Available online at www.sciencedirect.com

ScienceDirect Behavior Therapy 45 (2014) 67 – 82 www.elsevier.com/locate/bt

Repetitive Thought Impairs Sleep Quality: An Experience Sampling Study Keisuke Takano The University of Tokyo Shinji Sakamoto Nihon University Yoshihiko Tanno The University of Tokyo

Although previous research has suggested that presleep negative cognitive activities are associated with poor sleep quality, there is little evidence regarding the association between negative thoughts and sleep in real-life settings. The present study used experience sampling and long-term sleep monitoring with actigraphy to investigate the relationships among negative repetitive thought, mood, and sleep problems. During a 1-week sampling period, 43 undergraduate students recorded their thought content and mood eight times a day at semirandom intervals. In addition to these subjective reports, participants wore actigraphs on their wrists in order to measure sleep parameters. Analyses using multilevel modeling showed that repetitive thought in the evening was significantly associated with longer sleep-onset latency, decreased sleep efficiency, and reduced total sleep time. Furthermore, impaired sleep quality was significantly associated with reduced positive affect the next morning, and decreased positive affect was indirectly associated with increased repetitive thought in the evening.

This study was supported by grants from the Japan Society for the Promotion of Science (21-10591). We would like to express our gratitude for the assistance provided by the staff of the Self-Focus Research Project at Nihon University: Tomomi Hashimoto, Mari Inaba, Haruka Ishikura, Ryuichiro Kuki, Mayou Sohyama, and Rika Ueno. Address correspondence to Keisuke Takano, Ph.D., Department of Cognitive and Behavioral Science, Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan; e-mail: [email protected]. 0005-7894/45/76-91/$1.00/0 © 2013 Association for Behavioral and Cognitive Therapies. Published by Elsevier Ltd. All rights reserved.

These findings suggest the existence of a self-reinforcing cycle involving repetitive thought, mood, and impaired sleep quality, highlighting the importance of cognitive and emotional factors in enhancement and maintenance of good-quality sleep. Keywords: repetitive thought; self-focus; sleep; experience sampling; actigraphy

REPETITIVE THOUGHT, CONCEPTUALIZED as a “process of thinking attentively, repetitively or frequently about oneself and one’s world” (Segerstrom, Stanton, Alden, & Shortridge, 2003, p. 909), is known to have deleterious influences on mental health. Repetitive thought is a relatively broad concept that encompasses depressive rumination and anxious worry (Watkins, 2008). Empirical studies have shown that these negative and self-focused perseverative thoughts are cognitive risk factors for various psychological problems; for example, the tendency to ruminate predicts future onset, maintenance, and exacerbation of depressive disorders (Just & Alloy, 1997; Kuehner & Weber, 1999; Nolen-Hoeksema, 2000; Nolen-Hoeksema & Morrow, 1991; NolenHoeksema, Parker, & Larson, 1994). Furthermore, repetitive thought with negative content is a transdiagnostic phenomenon associated with a wide range of psychological and physiological maladjustments such as anxiety, substance abuse, eating and drinking problems, and sleep disturbances (Aldao, NolenHoeksema, & Schweizer, 2010; Harvey, Watkins, Mansell, & Shafran, 2004; Nolen-Hoeksema &

68

takano et al.

Harrell, 2002; Nolen-Hoeksema, Stice, Wade, & Bohon, 2007; Watkins, 2011). In the literature on sleep disturbance and insomnia, repetitive thought has increasingly received attention as one of the critical factors that contributes to the maintenance and exacerbation of sleep problems. Psychological models of insomnia have proposed that excessive cognitive activities such as rumination and worry trigger physiological arousal and emotional distress, leading to the perception of sleep problems and genuine sleep deficits (Harvey, 2002; Harvey, Tang, & Browning, 2005). Supporting the link between rumination and sleep, surveys on unselected samples indicate that trait rumination is cross-sectionally associated with impaired sleep quality and insomnia (Carney, Edinger, Meyer, Lindman, & Istre, 2006; Fernandez-Mendoza et al., 2010; Thomsen, Mehlsen, Christensen, & Zachariae, 2003) and prospectively predicts future increases in sleep problems (Takano, Iijima, & Tanno, 2012). Experimental studies found that worry induced in the laboratory was associated with prolonged sleep-onset latency (SOL) for naps (Gross & Borkovec, 1982; Tang & Harvey, 2004), and that stress-related rumination had a negative impact on nighttime sleep quality in individuals with high trait rumination (Guastella & Moulds, 2007; Zoccola, Dickerson, & Lam, 2009). Among patients with insomnia, those with high rumination exhibited lower sleep efficiency and quality (Carney, Harris, Moss, & Edinger, 2010). However, because these previous studies worked with repetitive thought as an experimental manipulation in a laboratory or conceptualized it as a stable, traitlike characteristic, little evidence exists for the association between sleep and repetitive thought as it occurs naturally in daily life. Daily life studies using diary techniques, or more intensive experience sampling methods (ESM; Csikszentmihalyi & Larson, 1987), have suggested that ruminative self-focus fluctuates over time in its association with moods, stress experiences, and avoidance (Dickson, Ciesla, & Reilly, 2012; Moberly & Watkins, 2008a, 2008b; Mor et al., 2010). With ESM, participants keep a portable digital device with them as they go about their regular daily activities and report their momentary feelings, thoughts, and environmental information in response to periodic prompts received from the device. The unique advantage of ESM is its ability to assess these moment-to-moment experiences in an individual’s daily life, thus improving ecological validity and reducing vulnerability to recall bias (Myin-Germeys et al., 2009). Utilizing the advantages of ESM, the present study primarily aimed to observe repetitive thought in daily life and to model the associations among repetitive thought, mood, and

sleep problems by testing the following three specific hypotheses. Our first hypothesis stated that repetitive thought would predict impairment of nighttime sleep quality. Thought-content analyses of presleep cognitions have suggested that worrisome thoughts are associated with sleep disturbances and insomnia (Harvey, 2000; Watts, Coyle, & East, 1994; Wicklow & Espie, 2000). Although there is no clear evidence of the association between daytime repetitive thought and nighttime sleep, it is reasonable to expect that repetitive thought occurring in the evening would have a stronger influence on sleep than that which occurred in the morning or afternoon. Data show that variables assessed at temporally proximal points have a greater association than those measured distally, particularly in cases when the variables in question have moderate to low stability (Cole & Maxwell, 2003). As the present study assessed the flow of thoughts over the course of the day, we were able to identify the time of day at which repetitive thought would have the most deleterious effect on sleep quality. Second, we hypothesized that disturbed sleep would negatively influence mood the following morning. The emotional effects of insufficient sleep and insomnia have been examined in a number of studies (Buysse et al., 2007; Durmer & Dinges, 2005); for example, sleep problems predicted future depression (Breslau, Roth, Rosenthal, & Andreski, 1996), and sleep deprivation was associated with mood disturbances such as increased fatigue or decreased vigor (Dinges et al., 1997). An ESM study showed that perceived sleep problems predicted reduced positive affect during daytime hours, although the adverse effect of impaired sleep was relatively weak for negative affect (Bower, Bylsma, Morris, & Rottenberg, 2010). These findings suggest the possibility that impaired sleep quality would contribute to mood disturbances the next day, particularly for positive moods. Third, we hypothesized that disturbed moods in the morning would contribute to further repetitive thoughts the following evening. This hypothesis was examined to determine whether there exists a selfreinforcing cycle, wherein sleep problems induced by repetitive thought contributed to mood disturbances the following day, resulting in further repetitive thought in the evening. Regarding the relationship between repetitive thought and mood problems, one previous longitudinal survey (Nolen-Hoeksema et al., 2007) and an ESM study (Moberly & Watkins, 2008a) suggested that negative moods predict future rumination, and vice versa. Considering this mutual relationship between mood and repetitive thought, it was expected that mood disturbances in the morning

69

repetitive thought and sleep would be associated with increased repetitive thought in the evening. Our further aim was to assess everyday sleep objectively, because many of the extant psychological studies that have investigated the association between repetitive thought and sleep have employed subjective measures of sleep disturbances. Although subjective complaints about sleep are important indicators of sleep problems and insomnia, subjective and objective sleep parameters often contradict each other (e.g., Lockley, Skene, & Arendt, 1999). Thus, we conducted ambulatory activity monitoring of sleep using an actigraphic monitor, a wristwatch-like device used to record momentary physical movement that can also be utilized to estimate sleep parameters. Actigraphic estimation of sleep–wake patterns shows good agreement with patterns measured using polysomnography (Morgenthaler et al., 2007). Furthermore, the use of actigraphy enabled us to monitor nighttime sleep for a few days or a few weeks, thus improving the reliability of sleep assessment. Previous cognitive studies have focused on only one night’s sleep or one nap (e.g., Tang & Harvey, 2004; Zoccola et al., 2009). Taken together, the present study examined the cyclic association among repetitive thought, nighttime sleep, and mood in a natural setting, using ESM to assess momentary repetitive thought and actigraphic sleep monitoring to test the following three hypotheses: First, we hypothesized that repetitive thought, particularly that occurring in the evening, would be associated with impaired nighttime sleep. Repetitive thought that occurs in the evening may be more strongly associated with sleep parameters than that which occurs in the morning or afternoon because of the temporal proximity between the variables (Cole & Maxwell, 2003). In examining this hypothesis, we controlled for covariates of physical, emotional, and social activities, because these variables potentially influence the association between repetitive thought and sleep. For example, acute exercise has been associated with small-to-moderate effects on various sleep parameters (Youngstedt, O’Connor, & Dishman, 1997), the induction of happy moods has been found to shorten SOL in a healthy sample (Talbot, Hairston, Eidelman, Gruber, & Harvey, 2009), and quality of daytime partner interaction has been associated with SOL and sleep efficiency during the ensuing night (Hasler & Troxel, 2010). Second, we tested whether nighttime sleep parameters would predict mood problems on the following day. Finally, we examined whether a disturbed mood in the morning recursively contributed to further repetitive thought in the evening.

Method participants A group of 49 undergraduate students (38 female; average age = 19.4, SD = 1.3) was recruited from introductory psychology courses across several universities in Japan. 1 All participants were Japanese. During the briefing session, each participant provided informed consent and received an explanation of the procedures for ESM and the usage of the actigraph. At the end of the 1-week course of ESM, participants received financial compensation and a report summarizing their personal results. measures During the briefing session, participants completed the Japanese version of the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977; Shima, Shikano, Kitamura, & Asai, 1985, for the Japanese version) as a baseline measure of depressive symptoms. The CES-D is a 20-item scale that measures depressive symptoms during the previous week. Each item is rated on a 4-point scale of frequency (less than 1 day = 0; 5–7 days = 3). The mean score on the CES-D was 15.0 (SD = 8.3). It had good internal consistency in the present study (α =.84). Fifteen participants had a CES-D score equal to or greater than 16, which is a cutoff point indicating significant depressive symptoms (Radloff, 1977; Shima et al., 1985). procedure We used a mobile-phone-based ESM to collect data from participants during the course of their daily activities (Takano & Tanno, 2010, 2011). We employed mobile phones as our ESM reporting device because the prevalence of the mobile phone is over 100% in Japan (some people have more than one phone; Ministry of Internal Affairs and Communications, Japan, 2012) and students are highly accustomed to using them. Each participant received eight e-mails on his or her mobile phone between 8 a.m. and 12 a.m. each day. The day was divided into eight intervals of 120 min; e-mails were sent once for each interval with semirandom timing. In order to avoid repetition in too short a 1

Daytime repetitive thought and activity data were previously reported by Takano, Sakamoto, and Tanno (in press). In that study, we found that momentary levels of repetitive thought were reduced when participants engaged in specific activities, such as eating or hobbies, although repetitive thought was not directly influenced by amount of physical activity. Furthermore, the influence of specific activities on reduced repetitive thought was particularly strong in individuals with higher levels of depression. In the current paper, we initially report the data of nighttime activity, used for calculation of sleep parameters, and the results of the associations among repetitive thought, mood, and sleep quality.

70

takano et al.

period, e-mails were not sent during the 15 min at the start and end of each interval. Each e-mail contained a URL directing the participant to an online questionnaire concerning his or her current thoughts. This was compatible with both smartphones and feature phones. When participants received the e-mail, they had to access the URL and complete the questionnaire within 30 min. Responses that were not completed within the time limit were not recorded in the system. Repetitive Thought For each ESM measurement, participants were asked to briefly record their current thoughts in several sentences in order to make them aware of their thinking. After noting the brief description, participants rated their thinking on a 7-point Likert scale based on three dimensions of repetitive thought: the extent to which the described thought was (a) unpleasant (unpleasantness); (b) related to the self, such as their feelings and problems (self-focus) 2; and (c) easy to control (controllability). The controllability item was reverse scored in order to reflect difficulties involved in controlling one’s thinking (uncontrollability). An aggregation of these three scores (unpleasantness, self-focus, and uncontrollability) was used as an indicator of repetitive thought. 3 Takano and Tanno (2011) showed that each of the items on this scale was significantly associated with trait rumination even after controlling for depressive symptoms, and that self-focused thinking tended to be perceived as unpleasant and uncontrollable, particularly for trait ruminators. Furthermore, these authors also indicated that selffocus, uncontrollability, and unpleasantness interacted to predict increased negative affect, suggesting that the combination of these three attributes of thinking may be particularly relevant to emotional problems. This momentary measure of repetitive thought showed good reliability (R = .95) in the present data, which is an index of averaged reliability over the time points with the times considered to be fixed (Shrout & Lane, 2012). 2 Our measure of self-focus (focusing on feelings and problems) follows the conceptualization of ruminative self-focus employed in previous ESM studies (Moberly & Watkins, 2008a, 2008b). In line with these studies, we included the factor of “problems,” because other researchers have proposed that ruminative thoughts arise from self-discrepancy (Martin & Tesser, 1996), and self-focus is conceptualized as a necessary component process in the sequence of self-regulation attempting to resolve problems and unattained goals (e.g., Carver & Scheier, 1990). 3 Although we used this scale as a measure of “ruminative selffocus” in our previous work (Takano & Tanno, 2011; Takano et al., in press), we decided to relabel it as “repetitive thought” in this study, because later analysis on thought content showed that this scale is responsive both to future worries and past rumination.

Negative and Positive Affect Negative and positive mood states were each assessed using three adjectives (i.e., “scared,” “afraid,” and “upset” for negative affect; “active,” “proud,” and “strong” for positive affect) selected from the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988). In the Japanese version of the PANAS, these items have the highest loadings for negative and positive affect factors (Sato & Yasuda, 2001). Each item was rated on a 7-point scale ranging from 1 (not at all) to 7 (very much). We used aggregations of each of these three items as indicators of negative and positive affect, which showed good reliability (R = .99 for both of the scales). Social Interaction Along with thinking and emotional variables, participants indicated whether they were interacting socially (a) with friends and (b) with family members when they received the e-mail. These social interaction variables were dummy coded: if participants were interacting with the target group, the variable was scored as 1; otherwise, it was scored as 0. Of 49 participants, data from 4 were excluded from the analysis for the following reasons: Two participants did not understand the ESM procedure, 1 participant marked the same response for most of the scales, and 3 participants’ data were not properly recorded: 1 participant did not wear the actigraphic monitor, 1 had a device that malfunctioned, and 1 did not record the bed-in and bed-out times. Thus, the final sample comprised 43 participants (33 female and 10 male). The mean response rate was 78.9% (SD = 12.3), and the mean response time was 8 min 17 sec (SD = 7 min 19 sec). The ratings for 1,899 occasions (78.9% of total 2,408 assessments) were recorded across participants. We divided a day into three phases (morning, afternoon, and evening) and calculated the average repetitive thought scores for each phase of each day. The average repetitive thought scores recorded in the morning (from 8 a.m. to 12 p.m.) were defined as repetitive thought in the morning. Similarly, the average scores recorded during the afternoon (from 12 p.m. to 8 p.m) and evening (from 8 p.m. to 12 a.m.) were defined as repetitive thought in the afternoon and in the evening, respectively. A phase was considered to be missing data (37 of total 903 phases) only if no response was recorded during that phase. For other momentary variables of contextual information, such as mood states, social interactions, and actigraph-defined physical activity (see below), we calculated average scores during the evening phase; we only examined evening scores of these variables in the following analyses because psychosocial activities that were temporally close to nighttime sleep

repetitive thought and sleep would have a greater impact on sleep quality. We also used only the controlling variables assessed in the evening because mood measures were highly autocorrelated among morning, afternoon, and evening (rs = .65–.70, for positive affect; rs =.48–.60, for negative affect), although autocorrelations of repetitive thought were moderate (rs =.31–.42). In order to avoid the risk of multicollinearity, we decided to select only the evening scores as control variables. Thus, we created a data set of seven observations per person with variables of repetitive thought at each phase and the evening as a control variable. Missing values were handled with listwise deletion.

materials Actigraphy An Actiwatch (Mini Mitter, Respironics Inc., Bend, OR) containing an accelerometer with a sensitivity of 0.05 g to record physical motion in all directions was used to estimate daytime physical activity levels and nighttime sleep parameters. Participants wore the actigraphs on their nondominant wrists. Motion data were collected in a 1-min epoch throughout the 8 nights (the night before starting ESM and 7 nights during ESM sampling). The daytime activity data were matched with subjective ESM reports and used as a measure of body movement. As our ESM measures focused on momentary psychosocial activities, we selected the physical activities most recent to each ESM probe in order to align them in terms of time scale. Mean physical activity 15 min prior to receiving each e-mail was calculated in order to represent the momentary levels of physical activity corresponding to reports of repetitive thought and other contextual variables measured by ESM. Although similar intervals of 5 (McCormick et al., 2009), 10 (Schwerdtfeger & Friedrich-Mai, 2009), 15, and 20 minutes (Leary, Donnan, MacDonald, & Murphy, 2000) were employed in previous studies using ambulatory activity monitoring, Leary et al. (2000) found that a 15-minute interval is best correlated with autonomic activities. The activity score was logarithmically transformed to correct for positive skew (Jones et al., 2006). Nighttime activity data were analyzed in order to estimate sleep parameters such as SOL, sleep efficiency, and total sleep time (TST). To record the time of getting into and out of bed, participants were asked to press an event marker button on the Actiwatch when they began trying to fall asleep and again when they got out of bed. From the level of observed activity, each epoch was determined as wake or sleep using the software algorithm. Sleep start time was established as the

71

first immobile period lasting for at least 10 min after getting into bed. SOL was defined as the number of minutes from bed-in time (recorded by an event marker button) to the sleep start time. Efficiency was the percentage of total time scored as sleep relative to the amount of time spent in bed. TST was the time between sleep start and sleep end time minus the time classified by the algorithm as awake during this interval. Previous studies have validated actigraph monitoring of activity as a measure of sleep with good rates of agreement with polysomnographic measurement in scoring sleep/wake time during sleep periods (Reid & Dawson, 1999; Sadeh, Sharkey, & Carskadon, 1994). Cole, Kripke, Gruen, Mullaney, and Gillin (1992) found that actigraph-defined sleep parameters were highly correlated with those measured by polysomnography (r = .90 for SOL; r = .71 for efficiency; r = .77 for TST). Actigraphy is highly sensitive in detecting sleep epochs, although its ability to detect wakefulness may be limited particularly in the first night of a laboratory sleep measurement, and for participants with disturbed sleep who lie quietly in bed for long periods of time (Kushida et al., 2001; Sadeh, Hauri, Kripke, & Lavie, 1995). Across all participants, 271 sleep reports (6.3 nights per participant on average out of 8 nights) were recorded. Matching these reports with those from ESM, the data of 217 days (collected from 7 nights during ESM sampling) were used to estimate the model for examining the associations between repetitive thought and sleep (the first hypothesis). Of the 217 matched reports, missing values were reported for time out of bed for 17 days; 200 observations were thus used to estimate sleep efficiency and TST. Two hundred seventeen matched sleep reports (collected from the night before starting ESM and the following 6 nights) were used to analyze the association between sleep and next-day moods (the second hypothesis).

statistical analysis We estimated multivariate random coefficient models to examinine our hypotheses because our data set had a three-level nested structure: the set of P sleep parameters (i.e., SOL, efficiency, and TST) were clustered within the day-level variables (i.e., ESM variables at each time of day), which were nested within the person-level variables (i.e., depression and gender differences). The lowest-level model was as follows: Y pti ¼

P X h¼1

D ph ðβh0i þ βh1i X 1ti þ … þ βhki X kti þ ehti Þ;

72 where D ph ¼

takano et al.



1 0

h¼p h ≠ p:

Thus, Ypti was the p-th sleep parameter for the t-th occasion within the i-th participant, and h indexed the set of sleep parameters. In order to represent the different sleep parameters as outcome variables, we had p dummy variables (i.e., Dph), defined for p = 1, . . . P. X’s were k number of day-level predictors including repetitive thought, mood, physical activity, and social interactions. Residuals were denoted as ehti. For the within-person variance structures (R matrix), the Kronecker product of unstructured and first-order autoregressed matrices was used; residuals between response variables (σ) were unstructured to be correlated with one another, and the residuals within a dependent variable were assumed to be autocorrelated in order to control for the sequential dependency of multiple assessment over the 7 days. The person-level structure was described as follows: βp0i ¼ βp00 þ βp01 W 1i þ βp02 W 2i þ u p0i βp1i ¼ βp10 þ u p1i … βpki ¼ βpk0 þ u pki : For each of the p response variables, both intercepts (βp0i) and slopes (βpki) were assumed to have a random effect (upki). We controlled for individual differences in depression and gender at this level (W1i and W2i). As a first step in model selection, we estimated a full model that included all random intercepts and random slopes varying at the person level. Because our model had a number of occasion-level predictors, only variance (but not covariance) components (τ) were assumed for the between-person variance structure (G matrix). If random effects did not converge to nonnegative variance, we reestimated the model with the intercept or slope as a fixed effect. To test associations between sleep quality and next-day mood (the second hypothesis), we estimated a similar multivariate multilevel model as that described above; positive and negative affect the next morning were set as Ypti, defined for p = 1, 2. For the day-level predictors (X1ti, . . . Xkti), this model included each of three sleep parameters and social interactions. For the person-level predictors, depressive symptoms and gender differences were also included as control variables. Within- and betweenperson variance matrices were structured the same as in the model for testing the first hypothesis. In order to test the concurrent and lagged associations between daytime repetitive thought

and mood as stated in the third hypothesis, we estimated a multilevel mediation model using the multivariate approach described below: Y pti ¼

P X h¼1

  Dph βh0i þ βh1i X ti þ βh2i Y ðp−2Þti þ βh3i Y ðp−1Þti þ ehti ;

where Ypti included repetitive thought in the morning, afternoon, and evening as defined by p = 1, 2, 3. The model included autoregressing parts of repetitive thought, Y(p-1)ti and Y(p-2)ti. In the case of (p – 1) b 1 or (p – 2) b 1, the corresponding coefficients (βp2i and/or βp3i) were fixed to be zero. Xti denoted moods in the morning, although our actual analyses specifically focused on positive moods. Residuals (epti) were specified for each of p cases, but the same (i.e., first-order autoregressed) structure was employed. Although we tested whether sleep parameters exhibited a day-dependent trend, such as a firstnight effect, we found no significant effect of day on each of the three sleep parameters (ps N .20). For mood measures, positive affect also showed no such day-dependent trend, though negative affect showed a weak but increasing trend over the sampling days (β = .17, SE = .09, t = 1.76, p = .08). Thus, we assumed the effect of day only for the model predicting negative affect in the test of the second hypothesis. Analyses were conducted using the SAS (Version 9) MIXED procedure and restricted maximum likelihood estimation.

Results On average, participants went to bed at 1:36 a.m. and rose at 8:10 a.m. Means, standard deviations, and ranges of ESM variables and other sleep parameters are presented in Table 1. Before conducting the main analyses, we compared thought contents between high and low repetitive thought scores in order to validate our momentary repetitive thought scale. We noted 26 thoughts with a score of 21 (7 on each of the three self-focus, unpleasantness, and uncontrollability items), and 17 thoughts with a score equal to or less than 5. The first author and a Ph.D. researcher who was blinded to the aims of the study independently categorized the high-scored thoughts into four topics (interrater concordance rate was 88%); of the 21 thoughts collected, 31% focused on frustration (e.g., It is really too bad that I have to write a paper today, though I’ve been so exhausted), 27% on regret or sorrow about the past (e.g., I cannot forget my former lover), 27% on worry (e.g., Will I get a job after university?), and 15% on physical complaints (e.g., My stomach is killing me). For the thoughts with low scores on the repetitive thought scale (concordance rate was 100%), 65% focused on

73

repetitive thought and sleep Table 1

Descriptive Statistics for Study Variables Variable

Repetitive thought Morning Afternoon Evening Evening variables (covariates) Physical activity Positive affect Negative affect Interaction with friends Interaction with family members Sleep parameters SOL (min) Efficiency (%) TST (hrs)

n

Mean (SD)

Range

217 217 217

13.5 13.2 13.1

(2.6) (2.7) (2.4)

6.00–18.00 4.00–20.00 5.33–21.00

217 217 217 217 217

5.0 8.2 5.0 0.2 0.2

(1.1) (3.5) (2.6) (0.3) (0.3)

0.55–7.08 3.00–19.00 3.00–14.00 0.00–1.00 0.00–1.00

271 252 252

11.0 81.5 5.4

(20.1) (10.1) (1.7)

0.00–218.00 8.23–99.34 0.68–9.33

Note. Physical activity was logarithmically transformed. ESM was collected for 7 days and sleep data was collected for 8 nights. SOL = sleep-onset latency; TST = total sleep time.

current concerns such as the landscape, weather, clothes, TV, and friends (e.g., The weather today is good), and 29% on time or schedule (e.g., We will be finished soon?). These results suggest that our repetitive thought measure was responsive to negative topics important to the participants, including past rumination or future worry, and that low scores on this scale indicated that participants’ attention was focused on transient environmental information that was lower valenced and less related to the self. We also tested the lagged correlations between repetitive thought and negative affect, considering the overlapping constructs between the items of repetitive thought (i.e., negative valence) and negative affect scales. A multilevel model was estimated in which negative affect measured at time (t + 1) was predicted by repetitive thought at time (t), controlling for the baseline levels of negative affect at time (t). The results showed that repetitive thought was a significant predictor of negative affect measured at the next occasion (β = .05, SE = .02, t = 2.10, p = .036), even controlling for the autocorrelation of negative affect (β = .36, SE = .04, t = 9.83, p b .001). Conversely, negative affect measured at time (t) significantly predicted repetitive thought at the (t + 1) time (β = .08, SE = .03, t = 3.14, p b .01), controlling for the current level of repetitive thought (β = .15, SE = .03, t = 5.94, p b .001).

associations between repetitive thought and sleep In order to test the association between repetitive thought at each time of the day and SOL, we conducted a multivariate multilevel model in which actigraphy-defined sleep parameters (i.e., SOL,

efficiency, TST) were predicted by levels of repetitive thought in the morning, afternoon, and evening, controlling for levels of physical activity, negative and positive affect, and social interaction during evening hours. We also controlled for the effect of gender differences as a dummy-coded gender variable (1 for female, 0 for male) and levels of depressive symptoms as measured by the CES-D. In predicting SOL, significant or near-significant random effects were found for repetitive thought in the evening (τ 2 = .09, z = 1.36, p = .09) and interaction with friends (τ 2 = .003, z = 1.98, p = .02). Random effects were assumed for physical activity (τ 2 = .67, z = 2.20, p = .01) in predicting sleep efficiency, and for repetitive thought in the afternoon (τ 2 = 184.2, z = 2.25, p = .01) in predicting TST. All other random components were fixed at zero, resulting in an improved model fit (–2LL = 3,808.4, AIC = 3,830.4) compared with the unselected full model (–2LL = 3,829.6, AIC = 3,849.6). Table 2 presents the estimated results for fixed effects. Only repetitive thought in the evening had a significant association with longer SOL (β = 1.02, t = 2.30, p b .05), worse sleep efficiency (β = –0.77, t = 2.59, p b .05), and reduced TST (β = –0.12, t = 2.33, p b .05), whereas repetitive thought at other times of the day had no effect on these sleep parameters (p N .10). Furthermore, social interaction with friends was associated with longer SOL (β = 9.90, t = 2.32, p b .05). We also estimated two models, which included repetitive thought in the morning (or afternoon) but excluded repetitive thought occurring at other times of the day, in order to determine to what extent repetitive thought in the evening masked the effect

74

takano et al.

Table 2

Parameter Estimates of the Multilevel Model of Repetitive Thought, Activity, and Other Variables Predicting Sleep Parameters SOL

Efficiency

TST

Estimates (SE)

t

Estimates (SE)

t

Estimates (SE)

t

9.97 (9.52) -5.22 (2.85) 0.07 (0.14)

1.05 1.83 0.47

92.54 (6.73) -0.64 (2.17) 0.07 (0.11)

13.76** 0.30 0.67

9.01 (1.12) -0.00 (0.37) 0.02 (0.02)

8.06** 0.01 1.29

0.10 (0.38) -0.43 (0.40) 1.02 (0.44)

0.26 1.09 2.30*

-0.36 (0.26) -0.14 (0.28) -0.77 (0.30)

1.39 0.51 2.59*

-0.06 (0.04) -0.08 (0.05) -0.12 (0.05)

1.40 1.61 2.33*

-0.80 (0.90) -0.55 (0.31) -0.10 (0.40) 9.90 (4.26) 0.51 (3.26)

0.89 1.75 0.24 2.32* 0.16

0.22 (0.62) 0.38 (0.22) 0.49 (0.28) -3.39 (2.26) -1.55 (2.28)

0.36 1.70 1.76 1.50 0.68

-0.06 -0.02 -0.00 -0.60 -0.15

0.57 0.41 0.07 1.62 0.39

Random effect (Within-person error: σ)

Estimates(SE)

z

Estimates (SE)

z

Estimates (SE)

z

SOL Efficiency TST Autocorrelation (ρ)

159.60 (18.84) -57.42 (9.81) -4.70 (1.51) 0.13 (0.06)

8.47** 5.85** 3.12** 1.94

71.76 (8.55) 5.77 (1.05)

8.39** 5.50**

1.92 (0.22)

8.56**

Fixed effect: β Intercept Gender Depression Repetitive thought Morning Afternoon Evening Evening variables (covariates) Physical activity Positive affect Negative affect Interaction with friends Interaction with family

(0.10) (0.04) (0.05) (0.37) (0.38)

Note. 617 observations (217 for SOL and 200 for efficiency and TST) were used for model estimation. SOL = Sleep onset latency; TST = Total sleep time; * p b .05. ** p b .01.Note. 617 observations (217 for SOL and 200 for efficiency and TST) were used for model estimation. SOL = sleep-onset latency; TST = total sleep time; * p b .05; ** p b .01.

of repetitive thought that occurred at earlier times of day because of their intercorrelations. Repetitive thought in the morning had a significant influence on efficiency (β = –0.54, SE = 0.24, t = 2.17, p b .05) and TST (β = –0.09, SE = 0.04, t = 2.18, p b .05), but not on SOL (p N .10) when repetitive thought in the afternoon and evening were excluded from the model. On the other hand, repetitive thought in the afternoon was associated with reduced TST (β = –0.12, SE = 0.04, t = 2.76, p b .01) but with neither SOL nor efficiency (ps N .10) if repetitive thought in the morning and evening were excluded. These results suggest that TST is associated with repetitive thought at any time of day, and that sleep efficiency is associated with afternoon-to-evening repetitive thought. SOL

is uniquely associated with evening repetitive thought. Next, we examined which components of repetitive thought (i.e., self-focus, unpleasantness, and uncontrollability) were separately predictive of sleep. The same multivariate random coefficient model was estimated with two exceptions: the evening repetitive thought variable was replaced with individual items of repetitive thought that were assessed in the evening. Furthermore, we simplified the model by excluding the gender difference variable because the original model did not converge. In order to avoid redundancies, we report the estimated fixed effects for the three target variables in Table 3. Self-focus was associated with decreased efficiency (β = –1.11, SE = 0.56, t = 2.00, p b .05) and TST (β = –0.18,

Table 3

Fixed Effects of Individual Items of Evening Repetitive Thought to Separately Predict Each Sleep Parameter SOL Estimates (SE)

Items in the evening repetitive thought Self-focus 1.25 (0.82) Unpleasantness 0.03 (0.84) Uncontrollability 1.72 (0.80)

Efficiency

TST

t

Estimates (SE)

t

Estimates (SE)

t

1.53 0.03 2.14*

-1.11 (0.56) -0.39 (0.59) -0.79 (0.56)

2.00* 0.66 1.41

-0.18 (0.09) -0.08 (0.09) -0.12 (0.09)

1.95 0.87 1.32

Note. SOL = Sleep onset latency; TST = Total sleep time; * p b .05.

75

repetitive thought and sleep SE = 0.09, t = 1.95, p = .053), whereas uncontrollability was associated with longer SOL (β = 1.72, SE = 0.80, t = 2.14, p b .05). Unpleasantness was not significantly associated with any of the sleep parameters. These results suggest that uncontrollability might contribute to difficulty of sleep initiation, and that thought content focused on the self is associated with sleep maintenance problems.

associations between sleep and next-day mood In order to test the association between sleep and next-day mood states, we estimated a multivariate random coefficient model in which positive and negative moods in the morning were predicted by the previous night’s sleep parameters, including SOL, efficiency, and TST, controlling for social interactions. In order to correct for the increasing trend of negative affect, the model of negative affect included the day variable. In predicting positive affect, we assumed random intercepts (τ 2 = 7.65, z = 3.85, p b .01) and random slopes for interactions with friends (τ 2 = 4.05, z = 1.40, p = .08) and with family members (τ 2 = 4.75, z = 0.83, p = .20). Similarly, we assumed random slopes for sleep efficiency (τ 2 = 0.00, z = 1.42, p = .08), interaction with friends (τ 2 = 4.99, z = 1.43, p = .08), and day (τ 2 = 0.20, z = 2.77, p b .01) in predicting negative affect. We decided to include these nonsignificant random effects in the model because they improved the goodness of fit (–2LL = 2,074.2, AIC = 2,094.2) compared with the model without these

assumptions (–2LL = 2,095.2, AIC = 2,107.2). As with the analyses described above, within-person residuals were structured as the Kronecker product of unstructured and autoregressed matrices. Table 4 shows the estimated fixed effects. Sleep efficiency and TST had significant effects on reduced positive affect (β = 0.06, t = 2.27, p b .05; β = –0.42, t = 3.29, p b .01, respectively) while SOL did not. For negative affect, none of the sleep parameters were significant predictors (ps N .20). We also tested the associations between sleep and next-day levels of repetitive thought with a univariate random coefficient model in which repetitive thought in the morning was predicted by the previous nights’ sleep parameters controlling for the same variables. The results showed a similar pattern as observed with negative affect: neither SOL (β = 0.00, SE = 0.01, t = 0.10, ns), efficiency (β = −0.01, SE = 0.02, t = 0.34, ns), nor TST (β = −0.01, SE = 0.13, t = 0.07) had a significant effect on repetitive thought. These results suggest that decreased sleep efficiency and increased TST are associated with decreased levels of positive affect the next morning, whereas sleep parameters had no significant influence on either negative affect or repetitive thought.

associations between daytime moods and repetitive thought To test whether decreased positive affect due to insufficient sleep contributed to further repetitive thought in the evening (the third hypothesis), we

Table 4

Parameter Estimates for the Multilevel Model of Sleep Parameters Predicting Next-Day Mood Positive affect

Fixed effect: β Gender Depression Sleep parameters SOL Efficiency TST Controlling variables Interaction with friends Interaction with family Day Random effect (Within-person error: σ) Positive affect Negative affect Autocorrelation (ρ)

Negative affect

Estimates (SE)

t

Estimates (SE)

t

–0.77 (1.10) –0.04 (0.06)

0.71 0.67

1.47 (0.78) 0.06 (0.04)

1.87 1.56

0.01 (0.01) 0.06 (0.03) –0.42 (0.13)

0.82 2.27* 3.29**

0.01 (0.01) 0.03 (0.02) –0.12 (0.14)

0.80 1.05 0.91

–0.22 (0.64) 0.12 (1.07) -

0.35 0.11 -

–0.01 (0.71) –0.25 (0.90) 0.15 (0.11)

0.02 0.28 1.38

Estimates (SE)

z

Estimates (SE)

z

4.64 (0.64) –0.53 (0.37) 0.15 (0.09)

7.21** 1.42 1.61

3.81 (0.51)

7.51**

Note. 434 observations (217 for each of positive and negative affect) across 43 participants were analyzed. SOL = sleep-onset latency; TST = total sleep time; * p b .05; ** p b .01.

76

takano et al.

modeled the time-dependent associations between repetitive thought and positive affect. A multivariate multilevel model was estimated in which repetitive thought at each time of day was autoregressed with a predictor of positive affect in the morning. We assumed a random effect for each intercept of the dependent variables (–2LL = 2,900.5, AIC = 2,918.5) but not for the slopes (–2LL = 2,900.1, AIC = 2,922.1). The results (Table 5) suggested that positive affect in the morning was associated with increased repetitive thought at the same assessment time (β = –0.12, SE = 0.06, t = 1.99, p b .05), but this did not have significant predictive power for repetitive thought that occurred in the afternoon and evening (ps N .10). However, repetitive thought was autocorrelated, which suggests that if one engaged in repetitive thought in the morning, he or she was more likely to continue ruminating in the afternoon (β = 0.26, SE = 0.07, t = 3.95, p b .01), and if it was afternoon, the repetitive thought would be continued until the evening (β = 0.12, SE = 0.06, t = 2.03, p b .05). These findings suggest that positive affect in the morning is indirectly associated with repetitive thought in the evening; reduced positive affect in the morning is concurrently associated with increased repetitive thought, which tends to be maintained over a day and results in increased repetitive thought in the evening due to the self-perpetuating nature of the process.

associations between repetitive thought and other sleep parameters Finally, we conducted a similar analysis using a multivariate random coefficient model on other sleep parameters (i.e., bed-in and bed-out time) as in the test of our first hypothesis. For bed-in time, repetitive thought in the morning had a significant association with delayed bedtime (β = 0.09, t = 2.33, p b .05), although repetitive thought at any other phase of the day had no significant effect. Furthermore, interaction with friends was also associated with later bed-in time (β = 1.11, t = 2.75, p b .01). None of the independent variables were significantly associated with bed-out time (p N .05).

Discussion In this study, we observed naturally occurring repetitive thought and moods at various times of the day in daily-life settings by using an ESM paradigm, and investigated their associations with objective measurements of sleep quality obtained by an actigraph. We tested three main hypotheses regarding the interrelationships between repetitive thought, moods, and sleep. First, we hypothesized that repetitive thought would have a significant influence on nighttime sleep quality, and that this

effect would be stronger for repetitive thought that occurred in the evening. Second, we proposed that impaired sleep quality would contribute to mood disturbances the next morning. Third, we hypothesized that disturbed moods in the morning would contribute to further repetitive thought in the evening. Our first set of analyses indicated that repetitive thought was associated with longer SOL, impaired sleep efficiency, and decreased TST at night. Furthermore, repetitive thought that occurred in the evening had specific associations with these sleep parameters, whereas repetitive thought at other times of the day had no association with these sleep parameters. This unique effect of evening repetitive thought (Table 2) suggests that the time when negative perseverative thinking occurs is important information in predicting sleep problems. One possible reason is that the variables measured in the evening were closer in time to nighttime sleep; thus, they were less influenced by decay of correlations resulting from the passage of time. Another possibility is that, because the evening is a quieter time that might be associated with fewer environmental distractors (such as social interactions) than midday (Moberly & Watkins, 2008a), individuals are more likely to turn their attention inward, and such self-focused processing is less likely to be interrupted. A previous ESM study found that individuals with depressive symptoms were more likely to engage in ruminative thinking at night (Takano & Tanno, 2011). Therefore, by enhancing negative emotional activity or arousing the autonomic nervous system (e.g., Pieper, Brosschot, van der Leeden, & Thayer, 2007), for example, evening repetitive thought might have a more pathological influence than that occurring at other times of day. Brosschot, Van Dijk, and Thayer (2007) suggested that the effect of worry on cardiac autonomic activity extends into the nocturnal sleep period. These findings are in line with the cognitive model of insomnia (Harvey, 2002), which proposes that the presleep excessive, negatively toned cognitive activity (such as worry and rumination) causes autonomic arousal and emotional distress, and consequently acts as an initial trigger for sleep disturbances. It is noteworthy that morning and afternoon repetitive thought were associated with reduced TST, and that afternoon repetitive thought was associated with decreased sleep efficiency, if the estimated models did not include repetitive thought in the evening as an independent variable. This suggests that although evening repetitive thought has the strongest effect on sleep, daytime repetitive thought might still indirectly influence nighttime sleep. Particularly TST, which shows significant

repetitive thought and sleep associations with repetitive thought at each time of the day, might be influenced by static and habitual factors linked to repetitive thought, such as personality (e.g., neuroticism) or lifestyle (e.g., eating or drinking habits; Cheng et al., 2012). On the other hand, SOL was associated only with repetitive thought in the evening, which suggests the possibility that SOL is more likely to be associated with acute arousal prior to sleep. However, in interpreting these results, it should be noted that our findings are based on a nonclinical student sample. An alternative hypothesis could be that daytime repetitive thought would have a significant effect on sleep if patients with depressive disorders or insomnia were examined, because those individuals show a stronger effect of repetitive thought than healthy individuals (Mor & Winquist, 2002). Our additional analyses focusing on the individual items of repetitive thought showed that self-focus was associated with reduced sleep efficiency and TST, whereas uncontrollability was associated with prolonged SOL. Interestingly, unpleasantness was not a significant predictor of any of the three sleep parameters, which suggests that the associations between repetitive thought and sleep might not merely rely on the negative emotions elicited by or associated with repetitive thought. Although previous studies have suggested that negative cognitive activity associated with the induction of worrisome thoughts contributes to sleep disturbances (Gross & Borkovec, 1982; Tang & Harvey, 2004), recent findings in Wuyts et al. (2012) suggest that using nonemotional cognitive tasks, such as a digit-span task and a Stroop task, to induce a cognitive load disturbs sleep even if emotional factors are eliminated. Our results are in line with these findings and further support the idea that, rather than negative valence, self-relevance and a perception of uncontrollability might be key factors in sleep-disturbing cognition. However, it is noteworthy that our measure of self-focus taps “problems” as well as moods and personalities, suggesting that the selffocus item per se might include aspects of both negative valence and self-orientation. It is possible that such a contamination may have influenced the associations among self-focus, unpleasantness, and sleep parameters. Thus, our findings should be interpreted in light of this measurement limitation. Our second analysis showed that decreased sleep efficiency was associated with decreased positive affect the following morning. This result is consistent with previous findings showing that impaired sleep quality was associated with emotional distress such as decreased positive affect (Bower et al., 2010) and increased mood disturbances the next day (Dinges et al., 1997). Considering that positive affect includes

77

“active” feelings, individuals might feel full of energy and recovered from their fatigue after experiencing good-quality sleep. On the other hand, increased TST was associated with reduced positive affect. Although this association was an unexpected result, it might reflect the aftereffect of oversleeping, which is typically seen in students’ weekend sleep and is associated with daytime sleepiness and mood problems (Jarrin, McGrath, Silverstein, & Drake, 2013; Saxvig, Pallesen, Wilhelmsen-Langeland, Molde, & Bjorvatn, 2012). Third, we investigated daytime associations between mood and repetitive thought by examining whether mood disturbances in the morning would enhance repetitive thinking at night. The results suggested that decreased positive affect in the morning was associated with an increased level of repetitive thought at the same assessment time, and that the high level of repetitive thought in the morning tended to persist throughout the afternoon and evening of that day. However, we failed to find a significant predictive power of positive affect directly on repetitive thought assessed at a later time of day, although we replicated the findings of Moberly and Watkins (2008), which showed that negative affect predicted increases in rumination at a later (i.e., 90-min) assessment point controlling for the current level of negative affect. This inconsistency may have occurred because the intervals between positive affect and repetitive thought that were employed in the present study were so long (morning vs. afternoon or evening) that the predictive power of positive affect might have been decayed. Thus, it is more likely that decreased positive affect is indirectly associated with nighttime repetitive thought owing to the selfperpetuating nature of repetitive thought, rather than that mood disturbances directly contribute to an increase in repetitive thought in the evening. Taken together, our results suggest that (a) sleep quality is impaired by repetitive thought that occurs in the evening; (b) disturbed sleep (particularly reduced sleep efficiency) is associated with reduced positive mood the next morning; and (c) reduced positive mood in the morning is concurrently associated with increased repetitive thought, which tends to continue until the evening. Such a vicious cycle among cognitive, emotional, and sleep problems might develop into the biased information processing that is characteristic of sleep disorders. Although the present study addresses self-focused repetitive thought in a nonclinical sample, researchers have argued that sleep-focused thinking is a problem specific to insomnia. Content analyses of presleep cognitive activity have shown that individuals with sleep disturbances and insomnia tend to ruminate about sleeplessness (e.g., thinking

78

takano et al.

about how difficult it is to fall asleep) and worry about daytime dysfunction (e.g., thinking that sleeplessness will interfere with work tomorrow) as well as to engage in other rehearsals and planning for personal and work-related matters (Wicklow & Espie, 2000). Recent experimental studies have revealed attentional biases toward sleep-related stimuli (e.g., pictures of bedroom objects or words such as “bedroom” or “pillow”) among patients with insomnia (Taylor, Espie, & White, 2003), and that this is a problem specific to insomnia but not of delayed sleep phase syndrome (MacMahon, Broomfield, & Espie, 2006; Marchetti, Biello, Broomfield, MacMahon, & Espie, 2006). As described in the model proposed by Espie, Broomfield, MacMahon, Macphee, and Taylor (2006), stressful experiences trigger psychological (rumination or worry) and physiological responses that inhibit normal sleep-related de-arousal and produce transient sleep disturbances. At this stage, self-focused rumination associated with a stressor and with depressed moods would be active, whereas insomnia symptoms per se would be unlikely to attract attention. However, if a stress-related cognition were resolved but insomnia symptoms persisted, the target of attention might shift from the original stressor to any persisting sleep disturbance, which might then contribute to chronic psychophysiological insomnia (Espie et al., 2006). We found that repetitive thought in the morning and afternoon had no direct effect on the sleep parameters. Thus, although people may engage in repetitive thought during the day, they may still experience good-quality sleep if they can disengage from their repetitive thinking by evening. From a practical standpoint, cognitive approaches for treating sleep disturbances should pay particular atten-

tion to interventions that target repetitive thought at night. Our findings suggest that repetitive thought need not be eliminated, but should be postponed from nighttime to the next morning or afternoon. This approach has been shown to be effective, because evidence indicates that postponing worry has an effect on somatic complaints and negative emotion (Jellesma, Verkuil, & Brosschot, 2009). However, repetitive thought in the morning is associated with that in the afternoon, and afternoon repetitive thought is associated with that in the evening (Table 5). This suggests that when people engage in repetitive thought in the morning, it is more likely to persist for the rest of the day. Thus, if ruminative individuals postpone their concern to the next day, they should be mindful to engage in some relaxing and distracting activities in order to disengage from repetitive thought during the evening. Our results also showed that repetitive thought in the morning was associated with a later bedtime. This is possibly because the day on which highintense repetitive thought was reported in the morning might be a busy day. Recoded repetitive thoughts in the morning included worries about exams or travel scheduled over the next few days. Thus, the association between repetitive thought in the morning and bedtime might reflect a need to stay up late because the people worrying may have many tasks to complete to prepare for upcoming important events, with which their mind had already been preoccupied that morning. Furthermore, we also found that social interaction in the evening was associated with longer SOL and later bed-in time. Nighttime interaction with friends might occur during a dinner or at a party, where people would be experiencing higher levels of mental and physical

Table 5

Estimates of the Multilevel Model of Positive Affect Predicting Repetitive Thought Observed at Each Time of the Day Repetitive thought Morning

Fixed effect (β) Intercept Positive affect Morning Repetitive thought Morning Afternoon Random effect (Within-person error: σ) Variance Autocorrelation

Afternoon

Evening

Estimates (SE)

t

Estimates (SE)

t

Estimates (SE)

t

14.47 (0.53)

27.25**

9.09 (1.07)

8.48**

10.45 (1.14)

9.17**

-0.12 (0.06)

1.99*

0.09 (0.06)

1.50

-0.01 (0.05)

0.09

0.26 (0.07) -

3.95**

0.09 (0.06) 0.13 (0.06)

1.41 2.03*

Estimates(SE)

z

Estimates (SE)

z

Estimates (SE)

z

5.15 (0.63) 0.13 (0.10)

8.23** 1.27

4.61 (0.54) 0.04 (0.11)

8.59** 0.42

3.49 (0.40) -0.17 (0.11)

8.78** 1.49

Note: 639 observations (213 each for repetitive thought in the morning, afternoon, and evening) were used.

repetitive thought and sleep arousal, possibly drinking alcohol, and would return home later than usual. Such increased arousal and distorted rhythm leads to taking more time to fall asleep. On the other hand, because interaction with family members had no significant influence on sleep parameters, typical interpersonal interactions might have no such effect on sleep problems. The present study has a number of limitations. First, our sample consisted solely of Japanese undergraduate students. Because functional differences may exist in self-focused thinking between clinical and nonclinical samples (Mor & Winquist, 2002), daytime repetitive thought might still have an influence on nighttime sleep among clinical samples with insomnia or depressive disorders. Furthermore, university students are known to have irregular sleep habits and lower sleep quality (Brown, Buboltz, & Soper, 2002). Although we believe that it is valuable to study risk factors for sleep problems in such a vulnerable population, our findings should be understood to have limited generalizability. Thus, it would be essential to examine whether the present results can be replicated in patient samples with clinically diagnosed depression, generalized anxiety disorder, and insomnia and/or other age groups. Second, the estimated person-level standard errors may be biased because of our limited sample size. Our ESM measurement frequency (56 times per 49 persons) was determined following the guideline of 30 observations per 30 individuals, as supported by simulation studies (Hox, 2010). However, we used a data set of on average 5 nights per 43 individuals after aggregating the occasion-level variables and excluding attrition. One simulation study suggested that the regression coefficients and standard errors are estimated without bias even in a small sample size of 5 cases per 30 individuals, although the standard errors of the second-level variances will be underestimated if the number of individuals is less than 100 (Maas & Hox, 2005). Replacement of missing observations with expectation–maximization procedures might be a possible approach to overcome this limitation (Smith, Borckardt, & Nash, 2012). Third, our estimation of sleep parameters may be influenced by sampling bias caused by a relatively large number of missing reports of bed-in and bed-out time. On average, records of sleep interval per participant were incomplete for 2–3 nights out of 8 nights. Certain special pre- or postsleep situations (e.g., being too tired to be conscious of falling asleep; being in a hurry in the morning because of oversleeping) could have prevented participants from recording their sleep intervals by pressing the event marker button on the actigraph. Future research could address this problem of missing values by using illuminometers, for example.

79

Fourth, we did not control for the use of substances such as caffeine, alcohol, nicotine, or psychiatric medications. These factors may have contaminated our findings by influencing sleep onset and maintenance. For example, the significant association between sleep and social interactions with friends might be confounded by drinking or smoking behaviors. Fifth, our sleep measurement solely relied on actigraphy; data from other measurement methods such as self-report sleep diaries were not available. Researchers recommend measuring sleep with multiple methods (Buysse, Ancoli-Israel, Edinger, Lichstein, & Morin, 2006); therefore, our results should be replicated utilizing other measurement approaches. Previous studies have suggested that excessive negative cognitive activities contribute to misperception of sleep (e.g., estimating SOL longer than the physiologically measured length); this is one of the cognitive characteristics of psychophysiological insomnia (Harvey & Tang, 2012; Tang & Harvey, 2004). Thus, if we had evaluated sleep quality using self-report measures, repetitive thought may have been associated with longer SOL and shorter TST than those observed in the present study. Sixth, this study includes methodological problems similar to those of other ESM studies: self-selection bias, attrition, and response bias on self-report questionnaires (Scollon, Kim-Prieto, & Diener, 2003). Participants who completed the study may have been more motivated and agreeable than those who decided not to participate or who dropped out of the study. In addition, our online reporting of thought content may have presented some of the same problems associated with self-report questionnaires such as social desirability, cognitive biases, and cultural norms, although retrospective recall bias was eliminated (Scollon et al., 2003). Seventh, as mentioned above, our measurement of self-focus tapped both aspects of negative valence and self-focused attention as it included participants’ focus on their problems as well as moods and personality. Although this conceptualization of selffocus is in line with previous theoretical and empirical studies (see Footnote 2), it is skewed toward negative self-focus. Thus, it will be important to replicate our results with measures that assess pure self-focus without contamination by negative emotions. Eighth, we did not find any significant influence of physical activity on sleep qualities, contrary to previous findings that suggest that acute exercise has a small-to-moderate effect on sleep parameters. One meta-analytic study identified various moderating variables that influenced these associations; for example, exercise performed in the late afternoon

80

takano et al.

enhanced sleep, whereas exercise close to bedtime decreased sleep quality (Youngstedt et al., 1997). In the present study, we controlled for level of physical activity occurring within 15 minutes of each evening assessment in our prediction of nighttime sleep, because it was expected that sleep could be impaired by increased physical activity. However, we did not find any influence of physical activity on sleep, possibly because the proximity between evening physical activity and sleep onset was not assessed systematically across individuals, and we did not measure the intensity and duration of physical activity, which are also potential moderators (Youngstedt et al., 1997). Another possible reason is that our ESM assessment focused on the momentary activities occurring when the ESM probe was emitted. Thus, our measurements did not capture activities over the whole sampling period; there were considerable periods of time when activities were not specified. Notwithstanding the above limitations, the present study contributes to our understanding of the role of negative cognition in sleep disturbance. Our data provide initial evidence regarding the deleterious effect of evening repetitive thought on nighttime sleep quality using the ESM paradigm and actigraphy for long-term sleep monitoring, which ensures ecological validity and reliability in observing the flow of thinking and state of sleep in real-world settings. These results highlight the importance of psychological factors in maintaining and enhancing goodquality sleep in everyday life. Conflict of Interest Statement The authors declare that there are no conflicts of interest.

References Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotion-regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review, 30, 217–237. http://dx.doi.org/10.1016/j.cpr.2009.11.004 Bower, B., Bylsma, L. M., Morris, B. H., & Rottenberg, J. (2010). Poor reported sleep quality predicts low positive affect in daily life among healthy and mood-disordered persons. Journal of Sleep Research, 19, 323–332. http://dx.doi.org/10.1111/ j.1365-2869.2009.00816.x Breslau, N., Roth, T., Rosenthal, L., & Andreski, P. (1996). Sleep disturbance and psychiatric disorders: A longitudinal epidemiological study of young adults. Biological Psychiatry, 15, 411–418. http://dx.doi.org/10.1016/0006-3223(95)00188-3 Brosschot, J. F., Van Dijk, E., & Thayer, J. F. (2007). Daily worry is related to low heart rate variability during waking and the subsequent nocturnal sleep period. International Journal of Psychophysiology, 63, 39–47. http://dx.doi.org/10.1016/ j.ijpsycho.2006.07.016 Brown, F. C., Buboltz, W. C., & Soper, B. (2002). Relationship of sleep hygiene awareness, sleep hygiene practices, and sleep quality in university students. Behavioral Medicine, 28, 33–38. http://dx.doi.org/10.1080/08964280209596396

Buysse, D. J., Ancoli-Israel, S., Edinger, J. D., Lichstein, K. L., & Morin, C. M. (2006). Recommendations for a standard research assessment of insomnia. Sleep, 29, 1155–1173. Buysse, D. J., Thompson, W., Scott, J., Franzen, P. L., Germain, A., Hall, M., ... Kupfer, D. J. (2007). Daytime symptoms in primary insomnia: A prospective analysis using ecological momentary assessment. Sleep Medicine, 8, 198–208. http://dx.doi.org/10.1016/j.sleep.2006.10.006 Carney, C. E., Edinger, J. D., Meyer, B., Lindman, L., & Istre, T. (2006). Symptom-focused rumination and sleep disturbance. Behavioral Sleep Medicine, 4, 228–241. http://dx.doi.org/ 10.1207/s15402010bsm0404_3 Carney, C. E., Harris, A. L., Moss, T., & Edinger, J. D. (2010). Distinguishing rumination from worry in clinical insomnia. Behaviour Research and Therapy, 48, 540–546. http://dx.doi.org/10.1016/j.brat.2010.03.004 Carver, C. S., & Scheier, M. F. (1990). Origins and functions of positive and negative affect: A control-process view. Psychological Review, 97, 19–35. http://dx.doi.org/10.1037/0033295X.97.1.19 Cheng, S. H., Shih, C. C., Lee, I. H., Hou, Y. W., Chen, K. C., Chen, K. T., . . . Yang, Y. C. (2012). A study on the sleep quality of incoming university students. Psychiatry Research, 30, 270–274. http://dx.doi.org/10.1016/j.psychres. 2011.08.011 Cole, A. D., & Maxwell, S. E. (2003). Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling. Journal of Abnormal Psychology, 112, 558–577. http://dx.doi.org/10.1037/0021843X.112.4.558 Cole, R. J., Kripke, D. F., Gruen, W., Mullaney, D. J., & Gillin, J. C. (1992). Automatic sleep/wake identification from wrist activity. Sleep, 15, 461–469. Csikszentmihalyi, M., & Larson, R. (1987). Validity and reliability of the experience sampling method. Journal of Nervous and Mental Disease, 175, 526–536. Dickson, K. S., Ciesla, J. A., & Reilly, L. C. (2012). Rumination, worry, cognitive avoidance, and behavioral avoidance: Examination of temporal effects. Behavior Therapy, 43, 629–640. http://dx.doi.org/10.1016/j.beth.2011.11.002 Dinges, D. F., Pack, F., Williams, K., Gillen, K. A., Powell, J. W., Ott, G. E., . . . Pack, A. I. (1997). Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep, 20, 267–277. Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive consequences of sleep deprivation. Seminars in Neurology, 25, 117–129. http://dx.doi.org/10.1055/s-2005-867080 Espie, C. A., Broomfield, N. M., MacMahon, K. M., Macphee, L. M., & Taylor, L. M. (2006). The attention-intentioneffort pathway in the development of psychophysiologic insomnia: A theoretical review. Sleep Medicine Reviews, 10, 215–245. http://dx.doi.org/10.1016/j.smrv.2006.03.002 Fernandez-Mendoza, J., Vela-Bueno, A., Vgontzas, A. N., Ramos-Platon, M. J., Olavarrieta-Bernardio, S., Bixler, E. O., & De la Cruz-Troca, J. J. (2010). Cognitive-emotional hyperarousal as a premorbid characteristic of individuals vulnerable to insomnia. Psychosomatic Medicine, 72, 397–403. http://dx.doi.org/10.1097/PSY.0b013e3181d75319 Gross, R. T., & Borkovec, T. D. (1982). Effects of a cognitive intrusion manipulation on the sleep-onset latency of good sleepers. Behavior Therapy, 13, 112–116. http://dx.doi.org/ 10.1016/S0005-7894(82)80054-3 Guastella, A. J., & Moulds, M. L. (2007). The impact of rumination on sleep quality following a stressful life event. Personality and Individual Differences, 42, 1151–1162. http://dx.doi.org/10.1016/j.paid.2006.04.028

repetitive thought and sleep Harvey, A. G. (2000). Pre-sleep cognitive activity in insomnia: A comparison of sleep-onset insomniacs and good sleepers. British Journal of Clinical Psychology, 39, 275–286. http://dx.doi.org/ 10.1348/014466500163284 Harvey, A. G. (2002). A cognitive model of insomnia. Behaviour Research and Therapy, 40, 869–893. http://dx.doi.org/10. 1016/S0005-7967(01)00061-4 Harvey, A. G., & Tang, N. K. (2012). (Mis)perception of sleep in insomnia: A puzzle and a resolution. Psychological Bulletin, 138, 77–101. http://dx.doi.org/10.1037/a0025730 Harvey, A. G., Tang, N. K. Y., & Browning, L. (2005). Cognitive approaches to insomnia. Clinical Psychology Review, 25, 593–611. http://dx.doi.org/10.1016/j.cpr.2005.04.005 Harvey, A. G., Watkins, E., Mansell, W., & Shafran, R. (2004). Cognitive behavioural processes across psychological disorders. Oxford, UK: Oxford University Press. Hasler, B. P., & Troxel, W. M. (2010). Couples’ nighttime sleep efficiency and concordance: Evidence for bidirectional associations with daytime relationship functioning. Psychosomatic Medicine, 72, 794–801. http://dx.doi.org/10.1097/PSY. 0b013e3181ecd08a Hox, J. J. (2010). Multilevel analysis: Techniques and applications (2 nd ed.). New York, NY: Routledge. Jarrin, D. C., McGrath, J. J., Silverstein, J. E., & Drake, C. (2013). Objective and subjective socioeconomic gradients exist for sleep quality, sleep latency, sleep duration, weekend oversleep, and daytime sleepiness in adults. Behavioral Sleep Medicine, 11, 144–158. http://dx.doi.org/10.1080/15402002.2011.636112 Jellesma, F. C., Verkuil, B., & Brosschot, J. F. (2009). Postponing worrisome thoughts in children: The effects of a postponement intervention on perseverative thoughts, emotions and somatic complaints. Social Science and Medicine, 69, 278–284. http://dx.doi.org/10.1016/j.socscimed.2009.04.031 Jones, H., Atkinson, G., Leary, A., George, K., Murphy, M., & Waterhouse, J. (2006). Reactivity of ambulatory blood pressure to physical activity varies with time of day. Hypertension, 47, 778–784. http://dx.doi.org/10.1161/ 01.HYP.0000206421.09642.b5 Just, N., & Alloy, L. B. (1997). The response styles theory of depression: Tests and an extension of the theory. Journal of Abnormal Psychology, 106, 221–229. http://dx.doi.org/ 10.1037/0021-843X.106.2.221 Kuehner, C., & Weber, I. (1999). Response to depression in unipolar depressed patients: An investigation of NolenHoeksema’s response styles theory. Psychological Medicine, 29, 1323–1333. Kushida, C. A., Chang, A., Gadkary, C., Guilleminault, C., Carrillo, O., & Dement, W. C. (2001). Comparison of actigraphic, polysomnographic, and subjective assessment of sleep parameters in sleep-disordered patients. Sleep Medicine, 2, 389–396. http://dx.doi.org/10.1016/S1389-9457(00)00098-8 Leary, A. C., Donnan, P. T., MacDonald, T. M., & Murphy, M. B. (2000). The influence of physical activity on the variability of ambulatory blood pressure. American Journal of Hypertension, 13, 1067–1073. http://dx.doi.org/10.1016/S0895-7061(00) 01186-9 Lockley, S. W., Skene, D. J., & Arendt, J. (1999). Comparison between subjective and actigraphic measurement of sleep and sleep rhythms. Journal of Sleep Research, 8, 175–183. http://dx.doi.org/10.1046/j.1365-2869.1999.00155.x Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling. Methodology, 1, 86–92. http://dx.doi.org/ 10.1027/1614-1881.1.3.86 MacMahon, K. M., Broomfield, N. M., & Espie, C. A. (2006). Attention bias for sleep-related stimuli in primary insomnia and delayed sleep phase syndrome using the dot-probe task. Sleep, 29, 1420–1427.

81

Marchetti, L. M., Biello, S. M., Broomfield, N. M., MacMahon, K. M., & Espie, C. A. (2006). Who is pre-occupied with sleep? A comparison of attention bias in people with psychophysiological insomnia, delayed sleep phase syndrome and good sleepers using the induced change blindness paradigm. Journal of Sleep Research, 15, 212–221. http://dx.doi.org/10.1111/j.1365-2869.2006.00510.x Martin, L. L., & Tesser, A. (1996). Some ruminative thoughts. In R. S. Wyer (Ed.), Ruminative thoughts: Advances in social cognition (Vol. 9, pp. 1–47). Hillsdale, NJ: Erlbaum. McCormick, B. P., Frey, G. C., Lee, C. T., Gajic, T., Stamatovic-Gajic, B., & Maksimovic, M. (2009). A pilot examination of social context and everyday physical activity among adults receiving community mental health services. Acta Physiatrica Scandinavica, 119, 243–247. http://dx.doi.org/10.1111/j.1600-0447.2008.01331.x Ministry of Internal Affairs and Communications, Japan. (2012). Numbers of subscriptions to telecommunication service. Retrieved April 30, 2013, from www.soumu.go.jp/ menu_news/s-news/01kiban03_02000096.html Moberly, N. J., & Watkins, E. R. (2008a). Ruminative self-focus and negative affect: An experience sampling study. Journal of Abnormal Psychology, 117, 314–323. http://dx.doi.org/ 10.1037/0021-843X.117.2.314 Moberly, N. J., & Watkins, E. R. (2008b). Ruminative self-focus, negative life events, and negative affect. Behaviour Research and Therapy, 46, 1034–1039. http://dx.doi.org/10.1016/ j.brat.2008.06.004 Mor, N., Doane, L. D., Adam, E. K., Mineka, S., Zinbarg, R. E., Griffith, J. W., ... Nazarian, M. (2010). Within-person variations in self-focused attention and negative affect in depression and anxiety: A diary study. Cognition and Emotion, 24, 48–62. http://dx.doi.org/10.1080/02699930802499715 Mor, N., & Winquist, J. (2002). Self-focused attention and negative affect: A meta-analysis. Psychological Bulletin, 128, 638–662. http://dx.doi.org/10.1037/0033-2909.128.4.638 Morgenthaler, T., Alessi, C., Friedman, L., Owens, J., Kapur, V., Boehlecke, B., ... Swick, T. J. (2007). Practice parameters for the use of actigraphy in the assessment of sleep and sleep disorders: An update for 2007. Sleep, 30, 519–529. Myin-Germeys, I., Oorschot, M., Collip, D., Lataster, J., Delespaul, P., & van Os, J. (2009). Experience sampling research in psychopathology: Opening the black box of daily life. Psychological Medicine, 39, 1533–1547. http://dx.doi.org/10.1017/S0033291708004947 Nolen-Hoeksema, S. (2000). The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. Journal of Abnormal Psychology, 109, 504–511. http://dx.doi.org/ 10.1037/0021-843X.109.3.504 Nolen-Hoeksema, S., & Harrell, Z. A. (2002). Rumination, depression, and alcohol use: Tests of gender differences. Journal of Cognitive Psychotherapy: An International Quarterly, 16, 391–403. http://dx.doi.org/10.1891/jcop.16.4.391.52526 Nolen-Hoeksema, S., & Morrow, J. (1991). A prospective study of depression and posttraumatic stress symptoms after a natural disaster: The 1989 Loma Prieta earthquake. Journal of Personality and Social Psychology, 61, 115–121. http://dx.doi.org/10.1037/0022-3514.61.1.115 Nolen-Hoeksema, S., Parker, L. E., & Larson, J. (1994). Ruminative coping with depressed mood following loss. Journal of Personality and Social Psychology, 67, 92–104. http://dx.doi.org/10.1037/0022-3514.67.1.92 Nolen-Hoeksema, S., Stice, E., Wade, E., & Bohon, C. (2007). Reciprocal relations between rumination and bulimic, substance abuse, and depressive symptoms in female adolescents. Journal of Abnormal Psychology, 116, 198–207. http://dx.doi.org/10.1037/0021-843X.116.1.198

82

takano et al.

Pieper, S., Brosschot, J. F., van der Leeden, R., & Thayer, J. F. (2007). Cardiac effects of momentary assessed worry episodes and stressful events. Psychosomatic Medicine, 69, 901–909. http://dx.doi.org/10.1097/PSY.0b013e31815a9230 Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. http://dx.doi.org/10.1177/ 014662167700100306 Reid, K., & Dawson, D. (1999). Correlation between wrist activity monitor and electrophysiological measures of sleep in a simulated shiftwork environment for younger and older subjects. Sleep, 22, 378–385. Sadeh, A., Hauri, P. J., Kripke, D. F., & Lavie, P. (1995). The role of actigraphy in the evaluation of sleep disorders. Sleep, 18, 288–302. Sadeh, A., Sharkey, K. M., & Carskadon, M. A. (1994). Activity-based sleep–wake identification: An empirical test of methodological issues. Sleep, 17, 201–207. Sato, A., & Yasuda, A. (2001). Development of the Japanese version of Positive and Negative Affect Schedule (PANAS) scales. Japanese Journal of Personality, 9, 138–139. Saxvig, I. W., Pallesen, S., Wilhelmsen-Langeland, A., Molde, H., & Bjorvatn, B. (2012). Prevalence and correlates of delayed sleep phase in high school students. Sleep Medicine, 13, 193–199. http://dx.doi.org/10.1016/j.sleep.2011.10.024 Schwerdtfeger, A., & Friedrich-Mai, P. (2009). Social interaction moderates the relationship between depressive mood and heart rate variability: Evidence from an ambulatory monitoring study. Health Psychology, 28, 501–509. http://dx.doi.org/ 10.1037/a0014664 Scollon, C. N., Kim-Prieto, C., & Diener, E. (2003). Experience sampling: Promises and pitfalls, strengths and weaknesses. Journal of Happiness Studies, 4, 5–34. Segerstrom, S. C., Stanton, A. L., Alden, L. E., & Shortridge, B. E. (2003). A multidimensional structure for repetitive thought: What’s on your mind, and how, and how much? Journal of Personality and Social Psychology, 85, 909–921. http://dx.doi.org/10.1037/0022-3514.85.5.909 Shima, S., Shikano, T., Kitamura, T., & Asai, M. (1985). New self-rating scales for depression. Clinical Psychiatry, 27, 717–723. Shrout, P. E., & Lane, S. P. (2012). Psychometrics. In M. R. Mehl, & T. S. Conner (Eds.), Handbook of research methods for studying daily life (pp. 302–320). New York, NY: Guilford Press. Smith, J. D., Borckardt, J. J., & Nash, M. R. (2012). Inferential precision in single-case time-series data streams: How well does the EM procedure perform when missing observations occur in autocorrelated data? Behavior Therapy, 43, 679–685. http://dx.doi.org/10.1016/j.beth.2011.10.001 Takano, K., Iijima, Y., & Tanno, Y. (2012). Repetitive thought and self-reported sleep disturbance. Behavior Therapy, 43, 779–789. http://dx.doi.org/10.1016/j.beth.2012.04.002 Takano, K., Sakamoto, S., & Tanno, Y. (in press). Ruminative self-focus in daily life: Associations with daily activities and depressive symptoms. Emotion. http://dx.doi.org/10.1037/ a0031867 Takano, K., & Tanno, Y. (2010). Concreteness of thinking and self-focus. Consciousness and Cognition, 19, 419–425. http://dx.doi.org/10.1016/j.concog.2009.11.010

Takano, K., & Tanno, Y. (2011). Diurnal variation in rumination. Emotion, 11, 1046–1058. http://dx.doi.org/ 10.1037/a0022757 Talbot, L. S., Hairston, I. S., Eidelman, P., Gruber, J., & Harvey, A. G. (2009). The effect of mood on sleep onset latency and REM sleep in interepisode bipolar disorder. Journal of Abnormal Psychology, 118, 448–458. http://dx.doi.org/10.1037/a0016605 Tang, N. K. Y., & Harvey, A. G. (2004). Effects of cognitive arousal and physiological arousal on sleep perception. Sleep, 27, 69–78. Taylor, L. M., Espie, C. A., & White, C. A. (2003). Attentional bias in people with acute versus persistent insomnia secondary to cancer. Behavioral Sleep Medicine, 1, 200–212. http:// dx.doi.org/10.1207/S15402010BSM0104_3 Thomsen, D. K., Mehlsen, M. Y., Christensen, S., & Zachariae, R. (2003). Rumination-relationship with negative mood and sleep quality. Personality and Individual Differences, 34, 1293–1301. http://dx.doi.org/10.1016/S0191-8869(02) 00120-4 Watkins, E. (2011). Dysregulation in level of goal and action identification across psychological disorders. Clinical Psychology Review, 31, 260–278. http://dx.doi.org/10.1016/ j.cpr.2010.05.004 Watkins, E. R. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134, 163–206. http://dx.doi.org/10.1037/0033-2909.134.2.163 Watson, D., Clark, I. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. http://dx.doi.org/10.1037/ 0022-3514.54.6.1063 Watts, F. N., Coyle, K., & East, M. P. (1994). The contribution of worry to insomnia. British Journal of Clinical Psychology, 33, 211–220. http://dx.doi.org/10.1111/j.2044-8260. 1994.tb01115.x Wicklow, A., & Espie, C. A. (2000). Intrusive thoughts and their relationship to actigraphic measurement of sleep: Towards a cognitive model of insomnia. Behaviour Research and Therapy, 38, 679–694. http://dx.doi.org/10.1016/S0005-7967(99) 00136-9 Wuyts, J., De Valck, E., Vandekerckhove, M., Pattyn, N., Bulckaert, A., Berckmans, D., ... Cluydts, R. (2012). The influence of pre-sleep cognitive arousal on sleep onset processes. International Journal of Psychophysiology, 83, 8–15. http://dx.doi.org/10.1016/j. ijpsycho.2011.09.016 Youngstedt, S. D., O’Connor, P. J., & Dishman, R. K. (1997). The effects of acute exercise on sleep: A quantitative synthesis. Sleep, 20, 203–214. Zoccola, P. M., Dickerson, S. S., & Lam, S. (2009). Rumination predicts longer sleep onset latency after an acute psychosocial stressor. Psychosomatic Medicine, 71, 771–775. http://dx.doi.org/10.1097/PSY.0b013e 3181ae58e8

R E C E I V E D : January 11, 2013 A C C E P T E D : September 18, 2013 Available online 27 September 2013