how much the association between predictors and sleep parameters differed across families. Details of these preliminary analyses are reported in the Supplemental Material available online. Unconditional models revealed little between-twin variance for media use (family-level intraclass correlations ≥ .97, indicating that, at most, 3% of the variance in media use was between twins). Thus, two-level models included the family-level cluster mean of media use as the focal predictor, and three-level models included day-level media use and family-level cluster means. Contextual effects of family-level effortful control were nonsignificant for all sleep parameters and trimmed from all models. Because we expected absolute levels of effortful control to be more meaningful in relation to children’s sleep than level of effortful control relative to one’s cotwin, we grand-mean-centered effortful control (for further discussion of centering decisions, see the Supplemental Material).

Person-level analyses

We used two-level analyses to examine main effects and interactions between average media use across the study week and person-level effortful control in relation to average sleep across the study week, while accounting for the nested structure of twins (Level 1) within families (Level 2). Two final sets of two-level models were estimated: one using the proportion score for media use and one using the weekly sum score (both grand-mean centered). In each case, random slope variance for the effect of effortful control on duration and efficiency was estimated, but random slope variance for latency and bedtime was nonsignificant and trimmed from the model. Cross-level interactions were tested for sleep parameters that had random slope variance; that is, we tested the extent to which the effect of effortful control on each sleep parameter differed as a function of media use.

Finally, although twin data are available, we chose not to fit quantitative genetic models as part of this study. We primarily focused on phenotypic associations between media use and sleep, taking advantage of the fact that our sample is one of only a few large samples in which objective sleep and diary data were extensively measured during this age range. In addition, the lack of twin-level variability in media use did not allow us to meaningfully answer questions about the heritability of media use or the extent to which it shares genetic and environmental variance with sleep.

Day-level analyses

Three-level analyses were used to examine the relation between day-level media use in the hour before bed and that night’s sleep (Level 1 associations) while accounting for nights nested within twin (Level 2) and twins nested within family (Level 3). Actigraphy and diary assessments were started on the same day, but diaries asked primary caregivers about the previous night’s activities before bed. Thus, diaries were lagged so that each night’s diary was matched with the previous night’s sleep, meaning that individuals with complete diary and actigraph data across the study week had 6 nights of data. Two sets of models were tested, the first examining the effect of any media use in the hour before bed (0 = no media use, 1 = at least one type of media) and the second examining the sum of different types of media in the hour before bed. Day-level media use was group-mean centered, and grand-mean-centered family-level cluster means were included at Level 3.

Correction for multiple testing

We corrected for multiple testing using Benjamini and Hochberg’s (1995) procedure for controlling the false-discovery rate (FDR). Rather than controlling the family-wise error rate (i.e., the probability of falsely rejecting even one null hypothesis), the FDR controls for the expected proportion of incorrectly rejected null hypotheses among a set of tests. The FDR is a sequential procedure, calculated in the following manner. First, for a series of m multiple comparisons, all observed p values are ordered from smallest to largest, and each is assigned a corresponding rank, i, from 1 to m. Second, for each ranked p value, a critical threshold is calculated using the equation pi ≤ (i/m)Q, where Q is the prespecified FDR (e.g., .05, comparable with α). Finally, the largest p value is found for which pi ≤ (i/m)Q, and the null hypothesis is rejected for this and all smaller p values. We used an FDR criterion (Q) of .05 and corrected for 32 multiple tests (8 day-level main effects, 16 week-level main effects, 8 possible interactions).

Results

Preliminary analyses

Frequency of each type of media use across the study week is reported in Table 1, and descriptive statistics and zero-order correlations for twin- and family-level study variables are presented in Table 2. Children in our sample slept an average of 8.10 hours (SD = 0.72). The mean proportion of study nights when children used at least one type of media was .71 (SD = .30), and children used an average of 0.97 types of media per night, with the most common type being television. Overall, 194 families (70.9% of the sample) were seen during the school year, whereas 80 families (29.1%) were seen during summer vacation or holidays. After sleep parameters were winsorized to 3 standard deviations above or below the mean, no variables except latency exceeded recommended cutoffs for skewness (±2.00) or kurtosis (±7.00; Muthén & Kaplan, 1985). Zero-order correlations are described in more detail in the Supplemental Material.

Table 1.

Frequency of Media Use Across the Study Week

Medium 0 nights (n) 1–2 nights (n) 3–4 nights (n) 5–6 nights (n) 7 nights (n) Number of nights
M SD
Desktop/laptop 430 (79.78%) 75 (13.91%) 26 (4.82%) 5 (0.93%) 3 (0.56%) 0.43 1.09
Video games 365 (67.71%) 119 (22.08%) 33 (6.12%) 19 (3.53%) 2 (0.37%) 0.71 1.37
Cell phone/tablet 264 (48.98%) 158 (29.31%) 61 (11.32%) 43 (7.98%) 13 (2.41%) 1.41 1.90
Television 49 (9.09%) 108 (20.04%) 124 (23.01%) 169 (31.35%) 89 (16.51%) 4.01 2.26
Any nontelevision media 180 (33.40%) 173 (32.10%) 107 (19.85%) 54 (10.02%) 25 (4.63%) 2.04 2.10
Any type of media 27 (5.01%) 63 (11.69%) 119 (22.08%) 183 (33.95%) 147 (27.27%) 4.77 2.09
Open in a new tab

Note: N = 539 individual twins with diary data.

Table 2.

Zero-Order Correlations and Descriptive Statistics for Twin- and Family-Level Variables

Variable 1 2 3 4 5 6 7 8 9 10 11 12
1. Sleep duration —
2. Sleep efficiency .65* —
3. Sleep latency −.22* −.13* —
4. Bedtime −.34* .04 −.02 —
5. Proportion of nights using media −.24* −.06 .09 .31* —
6. Mean number of media types −.27* −.04 .07 .36* .79* —
7. Effortful control .16* .12* .02 −.06 −.18* −.26* —
8. Age −.24* −.02 .00 .30* .18* .20* .06 —
9. Sex .16* .14* −.01 .01 −.03 −.08 .18* .01 —
10. Race .15* −.01 −.01 −.25* −.17* −.22* .02 −.16* −.03 —
11. Socioeconomic status .18* .13* −.10* −.15* −.28* −.28* .27* −.13* .04 .23* —
12. Vacation −.05 −.06 .04 .25* .15* .07 .01 −.04 −.03 −.05 .04 —
 N 481 481 481 481 539 539 541 545 547 547 531 547
 M 8.10 90.04 20.68 21.43 .71 0.97 3.29 8.45 .52 .58 −0.02 .29
 SD 1.09 7.77 22.16 1.10 .45 0.82 0.55 0.61 .50 .49 0.80 .45
 Minimum 4.73 65.21 0.00 18.37 0 0 1.80 6.97 0 0 −1.69 0
 Maximum 11.48 100.00 110.47 24.85 1 4 4.58 9.91 1 1 3.29 1
 Skewness −0.51 −1.16 2.27 0.64 −0.91 0.80 −0.12 −0.10 −0.07 −0.34 0.55 0.92
 Kurtosis 0.38 1.04 5.39 0.41 −1.17 0.79 −0.56 −0.07 −2.00 −1.89 0.51 −1.15
Open in a new tab

Note: Correlations are reported for person-level variables, but descriptive statistics are reported for variables at the smallest level of analysis (e.g., day level for sleep and media). N reflects the number of individuals with complete data, not the number of observations per individual. Bedtime is reported on a 24-hr scale; a mean of 21.45 corresponds to 9:26 p.m., and values greater than 24 correspond to times after midnight. Race is a binary variable coded 0 for all races and ethnicities other than non-Latino European American and 1 for non-Latino European American. Vacation status is a binary variable coded 0 for study weeks that took place during the school year and 1 for study weeks that took place during holidays or summer vacations. Socioeconomic status is a standardized mean composite of mother’s education, father’s education, and family income-to-needs ratio (2016 standards). Correlations were calculated in Mplus (Muthén & Muthén, 2012) using “Type = Complex” to account for twin dependence.

*

p < .05.

Average media use, effortful control, and sleep

Main effects and interactions of family-level media use and person-level effortful control were examined in two-level models (see Tables 3 and 4). Figure 1 depicts graphs and regions of significance for significant interactions, calculated using Preacher, Curran, and Bauer’s (2006) online utility for probing multilevel interactions. The majority of findings remained significant following FDR correction, and these are indicated in boldface in Table 3, with the largest p value that satisfied the pi ≤ (i/32) × .05 threshold corresponding to the main effect of average number of media types on sleep duration (p = .008).

Table 5.

Fixed Effects for Day-Level Analyses

Effect Any media use Sum of media use
b SE p 95% CI b SE p 95% CI
Sleep duration on daily media use 0.02 0.05 .746 [−0.08, 0.11] −0.04 0.03 .210 [−0.09, 0.02]
Sleep efficiency on daily media use −0.13 0.35 .707 [−0.82, 0.56] −0.15 0.21 .466 [−0.56, 0.26]
Sleep latency on daily media use −0.01 0.04 .78 [−0.09, 0.06] −0.04 0.02 .116 [−0.09, 0.01]
Bedtime on daily media use 0.08 0.03 .005 [0.02, 0.14] 0.10 0.02 .000 [0.07, 0.13]
Open in a new tab

Note: Parameter estimates for covariates are not listed. Boldface indicates that the finding remained significant following false-discovery-rate correction. CI = confidence interval.

Table 3.

Fixed Effects for Person-Level Analyses

Predictor and outcome Proportion of daily media use Mean number of media types
b SE p 95% CI b SE p 95% CI
Sleep duration
 Effortful control 0.14 0.06 .02 [0.02, 0.26] 0.14 0.06 .029 [0.01, .026]
 Family mean media use −0.39 0.13 .003 [−0.65, −0.13] −0.19 0.07 .008 [−0.33, −0.05]
 Family Mean Media Use × Effortful Control 0.53 0.20 .007 [0.14, 0.92]
Sleep efficiency
 Effortful control 0.70 0.52 .172 [−0.31, 1.72] 0.82 0.54 .13 [−0.23, 1.87]
 Family mean media use −0.64 1.10 .559 [−2.79, 1.51] 0.43 0.58 .46 [−0.71, 1.57]
 Family Mean Media Use × Effortful Control 5.28 1.73 .002 [1.89, 8.67] 1.72 0.81 .033 [0.14, 3.31]
Sleep latency
 Effortful control 0.04 0.05 .446 [−0.06, 0.14] 0.04 0.05 .390 [−0.06, 0.14]
 Family mean media use 0.18 0.13 .162 [−0.07, 0.43] 0.09 0.07 .176 [−0.04, 0.22]
Bedtime
 Effortful control −0.05 0.03 .13 [−0.11, 0.01] −0.05 0.03 .127 [−0.11, 0.01]
 Family mean media use 0.57 0.17 .001 [0.24, 0.90] 0.36 0.09 .000 [0.20, 0.53]
Open in a new tab

Note: Parameter estimates for covariates are not listed. In the person-level models for sleep duration and efficiency, the coefficient for effortful control represents the average slope coefficient, with random slope variance σ2u1 (see Table 4). For the mean number of media types, the interaction predicting duration was not significant, and thus, parameter estimates for the final main effect model are reported. Boldface indicates findings that remained significant following false-discovery-rate correction. CI = confidence interval.

Fig. 1.

Fig. 1.

Open in a new tab

Graphs and regions of significance for the interaction between media use and effortful control on (a) sleep duration and (b) sleep efficiency. In the top row, media use is the family-level cluster mean because of low variability in media use between twins. Media use is centered, so 0 on the x-axis represents the mean (M = .71, SD = .30). Thin, dashed lines represent the regions of significance for the relation between media use and each sleep parameter, conditional on effortful control (EC). The bottom row depicts the slope of the relation between media use and sleep for each interaction, with 95% confidence bands around the estimate, across values of EC. Where confidence bands include 0 (marked by the horizontal line), slopes are nonsignificant. Minimum and maximum values of EC on the x-axis correspond to minimum and maximum observed values in our sample (centered).

Table 4.

Variance Parameters for Person-Level Models

Variable and statistic Sleep duration Sleep efficiency Sleep latency Bedtime
Proportion of daily media use
 Residual Level 2 intercept variance, σ2u0 0.204 14.285 0.279 0.527
 Intercept-slope covariance, σ2u0u1 −0.055 −3.646
 Residual slope variance (effortful control), σ2u1 0.039 4.311
 Residual Level 1 intercept variance, σ2ε 0.212 12.559 0.130 0.035
Mean number of media types
 Residual Level 2 intercept variance, σ2u0 0.205 14.435 0.279 0.512
 Intercept-slope covariance, σ2u0u1 −0.06 −3.685
 Residual slope variance (effortful control), σ2u1 0.043 5.419
 Residual Level 1 intercept variance, σ2ε 0.214 12.424 0.130 0.035
Open in a new tab

A higher proportion of days with any media use was associated with lower mean sleep duration, b = −0.39, SE = 0.13, z = −2.96, p = .003, 95% confidence interval (CI) = [−0.65, −0.13], and with later bedtime, b = 0.57, SE = 0.17, z = 3.36, p = .001, 95% CI = [0.24, 0.90]; intercept (Level 2): pseudo R2 = .04. In other words, children who used media every night were predicted to sleep an average of 23.40 min (39% of an hour) less per night and go to bed 34.20 min later (57% of an hour) than children who never used media. There was no relation between proportion of nights that twins used media and sleep efficiency or latency.

However, findings for sleep duration and efficiency are qualified by significant cross-level interactions between proportion of media use and effortful control. For duration, the relation between media use and sleep was stronger when children were low in effortful control, b = 0.53, SE = 0.20, z = 2.68, p = .007, 95% CI = [0.14, 0.92]; slope: pseudo R2 = .11, intercept (Level 2): pseudo R2 = .02. Examining regions of significance showed that the regression of duration on media use became significant (p ≤ .05) at values of mean-centered effortful control lower than or equal to 0.25 (0.45 SD above the mean), with 64.72% of the sample falling within this region. The upper bound to the region of significance fell outside the range of the scale. In the bottom row of Figure 1, we present simple slopes for the relation between proportion of media use and sleep duration at values of effortful control 1 standard deviation below the mean (b = −0.68, SE = 0.19, z = −3.67, p < .001), at the mean (b = −0.39, SE = 0.13, z = −2.94, p = .003), and 1 standard deviation above the mean (b = −0.10, SE = 0.16, z = 0.65, p = .513).

Proportion of media use and effortful control also interacted in relation to sleep efficiency, b = 5.28, SE = 1.73, z = 3.05, p = .002, 95% CI = [1.89, 8.13]; slope: pseudo R2 = .24, intercept (Level 2): pseudo R2 = .02. Examining regions of significance revealed that higher media use became significantly related (p ≤ .05) to lower sleep efficiency at values of effortful control lower than or equal to −0.37 (0.67 SD below the mean), with 25.78% of the sample falling within this region. At the same time, higher media use was significantly related to higher sleep efficiency at values of effortful control greater than or equal to 0.70 (1.27 SD above the mean), with 11.33% of the sample falling within this region. In Figure 1, we also present simple slopes at 1 standard deviation below the mean (b = −3.50, SE = 1.51, z = −2.32, p = .020), at the mean (b = −0.62, SE = 1.10, z = 0.56, p = .572), and 1 standard deviation above the mean (b = 2.265, SE = 1.38, z = 1.64, p = .102) of effortful control. Because the upper bound to the region of significance falls above 1 standard deviation from the mean, a higher proportion of media use is predicted to be significantly related to higher sleep efficiency for a small but nonnegligible portion of our sample with high effortful control, but the simple slope calculated at 1 standard deviation above the mean is not statistically significant.

For the models examining media use as the average number of media types, main-effect findings were similar, with media use related to lower sleep duration, b = −0.19, SE = 0.07, z = −2.67, p = .008, 95% CI = [−0.33, −0.05]; intercept (Level 2): pseudo R2 = .05, and later bedtime, b = 0.36, SE = 0.09, z = 4.26, p < .001, 95% CI = [0.20, 0.53]; intercept (Level 2): pseudo R2 = .07. Specifically, for every additional media type that children used, on average, they were predicted to sleep 11.40 min less and go to bed 21.60 min later. There were no significant main effects of media on sleep latency or efficiency.

The interaction between media and effortful control on duration was not significant. For efficiency, there was a cross-level interaction between average number of media types and effortful control; specifically, higher mean number of media types was related to higher sleep efficiency. However, this interaction was weak (reducing Level 2 intercept variance by only 1% and not reducing random slope variance), and examining regions of significance showed that media use was related to efficiency for only the 2.74% of the sample with values of effortful control greater than or equal to 0.97. Further, this interaction was no longer significant after correction for multiple testing and is not interpreted further.

Finally, person-level effortful control was related to higher sleep duration (but no other sleep indicator) in both the proportion-score and sum-score models, but this relation was no longer significant after correction for multiple testing.

Day-to-day relations between media and sleep

Three-level models were used to examine the day-to-day effect of media use in the hour before bed and later sleep, with fixed effects reported in Table 5 (findings that remain significant following FDR correction are indicated in boldface). When examining media use as a binary variable (0 = no media use, 1 = any media use), we found no significant relation between day-level media use and sleep duration, efficiency, or latency (ps > .71). However, media use in the hour before bed predicted later bedtimes in both the model examining whether twins used any media, b = 0.08, SE = 0.03, z = 2.81, p = .005, 95% CI = [0.02, 0.14]; intercept (Level 1): pseudo R2 = .19, and the model examining the daily sum score, b = 0.10, SE = 0.02, z = 6.25, p < .001, 95% CI = [0.07, 0.13]; intercept (Level 1): pseudo R2 = .20.

Sensitivity analyses

Sensitivity analyses were conducted to examine consistency of findings across key analytical decisions. All significant findings held regardless of whether analyses included or excluded age, socioeconomic status, race, sex, vacation status, weekend, and bedtime resistance (primary-caregiver report on the Child Sleep Habits Questionnaire; Owens, Spirito, & McGuinn, 2000) as covariates, with two exceptions. Specifically, effortful control was no longer significantly related to sleep duration when age was individually dropped from the model using the average number of media types or when bedtime resistance was included as a covariate in either model. In addition, significant findings held across the use of nonwinsorized sleep outcomes and the inclusion of 13 twins who were initially excluded because of disability. When all sleep outcomes were included in a single path model rather than tested separately, the interaction between average number of media types and effortful control in relation to sleep efficiency was no longer significant, but no other conclusions were altered. When we examined effortful control at the subscale level, interactions held for Attentional Focusing but not Inhibitory Control or Activational Control, with the exception of a significant interaction between Activational Control and proportion of media use in relation to efficiency. Finally, we tested two alternate measures of sleep timing (wake time and midpoint, i.e., time halfway between bedtime and waking). Media use was related to midpoint at the day and week levels and to day-level wake time, regardless of measurement. However, although the average number of media types was significantly related to wake time, the proportion of nights using media was not.

Summary

In summary, main effects of family-level media use on bedtime and sleep duration were consistent across measurement of media use and key analytical decisions, with day-level media use also related to later bedtime. Interactions between weekly proportion of media use and effortful control explained approximately 11% of the between-family differences in the strength of the relation between children’s effortful control and their sleep duration (slope variance) and 24% of the between-family differences in the relation between effortful control and sleep efficiency. Regions of significance indicated that media use was associated with lower sleep duration only at low or moderate levels of effortful control and with lower sleep efficiency at low levels of effortful control. Unexpectedly, media use was also associated with higher efficiency at very high levels of effortful control. However, following correction for multiple testing, interactions were found using only the proportion of media use, and there were no main effects of latency or efficiency on sleep.

Discussion

We examined relations between children’s media use in the hour before bed and multiple sleep indicators, at both mean and day levels, and tested children’s effortful control as a moderator of mean-level relations between media use and sleep. Regardless of measurement, mean media use was associated with later bedtimes and lower sleep duration. Further, for both sleep duration and efficiency, effortful control moderated associations between sleep and the proportion of nights when media was used. For duration, this association was attenuated at higher levels of effortful control. For efficiency, a greater proportion of nightly media use was associated with lower efficiency for children with low effortful control but higher efficiency for children with high effortful control. For media use measured as the average number of types per night, a similar but weaker interaction was found for efficiency, but it did not survive correction for multiple testing, and media had no main effect on efficiency. Finally, most mean-level findings did not hold at the day level, with nightly media use before bed weakly related to later bedtime but to no other indicator.

Unexpectedly, variability in media use existed almost entirely at family and day levels, with negligible between-twin differences, potentially because of both measurement sensitivity and the nature of media use in families in middle childhood (e.g., prebedtime media as social activity or bedtime routine). Associations may reflect individual processes, despite our inability to detect between-twin differences, but may also reflect family processes (e.g., media-related routines), so caution in interpretation is merited.

In subjective sleep research, bedtime and sleep duration are the indicators most consistently associated with media use (Hale & Guan, 2015), but the few studies using actigraphy mostly reported no association between media use and duration (e.g., Chaput et al., 2014). This study is the first to our knowledge to combine the strengths of a large sample, wrist-worn actigraph data across at least 7 consecutive nights, and diary measurement of prebedtime media use, potentially enabling us to detect associations that past research could not. Further, findings highlight the importance of taking theoretically relevant individual differences into account and suggest that media use may be most detrimental to children already at risk for poorer sleep because of low self-regulation. These children may struggle more with switching from enjoyable activities to bedtime and down-regulating physiological and cognitive arousal associated with media use, leading to greater difficulty in establishing and maintaining healthy sleep routines. However, nonreplication across measurement merits caution, and the consistent finding of a main effect on bedtime suggests that media use may be associated with some sleep indicators regardless of effortful control.

Researchers have proposed that media use disrupts sleep through displacement of sleep time, heightened prebedtime arousal, and light-related alterations of circadian rhythm (LeBourgeois et al., 2017). Our finding that prebedtime media use was associated with bedtime but not sleep latency provides more evidence for time displacement than circadian rhythm disruption, although directly assessing light, rather than an indirect measure such as media use, may lead to different results. In addition, our finding that the proportion of media use was associated with lower efficiency at low or moderate levels of effortful control suggests that sleep disruptions go beyond bedtime delay because of time displacement. The unexpected relation between media use and higher efficiency for children with high effortful control is more difficult to interpret, but in consideration of the main effects on bedtime, highly regulated children may be able to compensate for less time in bed through more efficient sleep. However, replication is needed, because this interaction was significant only for the proportion of daily media use.

Interpretation of these inconsistent findings is limited by our measure of media use, which did not allow us to differentiate between higher screen time and media multitasking, which has been associated with lower sleep duration in childhood (Pea et al., 2012). However, our proportion score reflects consistency in media use across the week, whereas average number of types indicates intensity. Consistent but low-level media use and less consistent but more intense use may have different consequences for sleep, with the former being problematic only for individuals lower in self-regulation.

Finally, day-level media use was unassociated with sleep duration, efficiency, or latency. One explanation is that after analyses account for within-person and family consistency in media use, daily fluctuations around individual means are less relevant, even if a causal path from media use to sleep exists. Alternately, family-level associations may reflect the opposite direction of effect, with children who sleep poorly on one night being more likely to use media the next evening (e.g., Hart et al., 2017). Day-level findings support an association between day-level media use and later bedtime, although this association was weak, perhaps because children’s bedtimes are still largely determined by parents. A consistent pattern of prebedtime media use may have greater consequences for sleep as children transition into adolescence and experience later sleep onset, early school start times, and greater control over their own schedules (Carskadon, 2011). Indeed, Tavernier and colleagues (2017) reported day-level associations between working on the computer and adolescents’ sleep duration and efficiency.

Limitations and Future Directions

Generalizability of these findings is limited by a primarily European American and Latino twin sample. A measure of evening screen-time duration would capture individual differences better than binary indicators. We did not account for contextual differences (e.g., whether media was violent; Hale et al., 2018). We did not address bidirectional relations; future research should consider transactional relations between sleep and media at the day level and across the transition into adolescence.

Conclusions

Our findings extend research on media use to objective measurement of sleep in a large sample of children and suggest that media use is most detrimental for children low in effortful control. However, although effortful control is potentially malleable, media use before bedtime is a more easily targeted environmental factor, perhaps especially in childhood, when sleep habits are being established and parents have more ability to shape children’s sleep environments and routines. Biological and social underpinnings of sleep undergo major changes during puberty and the transition to adolescence (Dahl & Lewin, 2002), and early establishment of positive sleep and media-use habits may help youths maintain adequate sleep under these increasing constraints (El-Sheikh & Buckhalt, 2015). At the same time, day-level deviations from children’s average media use across the week were unrelated to sleep quantity or quality, suggesting that occasional, moderate media use is unlikely to be cause for concern, especially for well-regulated children.

Supplemental Material

Clifford_Supplemental_Material_rev – Supplemental material for Effortful Control Moderates the Relation Between Electronic-Media Use and Objective Sleep Indicators in Childhood
Click here for additional data file. (16.2KB, docx)

Supplemental material, Clifford_Supplemental_Material_rev for Effortful Control Moderates the Relation Between Electronic-Media Use and Objective Sleep Indicators in Childhood by Sierra Clifford, Leah D. Doane, Reagan Breitenstein, Kevin J. Grimm and Kathryn Lemery-Chalfant in Psychological Science

Footnotes

ORCID iD: Sierra Clifford Inline graphic https://orcid.org/0000-0003-0658-4986

Supplemental Material: Additional supporting information can be found at http://journals.sagepub.com/doi/suppl/10.1177/0956797620919432

Transparency

Action Editor: Brent W. Roberts

Editor: D. Stephen Lindsay

Author Contributions

S. Clifford developed the hypotheses, analyzed the data, and drafted the majority of the manuscript. L. D. Doane and K. Lemery-Chalfant designed and implemented the larger study from which data used in the current article were drawn. In addition, L. D. Doane and K. Lemery-Chalfant assisted in drafting the manuscript and provided critical revisions and statistical advice and assistance. K. J. Grimm provided statistical advice and assistance. R. Breitenstein oversaw collection, cleaning, and management of objective sleep data. All the authors approved the final manuscript for submission.

Declaration of Conflicting Interests: The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Funding: This research was supported by National Institute of Child Health and Human Development Grant No. R01HD079520.

Open Practices: Data and materials for this study have not been made publicly available, and the design and analysis plans were not preregistered.

References

  1. Acebo C., Sadeh A., Seifer R., Tzischinsky O., Wolfson A. R., Hafer A., Carskadon M. A. (1999). Estimating sleep patterns with activity monitoring in children and adolescents: How many nights are necessary for reliable measures? Sleep, 22, 95–103. doi: 10.1093/sleep/22.1.95 [DOI] [PubMed] [Google Scholar]
  2. Astill R. G., Van der Heijden K. B., Van IJzendoorn M. H., Van Someren E. J. (2012). Sleep, cognition, and behavioral problems in school-age children: A century of research meta-analyzed. Psychological Bulletin, 138, 1109–1138. doi: 10.1037/a0028204 [DOI] [PubMed] [Google Scholar]
  3. Benjamini Y., Hochberg Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society B: Methodological, 57, 289–300. doi: 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
  4. Carskadon M. A. (2011). Sleep in adolescents: The perfect storm. Pediatric Clinics, 58, 637–647. doi: 10.1016/j.pcl.2011.03.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carter B., Rees P., Hale L., Bhattacharjee D., Paradkar M. (2016). Association between portable screen-based media device access or use and sleep outcomes: A systematic review and meta-analysis. JAMA Pediatrics, 170, 1202–1208. doi: 10.1001/jamapediatrics.2016.2341 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chang A., Aesbach D., Duffy J. F., Czeisler C. A. (2015). Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. Proceedings of the National Academy of Sciences, USA, 112, 1232–1237. doi: 10.1073/pnas.1418490112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chaput J., Leduc G., Boyer C., Bélanger P., LeBlanc A. G., Borghese M. M., Tremblay M. S. (2014). Electronic screens in children’s bedrooms and adiposity, physical activity and sleep: Do the number and type of electronic devices matter? Canadian Journal of Public Health, 105(4), e273–e279. doi: 10.17269/cjph.105.4511 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chassiakos Y., Radesky J., Christakis D., Moreno M. A., Cross C. (2016). Children and adolescents and digital media. Pediatrics, 138(5), Article e20162593. doi: 10.1542/peds.2016-2593 [DOI] [PubMed] [Google Scholar]
  9. Chinoy E. D., Duffy J. F., Czeisler C. A. (2018). Unrestricted evening use of light-emitting tablet computers delays self-selected bedtime and disrupts circadian timing and alertness. Physiological Reports, 6(10), Article e13692. doi: 10.14814/phy2.13692 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dahl R. E. (1996). The regulation of sleep and arousal: Development and psychopathology. Development and Psychopathology, 8(1), 3–27. doi: 10.1017/S0954579400006945 [DOI] [Google Scholar]
  11. Dahl R. E., Lewin D. S. (2002). Pathways to adolescent health: Sleep regulation and behavior. Journal of Adolescent Health, 31, 175–184. doi: 10.1016/S1054-139X(02)00506-2 [DOI] [PubMed] [Google Scholar]
  12. Doane L. D., Breitenstein R. S., Beekman C., Clifford C., Smith T. J., Lemery-Chalfant K. (2019). Early life socioeconomic disparities in children’s sleep: The mediating role of the current home environment. Journal of Youth and Adolescence, 48(1), 56–70. doi: 10.1007/s10964-018-0917-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. El-Sheikh M., Buckhalt J. A. (2015). Moving sleep and child development research forward: Priorities and recommendations from the SRCD-sponsored forum on sleep and child development. Monographs of the Society for Research in Child Development, 80(1), 15–32. doi: 10.1111/mono.12142 [DOI] [PubMed] [Google Scholar]
  14. Fobian A. D., Avis K., Schwebel D. C. (2016). The impact of media use on adolescent sleep efficiency. Journal of Developmental and Behavioral Pediatrics, 37(1), 9–14. doi: 10.1097/DPB.0000000000000239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gregory A. M., Sadeh A. (2012). Sleep, emotional and behavioral difficulties in children and adolescents. Sleep Medicine Reviews, 16, 129–136. doi: 10.1016/j.smrv.2011.03.007 [DOI] [PubMed] [Google Scholar]
  16. Hale L., Guan S. (2015). Screen time and sleep among school-aged children and adolescents: A systematic literature review. Sleep Medicine Reviews, 21, 50–58. doi: 10.1016/j.smrv.2014.07.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hale L., Kirschen G. W., LeBourgeois M. K., Gradisar M., Garrison M. M., Montgomery-Downs H., . . . Buxton O. M. (2018). Youth screen media habits and sleep: Sleep-friendly screen behavior recommendations for clinicians, educators, and parents. Child and Adolescent Psychiatric Clinics of North America, 27, 229–245. doi: 10.1016/j.chc.2017.11.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Harbard E., Allen N. B., Trinder J., Bei B. (2016). What’s keeping teenagers up? Prebedtime behaviors and actigraphy-assessed sleep over school and vacation. Journal of Adolescent Health, 58, 426–432. doi: 10.1016/j.jadohealth.2015.12.011 [DOI] [PubMed] [Google Scholar]
  19. Harrex H. A., Skeaff S. A., Black K. E., Davison B. K., Haszard J. J., Meredith-Jones K., . . . Skidmore P. M. (2018). Sleep timing is associated with diet and physical activity levels in 9–11-year-old children from Dunedin, New Zealand: The PEDALS study. Journal of Sleep Research, 27(4), Article e12634. doi: 10.1111/jsr.12634 [DOI] [PubMed] [Google Scholar]
  20. Hart C. N., Hawley N., Davey A., Carskadon M., Raynor H., Jelalian E., . . . Wing R. R. (2017). Effect of experimental change in children’s sleep duration on television viewing and physical activity. Pediatric Obesity, 12, 462–467. doi: 10.1111/ijpo.12166 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. LeBourgeois M. K., Hale L., Chang A. M., Akacem L. D., Montgomery-Downs H. E., Buxton O. M. (2017). Digital media and sleep in childhood and adolescence. Pediatrics, 140(Suppl. 2), S92–S96. doi: 10.1542/peds.2016-1758J [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lemery-Chalfant K., Clifford C., McDonald K., O’Brien T. C. (2013). Arizona Twin Project: A focus on early resilience. Twin Research and Human Genetics, 16(1), 404–411. doi: 10.1017/thg.2012.107 [DOI] [PubMed] [Google Scholar]
  23. Matthews K. A., Pantesco E. J. (2016). Sleep characteristics and cardiovascular risk in children and adolescents: An enumerative review. Sleep Medicine, 18, 36–49. doi: 10.1016/j.sleep.2015.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Muller D., Signal L., Elder D., Gander P. (2017). Environmental and behavioural factors associated with school children’s sleep in Aotearoa/New Zealand. Journal of Pediatrics and Child Health, 53, 68–74. doi: 10.1111/jpc.13268 [DOI] [PubMed] [Google Scholar]
  25. Muthén B., Kaplan D. (1985). A comparison of some methodologies for the factor analysis of non-normal Likert variables. British Journal of Mathematical and Statistical Psychology, 38, 171–189. doi: 10.1111/j.2044-8317.1985.tb00832.x [DOI] [Google Scholar]
  26. Muthén L. K., Muthén B. O. (2012). Mplus user’s guide: Statistical analysis with latent variables (7th ed.). Los Angeles, CA: Author. [Google Scholar]
  27. Nixon G. M., Thompson J. M., Han D. Y., Becroft D. M., Clark P. M., Robinson E., . . . Mitchell E. A. (2008). Short sleep duration in middle childhood: Risk factors and consequences. Sleep, 31(1), 71–78. doi: 10.1093/sleep/31.1.71 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Owens J., Maxim R., McGuinn M., Nobile C., Msall M., Alario A. (1999). Television-viewing habits and sleep disturbance in school children. Pediatrics, 104(3), Article e27. doi: 10.1542/peds.104.3.e27 [DOI] [PubMed] [Google Scholar]
  29. Owens J. A., Spirito A., McGuinn M. (2000). The Children’s Sleep Habits Questionnaire (CSHQ): Psychometric properties of a survey instrument for school-aged children. Sleep, 23, 1043–1052. doi: 10.1093/sleep/23.8.1d [DOI] [PubMed] [Google Scholar]
  30. Pea R., Nass C., Meheula L., Rance M., Kumar A., Bamford H., . . . Zhou M. (2012). Media use, face-to-face communication, media multitasking, and social well-being among 8- to 12-year-old girls. Developmental Psychology, 48, 327–336. doi: 10.1037/a0027030 [DOI] [PubMed] [Google Scholar]
  31. Preacher K. J., Curran P. J., Bauer D. J. (2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437–448. doi: 10.3102/10769986031004437 [DOI] [Google Scholar]
  32. Rothbart M. K., Bates J. E. (2006). Temperament. In Damon W., Eisenberg N. (Eds.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (pp. 105–176). New York, NY: John Wiley. doi: 10.1002/9780470147658.chpsy0303 [DOI] [Google Scholar]
  33. Simonds J., Kieras J. E., Rueda M. R., Rothbart M. K. (2007). Effortful control, executive attention, and emotional regulation in 7–10-year-old children. Cognitive Development, 22, 474–488. doi: 10.1016/j.cogdev.2007.08.009 [DOI] [Google Scholar]
  34. Simonds J., Rothbart M. K. (2004, October). The Temperament in Middle Childhood Questionnaire (TMCQ): A computerized self-report instrument for ages 7–10. Poster session presented at the Occasional Temperament Conference, Athens, GA. [Google Scholar]
  35. Smith L. J., Gradisar M., King D. L., Short M. (2017). Intrinsic and extrinsic predictors of video-gaming behavior and adolescent bedtimes: The relationship between flow states, self-perceived risk-taking, device accessibility, parental regulation of media and bedtime. Sleep Medicine, 30, 64–70. doi: 10.1016/j.sleep.2016.01.009 [DOI] [PubMed] [Google Scholar]
  36. Tashjian S. M., Mullins J. L., Galaván A. (2019). Bedtime autonomy and cellphone use influence sleep duration in adolescents. Journal of Adolescent Health, 64, 124–130. doi: 10.1016/j.jadohealth.2018.07.018 [DOI] [PubMed] [Google Scholar]
  37. Tavernier R., Heissel J. A., Sladek M. R., Grant K. E., Adam E. K. (2017). Adolescents’ technology and face-to-face time use predict objective sleep outcomes. Sleep Health, 3, 276–283. doi: 10.1016/j.sleh.2017.04.005 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Clifford_Supplemental_Material_rev – Supplemental material for Effortful Control Moderates the Relation Between Electronic-Media Use and Objective Sleep Indicators in Childhood
Click here for additional data file. (16.2KB, docx)

Supplemental material, Clifford_Supplemental_Material_rev for Effortful Control Moderates the Relation Between Electronic-Media Use and Objective Sleep Indicators in Childhood by Sierra Clifford, Leah D. Doane, Reagan Breitenstein, Kevin J. Grimm and Kathryn Lemery-Chalfant in Psychological Science

Articles from Psychological Science are provided here courtesy of SAGE Publications

RESOURCES