Closer friends have more similar actigraphy-estimated sleep patterns

Venetia Jing Tong Kok; Clin K Y Lai; Annadata V Rukmini; Hana Yabuki; Joshua J Gooley

doi:10.1093/sleepadvances/zpaf071

. 2025 Oct 16;6(4):zpaf071. doi: 10.1093/sleepadvances/zpaf071

Closer friends have more similar actigraphy-estimated sleep patterns

Venetia Jing Tong Kok ¹, Clin K Y Lai ², Annadata V Rukmini ³, Hana Yabuki ⁴, Joshua J Gooley ^5,^✉

PMCID: PMC12640202 PMID: 41278219

Abstract

Study Objectives

Individuals in close relationships often show similarities in health lifestyle behaviors. Here, we tested associations between friendship closeness and sleep patterns in university students. We hypothesized that daily actigraphy-estimated sleep timing and duration would be more similar in closer friendships with evidence of positive covariation.

Methods

Friend pairs (n = 300; 150 pairs) wore actigraphy watches and completed daily sleep diaries for 2 weeks during the school semester. Friendships were classified as close (78 pairs) or casual (72 pairs) based on rankings completed by each friend. Daily friend-pair differences in actigraphy-estimated sleep timing (onset, offset, midpoint) and duration were calculated separately for non-school nights and school nights. Linear mixed models were used to test associations of friendship closeness (close vs. casual) with daily friend-pair differences in sleep, and to test for daily covariation of sleep in friend pairs, adjusting for covariates.

Results

On non-school nights, friend-pair differences in sleep timing were about 30 min smaller in close friends compared with casual friends. On school nights, friend-pair differences in sleep timing were associated with dyadic differences in chronotype and class start times, but not friendship closeness. Sleep timing and duration covaried positively in close friends but not casual friends, irrespective of whether students had next-day classes.

Conclusions

Actigraphy-estimated sleep was more similar in university students who were close friends. Daily covariation of sleep in close friendships suggests that students influenced each other’s sleep patterns. Close friendships, chronotype, and class start times play an important role in shaping social sleep behavior.

Statement of Significance

Close friendships play a key role in providing university students with emotional support to cope with stress and academic demands. However, these benefits may be undermined when friends who depend on one another share poor sleep patterns. We show that actigraphy-estimated sleep timing is more similar in university students who are close friends, and their daily sleep covaries in the same direction irrespective of whether they have next-day classes. These results suggest that students’ daily sleep timing and duration may be contingent upon the behavior of their close friends. Interventions to improve sleep health in university students may be more effective if they target close friend dyads or groups rather than individual students.

Keywords: friends, university students, dyads, sleep timing, social, school start time

Introduction

Sleep is influenced by our social environment and also impacts the quality of interpersonal relationships [1]. Efforts to improve sleep health should therefore consider the social context in which sleep problems arise. Sleep difficulties often emerge or intensify during the transition from adolescence to adulthood [2] and may contribute to the onset of mental health problems [3–5]. Close friendships play a key role in providing emotional support to manage stress [6]; however, some of these benefits may be undermined when friends who depend on one another share poor sleep patterns. Sleep timing and duration tend to be more similar among individuals who are socially connected due to the tendency to form relationships with others with similar characteristics (i.e. homophilic selection) and effects of peer influence [7, 8]. This has important implications for whether sleep health interventions in emerging adults should target close friend dyads or groups, rather than individuals, to achieve shared behavior change.

Emerging adults who enter university encounter marked changes in their social environment that may influence their sleep. University students have greater autonomy compared with high school students, yet they typically have fewer family and financial responsibilities compared with full-time working adults. With fewer structured social demands, university students may choose to go to bed later and/or wake up later compared with high school [9, 10]. Given that university students are often living apart from their family and high school friends during this period, they must build new networks of social support, primarily through friendships with peers at university. These friendships play an important role in the psychological adjustment of university students to cope with emotional and academic demands [11]. However, qualitative studies indicate that university students may forgo sleep in favor of socializing with friends, often due to fear of missing out on late-night fun [12, 13]. Consequently, sleep timing and sleep duration in university students may depend on how their social time budget is managed with friends.

The degree to which daily sleep timing is similar and/or covaries in friends is poorly understood. The closeness of the relationship may be an important factor given that closer friends spend more time with one another [14]. However, studies on friendship social networks have not tested whether stronger social ties (i.e. closer friendships) predict more similar sleep behavior. Prior studies of friends have also relied on self-reported measures of sleep quality, sleep duration, and sleep timing [7, 15, 16], rather than assessing daily objective sleep parameters using actigraphy. There are also conflicting findings regarding whether individuals in close relationships have more similar chronotypes [17–19]. Sleep behavior may be more similar among friends when their social activities and sleep are not constrained by early class start times [20]; however, past studies have not tested whether social influences on sleep may differ according to whether students have classes on the next day. The present study in university students was performed to address these knowledge gaps by measuring daily actigraphy-estimated sleep parameters in friend pairs in whom data were collected for friendship closeness, chronotype, and class schedules.

The goal of our study was to test whether actigraphy-estimated sleep timing and duration are more similar in close friends compared with casual friends. The research was motivated by the expectation that sleep characteristics are more likely to be shared across close social ties due to homophilic selection and peer influence. This is because closer friendships are characterized by deeper emotional bonds and more frequent interactions compared with casual and less intimate relationships [14, 21]. We studied daily sleep parameters simultaneously in friend pairs using actigraphy and then sorted the data by non-school nights and school nights. Linear mixed models were used to test associations of friendship closeness with friend-pair differences in sleep timing and duration and to test for daily covariation of these sleep parameters in friends. On non-school nights, we hypothesized that friend-pair differences in sleep timing and duration would be smaller for close friends compared with casual friends, i.e. sleep parameters would be more similar in close friendships. We also hypothesized that sleep timing on non-school nights would positively covary in close friends, such that students would go to bed later and obtain shorter sleep on nights when their close friends showed the same behavior. On school nights, we hypothesized that associations between friendship closeness and actigraphy-estimated sleep would be weaker because sleep timing and sleep duration would be predominantly influenced by students’ class start times.

Materials and Methods

Ethics statement

University students provided written informed consent to participate in the research. The study protocol was approved by the National University of Singapore (NUS) Institutional Review Board (NUS-IRB-2020-75). The research adhered to ethical guidelines for human research in the 2024 Revision of the Declaration of Helsinki.

Participant recruitment

Undergraduate students aged 18 years and older were recruited as friend pairs from NUS, where most Bachelor’s degree programs span 4 years. Students were recruited by advertising on university digital platforms, placing flyers on campus billboards, and setting up in-person recruitment booths at approved campus locations. There were no pre-study exclusionary criteria. Among 188 friend pairs (n = 376) who consented to participate in the research, 13 friend pairs were discontinued (n = 26) because at least one individual decided to withdraw from the study or was withdrawn by the researchers due to non-compliance. Another 25 friend pairs (n = 50) completed the study, but their data were excluded from analyses because at least one individual in the friend pair had missing actigraphy data due to technical problems and/or removals of the actigraphy watch (<12 nights of valid data). The final dataset used for analyses included 300 students (150 friend pairs) with 3809 nocturnal sleep recordings and daily diaries.

Pre-study questionnaires

Students in each friend pair completed online questionnaires (Qualtrics, Seattle, WA, USA) independently to collect information on their demographics, friendship characteristics, and chronotype. Demographic information included age (in years), biological sex (male, female), self-identified ethnicity (Chinese, Malay, Indian, Mixed/others), class year of enrollment at the university (1st, 2nd, 3rd, 4th, 5th), and location of residence (on campus, off campus).

Participants were asked to provide information on the duration and closeness of their friendship, including when they first met their friend (before or after enrolling at the university), the length of the friendship (<1 year, 1 to <3 years, 3 to <5 years, >5 years), and whether they were enrolled in the same degree program (yes, no). Students independently ranked the relative closeness of their friendship by responding to the question, “Out of all of your friends, how would you rank this friend?” Response options included “Not my top 10,” “One of my top 10 friends,” “One of my top 5 friends,” “Second best friend,” or “Best friend.” Participants could select only one response.

The degree of voluntary interdependence in the friendship was evaluated using 5 items from the Acquaintances Description Form-Revised (ADF-F2) [22, 23]. Voluntary interdependence is a measure of relationship strength that refers to whether an individual commits time to interacting with the other person apart from factors that are external to the relationship. The five voluntary interdependence items on the ADF-F2 describe scenarios related to initiating contact when not hearing from a friend for several days, arranging one’s schedule to have free time together, changing plans to spend time together, initiating contact to plan leisure activities together, and doing activities that might not be particularly interesting just to spend time with the friend. Participants indicated the likelihood that they would initiate contact or re-arrange their activities using a 7-point scale with options ranging from “Definitely not” (0) to “Definitely; no doubt about it” (6). Participants also indicated whether they would arrange leisure time or do activities that may be uninteresting using a 7-point scale with options ranging from “Never” (0) to “Always, invariably, without exception” (6). Scores on individual items were summed to derive a global measure of friendship strength ranging from 0 to 30, with higher scores indicating greater voluntary interdependence.

Chronotype was assessed using the Morningness-Eveningness Questionnaire (MEQ) [24]. The MEQ is a 19-item instrument that measures individual preferences for the timing of sleep and waking activities. The MEQ comprises Likert-type items and time-scale items that are scored and summed to derive a total MEQ score ranging from 16 to 86. Higher MEQ scores indicate more of a morningness preference, whereas lower scores indicate more of an eveningness preference.

Monitoring of sleep–wake patterns

Sleep patterns for a given friend pair were monitored during the same 2-week period using actigraphy watches (ActiGraph wGT3X-BT; ActiGraph, Pensacola, FL, USA). Data were collected during the school term ranging from semester 2 of the 2020–2021 academic year to semester 2 of the 2021–2022 academic year. Students’ sleep patterns were assessed only during periods when they were taking regularly scheduled classes. Data were not collected during reading week (a 1-week mid-semester break) or during major examination periods (midterm exams or final exams). Students were instructed to wear the actigraphy device on their non-dominant wrist at all times, except during activities that could damage the watch. The actigraphy watch used an accelerometer to record wrist movements and a photo-sensor to detect light at a sampling frequency of 30 Hz. Students completed daily sleep diaries after waking up and before going to bed using Qualtrics online surveys (see details below), and they sent the time-stamped text messages “sleep” when going to bed and “wake” upon awakening (Telegram app, Dubai, United Arab Emirates).

Daily sleep diaries

Students were instructed to complete a wake-up time diary about an hour after waking up for the day. Bedtime was recorded with the question “What time did you go to bed last night? (i.e. the time you tried to fall asleep)”. Wake-up time was recorded with the question “What time did you wake up today? Your wake-up time is the last time that you woke up without going back to sleep.” Responses were entered in clock time format (hh:mm am/pm). Students were also instructed to complete a bedtime diary between 9:00 pm and the time that they went to sleep. Students indicated whether they had any classes earlier that day by responding to a yes/no question. If they said yes, they were asked to indicate the start time of their first class of the day (hh:mm am/pm format) and whether they attended the class.

Friendship classification

Friendships were classified as either casual or close for each friend pair based on their independent rankings of the relative closeness of their friendship. Friend pairs were defined as close in instances where both individuals ranked the other friend as “One of my top 5 friends,” “Second best friend,” or “Best friend” (78 pairs). By comparison, friend pairs were defined as casual if either individual ranked the other friend as “Not my top 10” or “One of my top 10 friends” (72 pairs). This friendship classification scheme was based on theoretical and practical considerations. From a theoretical standpoint, our method of defining close friend pairs aligned with prior work where the innermost layer of a personal social network typically involves about five individuals [25]. This “support clique” layer represents the closest relationships that account for the greatest proportion of a person’s social time budget and emotional closeness compared with other layers of the social network. Our requirement that both friends in a close friend pair had to rate the other as within their top 5 friends is also consistent with social frameworks that emphasize reciprocity in relationship closeness [14]. From a practical standpoint, classifying friend pairs into two groups (close and casual) ensured that there was a sufficient number of friend pairs in each group to conduct meaningful statistical comparisons of actigraphy-estimated sleep parameters.

Determination of actigraphy-estimated sleep parameters

Actigraphy data were analyzed in 60-s epochs using ActiLife software (version 6.13.4; ActiGraph, Pensacola, FL, USA). Data were plotted as actograms with nocturnal time-in-bed intervals marked by the researchers using students’ daily diaries and time-stamped telegram text messages. Epochs within these intervals were scored as sleep or wakefulness using the Sadeh algorithm for sleep–wake detection [26]. These data were used to determine the daily actigraphy-estimated sleep onset and sleep offset (hereafter referred to as “sleep onset” and “sleep offset,” respectively) for each nocturnal sleep episode. The daily sleep midpoint was defined as the halfway point between sleep onset and sleep offset, while the daily sleep duration was defined as the time interval from sleep onset to sleep offset.

The primary dyadic outcome variables in our study were the friend-pair differences in sleep onset, sleep offset, sleep midpoint, and sleep duration. Separate analyses were performed for non-school nights (no classes the next day) and school nights (at least one class on the next day) where friend-pair differences were calculated using either aggregated data or daily sleep data. Analyses of aggregated data were performed to test the overall degree of similarity in sleep timing and duration within friend pairs. In these analyses, the median sleep parameters (onset, offset, midpoint, duration) for each student were calculated over the 2-week actigraphy recording, and then the absolute differences were determined for each friend pair. We chose to use the median as a measure of central tendency for each student because it is less susceptible to outliers or skewed data compared with the mean, especially when there are few data points. Analyses of daily data assessed the absolute differences for sleep variables that were calculated on a nightly basis for a given friend pair. In the daily friend-pair analyses, non-school nights were defined as nights where both students did not have any classes the next day, whereas school nights were defined as nights where at least one student in the friend pair had classes the next day.

Statistical comparisons

Data analyses and visualizations were performed using R (version 4.4.1), Python (version 3.8.8), and GraphPad Prism 10 (version 10.0.3; GraphPad Software, San Diego, CA). R and Python packages used in each analysis are indicated in the following sections. The threshold for statistical significance was p < .05 in all analyses.

Friend-pair characteristics were compared between close and casual friendship groups by performing chi-square tests (GraphPad Prism 10). Categorical dependent variables included friend-pair differences in age (<1 year, 1–2 years, >2 years), biological sex (female–female, male–male, female–male), self-identified ethnicity (same, different), university class year (same, different), whether students resided on campus (yes–yes, no–no, yes–no), degree program of enrollment (same, different), the difference in chronotype score on the MEQ (0–5, 6–10, 11–15, >15), when they first met each other (before entering university, after entering university), and the length of the relationship (<1 year, 1 to <3 years, 3 to <5 years, >5 years). Friend-pair differences in voluntary interdependence scores on the ADF-F2 were compared between close and casual friendship groups by performing a Student’s unpaired t-test and calculating the standardized effect size (Cohen’s d with 95% CIs for unpaired samples).

Estimation statistics were used to compare friend-pair differences in median actigraphy-estimated sleep (onset, offset, midpoint, duration) between casual and close friendship groups. Comparisons were performed separately for non-school nights (casual, 72 pairs; close, 78 pairs) and school nights (casual, 69 pairs; close, 78 pairs). There were missing data for three casual friend pairs on school nights because some students reported having no scheduled classes during the 2-week actigraphy recording. Simple effect sizes (mean difference in hours) and standardized effect sizes (Cohen’s d) with 95% CIs were measured by bootstrap resampling with 5000 samples. Comparisons of friend-pair differences between close and casual friendship groups were also performed using a two-sided permutation t-test. Analyses and visualizations were performed using the “dabest” package (version 2023.2.14) in Python [27].

Models for predicting daily friend-pair differences in actigraphy-estimated sleep parameters

Linear mixed models were used to test the associations of friendship closeness with daily friend-pair differences in sleep onset, sleep offset, sleep midpoint, and sleep duration, adjusting for friend-pair characteristics (non-school nights: 149 friend pairs, 734 paired nights; school nights: 144 friend pairs, 703 paired nights). Data were included in the models for non-school nights only if both students in the friend pair did not have any classes scheduled on the next day. One friend pair was excluded from analyses for both non-school nights and school nights because one student did not complete the MEQ (listwise deletion was used for covariates). Data were included in the models for school nights if at least one student in the friend pair had a class on the next day, irrespective of whether he/she attended the class. Five friend pairs were excluded from analyses of school nights because one student in the friend pair did not have scheduled classes during the 2-week assessment (n = 3) or had missing sleep data on school nights (n = 2).

In all linear mixed models for predicting daily dyadic differences in actigraphy-estimated sleep, closeness of the friendship (close, casual) was included as a fixed-effect factor, and friend pair was included as a random-effect factor. Covariates included friend-pair differences in age (<1 year, 1 to 2 years, >2 years), biological sex (female–male, male–male, female–female), ethnicity (same, different), on-campus residence (yes–yes, yes–no, no–no), class year (same, different), degree program (same, different), and chronotype (difference in MEQ score: 0–5, 6–10, 11–15, >15). Statistical models for school nights also included friend-pair differences in students’ first class start time on the next day. Daily class start time data were binned into categories, where a given student could report not attending a class (no class: either a non-school day or they did not attend their first scheduled class), attending an early morning class (early am: before 10:00 am), attending a late morning class (late morning: from 10:00 am to <12:00 pm), or attending an afternoon class (pm: 12:00 pm or later). Therefore, subcategories for first class of the day in the model included all possible friend-pair combinations of no class, early am, late am, and pm (10 possible combinations). Reference categories in the statistical models were chosen based on the most common friend-pair characteristics in the sample: an age difference of 1–2 years, female–female, same ethnicity, no–no for living on campus, same university class year, different university degree program, a difference in chronotype score of 0–5, and a pm–pm first class of the day (for the school night models only). Linear mixed models were implemented using the “lme4” (version 1.1.31) and “lmerTest” (version 3.1.3) packages in R [28, 29]. Model assumptions including linearity, homogeneity of variance, collinearity, and normality of residuals were checked with the “performance” package (version 0.12.4) [30]. Tukey’s method was used for post-hoc multiple comparison tests using the “emmeans” package (version 1.10.5) [31]. Plots were generated using the “gridExtra” (version 2.3) and “ggplot2” (version 3.5.1) packages in R [32, 33].

Models for predicting daily actigraphy-estimated sleep parameters in individual students

Linear mixed models were also used to test for daily covariation of actigraphy-estimated sleep parameters in friend pairs, where data were analyzed separately for students in close friendships and casual friendships. For a given individual and sleep outcome variable (onset, offset, midpoint, duration), the corresponding sleep variable of his/her friend on the same night was included as a continuous fixed-effect factor, and individual student was included as a random-effect factor. Covariates included age in years (continuous variable), biological sex (female, male), ethnicity (Chinese, non-Chinese), on-campus residence (yes, no), class year (Y1, Y2, Y3, Y4+), and chronotype score on the MEQ (continuous variable). Statistical models for school nights also included students’ first class start time of the day (Did not attend the class, early am, late am, pm). Reference categories for categorical covariates were female, Chinese, off-campus residence, and Y1 student. The statistical models were used to estimate the degree to which students’ daily actigraphy-estimated sleep differed for a 1-h change in their friend’s sleep (i.e. the β regression coefficient with 95% CIs). Models were implemented in R with multiple comparison tests performed using Tukey’s test [28, 29, 31].

Results

Participant demographics and friendship characteristics

Our study sample of 300 university students comprised predominantly young adults (M ± SD = 21.17 years ±1.92 years; range = 18–38 years, only one participant >27 years), most of whom were women (69.7 per cent) and identified as Chinese (93.0 per cent). More than half of students lived off campus (60.7 per cent) and there was broad representation across class-year of enrollment (Y1, 32.3 per cent; Y2, 23.7 per cent; Y3, 24.0 per cent; Y4 or Y5, 20.0 per cent). Most friend pairs in our sample had an age difference of ≤2 years (<1 year, 40.7 per cent; 1–2 years, 46.7 per cent; >2 years, 12.7 per cent), were same-sex pairs (female–female, 56.0 per cent; male–male, 16.7 per cent; female–male, 27.3 per cent), and identified as the same ethnicity (92.7 per cent) (Table 1). Friend pairs were more likely to live off campus (50.7 per cent) compared with on campus (29.3 per cent) or being split between on-campus and off-campus locations (20.0 per cent), and they were usually enrolled in the same class year of university (80.0 per cent). The distribution of friend-pair characteristics for age, sex, ethnicity, location of residence, and class year did not differ statistically between casual and close friendship groups (χ² < 4.50, p > .05 for all comparisons) (Table 1). Friend-pair differences in chronotype score on the MEQ did not differ between casual and close friendship groups when treated as a categorical variable (χ²(3) = 1.46, p = .692) (Table 1), or as a continuous variable (t(297) = 0.83, p = .406; Cohen’s d = 0.10, 95 per cent CI = −0.13 to 0.32).

Table 1.

Friend-pair characteristics

Characteristic	All friends (150 pairs)	Casual friends (72 pairs)	Close friends (78 pairs)	Chi-square	P
Age difference, n (%)
<1 year	61 (40.7)	23 (31.9)	38 (48.7)	χ²(2) = 4.42	.110
1–2 years	70 (46.7)	39 (54.2)	31 (39.7)
>2 years	19 (12.7)	10 (13.9)	9 (11.5)
Sex, n (%)
Female–female	84 (56.0)	41 (56.9)	43 (55.1)	χ²(2) = 0.46	.795
Male–male	25 (16.7)	13 (18.1)	12 (15.4)
Female–male	41 (27.3)	18 (25.0)	23 (29.5)
Ethnicity, n (%)
Same	139 (92.7)	64 (88.9)	75 (96.2)	χ²(1) = 2.91	.088
Different	11 (7.3)	8 (11.1)	3 (3.8)	χ²(1) = 2.91	.088
University year, n (%)
Same	120 (80.0)	57 (79.2)	63 (80.8)	χ²(1) = 0.06	.806
Different	30 (20.0)	15 (20.8)	15 (19.2)	χ²(1) = 0.06	.806
Degree program, n (%)
Same	50 (33.3)	23 (31.9)	27 (34.6)	χ²(1) = 0.12	.729
Different	100 (66.7)	49 (68.1)	51 (65.4)	χ²(1) = 0.12	.729
Live on campus, n (%)
Yes–yes	44 (29.3)	22 (30.6)	22 (28.2)	χ²(2) = 2.86	.239
No–no	76 (50.7)	32 (44.4)	44 (56.4)
Yes–no	30 (20.0)	18 (25.0)	12 (15.4)
Difference in MEQ score, n (%)
0–5	46 (30.9)	21 (29.2)	25 (32.5)	χ²(3) = 1.46	.692
6–10	33 (22.1)	19 (26.4)	14 (18.2)
11–15	26 (17.4)	12 (16.7)	14 (18.2)
>15	44 (29.5)	20 (27.8)	24 (31.2)
Missing data	1 (0.7)	0 (0.0)	1 (1.3)

Open in a new tab

MEQ, Morningness-Eveningness Questionnaire.

Next, we compared friendship characteristics between casual and close friendship groups to confirm that they represented distinct relationship groups. Most close friends reported that they met prior to entering university (59.0 per cent) and their relationship duration was at least 3 years (56.4 per cent: <1 year, 17.3 per cent; 1 to <3 years, 26.3 per cent; 3 to <5 years, 20.5 per cent; ≥5 years, 35.9 per cent). By comparison, most casual friends reported that they first met after entering university (64.6 per cent) and their relationship duration was <3 years (70.2 per cent: <1 year, 27.8 per cent; 1 to <3 years, 42.4 per cent; 3 to <5 years, 13.2 per cent; ≥5 years, 16.7 per cent). Statistically significant differences were detected between casual and close friendship groups for whether they met before entering university (χ²(1) = 16.66, p < .001), and the duration of their relationship (χ²(3) = 22.11, p < .001). Close friends also scored significantly higher than casual friends on voluntary interdependence on the ADF-F2 (M ± SD: casual, 16.60 ± 5.42; close, 20.99 ± 5.95; t(298) = 6.66, p < .001) with a moderately large effect size (Cohen’s d = 0.77, 95% CI = 0.53–1.00).

Associations of friendship closeness with sleep patterns on non-school nights

On non-school nights, students in casual (n = 144) and close (n = 156) friendship groups went to sleep close to 2:00 am (M ± SD: casual, 2:19 am ± 1.53 h; close, 2:01 am ± 1.47 h) and woke up close to 9:30 am (M ± SD: casual, 9:46 am ± 1.72 h; close, 9:28 am ±1.52 h), resulting in an average nocturnal sleep duration of about 7.4 h (M ± SD: casual, 7.37 h ± 1.09 h; close, 7.40 h ± 0.93 h). However, friend-pair differences in median sleep timing were about 30–40 min smaller for close versus casual friendship groups for sleep onset (mean difference = −0.56 h, 95% CI = −0.98 to −0.12 h; Cohen’s d = −0.42, 95% CI = −0.76 to −0.03; p = .010) (Table 2 and Figure 1A) and sleep offset (mean difference = −0.65 h, 95% CI = −1.14 to −0.15 h; Cohen’s d = −0.43, 95% CI = −0.77 to −0.05; p = .011) (Table 2 and Figure 1B). Consequently, friend-pair differences in sleep midpoint were about 35 min smaller for close versus casual friends (mean difference = −0.60 h, 95% CI = −1.01 to −0.15 h; Cohen’s d = −0.45, 95% CI = −0.78 to −0.04; p = .006) (Table 2 and Figure 1C), but friend-pair differences in sleep duration were comparable between groups (mean difference = −0.20, 95% CI = −0.47 to 0.09 h; Cohen’s d = −0.22, 95% CI = −0.53 to 0.12; p = .173) (Table 2 and Figure 1D).

Table 2.

Friend-pair differences in median actigraphy-estimated sleep parameters

Sleep variable	Within-group friend-pair differences in sleep		Between-group comparisons of friend-pair differences (close minus casual)
	Casual friends (M ± SD)	Close friends (M ± SD)	Mean difference in hours (95% CI)	Cohen’s d (95% CI)	P
Non-school nights
Sleep onset (h)	1.75 ± 1.38	1.18 ± 1.28	−0.56 (−0.98 to −0.12)	−0.42 (−0.76 to −0.03)	.010
Sleep offset (h)	1.97 ± 1.57	1.32 ± 1.48	−0.65 (−1.14 to −0.15)	−0.43 (−0.77 to −0.05)	.011
Sleep midpoint (h)	1.78 ± 1.38	1.18 ± 1.29	−0.60 (−1.01 to −0.15)	−0.45 (−0.78 to −0.04)	.006
Sleep duration (h)	1.21 ± 0.94	1.01 ± 0.85	−0.20 (−0.47 to 0.09)	−0.22 (−0.53 to 0.12)	.173
School nights
Sleep onset (h)	1.68 ± 1.11	1.32 ± 1.27	−0.36 (−0.72 to 0.04)	−0.30 (−0.65 to 0.04)	.076
Sleep offset (h)	1.60 ± 1.38	1.24 ± 1.13	−0.36 (−0.78 to 0.02)	−0.29 (−0.61 to 0.03)	.084
Sleep midpoint (h)	1.59 ± 1.10	1.18 ± 1.12	−0.41 (−0.75 to −0.05)	−0.37 (−0.71 to −0.03)	.028
Sleep duration (h)	1.05 ± 0.92	0.94 ± 0.80	−0.11 (−0.41 to 0.14)	−0.13 (−0.46 to 0.18)	0.435

Open in a new tab

Friend-pair differences in actigraphy-estimated sleep on non-school nights. Median sleep variables on non-school nights were measured over a 2-week period in university students who were recruited as friend pairs (n = 300, 150 pairs). Friendships were classified as casual (72 pairs) or close (78 pairs) based on rankings provided by students. Friend-pair differences are shown for actigraphy-estimated (A) sleep onset, (B) sleep offset, (C) sleep midpoint, and (D) sleep duration. Slope graphs show the absolute difference for each friend pair for casual (left panels) and close (middle panels) friendships. In each plot, friend A was defined as the individual in the friend pair with earlier sleep timing (onset, offset, midpoint) or shorter sleep duration so that the absolute difference could be calculated relative to friend B. Gray lines connect sleep data for each friend pair. Simple effect sizes are shown on the right as a bootstrap sampling distribution with the mean difference in hours and 95% CIs. Box plots show the median and interquartile range, with whiskers showing the 5th and 95th percentiles. Comparisons of friend-pair differences for each sleep variable are shown for close versus casual friendships (right panels). Jittered dot plots show the distribution of friend-pair differences in casual friend pairs and close friend pairs for each sleep variable. Between-group effect sizes are shown on the right for friend-pair differences in sleep for close friends relative to casual friends. Negative values indicate that friend-pair differences were smaller in close friendships.

Linear mixed models that adjusted for friend-pair characteristics (149 friend pairs, 734 paired nights) provided additional evidence that friend-pair differences in daily sleep timing on non-school nights were about 30 min smaller in close friends versus casual friends for sleep onset (β = −0.48 h, 95% CI = −0.86 to −0.11 h, p = .016), sleep offset (β = −0.50 h, 95% CI = −0.95 to −0.05, p = .039), and sleep midpoint (β = −0.54 h, 95% CI = −0.94 to −0.15 h, p = .011) (Figure 2 A and Supplementary Table S1). Friend-pair differences in these sleep parameters were also about an hour larger in instances where friends differed strongly by chronotype versus having a similar chronotype (MEQ score difference of >15 points versus 0–5 points; Tukey’s test, p < .05 for each comparison) (Figure 2 A and Supplementary Table S1). Other friend-pair characteristics including differences in age, sex, ethnicity, living on campus, class year, and degree program were not statistically associated with sleep timing (Supplementary Table S1). Friendship closeness (close vs. casual) was not statistically associated with friend-pair differences in sleep duration on non-school nights (β = −0.20 h, 95% CI = −0.45 to 0.06 h, p = .145) (Figure 2 A and Supplementary Table S1). All covariates that were entered into the model were not statistically associated with daily friend-pair differences in sleep duration (Supplementary Table S1).

Associations of friendship closeness with actigraphy-estimated sleep on non-school nights. (A) Linear mixed models were used to test associations of friendship closeness with daily friend-pair differences in actigraphy-estimated sleep on non-school nights. Sleep patterns were measured during the same 2-week period in university students who were recruited as friend pairs. Friendship closeness (close vs. casual) was entered as a fixed-effect factor and friend pair was entered as a random-effect factor. Covariates included friend-pair differences in age, biological sex, ethnicity, location of residence, class year of enrollment, degree program, and chronotype. The estimated difference in each sleep variable (i.e. the β regression coefficient) with 95% CIs is shown for each item relative to the reference (vertical dotted line). Positive values indicate that friend-pair differences were larger, whereas negative values indicate that friend-pair differences were smaller. Forest plots show results for friendship closeness, friend-pair differences in chronotype scores on the Morningness-Eveningness Questionnaire (MEQ), and location of residence. Full results of the linear mixed models (i.e. including all covariates) are provided in Supplementary Table S1. (B) Separate linear mixed models tested whether students’ daily actigraphy-estimated sleep covaried with their friend’s sleep on non-school nights. The friend’s sleep variable was entered as a fixed-effect predictor and individual student was entered as a random-effect factor. Forest plots show the estimated difference for each sleep variable for a 1-h change in behavior of the close or casual friend (β coefficient with 95% CIs). Positive values indicate covariation of sleep in the same direction as the friend. Full results of the statistical models with covariates are shown in Supplementary Table S2 (close friends) and Supplementary Table S3 (casual friends).

Next, we tested daily covariation of actigraphy-estimated sleep in friend pairs on non-school nights. In close friendships, daily sleep timing and sleep duration covaried in the same direction (Figure 2 B and Supplementary Table S2). Students went to sleep and woke up several minutes later for each 1-h delay in their close friend’s sleep timing (sleep onset: β = 0.13 h, 95% CI = 0.07 to 0.19 h, p < .001; sleep offset, β = 0.12 h, 95% CI = 0.06 to 0.18 h, p < .001); hence, sleep midpoint also covaried in the same direction by about 10 min (sleep midpoint: β = 0.15 h, 95% CI = 0.09 to 0.21 h, p < .001). Additionally, students obtained about 5 min more sleep for each 1-h increase in sleep duration in their close friend (sleep duration: β = 0.08 h, 95% CI = 0.02 to 0.14 h, p = .014). By comparison, sleep parameters did not statistically covary between casual friends for sleep timing or sleep duration (sleep onset: β = 0.01 h, 95% CI = −0.06 to 0.07 h, p = .854; sleep offset, β = 0.05 h, 95% CI = −0.02 to 0.11 h, p = .155; sleep midpoint: β = 0.05 h, 95% CI = −0.01 to 0.11 h, p = .096; sleep duration, β = −0.005 h, 95% CI = −0.07 to 0.06 h, p = .889) (Figure 2 B and Supplementary Table S3). Students in close and casual friendships went to sleep earlier if they had an earlier chronotype (i.e. higher MEQ score), whereas other covariates were not statistically associated with sleep timing or sleep duration on non-school nights (Supplementary Table S2 and S3).

Associations of friendship closeness with sleep patterns on school nights

On school nights, students in casual (n = 141) and close (n = 156) friendship groups went to sleep at about 2:00 am (M ± SD: casual, 2:10 am ± 1.53 h; close, 1:51 am ± 1.52 h) and woke up at about 9:00 am (M ± SD: casual, 9:00 am ±1.57 h; close, 8:44 am ± 1.37 h), resulting in a little less than 7 h of nocturnal sleep on average (M ± SD: casual, 6.77 h ± 1.04 h; close, 6.86 h ± 0.99 h). Friend-pair differences in students’ median sleep timing tended to be smaller for close versus casual friendship groups, but did not reach statistical significance for sleep onset (mean difference = −0.36 h, 95% CI = −0.72 to 0.04 h; Cohen’s d = −0.30, 95% CI = −0.65 to 0.04; p = .076) (Table 2 and Figure 3 A) or sleep offset (mean difference = −0.36 h, 95% CI = −0.78 to 0.02 h; Cohen’s d = −0.29, 95% CI = −0.61 to 0.03; p = .084) (Table 2, Figure 3 B). However, friend-pair differences in sleep midpoint were about 25 min smaller for close versus casual friends and reached statistical significance in the unadjusted analysis (mean difference = −0.41 h, 95% CI = −0.75 to −0.05 h; Cohen’s d = −0.37, 95% CI = −0.71 to −0.03; p = .028) (Table 2 and Figure 3 C). Friend-pair differences in sleep duration did not differ between casual and close friendship groups (mean difference = −0.11, 95% CI = −0.41 to 0.14 h; Cohen’s d = −0.13, 95% CI = −0.46 to 0.18; p = .435) (Table 2 and Figure 3 D).

Friend-pair differences in actigraphy-estimated sleep on school nights. Median sleep variables on school nights were measured over a 2-week period in university students who were recruited as friend pairs (n = 300, 150 pairs). Friendships were classified as casual (69 pairs; 3 pairs had missing data) or close (78 pairs) based on rankings provided by students. Friend-pair differences are shown for actigraphy-estimated (A) sleep onset, (B) sleep offset, (C) sleep midpoint, and (D) sleep duration. Slope graphs show the absolute difference for each friend pair for casual (left panels) and close (middle panels) friendships. In each plot, friend A was defined as the individual in the friend pair with earlier sleep timing (onset, offset, midpoint) or shorter sleep duration so that the absolute difference could be calculated relative to friend B. Gray lines connect sleep data for each friend pair. Simple effect sizes are shown on the right as a bootstrap sampling distribution with the mean difference in hours and 95% CIs. Box plots show the median and interquartile range, with whiskers showing the 5th and 95th percentiles. Comparisons of friend-pair differences for each sleep variable are shown for close versus casual friendships (right panels). Jittered dot plots show the distribution of friend-pair differences in casual friend pairs and close friend pairs for each sleep variable. Between-group effect sizes are shown on the right for friend-pair differences in sleep for close friends relative to casual friends. Negative values indicate smaller friend-pair differences in close friend pairs.

In linear mixed models that adjusted for friend-pair characteristics (144 friend pairs, 703 paired nights), friend-pair differences in actigraphy-estimated sleep parameters tended to be smaller in close friends but did not differ statistically relative to casual friends for sleep onset (β = −0.29 h, 95% CI = −0.62 to 0.04 h, p = .102), sleep offset (β = −0.31 h, 95% CI = −0.62 to −0.01, p = .062), sleep midpoint (β = −0.30 h, 95% CI = −0.60 to −0.01 h, p = .059), and sleep duration (β = −0.14 h, 95% CI = −0.41 to 0.13 h, p = .323) (Figure 4A and Supplementary Table S4). However, friend-pair differences in sleep timing on school nights were about 45–60 min greater when friends had strongly divergent chronotype scores (MEQ score difference of >15 points versus 0–5 points; Tukey’s test, p < .05 for each comparison) (Figure 4 A and Supplementary Table S4). Friend-pair differences in sleep onset were also about 30 min greater when one student lived on campus and the other lived off campus, as compared with both students living on campus (Tukey’s test, p < .05). Friend-pair differences in sleep timing and sleep duration were strongly associated with differences in students’ first class start time of the day. When friends had similar morning class start times, they exhibited smaller differences in their sleep offset (early am: β = −1.20 h, 95% CI = −1.64 to −0.76 h, p < .001; late am: β = −0.59 h, 95% CI = −1.02 to −0.15 h, p = .009) and sleep midpoint (early am: β = −0.75 h, 95% CI = −1.13 to −0.36 h, p < .001), as compared with days when both friends had their first class of the day in the afternoon (Tukey’s test, p < .05 for all comparisons). Conversely, when friends had divergent class start times where one student had an early morning class and the other had his/her first class in the afternoon, they exhibited larger differences in sleep offset (β = 0.48 h, 95% CI = 0.13 to 0.85, p = .010) and sleep duration (β = 0.52 h, 95% CI = 0.19 to 0.86 h, p = .002), as compared with both friends having their first class in the afternoon (Tukey’s test, p < .05 for both comparisons) (Figure 4 A and Supplementary Table S4).

Associations of friendship closeness with actigraphy-estimated sleep on school nights. (A) Linear mixed models were used to test associations of friendship closeness with daily friend-pair differences in actigraphy-estimated sleep on school nights. Sleep patterns were measured during the same 2-week period in university students who were recruited as friend pairs. Friendship closeness (close vs. casual) was entered as a fixed-effect factor and friend pair was entered as a random-effect factor. Covariates included friend-pair differences in age, biological sex, ethnicity, location of residence, class year of enrollment, degree program, chronotype, and first class start time. The estimated difference in each sleep variable (i.e. the β regression coefficient) with 95% CIs is shown for each item relative to the reference (vertical dotted line). Positive values indicate that friend-pair differences were larger, whereas negative values indicate that friend-pair differences were smaller. Forest plots show results for friendship closeness, friend-pair differences in chronotype scores on the Morningness-Eveningness Questionnaire (MEQ), location of residence, and first class start time. Full results of the linear mixed models (i.e. including all covariates) are provided in Supplementary Table S4. (B) Separate linear mixed models tested whether students’ daily actigraphy-estimated sleep covaried with their friend’s sleep on school nights. The friend’s sleep variable was entered as a fixed-effect predictor and individual student was entered as a random-effect factor. Forest plots show the estimated difference for each sleep variable for a 1-h change in behavior of the close or casual friend (β coefficient with 95% CIs). Positive values indicate covariation of sleep in the same direction as the friend. Full results of the statistical models with covariates are shown in Supplementary Table S5 (close friends) and Supplementary Table S6 (casual friends).

Analyses of daily actigraphy-estimated sleep parameters on school nights showed that sleep timing and sleep duration in individual students covaried in the same direction as their close friend (Figure 4 B and Supplementary Table S5). Students’ daily sleep timing shifted later by several minutes for each 1-h delay in their close friend’s sleep behavior (sleep onset: β = 0.16 h, 95% CI = 0.10 to 0.22 h, p < .001; sleep offset, β = 0.12 h, 95% CI = 0.07 to 0.17 h, p < .001; sleep midpoint: β = 0.14 h, 95% CI = 0.08 to 0.19 h, p < .001), and their daily sleep duration increased by about 8 min for each 1-h increase in sleep duration in their close friend (sleep duration: β = 0.14 h, 95% CI = 0.08 to 0.19 h, p < .001). By comparison, sleep timing and duration did not statistically covary between casual friends (sleep onset: β = −0.01 h, 95% CI = −0.07 to 0.05 h, p = .796; sleep offset, β = 0.004 h, 95% CI = −0.05 to 0.05 h, p = .882; sleep midpoint: β = −0.001 h, 95% CI = −0.05 to 0.05 h, p = .959; sleep duration, β = 0.002 h, 95% CI = −0.05 to 0.06 h, p = .947) (Figure 4 B and Supplementary Table S6). Earlier chronotype was associated with an earlier sleep onset, earlier sleep offset, and longer sleep duration in close and casual friendship groups (Supplementary Tables S5 and S6). Early morning classes were also associated with earlier wake-up times and a decrease in sleep duration by about an hour in students in close and casual friendship groups (early am class start time: close friends, β = −1.03 h, 95% CI = −1.25 to −0.81 h; casual friends, β = −1.37 h, 95% CI = −1.60 to −1.12 h; Tukey’s test, p < .001 for comparisons with pm class start times). Other covariates including age, sex, ethnicity, location of residence, and class year were not associated with sleep timing or sleep duration for students in close or casual friendship groups (Supplementary Tables S5 and S6).

Discussion

Our study provides evidence that actigraphy-estimated sleep timing is more similar in university students who are close friends. Daily friend-pair differences in sleep timing were smaller in close friendships compared with casual friendships only when students did not have classes on the following day. Nonetheless, sleep timing and sleep duration covaried in close friends on both non-school nights and school nights. These results suggest that students in close friendships may influence each other’s sleep patterns. Our findings show that university students’ close social ties predicted their sleep patterns, adjusting for demographic characteristics, chronotype, and class start times. Interventions for improving sleep health in university students should therefore consider both individual factors and social influences on sleep–wake behavior.

Actigraphy-estimated sleep timing was more similar in closer friendships

Our study shows that relationship closeness may be an important determinant of dyadic sleep timing in friends. On non-school nights, friend-pair differences in actigraphy-estimated sleep onset, sleep offset, and sleep midpoint were about 30 min smaller in close friends compared with casual friends. These results could be explained by stronger effects of homophilic selection and/or peer influence on sleep patterns in close friendships compared with casual friendships [25, 34]. Our findings are consistent with prior studies of close dyadic relationships in which sleep timing was correlated within bedpartners and within parent-adolescent dyads [17, 35, 36]. Prior analyses of adolescent friendship social networks also showed that individuals with self-reported short sleep were more likely to have friends with self-reported short sleep [7]. In that study, longitudinal evidence was also provided that self-reported short sleep spread through peer influence within the social network, similar to other health behaviors including diet, physical activity, and substance use [37–41]. Building on previous work, our findings suggest that effects of peer influence on sleep among friends may be greatest for student’s closest relationships.

We found that friend-pair differences in chronotype score did not differ between close and casual friendships. These results suggest that morningness-eveningness preference was not a key determinant of whether students formed close friendships. Our findings are consistent with prior work in adolescents showing that the likelihood of best-friend dyads having the same chronotype (morning type, neutral type, or evening type) was no better than chance [19]. As expected, objective sleep timing in our study was associated with self-rated chronotype [42], where sleep timing was earlier in students who scored higher on morningness. Accordingly, greater differences in chronotype between friends were associated with larger differences in sleep onset, sleep offset, and sleep midpoint, irrespective of whether they had classes the next day. Nonetheless, closer friendships predicted smaller friend-pair differences in sleep timing on non-school nights, adjusting for friend-pair differences in chronotype. Together, these results suggest that effects of peer influence on sleep timing may compete with students’ chronotype preference.

Friend-pair differences in actigraphy-estimated sleep on school nights were strongly associated with students’ first class start time. Consistent with our previous work [20], university students slept about an hour less when they had early morning classes because they woke up earlier than usual. Sleep offset and sleep duration were therefore more similar within friend pairs when both friends had an early morning class on the same day. This presumably occurred because early class start times restrict the range of wake-up times compared with waking up naturally [43]. Sleep offset and sleep duration diverged markedly within friend pairs when one friend had an early morning class and the other had either an afternoon class start time or no classes. These results show that class start times imposed strong social constraints on sleep behavior that may have attenuated the association of friendship closeness with sleep timing that was otherwise observed on non-school nights.

Daily actigraphy-estimated sleep parameters covaried in close friends

We found that sleep timing and sleep duration covaried positively in close friends but not casual friends. Covariation of sleep parameters in close friendships was observed on both non-school nights and school nights, indicating that these effects were robust to individual differences in class schedules. Our findings suggest a role for peer influence, where sleep patterns of individual students were contingent upon the behavior of their close friends. These results are consistent with previous work in university students in whom sleep patterns were measured using consumer-wearable devices, and personal social networks were derived from communicative interactions (i.e. voice calls and text messages) [8]. In that study, students’ daily bedtime, rise time, and total sleep time covaried positively with the average behavior of their social contacts. The strength of the associations was comparable to our findings, where sleep timing and sleep duration covaried by several minutes for every 1-h difference in sleep timing/duration of a students’ social contacts. This degree of covariation is likely meaningful given that the timing of sleep in university students can vary by several hours over a typical week [20]. It should also be highlighted that daily covariation of sleep was calculated for all nights irrespective of whether friends spent time together. This approach likely underestimates day-specific effects of peer influence on sleep that are directly attributable to social activities. Future studies should assess the effects of peer influence on sleep behavior in instances when direct social contact between friends can be confirmed.

The strength of covariation in actigraphy-estimated sleep that we observed in close friends is consistent with findings from other close dyadic relationships involving individuals residing in the same household. A study of cohabiting couples that measured sleep by daily diaries found that sleep duration covaried positively by several minutes for every 1-h increase in their partner’s sleep duration [44]. Similarly, a study of parent–adolescent dyads found that daily self-reported sleep timing was predicted by the other family member’s sleep patterns, with positive covariation of bedtimes and wake-up times by several minutes for each 1-h change in sleep behavior of the dyadic partner [36]. Together with our findings, these studies highlight how close relationships with friends, romantic partners, and family members may influence daily sleep timing within personal social networks [7].

Limitations and considerations

Our cross-sectional study was designed to test daily covariation of actigraphy-estimated sleep in friend dyads. Our study was not designed to assess the longitudinal spread of behavior across social ties that may take place over longer time scales [7, 8, 38]. Relatedly, our study included only friend dyads rather than assessing behavior across each student’s friendship social network. We also did not collect detailed information on students’ home environment (e.g. shared bedroom; number of friends or family), daily time-use (e.g. academic work, cocurricular activities, extra-curricular activities, paid work), or their daily interactions with friends, which may have influenced covariation of actigraphy-estimated sleep. This limitation could be addressed by asking students to report in their daily diaries when their sleep was influenced by work or academic commitments, as well as social interactions and the persons with whom they were spending time.

Our definition of close friendships was based on instances where both individuals in the dyad ranked each other as within their top 5 friends. Using this classification approach, we were able to show that actigraphy-estimated sleep was more similar in close friendships compared with casual friendships. It is possible that sleep timing would be most similar in reciprocal best-friend pairs and/or in friends who spend the most time together near bedtime. Future work could verify and quantify social interactions by contact tracing, e.g. using proximity sensors to detect when students are spending time in the same physical location [45]. It is also important to test whether our findings are generalizable to student populations in other sociocultural contexts. At the university where our research was performed, most students reside off campus (typically at home), and those who stay on campus usually have individual bedrooms. Peer influence on sleep may be stronger in residential colleges and dormitories where room sharing is more common and opportunities for in-person social activities are greater.

Implications and future directions

Our findings suggest that university students may face competing social pressures on their sleep arising from their drive to socialize and their class timetable. On the one hand, university students may delay or sacrifice sleep to engage in social activities, driven by factors such as fear of missing out and increased social and technological distractions [12]. On the other hand, students’ social time is constrained by their class start times and academic workload [20, 46]. Our findings shed light on how students may navigate these competing demands. Close friends may have more similar sleep timing on non-school nights because they have greater freedom to coordinate their sleep patterns to facilitate social activities. Socializing may therefore take precedence in determining sleep timing when behavior is relatively unconstrained, while structured commitments like school schedules may restrict opportunities for close friends to align their sleep patterns.

Our findings have implications for sleep health interventions in university students. We and others have shown that close social ties are associated with similarities in sleep timing and duration [7, 8, 35, 36, 44, 47]. Therefore, efforts to improve sleep may be most effective when tailored to include a corecipient or supporter who is in a close relationship with the student to achieve shared behavior change. Dyadic interventions in friends have proven effective for other health lifestyle behaviors, including increasing physical activity and reducing sedentary behavior [48]. Notably, interventions with shared dyadic target goals in friends were associated with larger effect sizes compared with interventions that targeted individuals [48]. Additionally, dyadic and group-based interventions are effective in improving diet and weight loss outcomes [49]. In principle, similar strategies can be implemented to improve sleep in close friendships by harnessing the power of peer support and influence.

Conclusion

Social factors have a strong influence on university students’ actigraphy-estimated sleep. Sleep timing was more similar in close friends compared with casual friends when students did not have classes the next day. The association between friendship strength and sleep timing was weaker on school nights when students’ sleep was constrained by their first class of the day. Daily sleep timing and sleep duration covaried in close friends but not casual friends, indicating that friendship strength is an important determinant of whether students’ sleep is influenced by peers. Our findings highlight the role that friendships play in social sleep patterns in university students. Future work in universities should assess whether dyadic or group-based sleep health interventions are more effective when coadministered to close friends rather than to individual students.

Supplementary Material

20251006_FRAN_Supplementary_zpaf071

20251006_fran_supplementary_zpaf071.pdf^{(138.4KB, pdf)}

Acknowledgments

We thank research staff and students in the Chronobiology and Sleep Laboratory for carrying out the research.

Contributor Information

Venetia Jing Tong Kok, Neuroscience and Behavioural Disorders Programme, Duke-NUS Medical School, Singapore, Singapore.

Clin K Y Lai, Neuroscience and Behavioural Disorders Programme, Duke-NUS Medical School, Singapore, Singapore.

Annadata V Rukmini, Neuroscience and Behavioural Disorders Programme, Duke-NUS Medical School, Singapore, Singapore.

Hana Yabuki, Neuroscience and Behavioural Disorders Programme, Duke-NUS Medical School, Singapore, Singapore.

Joshua J Gooley, Neuroscience and Behavioural Disorders Programme, Duke-NUS Medical School, Singapore, Singapore.

Author contributions

Venetia Jing Tong Kok (Data curation [lead], Formal analysis [lead], Investigation [lead], Methodology [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Clin K.Y. Lai (Conceptualization, Formal analysis, Investigation, Methodology, Writing—review & editing [supporting]), A.V. Rukmini (Formal analysis, Investigation, Methodology, Project administration, Writing—review & editing [supporting]), Hana Yabuki (Formal analysis, Investigation, Methodology, Writing—review & editing [supporting]), Joshua James Gooley (Conceptualization [lead], Funding acquisition [lead], Investigation [supporting], Methodology [equal], Project administration [lead], Resources [lead], Supervision [lead], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal])

Funding

The work was supported by the Ministry of Education, Singapore (MOE2019-T2-2-074). The funder had no role in conducting the research.

Disclosure statement

Financial disclosure: The authors have no financial conflicts of interest to disclose.

Non-financial disclosure: The authors have no non-financial conflicts of interest to disclose.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

1. Gordon AM, Carrillo B, Barnes CM. Sleep and social relationships in healthy populations: a systematic review. Sleep Med Rev. 2021;57:101428. 10.1016/j.smrv.2021.101428 [DOI] [PubMed] [Google Scholar]
2. Crowley SJ, Wolfson AR, Tarokh L, Carskadon MA. An update on adolescent sleep: new evidence informing the perfect storm model. J Adolesc. 2018;67(1):55–65. 10.1016/j.adolescence.2018.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
3. Lovato N, Gradisar M. A meta-analysis and model of the relationship between sleep and depression in adolescents: recommendations for future research and clinical practice. Sleep Med Rev. 2014;18(6):521–529. 10.1016/j.smrv.2014.03.006 [DOI] [PubMed] [Google Scholar]
4. Gradisar M, Kahn M, Micic G, et al. Sleep's role in the development and resolution of adolescent depression. Nat Rev Psychol. 2022;1(9):512–523. 10.1038/s44159-022-00074-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
5. Loke YM, Lim S, Rukmini AV, Chen P, Wang JCK, Gooley JJ. Development and testing of the Sleep Health And Wellness Questionnaire (SHAWQ) in adolescents and university students: composite SHAWQ scores are associated with sleep problems, depression symptoms, and academic performance. Front Sleep. 2023;2:1188424. 10.3389/frsle.2023.1188424 [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Adams RE, Santo JB, Bukowski WM. The presence of a best friend buffers the effects of negative experiences. Dev Psychol. 2011;47(6):1786–1791. 10.1037/a0025401 [DOI] [PubMed] [Google Scholar]
7. Mednick SC, Christakis NA, Fowler JH. The spread of sleep loss influences drug use in adolescent social networks. PloS One. 2010;5(3):e9775. 10.1371/journal.pone.0009775 [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Wang C, Mattingly S, Payne J, Lizardo O, Hachen DS. The impact of social networks on sleep among a cohort of college students. SSM Population Health. 2021;16:100937. 10.1016/j.ssmph.2021.100937 [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Urner M, Tornic J, Bloch KE. Sleep patterns in high school and university students: a longitudinal study. Chronobiol Int. 2009;26(6):1222–1234. 10.3109/07420520903244600 [DOI] [PubMed] [Google Scholar]
10. Lund HG, Reider BD, Whiting AB, Prichard JR. Sleep patterns and predictors of disturbed sleep in a large population of college students. J Adolesc Health. 2010;46(2):124–132. 10.1016/j.jadohealth.2009.06.016 [DOI] [PubMed] [Google Scholar]
11. Wise RA, King AR. Family environment as a predictor of the quality of college students' friendships. J Family Issues. 2007;29(6):828–848. 10.1177/0192513X07309461 [DOI] [Google Scholar]
12. Adams SK, Williford DN, Vaccaro A, Kisler TS, Francis A, Newman B. The young and the restless: socializing trumps sleep, fear of missing out, and technological distractions in first-year college students. Int J Adolesc Youth. 2017;22(3):337–348. 10.1080/02673843.2016.1181557 [DOI] [Google Scholar]
13. Foulkes L, McMillan D, Gregory AM. A bad night's sleep on campus: an interview study of first-year university students with poor sleep quality. Sleep Health. 2019;5(3):280–287. 10.1016/j.sleh.2019.01.003 [DOI] [PubMed] [Google Scholar]
14. Rubin KH, Bukowski WM, Parker JG. Peer Interactions, Relationships, and Groups. Handbook of child psychology: Social, emotional, and Personality Development. Vol. 3. 6th ed. Hoboken, NJ, USA: John Wiley & Sons, Inc.; 2006:571–645. [Google Scholar]
15. Li X, Kawachi I, Buxton OM, Haneuse S, Onnela JP. Social network analysis of group position, popularity, and sleep behaviors among U.S. adolescents. Soc Sci Med. 2019;232:417–426. 10.1016/j.socscimed.2019.05.026 [DOI] [PubMed] [Google Scholar]
16. Palmer CA, Powell SL, Deutchman DR, Tintzman C, Poppler A, Oosterhoff B. Sleepy and secluded: sleep disturbances are associated with connectedness in early adolescent social networks. J Res Adolesc. 2022;32(2):756–768. 10.1111/jora.12670 [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Randler C, Kretz S. Assortative mating in morningness-eveningness. Int J Psychol: J int psychol. 2011;46(2):91–96. 10.1080/00207594.2010.518237 [DOI] [Google Scholar]
18. Jocz P, Stolarski M, Jankowski KS. Similarity in chronotype and preferred time for sex and its role in relationship quality and sexual satisfaction. Front Psychol. 2018;9:443. 10.3389/fpsyg.2018.00443 [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Andershed A-K. In Sync with Adolescence: The Role of Morningness-Eveningness in Development. New York, NY, USA: Springer Science + Business Media; 2005. [Google Scholar]
20. Yeo SC, Lai CKY, Tan J, et al. Early morning university classes are associated with impaired sleep and academic performance. Nat Hum Behav. 2023;7(4):502–514. 10.1038/s41562-023-01531-x [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Giletta M, Scholte RH, Burk WJ, et al. Similarity in depressive symptoms in adolescents' friendship dyads: selection or socialization? Dev Psychol. 2011;47(6):1804–1814. 10.1037/a0023872 [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Lea M. Factors underlying friendship: an analysis of responses on the acquaintance description form in relation to Wright's friendship model. J Soc Pers Relat. 1989;6(3):275–292. 10.1177/0265407589063003 [DOI] [Google Scholar]
23. Wright PH. A model and a technique for studies of friendship. J Exp Soc Psychol. 1969;5(3):295–309. 10.1016/0022-1031(69)90055-9 [DOI] [Google Scholar]
24. Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol. 1976;4(2):97–110 [PubMed] [Google Scholar]
25. Sutcliffe A, Dunbar R, Binder J, Arrow H. Relationships and the social brain: integrating psychological and evolutionary perspectives. Br J Psychol. 2012;103(2):149–168. 10.1111/j.2044-8295.2011.02061.x [DOI] [PubMed] [Google Scholar]
26. Sadeh A, Sharkey M, Carskadon MA. Activity-based sleep-wake identification: an empirical test of methodological issues. Sleep. 1994;17(3):201–207. 10.1093/sleep/17.3.201 [DOI] [PubMed] [Google Scholar]
27. Ho J, Tumkaya T, Aryal S, Choi H, Claridge-Chang A. Moving beyond P values: data analysis with estimation graphics. Nat Methods. 2019;16(7):565–566. 10.1038/s41592-019-0470-3 [DOI] [PubMed] [Google Scholar]
28. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67(1):1–48. 10.18637/jss.v067.i01 [DOI] [Google Scholar]
29. Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in linear mixed effects models. J Stat Softw. 2017;82(13):1–26. 10.18637/jss.v082.i13 [DOI] [Google Scholar]
30. Lüdecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D. performance: an R package for assessment, comparison and testing of statistical models. J Open Sour Softw. 2021;6(60):3139. 10.21105/joss.03139 [DOI] [Google Scholar]
31. Lenth RV, Piaskowski J. emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 2.0.0. 2025. https://rvlenth.github.io/emmeans/
32. Auguie B, Antonov A. gridExtra-package: Miscellaneous Functions for “Grid” Graphics. R package version 2.3. 2017. 10.32614/CRAN.package.gridExtra [DOI] [Google Scholar]
33. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag; New York; 2016, 10.1007/978-3-319-24277-4 [DOI] [Google Scholar]
34. Adams J, Lawrence EM, Goode JA, Schaefer DR, Mollborn S. Peer network processes in adolescents' health lifestyles. J Health Soc Behav. 2022;63(1):125–141. 10.1177/00221465211054394 [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Gunn HE, Buysse DJ, Hasler BP, Begley A, Troxel WM. Sleep concordance in couples is associated with relationship characteristics. Sleep. 2015;38(6):933–939. 10.5665/sleep.4744 [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Fuligni AJ, Tsai KM, Krull JL, Gonzales NA. Daily concordance between parent and adolescent sleep habits. J Adolesc Health. 2015;56(2):244–250. 10.1016/j.jadohealth.2014.09.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Crandall CS. Social contagion of binge eating. J Pers Soc Psychol. 1988;55(4):588–598. 10.1037//0022-3514.55.4.588 [DOI] [PubMed] [Google Scholar]
38. Christakis NA, Fowler JH. Social contagion theory: examining dynamic social networks and human behavior. Stat Med. 2013;32(4):556–577. 10.1002/sim.5408 [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Aral S, Nicolaides C. Exercise contagion in a global social network. Nat Commun. 2017;8(1):14753. 10.1038/ncomms14753 [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Carrell SE, Hoekstra M, West JE. Is poor fitness contagious? Evidence from randomly assigned friends. J Public Econ. 2011;95(7-8):657–663. 10.1016/j.jpubeco.2010.12.005 [DOI] [Google Scholar]
41. Ennett ST, Bauman KE, Hussong A, et al. The peer context of adolescent substance use: findings from social network analysis. J Res Adolesc. 2006;16(2):159–186. 10.1111/j.1532-7795.2006.00127.x [DOI] [Google Scholar]
42. Roveda E, Vitale J, Montaruli A, Galasso L, Carandente F, Caumo A. Predicting the actigraphy-based acrophase using the Morningness-Eveningness Questionnaire (MEQ) in college students of North Italy. Chronobiol Int. 2017;34(5):551–562. 10.1080/07420528.2016.1276928 [DOI] [PubMed] [Google Scholar]
43. Charoenthammanon RS, Gooley JJ. The adolescent circadian clock entrains to social time rather than sun time. Curr Biol. 2025;35(9):2100–2111.e5. 10.1016/j.cub.2025.03.059 [DOI] [PubMed] [Google Scholar]
44. Lee S, Martire LM, Damaske SA, et al. Covariation in couples' nightly sleep and gender differences. Sleep Health. 2018;4(2):201–208. 10.1016/j.sleh.2017.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
45. Sekara V, Lehmann S. The strength of friendship ties in proximity sensor data. PloS One. 2014;9(7):e100915. 10.1371/journal.pone.0100915 [DOI] [PMC free article] [PubMed] [Google Scholar]
46. Yeo SC, Tan J, Lo JC, Chee MWL, Gooley JJ. Associations of time spent on homework or studying with nocturnal sleep behavior and depression symptoms in adolescents from Singapore. Sleep Health. 2020;6(6):758–766. 10.1016/j.sleh.2020.04.011 [DOI] [PubMed] [Google Scholar]
47. Walters EM, Phillips AJK, Mellor A, et al. Sleep and wake are shared and transmitted between individuals with insomnia and their bed-sharing partners. Sleep. 2020;43(1). 10.1093/sleep/zsz206 [DOI] [Google Scholar]
48. Carr RM, Prestwich A, Kwasnicka D, et al. Dyadic interventions to promote physical activity and reduce sedentary behaviour: systematic review and meta-analysis. Health Psychol Rev. 2019;13(1):91–109. 10.1080/17437199.2018.1532312 [DOI] [PubMed] [Google Scholar]
49. John JC, Ho J, Raber M, Basen-Engquist K, Jacobson L, Strong LL. Dyad and group-based interventions in physical activity, diet, and weight loss: a systematic review of the evidence. J Behav Med. 2024;47(3):355–373. 10.1007/s10865-023-00457-z [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

20251006_FRAN_Supplementary_zpaf071

20251006_fran_supplementary_zpaf071.pdf^{(138.4KB, pdf)}

Data Availability Statement

The data underlying this article will be shared on reasonable request to the corresponding author.

[ref1] 1. Gordon AM, Carrillo B, Barnes CM. Sleep and social relationships in healthy populations: a systematic review. Sleep Med Rev. 2021;57:101428. 10.1016/j.smrv.2021.101428 [DOI] [PubMed] [Google Scholar]

[ref2] 2. Crowley SJ, Wolfson AR, Tarokh L, Carskadon MA. An update on adolescent sleep: new evidence informing the perfect storm model. J Adolesc. 2018;67(1):55–65. 10.1016/j.adolescence.2018.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref3] 3. Lovato N, Gradisar M. A meta-analysis and model of the relationship between sleep and depression in adolescents: recommendations for future research and clinical practice. Sleep Med Rev. 2014;18(6):521–529. 10.1016/j.smrv.2014.03.006 [DOI] [PubMed] [Google Scholar]

[ref4] 4. Gradisar M, Kahn M, Micic G, et al. Sleep's role in the development and resolution of adolescent depression. Nat Rev Psychol. 2022;1(9):512–523. 10.1038/s44159-022-00074-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref5] 5. Loke YM, Lim S, Rukmini AV, Chen P, Wang JCK, Gooley JJ. Development and testing of the Sleep Health And Wellness Questionnaire (SHAWQ) in adolescents and university students: composite SHAWQ scores are associated with sleep problems, depression symptoms, and academic performance. Front Sleep. 2023;2:1188424. 10.3389/frsle.2023.1188424 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref6] 6. Adams RE, Santo JB, Bukowski WM. The presence of a best friend buffers the effects of negative experiences. Dev Psychol. 2011;47(6):1786–1791. 10.1037/a0025401 [DOI] [PubMed] [Google Scholar]

[ref7] 7. Mednick SC, Christakis NA, Fowler JH. The spread of sleep loss influences drug use in adolescent social networks. PloS One. 2010;5(3):e9775. 10.1371/journal.pone.0009775 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref8] 8. Wang C, Mattingly S, Payne J, Lizardo O, Hachen DS. The impact of social networks on sleep among a cohort of college students. SSM Population Health. 2021;16:100937. 10.1016/j.ssmph.2021.100937 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref9] 9. Urner M, Tornic J, Bloch KE. Sleep patterns in high school and university students: a longitudinal study. Chronobiol Int. 2009;26(6):1222–1234. 10.3109/07420520903244600 [DOI] [PubMed] [Google Scholar]

[ref10] 10. Lund HG, Reider BD, Whiting AB, Prichard JR. Sleep patterns and predictors of disturbed sleep in a large population of college students. J Adolesc Health. 2010;46(2):124–132. 10.1016/j.jadohealth.2009.06.016 [DOI] [PubMed] [Google Scholar]

[ref11] 11. Wise RA, King AR. Family environment as a predictor of the quality of college students' friendships. J Family Issues. 2007;29(6):828–848. 10.1177/0192513X07309461 [DOI] [Google Scholar]

[ref12] 12. Adams SK, Williford DN, Vaccaro A, Kisler TS, Francis A, Newman B. The young and the restless: socializing trumps sleep, fear of missing out, and technological distractions in first-year college students. Int J Adolesc Youth. 2017;22(3):337–348. 10.1080/02673843.2016.1181557 [DOI] [Google Scholar]

[ref13] 13. Foulkes L, McMillan D, Gregory AM. A bad night's sleep on campus: an interview study of first-year university students with poor sleep quality. Sleep Health. 2019;5(3):280–287. 10.1016/j.sleh.2019.01.003 [DOI] [PubMed] [Google Scholar]

[ref14] 14. Rubin KH, Bukowski WM, Parker JG. Peer Interactions, Relationships, and Groups. Handbook of child psychology: Social, emotional, and Personality Development. Vol. 3. 6th ed. Hoboken, NJ, USA: John Wiley & Sons, Inc.; 2006:571–645. [Google Scholar]

[ref15] 15. Li X, Kawachi I, Buxton OM, Haneuse S, Onnela JP. Social network analysis of group position, popularity, and sleep behaviors among U.S. adolescents. Soc Sci Med. 2019;232:417–426. 10.1016/j.socscimed.2019.05.026 [DOI] [PubMed] [Google Scholar]

[ref16] 16. Palmer CA, Powell SL, Deutchman DR, Tintzman C, Poppler A, Oosterhoff B. Sleepy and secluded: sleep disturbances are associated with connectedness in early adolescent social networks. J Res Adolesc. 2022;32(2):756–768. 10.1111/jora.12670 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref17] 17. Randler C, Kretz S. Assortative mating in morningness-eveningness. Int J Psychol: J int psychol. 2011;46(2):91–96. 10.1080/00207594.2010.518237 [DOI] [Google Scholar]

[ref18] 18. Jocz P, Stolarski M, Jankowski KS. Similarity in chronotype and preferred time for sex and its role in relationship quality and sexual satisfaction. Front Psychol. 2018;9:443. 10.3389/fpsyg.2018.00443 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref19] 19. Andershed A-K. In Sync with Adolescence: The Role of Morningness-Eveningness in Development. New York, NY, USA: Springer Science + Business Media; 2005. [Google Scholar]

[ref20] 20. Yeo SC, Lai CKY, Tan J, et al. Early morning university classes are associated with impaired sleep and academic performance. Nat Hum Behav. 2023;7(4):502–514. 10.1038/s41562-023-01531-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref21] 21. Giletta M, Scholte RH, Burk WJ, et al. Similarity in depressive symptoms in adolescents' friendship dyads: selection or socialization? Dev Psychol. 2011;47(6):1804–1814. 10.1037/a0023872 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref22] 22. Lea M. Factors underlying friendship: an analysis of responses on the acquaintance description form in relation to Wright's friendship model. J Soc Pers Relat. 1989;6(3):275–292. 10.1177/0265407589063003 [DOI] [Google Scholar]

[ref23] 23. Wright PH. A model and a technique for studies of friendship. J Exp Soc Psychol. 1969;5(3):295–309. 10.1016/0022-1031(69)90055-9 [DOI] [Google Scholar]

[ref24] 24. Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol. 1976;4(2):97–110 [PubMed] [Google Scholar]

[ref25] 25. Sutcliffe A, Dunbar R, Binder J, Arrow H. Relationships and the social brain: integrating psychological and evolutionary perspectives. Br J Psychol. 2012;103(2):149–168. 10.1111/j.2044-8295.2011.02061.x [DOI] [PubMed] [Google Scholar]

[ref26] 26. Sadeh A, Sharkey M, Carskadon MA. Activity-based sleep-wake identification: an empirical test of methodological issues. Sleep. 1994;17(3):201–207. 10.1093/sleep/17.3.201 [DOI] [PubMed] [Google Scholar]

[ref27] 27. Ho J, Tumkaya T, Aryal S, Choi H, Claridge-Chang A. Moving beyond P values: data analysis with estimation graphics. Nat Methods. 2019;16(7):565–566. 10.1038/s41592-019-0470-3 [DOI] [PubMed] [Google Scholar]

[ref28] 28. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67(1):1–48. 10.18637/jss.v067.i01 [DOI] [Google Scholar]

[ref29] 29. Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in linear mixed effects models. J Stat Softw. 2017;82(13):1–26. 10.18637/jss.v082.i13 [DOI] [Google Scholar]

[ref30] 30. Lüdecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D. performance: an R package for assessment, comparison and testing of statistical models. J Open Sour Softw. 2021;6(60):3139. 10.21105/joss.03139 [DOI] [Google Scholar]

[ref31] 31. Lenth RV, Piaskowski J. emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 2.0.0. 2025. https://rvlenth.github.io/emmeans/

[ref32] 32. Auguie B, Antonov A. gridExtra-package: Miscellaneous Functions for “Grid” Graphics. R package version 2.3. 2017. 10.32614/CRAN.package.gridExtra [DOI] [Google Scholar]

[ref33] 33. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag; New York; 2016, 10.1007/978-3-319-24277-4 [DOI] [Google Scholar]

[ref34] 34. Adams J, Lawrence EM, Goode JA, Schaefer DR, Mollborn S. Peer network processes in adolescents' health lifestyles. J Health Soc Behav. 2022;63(1):125–141. 10.1177/00221465211054394 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref35] 35. Gunn HE, Buysse DJ, Hasler BP, Begley A, Troxel WM. Sleep concordance in couples is associated with relationship characteristics. Sleep. 2015;38(6):933–939. 10.5665/sleep.4744 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref36] 36. Fuligni AJ, Tsai KM, Krull JL, Gonzales NA. Daily concordance between parent and adolescent sleep habits. J Adolesc Health. 2015;56(2):244–250. 10.1016/j.jadohealth.2014.09.013 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref37] 37. Crandall CS. Social contagion of binge eating. J Pers Soc Psychol. 1988;55(4):588–598. 10.1037//0022-3514.55.4.588 [DOI] [PubMed] [Google Scholar]

[ref38] 38. Christakis NA, Fowler JH. Social contagion theory: examining dynamic social networks and human behavior. Stat Med. 2013;32(4):556–577. 10.1002/sim.5408 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref39] 39. Aral S, Nicolaides C. Exercise contagion in a global social network. Nat Commun. 2017;8(1):14753. 10.1038/ncomms14753 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref40] 40. Carrell SE, Hoekstra M, West JE. Is poor fitness contagious? Evidence from randomly assigned friends. J Public Econ. 2011;95(7-8):657–663. 10.1016/j.jpubeco.2010.12.005 [DOI] [Google Scholar]

[ref41] 41. Ennett ST, Bauman KE, Hussong A, et al. The peer context of adolescent substance use: findings from social network analysis. J Res Adolesc. 2006;16(2):159–186. 10.1111/j.1532-7795.2006.00127.x [DOI] [Google Scholar]

[ref42] 42. Roveda E, Vitale J, Montaruli A, Galasso L, Carandente F, Caumo A. Predicting the actigraphy-based acrophase using the Morningness-Eveningness Questionnaire (MEQ) in college students of North Italy. Chronobiol Int. 2017;34(5):551–562. 10.1080/07420528.2016.1276928 [DOI] [PubMed] [Google Scholar]

[ref43] 43. Charoenthammanon RS, Gooley JJ. The adolescent circadian clock entrains to social time rather than sun time. Curr Biol. 2025;35(9):2100–2111.e5. 10.1016/j.cub.2025.03.059 [DOI] [PubMed] [Google Scholar]

[ref44] 44. Lee S, Martire LM, Damaske SA, et al. Covariation in couples' nightly sleep and gender differences. Sleep Health. 2018;4(2):201–208. 10.1016/j.sleh.2017.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref45] 45. Sekara V, Lehmann S. The strength of friendship ties in proximity sensor data. PloS One. 2014;9(7):e100915. 10.1371/journal.pone.0100915 [DOI] [PMC free article] [PubMed] [Google Scholar]

[ref46] 46. Yeo SC, Tan J, Lo JC, Chee MWL, Gooley JJ. Associations of time spent on homework or studying with nocturnal sleep behavior and depression symptoms in adolescents from Singapore. Sleep Health. 2020;6(6):758–766. 10.1016/j.sleh.2020.04.011 [DOI] [PubMed] [Google Scholar]

[ref47] 47. Walters EM, Phillips AJK, Mellor A, et al. Sleep and wake are shared and transmitted between individuals with insomnia and their bed-sharing partners. Sleep. 2020;43(1). 10.1093/sleep/zsz206 [DOI] [Google Scholar]

[ref48] 48. Carr RM, Prestwich A, Kwasnicka D, et al. Dyadic interventions to promote physical activity and reduce sedentary behaviour: systematic review and meta-analysis. Health Psychol Rev. 2019;13(1):91–109. 10.1080/17437199.2018.1532312 [DOI] [PubMed] [Google Scholar]

[ref49] 49. John JC, Ho J, Raber M, Basen-Engquist K, Jacobson L, Strong LL. Dyad and group-based interventions in physical activity, diet, and weight loss: a systematic review of the evidence. J Behav Med. 2024;47(3):355–373. 10.1007/s10865-023-00457-z [DOI] [PubMed] [Google Scholar]

PERMALINK

Closer friends have more similar actigraphy-estimated sleep patterns

Venetia Jing Tong Kok

Clin K Y Lai

Annadata V Rukmini

Hana Yabuki

Joshua J Gooley

Roles

Abstract

Study Objectives

Methods

Results

Conclusions

Statement of Significance

Introduction

Materials and Methods

Ethics statement

Participant recruitment

Pre-study questionnaires

Monitoring of sleep–wake patterns

Daily sleep diaries

Friendship classification

Determination of actigraphy-estimated sleep parameters

Statistical comparisons

Models for predicting daily friend-pair differences in actigraphy-estimated sleep parameters

Models for predicting daily actigraphy-estimated sleep parameters in individual students

Results

Participant demographics and friendship characteristics

Table 1.

Associations of friendship closeness with sleep patterns on non-school nights

Table 2.

Figure 1.

Figure 2.

Associations of friendship closeness with sleep patterns on school nights

Figure 3.

Figure 4.

Discussion

Actigraphy-estimated sleep timing was more similar in closer friendships

Daily actigraphy-estimated sleep parameters covaried in close friends

Limitations and considerations

Implications and future directions

Conclusion

Supplementary Material

Acknowledgments

Contributor Information

Author contributions

Funding

Disclosure statement

Data availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases