Tired,angry,and unhappy with us: Poor sleep quality predicts increased anger and worsened perceptions of relationship quality

Abstract

Existing research demonstrates that poor sleep is associated with lower perceptions of relationship quality. Poor sleep also predicts more intense experiences of negative affect, anger in particular. Greater anger is also tied to worse relationship outcomes. The current research explored the interplay among these factors across three studies: one correlational, one longitudinal, and one quasi-experiment (Total N = 695). We hypothesized that poorer sleep quality would predict worsened perceived relationship quality and increased anger. We also hypothesized that increased anger would account for the association between poorer sleep and reduced perceived relationship quality. Our hypotheses were supported.

Keywords

Affect anger relationship quality sleep

Introduction

Romantic relationships are the cornerstone of many people’s lives (e.g., Finkel et al., 2017). Perceiving one’s relationship to be high quality – characterized by satisfaction, commitment, trust, passion, and love (e.g., Fletcher et al., 2000) – predicts positive outcomes, including better emotional regulation (Beckes & Coan, 2011), healthy lifestyle (Roberson et al., 2018), and improved immune and cardiovascular functioning (e.g., Loving & Slatcher, 2013). As such, having a clear understanding of what accounts for people’s perceptions that their relationship is high, versus low, quality has been a central task of relationship scientists for decades (e.g., Berscheid, 1999; Reis, 2007).

How intrapersonal features of the people in a relationship (e.g., Feeney & Fitzgerald, 2019) and interpersonal dynamics of the relationship itself (e.g., Gottman, 2014) predict relationship perceptions is increasingly well understood. However, the impact of situational and contextual features has received less empirical attention, at least until the past decade (e.g., Finkel et al., 2017). For instance, life transitions (e.g., Alhainen et al., 2020), increased stress (Neff & Karney, 2007), and reduced socioeconomic status (Cho et al., 2020) can all contribute to people feeling more negatively about their romantic relationship. These factors, although extrinsic to the relationship, exert powerful influences on people’s overarching views regarding their romantic bonds.

One key factor that may influence perceptions of relationship quality is sleep. Sleep is often studied in the context of interpersonal relationships for several reasons. From an evolutionary perspective, sleeping alongside trusted others is believed to have served a protective function for our ancestors, while for current day co-sleeping couples, the sleep of one partner may directly affect the sleep of the other (Richter et al., 2016; Troxel, 2010). In addition, sleep and relationship problems often co-exist. For instance, sleep problems in one partner are associated with decreased marital happiness in the other (Troxel et al., 2007), and romantic couples report more conflict and negative affect on days after one of them has slept poorly (Gordon & Chen, 2014).

Although a growing body of research consistently supports a positive link between quality of sleep and romantic relationship satisfaction (Gordon et al., 2021), data regarding the processes by which the two are related are mixed. Few studies have examined the potential mechanisms connecting the two or utilized experimental designs. The current research sought to address several of these issues across three studies. We first tested the hypothesis that poor sleep quality would predict worse perceptions of relationship quality. We then expanded on this idea to examine the temporal associations between sleep and relationship quality across time. We also examined increased experiences of anger as one mechanism that accounts for the association between sleep and perceived relationship quality. Our general prediction was that poor sleep quality can reduce people’s perceptions that their romantic relationship is high quality through the mechanism of increased anger.

Sleep quality

An important dimension of sleep health is sleep quality, or subjective satisfaction with one’s sleep (Buysse, 2014). Poor sleep quality can encompass a variety of complaints, from minor sleep disturbances to a formal insomnia diagnosis, but generally includes dissatisfaction with sleep continuity or depth (Grandner, 2019). Indeed, endorsement of at least one sleep complaint (i.e., difficulty falling asleep) is relatively common among American adults, with prevalence estimates of about 15%–25% depending on the type of assessment (Grandner, 2019).

Quality sleep plays an important role in people’s cognitive and affective functioning, which may, in turn, contribute to maintaining optimal relationships with others. For instance, sleep deprivation can result in worsened problem-solving skills, decreases in empathic understanding, poor affect regulation, decreases in positive emotions, and increases in negative emotions (e.g., Palmer & Alfano, 2016; Van Der Helm et al., 2010; Yoo et al., 2007). Beyond sleep deprivation, more subtle and naturally occurring shifts in sleep quality are also associated with changes in affective functioning. Poor sleep quality correlates strongly with depression and anxiety in both clinical and non-clinical populations (Tsuno et al., 2005), and may also predict increases in mental health problems over time (O’Leary et al., 2017). Although daily mood and sleep tend to be related in a bidirectional manner, several prospective diary studies show that nightly fluctuations in sleep quality are stronger predictors of next-day dips in mood or affect, rather than vice versa (Simor et al., 2015; Triantafillou et al., 2019).

Central to the current research, those who chronically experience poor sleep are particularly susceptible to experiences of anger (Hisler & Krizan, 2017). Anger is a discrete emotion that arises in response to a real or perceived threat, during which the amygdala responds to activate the sympathetic nervous system. Lack of sleep has been shown to increase amygdala reactivity when experiencing anger (Saghir et al., 2018), as well as alter amygdala reactivity to threatening stimuli via weakened connections with the prefrontal cortex; an area of the brain which can regulate/inhibit anger responses (Yoo et al., 2007). On the other hand, sufficient sleep may increase the functional connectivity of the amygdala and prefrontal cortex (Motomura et al., 2017). A 2021 meta-analysis linked sleep quality to increases on several metrics of aggression, including anger and hostility, across a variety of populations (Saghir et al., 2018; Van Veen et al., 2021). Thus, across a variety of indices, poor sleep predicts greater reactivity to anger inducing stimuli as well as enhanced mean level experiences and expressions of anger.

The current research sought to consider how enhanced anger experienced after poor sleep relates to relationship perceptions. Although poor affective regulation and heightened experiences of negative affect predict relationship outcomes (e.g., Gottman, 2014; Velotti et al., 2016), anger especially is associated with poor relationship quality (Renshaw et al., 2010). Research examining different components of negative affect found that, although the emotions of both anger and sadness predicted people’s own feelings of relationship satisfaction, only anger predicted relationship satisfaction for both members of the couple. Anger is also critical in predicting the interaction patterns in romantic couples, both during conflict and otherwise, with greater anger being associated with increased negative affect reciprocity (e.g., Slep et al., 2021). As negative affect reciprocity is one of the most robust predictors of relationship outcomes (e.g., Gottman, 2014), anger may be an especially important factor in the association between sleep and relationship quality.

Overview of the current research

The current research sought to explore the dynamics between sleep quality, anger, and relationship quality. We hypothesized that poor sleep quality would predict lower perceived relationship quality. Furthermore, we hypothesized that poor sleep quality would predict increased feelings of anger, and that these feelings of anger would account for the link between poor sleep quality and lower perceived relationship quality.

We sought to test these ideas across three studies. In Study 1, we examined our predictions in a sample of romantically involved adults using well-validated measures of our key variables. In Study 2, we examined the temporal associations between these variables by examining how changes in sleep quality over time predicted changes in feelings of anger and perceptions of relationship quality in a longitudinal study of dating and married couples. Finally, in Study 3, we experimentally induced varying affective states among romantically involved undergraduates. We predicted that poor sleepers would report decreased perceptions of relationship quality when exposed to an experimental anger induction, but not other affect inductions.

Transparency and openness

For each of the studies in this manuscript, we report how we determined our sample size, all data exclusions (if any), and all measures and manipulations for each study. All research materials, bivariate correlations between all key measures, and analytic code are available in the online supplement to this manuscript. Copies of all materials, correlations, code, as well as data for Study 1 and 3, are also available at https://osf.io/8bwxk/files/?view_only=dabf37e6cfb44cd2bbcfca9c8cf177c9. Data for Study 2 cannot be made publicly available due to privacy restrictions. Study 3 was pre-registered (https://osf.io/7q2s5/?view_only=0a55d7766dbe4dcf851f1f90082b60ea).

Study 1

Method

Participants

Participants were 209 romantically involved non-student adults (115 identified as cisgender female, 93 identified as cisgender male, and 1 preferred not to say), recruited from the United States and United Kingdom via Prolific.com. Drawing from past literature examining both sleep and anger, as well as sleep and relationship dynamics, we determined sample size using G*Power 9.1, where α = .05 and desired power = .80, with a small, estimated effect size of d = .2 for two-tailed effect with two regression predictors (e.g., Gordon & Chen, 2014; Van Veen et al., 2021). G*Power estimated that approximately 100 participants would be required to detect an effect. We aimed to collect 200 participants to account for data quality issues. A total of 211 participants ended up completing the study before the survey link was closed. Data were discarded if they were incomplete or if participants had completed the study in less than 5 minutes (half the projected completion time for the study). Incomplete data from two participants were discarded, leaving 209 participants.

Participants were an average of 43.65 years old (Med = 42, SD = 23.34). One hundred and twenty-four participants reported being in a committed dating relationship, 49 were married, 23 were engaged, and 13 reported being in a civil union or partnership. Relationship length, sexual orientation, race/ethnicity, class, and disability information were not recorded.

Procedure

Participants completed all measures in a single online session using a device of their choosing.

Measures

Sleep quality

Participants completed the Pittsburgh Sleep Quality Index (Buysse et al., 1989). This well-validated measure of sleep quality across the past month consists of 19 self-rated items (5 additional “bed-partner” rated items were omitted), yielding a global sleep quality score between zero and 21, with lower numbers indicating better quality sleep (M = 6.80, SD = 2.05, α = .87; “Over the past month, how would you rate your sleep quality overall?”; 1 = very good/4 = very bad).

Anger

Participants completed the State Hostility Scale (Anderson et al., 1995). This 35-item measure assesses current feelings of general anger (M = 2.07, SD = .64, α = .97; “I feel angry.”; 1 = strongly disagree/5 = strongly agree).

Perceived relationship quality

Participants completed the Perceived Relationships Quality Scale (PRQS; Fletcher et al., 2000). The PRQS includes six 3-item subscales assessing participants’ love (“How much do you love your partner?), intimacy (How intimate do you feel with your partner?”), passion (“How lustful do you feel toward your partner?), trust (“How much do you trust your partner?”, satisfaction (“How satisfied are you with your partner?”, and commitment (“How committed are you to your partner?”; all 1 = not at all/7 = extremely) in their relationship.

Recent research indicates that these dimensions are all reasonable assessments of different aspects of relationship quality (Finkel et al., 2013). Indeed, these measures were all highly correlated in the current dataset; all Pearson’s r’s > .45 p’s < 001. Thus, we averaged them into a single index of perceived relationship quality (M = 5.97, SD = 1.05, α = .96).

Results

To test primary hypotheses, we ran a series of three regression models in SAS 9.4. See Figure 1. First, we predicted participants’ perceived relationship quality from their sleep quality (Model 1). Next, we predicted participants’ reported anger from their sleep quality (Model 2). Finally, we examined whether participants’ reported anger would mediate the hypothesized relationship between sleep quality and perceptions that they were in a higher, versus lower, quality relationship (Model 3). We tested mediation using a bias-corrected bootstrapping approach. All variables were standardized prior to analyses (M = 0, SD = 1) in order to ease interpretation of effect sizes. Additional versions of each model were also explored, controlling for the main effect and potential moderation of participant age and participant gender. In each of these models, the effects discussed below remained robust and not meaningfully altered. No main or moderation effects of any covariate emerged; these analyses are not discussed further.

Figure 1.

Study 1 mediational pathways.

Model 1

This model predicted participants’ perceived relationship quality from their sleep quality over the past month. Sleep quality was marginally, but not significantly associated with relationship quality (Table 1).

Table 1.

Study 1, model 1.

Effect	β	Df	t value	p value	95% CI
Intercept	.00012	207	.00	.99	−.14, .14
Sleep quality (SQ)	−.12	207	−1.79	.07	−.26, .01

Model 2

This model predicted participants’ current feelings of anger from their sleep quality over the past month. As predicted, sleep quality was associated with general anger, such that worse sleep quality predicted greater feelings of anger (Table 2).

Table 2.

Study 1, model 2.

Effect	β	Df	t value	p value	95% CI
Intercept	.0002	207	.00	.99	−.13, .13
SQ	.37	207	5.72	<.001	.24, .50

Model 3

This model examined whether the association between perceived relationship quality and sleep quality was mediated by feelings of anger. Thus, we conducted a regression predicting perceived relationship quality from anger and sleep quality, simultaneously (Table 3).

In this model, greater anger was associated with more negative relationship quality. Furthermore, the direct association between sleep quality and perceived relationship quality was no longer marginal. A bias-corrected bootstrapping approach to test for significant mediation (based on 5,000 resamples), produced a 95% confidence interval that did not contain zero (95% CI −.002, −.09). Thus, anger mediated the direct association between sleep quality and perceived relationship quality.

Table 3.

Study 1, model 3.

Effect	β	Df	t value	p value	95% CI
Intercept	.0007	207	.00	.99	−.13,.13
Anger	−.24	207	−3.23	.001	−.38, −.09
SQ	−.04	207	−.50	.62	−.18, .11

Study 2

The findings from Study 1 supported our prediction that poor sleep would be associated with worse relationship quality perceptions through the mechanism of increased anger. That said, Study 1’s data were cross-sectional, which limits the conclusions that can be drawn regarding mediation. Study 2 addressed this limitation by examining temporal changes in our variables of interest. Study 2 also investigated several other nuances of our predictions including the cyclical nature of poor sleep and negative relationship outcomes, the dyadic nature of sleep, as well as other potential affective sequelae associated with sleep.

Method

Participants

Participants consisted of two subsamples recruited from the Chicago metro area, who experienced nearly identical study procedures (N = 134 couples). Within the larger study, one subsample consisted of both members of 75 undergraduate dating couples, that consisted of one female identifying and one male identifying member. The second subsample consisted of both members of 59 married couples, also consisting of one partner who identified as female and one who identified as male. Both sets of participants took part in the current study in 2009 and were then followed longitudinally over time.¹ Sample size and measures were pre-determined based on theoretical considerations and practical concerns. Although we were unable to conduct a priori power calculations, given the effects from Study 1, a sample of 268 individuals nested within couples should be sufficient to detect small effects of d = .2, with α = .05 and a desired power of .80.

Among the dating couples, one couple was excluded from analyses as they did not meet initial inclusion criteria (final subsample n = 148 individuals). Participants were an average of 20.46 years old (Med = 19, SD = 1.71), 67.6% White (2.01% Black, 24.0% Asian, 4.51% Latinx, 1.13% Native American, .75% Other), were currently enrolled as students at a local university, and had been romantically involved in their relationship for an average of 16.80 months (SD = 13.73). Disability status information was not collected.

Participants in the married subsample were an average of 38.75 years old (Med = 35.5, SD = 13.19), were 84.75% White (4.24% Black, 5.08% Asian, 4.24% Latinx, 1% Native American, .7% Other), were highly educated (48.3% had undergraduate degrees, 40.68% had graduate degrees, 10.02% had high school diplomas), were living in the local area, and had been romantically involved in their current relationship for an average of 159.79 months (SD = 142.58). Disability status information was not collected.

No married couples divorced, and only 14 of the dating couples broke up over the course of the study. For couples who broke up, their data were not used in any analyses after the reported end of their relationship. These participants did not vary meaningfully from those whose relationships remained intact in terms of either demographics or key study measures. Additionally, the married subsample represents approximately half of 120 married couples that were originally recruited (e.g., Finkel et al., 2013). The other 61 couples were excluded from the current analyses as they received an intervention treatment halfway through the study aimed at bolstering relationship quality, the outcome variable of interest in the current research (Finkel et al., 2013). The 59 couples included here were the control group and did not receive the intervention.

Procedure

Participants in both subsamples completed measures during an online intake session. Relevant to the current study, participants completed measures of demographics at this time. Participants in both subsamples completed a series of 6 follow-up assessments, which included measures reflecting on their past month’s experience of sleep quality, general anger, and relationship quality at each session. For the dating subsample, assessment waves occurred monthly for 6 months; for the married subsample, assessment waves occurred every 4 months for 2 years. This variation in assessment timing was the only procedural difference between the two subsamples.

For both subsamples, participants’ whose relationships ended during the course of the study continued to provide data for the larger study; however, they were no longer included in the current analyses from that assessment forward.

Measures

Sleep quality

At each follow-up assessment, participants in both subsamples completed a 1-item measure on sleep quality, taken from the Pittsburgh Sleep Quality scale, (M_{acrossassessments} = 3.50, SD = .99; “During the past month, how would you rate your overall sleep quality?”; 0 = extremely poor/5 = excellent; adapted from Buysse et al., 1989). They also completed a 1-item measure assessing their average sleep quantity (Range = 0–12, M_{acrossassessments} = 7.06, SD = 1.13; “During the past month, how many hours have you slept per night, on average?”). Although the primary interest in this study was perceptions of sleep quality, this measure was included for use as a covariate in auxiliary analyses.

Anger

At each follow-up assessment, participants in both subsamples also completed a novel 3-item measure of their recent feelings of anger. The key measure of interest in the present study was designed to be a general one, tapping specifically into anger that was not directed toward participants’ partners (M_{acrossassessments} = 2.94, SD = 1.31; α = .95; “I feel angry with things in my life (other than my romantic partner) these days.”; 1 = not at all/7 = extremely).

There was also an analogous 3-item measure assessing recent feelings of anger toward the partner (M_{acrossassessments} = 2.94, SD = 1.31; α = .95; “I feel angry with my partner these days.”). This was not the primary measurement of anger in the present work. We were concerned about conceptual overlap between high levels of partner directed anger and low reports of relationship quality (see the bivariate correlations table in the supplementary materials). Thus, general experiences of anger and partner-specific experiences of anger were separated for analysis. General anger was analyzed as the primary predictor and partner-directed anger as a covariate in auxiliary analyses.

Perceived relationship quality

At each follow-up assessment, participants in both subsamples completed the same well-validated measure of their perceptions of relationship quality used in Study 1. Specifically, they completed an abbreviated version of the Perceived Relationships Quality Scale (PRQS; Fletcher et al., 2000). The abbreviated PRQS included four of the previously employed 3-item subscales assessing participants’ love, intimacy, passion, and trust in their relationship. Commitment (“I am highly committed to maintaining my relationship with my partner.”) and satisfaction (“I feel satisfied with our relationship; both 1 = not at all/7 = extremely) were assessed via the Investment Model Scale, so those subscales were omitted from the PRQS. As in Study 1, these measures were all highly correlated (Pearson’s r’s > .43 p’s < 001); thus, we averaged them into a single index of relationship quality (M_{acrossassessments} = 5.93, SD = .96; α = .94).

Results

Analytic strategy

Data from the dating and married subsamples were examined in a single dataset.² We employed multi-level modeling a using maximum likelihood approach, given the size of our sample, in SAS 9.4 to account for the non-independence in the data (partners nested within 134 couples across time, Kashy & Cook, 2020; Snijders & Bosker, 2012). Due to variability in time between follow-up assessments in the data, we estimated both random intercepts and random slopes in our models for time. We did not estimate random effects for our key predictor variables, as our predictions focused on describing the average tendencies within our data, rather than variability. Our couples were distinguishable dyads based on identified gender as noted above. All variables were standardized prior to analyses (M = 0, SD = 1) to ease interpretation of effect sizes.

To test the primary hypotheses, we utilized a regressed change approach (Cohen et al., 2003) within our analysis. This approach predicts the current assessment’s outcome variable from the current assessment’s predictor variables, as well as the previous assessment’s outcome and predictor variables. The inclusion of these covariates removes all correlations between current and previous assessments’ outcomes and predictor variables (Cohen et al., 2003), allowing conclusions regarding assessment-to-assessment change. This approach is commonly used to study changes over time in relational variables. Thus, this study explored changes in sleep quality from one assessment to the next predicting changes in perceived relationship quality from one assessment to the next (Model 1), and changes in general anger from one assessment to the next (Model 2). For example, if a person reported sleeping worse this assessment compared to last assessment, it was predicted that they would perceive their relationship less positively and would feel more anger, in general, at this assessment compared to last assessment.

Finally, this study also investigated whether changes in general anger across assessments would mediate the proposed association between changes in sleep quality and changes in perceived relationship quality (Model 3). In order to examine the proposed mediational pathway in the nested data, we employed the MEDMC web utility for testing mediation in multi-level models hosted by Selig & Preacher (2008; http://quantpsy.org/medmc/medmc.htm). The MEDMC utility uses a maximum likelihood approach, with Monte Carlo 95% confidence intervals, based on 10,000 resamples.

For all models discussed below, each analysis also was run separately for the dating and married subsamples; effects did not vary substantively. Additional versions of each model were also explored, controlling for the main effect and potential moderation of variables: relationship length, participant age, and participant gender. In each of these models, the effects discussed below remained robust and not meaningfully altered. No moderation effects of any covariate emerged; these analyses are not discussed further.

Primary results

See Figure 2.

Figure 2.

Study 2 mediational pathways.

Model 1

This model predicted changes in perceived relationship quality between assessments from changes in sleep quality between assessments. As predicted, changes in sleep quality were associated with changes in perceived relationship quality, such that worse sleep quality predicted more negative perceptions of relationship quality (Table 4).³

Table 4.

Study 2, model 1.

Effect	β	Df	t value	p value	95% CI
Intercept	.17	104	.91	.36	−.04, .12
Current SQ	.07	213	4.21	<.001	.03, .09
Previous SQ	.01	128	1.34	.18	−.01, .05
Previous perceived relationship quality (PRQ)	−.03	246	−1.58	.11	−.06, .01

Model 2

This model predicted changes in general anger between assessments from changes in sleep quality between assessments. As predicted, changes in sleep quality were associated with changes in anger, such that worse sleep quality predicted increased feelings of anger (Table 5).⁴

Table 5.

Study 2, model 2.

Effect	β	Df	t value	p value	95% CI
Intercept	.004	101	.08	.93	−.09, .10
Current SQ	−.17	213	−7.89	<.001	−.21, −.13
Previous SQ	.001	185	.09	.93	−.04, .04
Previous anger (A)	.05	215	2.28	.02	.01, .09

Model 3

This model examined whether the changes in perceived relationship quality associated with changes in sleep quality were mediated by changes in anger. Thus, we conducted a regression predicting changes in perceived relationship quality from changes in anger and changes in sleep quality, simultaneously (Table 6).

Table 6.

Study 2, model 3.

Effect	β	Df	t value	p value	95% CI
Intercept	.15	102	3.14	.002	.05, .24
Current SQ	.04	211	2.48	.01	−.01, .07
Previous SQ	.01	156	.88	.38	−.02, .04
Current A	−.23	252	−13.68	<.001	−.26, −.19
Previous A	−.03	212	−2.46	.01	−.07, −.01
Previous PRQ	.04	276	2.10	.04	.002, .07

In this mediational model, as predicted, increased anger was associated with more negative perceptions of relationship quality. Furthermore, the direct association between changes in sleep quality and changes in perceived relationship quality was no longer statistically significant. Employing the Monte Carlo approach to test for significant mediation as outlined above (a maximum likelihood approach based on 10,000 resamples), produced a 95% confidence interval for the indirect effect that did not contain zero (95% CI -.03, −.005). Thus, changes in anger mediated the direct association between changes in sleep quality and changes in perceived relationship quality.

One issue with our primary mediation is that it does not disentangle effects in our models that exist within-person versus between-person (Zhang et al., 2009). As such, we re-ran our primary models using the SPSS macro MLMED 2.0 (Rockwood & Hayes, 2017), which estimates both the within and between person effects for Models 2 and 3 in our primary analyses. Due to the intricate nature of our nesting (people within couples within time), we had to run two separate models – one for male identified partners and one for female identified partners – as MLMED 2.0 can only handle 2-level data structures. We did not predict differences based on gender, this simply allowed us to simplify our data structure for analysis. This approach has been used in previous research to disentangle within and between person effects in complex multi-level datasets (Carswell, et al., 2021).

The test for mediation, again based on a Monte Carlo bootstrapping approach as in our primary models, indicated that the pathway between anger and relationship quality was only significant at the within-person level for both male and female identified partners in our dataset. Furthermore, changes in anger significantly mediated the association between changes in sleep and changes in relationship quality at the within-person (95% CI_male −.01, −.001/95% CI_female −.01, −.003), but not the between-person (95% CI_male −.04,.13/95% CI_female −.02, .12) level. Based on these effects, our hypotheses were supported, such that people who experienced worsening changes in their sleep quality across a given month, compared to a different monthly assessment for the same individuals, experienced increasing anger and, thus, reductions in their perceived relationship quality. Our mediation pathway thus seems to be more about fluctuations within people, rather than differences across people.

Auxiliary results

Three additional, exploratory analyses were run to examine the more nuanced aspects of the association between sleep quality and relationship quality. The general analytic strategy for these additional analyses was the same as for the primary models.

Temporal effects of sleep and perceived relationship quality

Many researchers suggest a reciprocal relationship between sleep quality and relationship outcomes (Hasler & Troxel, 2010). As such, alternate versions of the current study’s three primary models were assessed utilizing sleep quality as the outcome and perceived relationship quality as the predictor.

We first examined whether changes in perceived relationship quality predicted changes in sleep quality (Model 1). Perhaps unsurprisingly, monthly changes in perceived relationship quality were associated with changes in sleep such that greater relationship quality predicted better sleep, β = .15, t (229) = 4.58, p < .001, 95% CI (.08, .21).

We next examined whether changes in anger predicted changes in sleep quality (Model 2). Monthly changes in anger were also associated with changes in sleep, such that increases in anger predicted worse sleep quality, β = −.21, t (246) = -8.52, p < .001, 95% CI (−.27, −.17). These significant reversals of the temporal patterning of our primary effects provide support for the cyclical relationship between sleep and emotional reactivity.

Despite these associations, we found that changes in anger did not mediate the association between changes in perceived relationship quality and changes in sleep (Model 3). The direct association between changes in sleep quality and changes in perceived relationship quality was still marginally significant, β = .06, t (224) = 1.83, p = .07, 95% CI (−.004, 0.13), and the Monte Carlo analysis for significant mediation revealed a 95% confidence interval for the indirect effect that contained zero (95% CI -.06, .008). This lack of mediation demonstrates that poorer sleep over time predicts decreases in perceived relationship quality through the mediating factor of increased anger (i.e., our primary analyses tested above); however, poorer perceived relationship quality over time does not predict decreases in sleep quality through the same pathway.

Sleep as a dyadic process

The sleep quality of individuals in relationships is dyadic in nature (Gunn et al., 2014). If one partner snores, for example, that person might disrupt not only their own sleep, but their partner’s sleep as well. Therefore, our second auxiliary analysis examined the dyadic processes at play between sleep and relationship quality from the perspective of the Actor-Partner Interdependence Model (APIM; Kenny et al., 2020).

For these APIM models, we predicted monthly changes in perceived relationship quality from changes in own sleep quality (the actor effect), changes in partner sleep quality (the partner effect), and the interaction between the two (the relationship effect). Effectively, we repeated our primary analysis (Model 1), including both partners’ current and previous months sleep quality and their interactions.

Results were consistent with our primary Model 1 and indicated that monthly changes in one’s own sleep quality marginally predicted monthly changes in one’s perceptions of relationship quality, such that poorer sleep quality predicted decreases in perceived relationship quality, β = .08, t (367) = 1.77, p = .07, 95% CI (−.004, 0.07). Changes in partner’s sleep quality, β = .04, t (251) = 1.52, p = .10, 95% CI (−.02, 0.18) was also marginally associated with relationship quality; however, the interaction between own and partner’s sleep quality was not, β = −.01, t (338) = -.75, p = .45, 95% CI (−.04, 0.02). Based on these effects, it seems that a person’s own sleep quality is more closely tied to their perceptions of their relationship than their partner’s sleep quality.

The role of other affective states

Given the role anger plays in the association between sleep quality and perceived relationship quality, it seemed important to test whether or not this mediation might generalize to other negative affective states as well, specifically depression. Further, poor sleep quality has been associated with increased risk of depression (Ford & Kamerow, 1989). As such, non-clinical depressive symptomology (Straus et al., 1999, M_{acrossfollowups} = 2.19, SD = .95, α = .88) was tested as an alternative mediator in our model. We predicted that depressive symptoms would not mediate the association between our key variables due to findings suggesting that anger is an especially important state for determining relationship outcomes (Renshaw et al., 2010).

To test these predictions, we repeated our primary analyses examining depression, rather than anger, as a potential mediator. Specifically, this study explored whether changes in sleep predicted changes in non-clinical depressive symptomology (Model 2), as well as whether changes in depressive symptomology would mediate the association between changes in perceived relationship quality and changes in sleep (Model 3).

Consistent with previous research, monthly increases in non-clinical depressive symptomology were predicted by monthly worsening of sleep, β = −.18, t (224) = -8.65, p < .001, 95% CI (−.23, −.14). When entered as a potential mediator between sleep quality and perceived relationship quality, increases in depressive symptoms did predict poorer perceptions of relationship quality, β = −.27, t (249) = −9.59, p < .001, 95% CI (−.32, −.21), but the significant direct association between poorer sleep and poorer perceptions of relationship quality remained marginally significant, β = .03, t (291) = 1.57, p = .10, 95% CI (−.009, .06). The Monte Carlo estimation of the indirect effect of poorer sleep on poorer relationship quality perceptions through changes in depressive symptoms contained zero, 95% CI(−.004, .007) indicating that depressive symptoms did not mediate the association between sleep and relationship perceptions.

Furthermore, this mediational model was repeated to include both monthly changes in anger (the original predicted mediator) and changes in depressive symptomology. In this stringent analysis, changes in anger still significantly mediated the association between relationship and sleep quality (95% CI [−.01, −.007]), while changes in depressive symptoms still did not (95% CI [−.011, .012]).

Study 3

The findings from the previous studies support predictions that poor sleep would be associated with worse relationship quality perceptions through increased feelings of anger. Study 3 further expands upon these findings to examine whether anger connected sleep and relationship quality via baseline effects (i.e., I am tired so I feel more angry at baseline, and thus perceive my relationship as worse.), reactivity effects (i.e., I am tired so I am more reactive to anger-related stimuli, and thus perceive my relationship as worse.), or some hybrid of the two. Studies 1 and 2 established baseline effects for our key mechanism; however, emotions are typically experienced in response to discreet stimuli, so Study 3 aimed to begin examining whether poor sleep increased reactivity to anger inducing stimuli, in addition to the baseline effects shown in the previous studies. Thus, Study 3 included an experimental manipulation of affect. Specifically, we assessed people’s sleep quality over the previous month before exposing them to an induction of anger, distress/sadness, positive affect, or a no affect manipulation control condition. We then assessed their affective state and perceptions of relationship quality. As stated in the pre-registration for this study, we examined whether poor sleep would predict enhanced reactivity to anger-inducing stimuli, and whether greater anger (induced reactivity or baseline) would predict lower relationship quality among poorly rested, compared to well-rested, people. We predicted that poor sleep would be associated with greater anger, and that greater induced anger would interact with poor sleep to predict reduced perceptions of relationship quality. This mediation-by-moderation prediction tests our key pathways in a slightly different way than the previous studies, but still focuses on anger as a key mechanism between poor sleep and relationship quality.

Method

Participants

Participants were 218 romantically involved college students (146 who identified as cisgender females, 72 who identified as cisgender males), recruited via a departmental subject pool at a university in Southeastern Pennsylvania. We determined sample size using G*Power 9.1, where α = .05 and desired power = .80, with a medium estimated effect size of d = .5, given the strength of the findings from the previous two studies, for two regression predictors and a two-tailed effect. G*Power estimated that 150 participants would be required to detect our effects. We aimed to collect 200 participants to account for data quality issues. We collected data for three academic semesters, allowing as many qualified participants as were interested to complete the survey. At the conclusion of the third semester, we had recruited 243. Data from 25 participants were discarded due to them completing the study multiple times or failing to complete the key affect manipulation, resulting in 218 participants.

` Participants were an average of 19.31 years old (Med = 19, SD = 2.80). One hundred and seventy-six participants reported being in a committed dating relationship, 37 reported being in a casual dating relationship, 1 was married, 1 was engaged, and 3 reported being in a civil union or partnership. One hundred and four participants reported being involved in their current relationship for more than one year, 44 had been involved in their current relationship for between 6 months and a year, 37 had been involved in their current relationship for between 3 and 6 months, and 32 had been involved in their current relationship for less than three months. Information on sexual orientation, race/ethnicity, class, and disability status was not recorded.

Procedure

Participants completed all study measures in a single online session. Participants first completed all demographic and sleep quality measures before being randomly assigned to experience one of four affect manipulation conditions: anger, distress/sadness, positive affect, and a no manipulation control condition.

For each of the affect manipulations, participants were asked to respond to a hypothetical situation designed to induce the relevant affective state. Participants were asked to read the prompt, imagine their affective reaction to the hypothetical situation, and then write at least one sentence, or for at least 30 seconds, regarding their response. All participants included completed the prompt as instructed. The control condition did not complete writing tasks but completed survey measures.

After completing the affect manipulation, participants completed a measure of state affect, followed by measures of perceived relationship quality.

Materials and measures

Sleep quality

Participants completed the same 1-item measures of sleep quality (M_{acrossassessments} = 3.32, SD = .81; adapted from Buysse et al., 1989) and quantity (Range = 0–12, M_{acrossassessments} = 6.99, SD = 1.03) used in Study 2.⁵ Consistent with Study 2, the sleep quantity measure was included as a covariate as an alternate method of sleep assessment.

Affect manipulation

Participants were randomly assigned to experience one of four affect manipulation conditions: anger, distress/sadness, positive affect, or a no manipulation control. All of the conditions began with the same prompt: “Imagine you have been preparing for a final exam for several weeks. You were feeling prepared for the exam and decided to take last night (the night before the exam) off from studying to visit your best friend who was in town.”

In the anger induction condition, participants were asked to imagine that on exam day, a series of mishaps transpired (i.e., having coffee spilled on them, walking to their car only to realize they’ve left their keys in the classroom) designed to elicit inconvenience, anger, and annoyance.

In the distress/sadness condition, participants were asked to imagine several sad occurrences in their immediate circumstance (i.e., their friend recently lost a close other and COVID-19 deaths throughout the world) designed to elicit sadness and distress.

In the positive affect condition, participants were asked to imagine that on exam day, a series of positive events transpired, (i.e., their friend orders their favorite food, good weather, cheery neighbors, finding money on the ground) designed to elicit a positive state.

In each of these conditions, the full text for which can be found in the online supplemental materials, participants were asked to consider how each scenario would make them feel before writing for at least 30 seconds, or one sentence, about their response to the situation. In the no manipulation control condition, participants completed no task, just the other measures for the study.

State affect

Participants completed a measure of their current affective state, after the affect manipulation, or immediately following demographic measures for the control condition. To measure state affect, the current study utilized the Positive and Negative Affect Schedule (PANAS; Watson et al., 1988). This measure asks participants to rate a series of 20 affective terms, 10 positive and 10 negative, for how much they were feeling each at the present moment (1 = not very much, 5 = very much).

To calculate indices for different aspects of participants’ affective state, we averaged responses to the terms “hostility,” and “irritable,” to create an anger subscale (M = 1.84 SD = .84, α = .60). We averaged responses to “distressed,” and “upset,” to create a distress subscale (M = 2.17, SD = 1.00, α = .67). Finally, we averaged responses to the positive affect terms from the measure to create a positive mood subscale (interested, excited, strong, enthusiastic, proud, alert, attentive, inspired, active, determined; M = 2.91, SD = .79, α = .89). In creating these subscales, we had a mood manipulation check for each of our primary mood manipulation conditions.

Perceived relationship quality

Participants completed the same measure of perceived relationship quality used in Study 1 (Fletcher et al., 2000). As in previous studies, we averaged them into a single index of relationship quality (M = 5.94, SD = .99, α = .95).

Dispositional anger

Participants completed Buss and Perry’s (1992) widely used 29-item measure of dispositional tendencies toward anger and aggression as a covariate (e.g., “Some of my friends think I am a hothead.” 1 = extremely uncharacteristic of me, 7 = extremely characteristic of me; M = 5.94, SD = .99, α = .91).⁶

Results

To test primary hypotheses, we ran a series of ANOVA models in SPSS 27. First, we conducted a pre-registered MANOVA predicting participants’ angry, distress, and positive mood from their assigned affect manipulation condition (Manipulation Check).

Next, we conducted a pre-registered ANCOVA predicting participants’ relationship quality perceptions from their assigned mood manipulation, their reported sleep quality, and the interaction of the two (Perceived Relationship Quality). We followed up this analysis with a series of regressions in which we examined whether sleep quality predicted relationship quality in each of the mood manipulation conditions.

Next, we conducted a regression in which we predicted participants perceived relationship quality from their sleep quality, their experienced emotions (anger, distress, and positive), and sleep’s interaction with each emotion, within the anger condition. This analysis, which was not pre-registered, allowed us to examine whether our pre-registered predicted effects in the anger induction condition were due to participants’ actual experiences of anger in the study.

Finally, as an auxiliary analysis, we conducted a series of MANCOVAs in which we predicted each affective state (anger, distress, or positive) from manipulation condition, sleep quality, and their interaction (Affect Auxiliary Analysis). Please note, this analysis was not pre-registered. Essentially, we examined whether poor sleep would predict people experiencing more intense emotional states in general, as well as whether they were more reactive to our affect manipulation.

Additional versions of each model were also explored, controlling for the main effect and potential moderation of and by variables: relationship length, participant age, participant gender, and dispositional anger. In each of these models, the effects discussed below were not meaningfully altered; these analyses are not discussed further.

Manipulation check

We first conducted a MANOVA predicting anger, distress, and positivity, respectively, from our affect manipulation. We predicted that our manipulation would, in fact, result in participants experiencing more of the target affective state in the relevant condition. The main effect of condition was significant for all three affective states (Table 7).⁷

Table 7.

Study 3: Manipulation check.

Effect of manipulation on state affect	Df	F Value	p value	Partial eta squared
Anger	3, 214	2.64	.05	.036
Distress	3, 214	4.73	.003	.063
Positive affect	3, 214	6.14	<.001	.079

Planned contrasts demonstrated that, for anger, participants in the anger condition experienced more anger (M = 2.10, SD = .99) than participants in the positive affect condition (M = 1.67, SD = .75), t (214) = 2.67, p = .009, 95% CI (.11, .76), and marginally more anger than those in the control conditions (M = 1.80, SD = .79), t (214) = 1.67, p = .09, 95% CI (−.05, .66), but not more anger than participants in the distress/sadness condition (M = 1.86, SD = .78), t (214) = 1.37, p = .17, 95% CI (−.11, .59).

For state distress, participants in the distress/sadness condition experienced more state distress (M = 2.41, SD = .92) than participants in the positive affect (M = 1.92, SD = .86), t (214) = 3.08, p = .003, 95% CI (.18, .82), or control conditions (M = 1.96, SD = .96), t (214) = 2.46, p = .015, 95% CI (.09, .82), but not more distress than participants in the anger condition (M = 2.45, SD = 1.19), t (214) = -.20, p = .84, 95% CI (−.46, .38). For state positive affect, participants in the positive affect condition experienced more positive affect (M = 3.18, SD = .75) than participants in the anger (M = 2.62, SD = .83), t (214) = 3.80, p < .001, 95% CI (.27, .85), or distress/sadness conditions, (M = 2.75, SD = .65), t (214) = 3.31, p = .001, 95% CI (.17, .68), but not more positive affect than participants in the control condition (M = 3.01, SD = .83), t (214) = .23, p = .178, 95% CI (−.11, .46).

Relationship quality

To test our primary hypotheses that poor sleep would predict worse perceptions of relationship quality, and that this would be driven by the mechanism of increased anger, we next conducted an ANCOVA predicting participants’ relationship quality from their affect manipulation condition, sleep quality, and the interaction between the two (Table 8). A non-significant interaction between affect manipulation condition and reported sleep quality emerged. There were no significant main effects.

Table 8.

Study 3: Perceived relationship quality.

Effect of manipulation on perceived relationship quality	Df	F Value	p value	Partial eta squared
Manipulation	3, 209	1.96	.116	.02
SQ	1, 209	.11	.744	.001
Manipulation × SQ	3, 209	2.16	.093	.03

We next examined our planned tests of simple effects. These revealed that sleep quality was significantly related to perceived relationship quality in the anger condition, B = .42, t (47) = 2.44, p = .019, 95% CI (.07, .76), such that poorer sleep quality when induced to experience anger was associated with worse perceptions of relationship quality. Sleep quality was not significantly associated with perceived relationship quality in the distress/sadness, positive affect, or control conditions, all B’s<|.21|, p’s > .39. Thus, poor sleep predicted worse perceptions of relationship quality only among participants who were induced to experience an angry state in the laboratory. These findings extend the effects from Studies 1 and 2.

We ran a follow-up to this analysis wherein we predicted relationship quality among participants in the anger induction condition from sleep quality, experienced anger, distress, positive emotions, and each emotional state’s interaction with sleep in a single regression (Table 9). In this analysis, which was not pre-registered, a main effect of sleep quality again emerged such that worse sleep was associated with worse perceptions of relationship quality. This main effect was qualified by a significant interaction between experienced anger and sleep such that poorly rested people who experienced greater anger in the study perceived their relationships as worse. This finding is in line with our hypotheses.

Table 9.

Effect of sleep and state affect within anger induction.

Effect	β	Df	t value	p value	95% CI
Intercept	.13	47	1.11	.26	−.11, .34
Sleep quality	.31	47	2.59	.01	.07, .56
Anger	.02	47	.93	.92	−.45, .41
Distress	.05	47	.24	.81	−.38, .49
Positive affect	−.07	47	−.52	.61	−.34, .20
SQ × anger	.38	47	2.07	.045	.01, .50
SQ × distress	.57	47	3.39	.002	−.23, .92
SQ × positive affect	−.25	47	−1.90	.06	−.79, .02

There was also a significant interaction between sleep and distress. This effect was not predicted, but emerged such that those who were poorly rested and experienced greater distress perceived their relationships as worse. Positive affect did not significantly interact with sleep. These findings suggest, as in Study 2, that other affective states may also play a role in sleep quality’s ties to perceived relationship quality. Although the present research focuses on anger, these other states are deserving of greater attention in future work.

Affect auxiliary analysis

We next conducted a MANCOVA predicting anger, distress, and positivity, respectively, from our affect manipulation condition, sleep quality, and their interaction (Table 10). This analysis explored whether poor sleep predicted greater reactivity to our affect inductions. Although sleep quality did not interact with condition to predict any of the affective states, sleep quality did exert a significant main effect on all three.

Table 10.

Study 3: Affect auxilliary analysis.

Effects on state affect	Df	F Value	p value	Partial eta squared
Anger
Manipulation	3, 209	.85	.467	.012
SQ	1, 209	8.86	.003	.041
Manipulation × SQ	3, 209	.92	.433	.013
Distress
Manipulation	3, 209	2.13	.098	.030
SQ	1, 209	4.63	.032	.022
Manipulation × SQ	3, 209	.94	.424	.013
Positive affect
Manipulation	3, 209	1.03	.382	.015
SQ	1, 209	5.53	.020	.026
Manipulation × SQ	3, 209	.43	.729	.006

Further examination of these effects demonstrated that, in line with previous research, better sleep quality was associated with less anger, B = −.21, t (216) = −3.05, p = .003, 95% CI (−.35, −.08), less state distress, B = −.19, t (216) = −2.24, p = .026, 95% CI (−.35, −.02), and greater state positive affect, B = .14, t (216) = 2.09, p = .038, 95% CI (.008, .27).⁸ These findings suggest that poorly rested participants were not more reactive to affect manipulations compared to well-rested participants, as some existing research might suggest (e.g., Saghir et al., 2018; Yoo et al., 2007); however, they did experience a higher baseline of negative affect, including anger and less positive affect, which replicates other findings in the literature linking sleep and emotions (e.g., Hisler & Krizan, 2017).

Taken together, the effects of Study 3 are mixed. The pre-registered analyses provide some support that increased reactivity to anger inducing stimuli may link poor sleep and worse perceptions of relationship quality. However, poor sleep was also associated with greater baseline anger across the emotion induction conditions, suggesting that poor sleep may also simply increase anger in general.

General discussion

Poor sleep is associated with lower relationship quality (e.g., Troxel et al., 2007). Poor sleep also predicts more intense experiences of negative affect, anger in particular (e.g., Hisler & Krizan, 2017); and greater anger is also tied to worse relationship outcomes (e.g., Gottman, 2014). The current research explored the interplay among these factors across three multi-modal studies.

Our findings generally supported our hypotheses that poor sleep quality would be associated with worse perceived relationship quality and increased feelings of anger. Furthermore, increased anger accounted for the association between sleep quality and perceived relationship quality. Poor sleep predicts people feeling angrier, and these intensified feelings of anger predict more negative perceptions about their romantic relationships.

We also examined several alternative models that could account for these predicted effects. Auxiliary analyses supported existing literature showing a cyclical association between sleep and relationship quality (e.g., Troxel et al., 2007); however, increased anger was not a mediator in this case (Study 2). The current research also supported an individual, rather than dyadic, approach to examining the interplay between sleep and relationship outcomes, in addition to ruling out sadness (indexed via non-clinical depressive symptomology) as an alternative mediator of the association between poor sleep and worse perceived relationship quality (Study 2). Future research should endeavor to expand upon these exploratory findings to examine, with greater specificity, the temporal patterning of sleep-related outcomes over time, as well as the dyadic versus individual nature of various sleep—related outcomes.

Our findings also suggest that anger, rather than sadness, may be especially likely to predict worse perceived relationship quality during times of poor sleep (Studies 2/3). Our findings do suggest that sadness is an important consequence of poor sleep and may be implicated in relationship functioning in its own right, but that sadness may not be a mechanism linking poor sleep and relationship perceptions. Future research should seek to investigate these nuances. Many emotion theorists argue that anger functions to protect oneself from vulnerability (Janocha et al., 2018), which may be at odds with key romantic relationship maintenance behaviors, such as sacrifice (Zoppolat et al., 2020) and closeness (Aron, 2003). Thus, the self-protective goals associated with the experience of anger may inhibit positive relationship perceptions. Sadness on the other hand, may not mediate the association between sleep and relationship quality perceptions because its functional roots are more conducive to relationship promoting goals, such as sacrifice, compromise, and closeness (Wolpert, 2008). Further supporting this idea, in the context of relationship conflict dynamics, it is anger, not sadness, that predicts couples entering into a cycle of negative affect reciprocity which robustly predicts divorce risk (e.g., Gottman, 2014). As mentioned previously, romantic couples report increased conflict and increased negative affect during conflict after one individual has slept poorly (Gordon & Chen, 2014). People may experience both increased sadness and increased anger in the wake of poor sleep; yet, when it comes to predicting relationship quality in the wake of poor sleep, anger may play a unique role.

It is plausible that poor sleep may predispose couples to either engage in more hostile attributions, increasing the likelihood of conflict, or feel angrier during conflict, thus exacerbating existing negative patterns. Given the importance of relationship quality perceptions for both relational outcomes and general well-being (e.g., Finkel et al., 2017), understanding the influences of relationship quality and targeting intervention strategies addressing these is crucial. There are periods in a relationship where detriments to sleep quality are more likely, such as during the transition into parenthood (Da Costa et al., 2021), retirement (Alhainen et al., 2020), menopause (Polo-Kantola, 2007), or periods of increased stress (Sun Han et al., 2012). During these times when poor sleep is often unavoidable, targeting marital interventions to address anger awareness and emotional regulation may be useful (e.g., Finkel et al., 2013; Gross & Jazaieri, 2014).

Finally, although emotional regulation, specifically anger, is an important point of intervention in sleep deprived couples, it is crucial to consider improving sleep quality as critical when aiming to preserve relationship quality. For instance, focusing on improving sleep hygiene through establishing daily routines such as reducing caffeine, incorporating relaxation practices, eliminating distractions, and/or regulating bedroom temperature may all help improve sleep quality (Brown et al., 2010; Suni & Vyas, 2023). In

Strengths and limitations

Despite these interesting findings, there were several limitations in the current work. In particular, the findings from Study 3 in the current work were mixed. Future research is necessary to further distinguish whether poor sleep is associated with greater anger due to baseline increases in anger, greater reactivity to anger inducing stimuli, or both. The current work suggests that both pathways may exist and function to link sleep to relationship quality, but additional examination is necessary.

There were also some limitations in our measurements of sleep and affect across studies. Regarding our measurements of sleep quality, we utilized the PSQI in Study 1 and a 1-item measure derived from the PSQI in Studies 2 and 3. Previous work has shown that the PSQI in general, including the single-item PSQI component of “subjective sleep quality,” which we relied on here, discriminates between poor and good sleepers (Mollayeva et al., 2016). That said, physiological components of sleep quality, like sleep architecture, can only be captured using polysomnography.

Furthermore, this measurement asks participants about sleep quality over the past month. Inquiring about monthly sleep quality, rather than the previous night, may result in decreased accuracy of estimates and may not capture the impact of previous night’s sleep on day-to-day outcomes, such as those in Study 3. Future research may consider using a daily diary approach to capture sleep, and incorporating behavioral and physiological sleep measurement tools.

In terms of affect, in Study 3 the measures of anger and distress derived from the PANAS might not have effectively captured our affect of interest. Although imperfect, we view our attempts to experimentally manipulate affect as a strength of the current work and supportive of our hypothesis; however, the strength of any causal conclusions is limited.

Additionally limiting was the concurrent nature of the constructs examined. This limits the assumptions that can be made regarding temporal patterning, including the bi-directional relationship between sleep and perceptions of relationship quality. Our initial attempts to examine the bi-directional nature of sleep and relationship quality are a strength of the current work; however, longitudinal studies using varied temporal analytic approaches are important to further elucidate the nature of this association.

Finally, Study 1 participants were recruited through Prolific, which may be a limitation to data quality. We believe that our various types of samples (Prolific.co, a university and community sample, and a student sample) provide important information about our effects across different age and relationship stages. However, any given sampling technique can be prone to problematic data and/or bias.

Despite these limitations, the current research possessed many strengths that lend confidence to our interpretations of the present findings. In particular, the current sample was diverse in terms of participants’ ages, relationship characteristics, and sleep experiences. This allowed us to generalize our findings to dating and marital relationships, across a diverse sample. It is likely that the causes of poor sleep are different in college aged dating couples compared to mid-life married couples, and yet effects remained robust across both groups.

Additionally, our ability to examine changes over time represents another strength. Future research would benefit from expanding this temporal approach to examine whether chronically poor sleep quality contributes to people’s risk of breakup and/or divorce. It is possible that the decrements in people’s perceived relationship quality that are associated with poor sleep may be recoverable over the short term, as our month-to-month data suggest, but that more chronic poor sleep may predict consistent and problematic reductions in people’s perceptions of their relationship.

Conclusion

The current findings demonstrate that poor sleep quality – which is common -predicts how people experience affective states and thus view their relationships. Future research should aim to expand upon these findings as well as identify intervention points for couples that could either help them improve their sleep quality or weather the storms of poor sleep while protecting perceptions of their relationship.

Supplemental Material

Supplemental Material - Tired, angry, and unhappy with us: Poor sleep quality predicts increased anger and worsened perceptions of relationship quality

Supplemental Material for Tired, angry, and unhappy with us: Poor sleep quality predicts increased anger and worsened perceptions of relationship quality by Alexis Audigier, Sara Glass, Erica B Slotter and Elizabeth Pantesco in Journal of Social and Personal Relationships

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the data for Study 2 were taken from a larger National Science Foundation funded grant (#0719780).

Ethical approval

This research was conducted in accordance with APA ethical guidelines and was approved by the Institutional Review Boards at all relevant institutions.

Informed consent

All participants provided informed consent before taking part.

Open research statement

As part of IARR’s encouragement of open research practices, all research materials, bivariate correlations between all key measures, and analytic code for Studies 1–3 are available in the online supplement to this manuscript. Copies of all materials, correlations, code, as well as data for Study 1 and 3, are also available at https://osf.io/8bwxk/files/?view_only=dabf37e6cfb44cd2bbcfca9c8cf177c9. Data for Study 2 cannot be made publicly available due to privacy restrictions, but are available upon request from the corresponding author. Study 3 was pre-registered using Open Science Framework (). We, the authors (Ms. Audigier, Ms. Glass, Dr.’s Slotter & Pantesco) pre-registered the method, measures, and data analytic plan prior to any data collection. Any deviations from the pre-registration are clearly noted in the manuscript.

ORCID iD

Erica B. Slotter

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Alhainen

Myllyntausta

Pentti

Vahtera

Stenholm

(2020). Concurrent changes in sleep and physical activity during the transition to retirement: A prospective cohort study. Sleep Medicine, 68, 35–41. https://doi.org/10.1016/j.sleep.2019.09.009

Anderson

C. A.

Deuser

W. E.

DeNeve

K. M.

(1995). Hot temperatures, hostile affect, hostile cognition, and arousal: Tests of a general model of affective aggression. Personality and social psychology bulletin, 21(5), 434–448.

Aron

(2003). Self and close relationships. In Leary

M. R.

Tangney

J. P.

(Eds.), Handbook of self and identity (pp. 442–461). Guilford Press.

Bassett

Lupis

Gianferante

Rohleder

Wolf

(2015). Sleep quality but not sleep quantity effects on cortisol responses to acute psychosocial stress. The International Journal on the Biology of Stress, 18(6), 638. https://doi.org/10.3109/10253890.2015

Beckes

Coan

J. A.

(2011). Social baseline theory: The role of social proximity in emotion and economy of action. Social and Personality Compass, 5(12), 976–988. https://doi.org/10.1111/j.1751-9004.2011.00400.x

Berscheid

(1999). The greening of relationship science. American Psychologist, 54(4), 260–266. https://doi.org/10.1037//0003-066x.54.4.260

Brennan

K. A.

Clark

C. L.

Shaver

P. R.

(1998). Self-report measurement of adult attachment: An integrative overview. In Simpson

J. A.

Rholes

W. S.

(Eds.), Attachment theory and close relationships (pp. 46–76). Guilford Press.

Brown

Buboltz

Soper

(2010). Relationship of sleep hygiene awareness, sleep hygiene practices, and sleep quality in university students. Behavioral Medicine, 28(1), 33–39. https://doi.org/10.1080/08964280209596396

Buss

A. H.

Perry

(1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63(3), 452–459. https://doi.org/10.1037//0022-3514.63.3.452

10.

Buysse

D. J.

(2014). Sleep health: Can we define it? Does it matter? Sleep, 37(1), 9–17. https://doi.org/10.5665/sleep.3298

11.

Buysse

D. J.

Reynolds

C. F.

III Monk

T. H.

Berman

S. R.

Kupfer

D. J.

(1989). The Pittsburgh sleep quality index: A new instrument for psychiatric practice and research. Psychiatry Research, 28(2), 193–213. https://doi.org/10.1016/0165-1781(89)90047-4

12.

Carswell

Muise

Harasymchuk

Horne

Visserman

M. L.

Impett

E. A.

(2021). Growing desire or growing apart? Consequences of personal self-expansion for romantic passion. Journal of Personality and Social Psychology, 121(2), 354–377. https://doi.org/10.1037/pspi0000357

13.

Cho

Impett

Campos

Chen

Keltner

(2020). Socioeconomic inequality undermines relationship quality in romantic relationships. Journal of Social and Personal Relationships, 37(5), 1722–1742. https://doi.org/10.1177/0265407520907969

14.

Cohen

West

S. G.

Aiken

L. S.

(2003). Applied multiple regression/correlation analysis for the behavioral sciences. Lawrence Erlbaum Associates.

15.

Da Costa

Lai

Zelkowitz

(2021). A prospective study on the course of sleep distrubances in first-time fathers during the transition to parenthood. Infant Mental Health, 42(2), 1–11. https://doi.org/10.1002/imhj.21911

16.

DeWall

C. N.

Anderson

Bushman

(2011). The general aggression model: Theoretical extensions to violence. Psychology of Violence, 1(3), 245–258. https://doi.org/10.1037/a0023842

17.

Donnellan

M. B.

Oswald

F. L.

Baird

B. M.

Lucas

R. E.

(2006). The mini-IPIP scales: Tiny-yet-effective measures of the Big five factors of personality. Psychological Assessment, 18(2), 192–203. https://doi.org/10.1037/1040-3590.18.2.192

18.

Feeney

Fitzgerald

(2019). Attachment, conflict and relationship quality: Laboratory-based and clinical insights. Current Opinion in Psychology, 25, 127–131. https://doi.org/10.1016/j.copsyc.2018.04.002

19.

Finkel

E. J.

Simpson

J. A.

Eastwick

P. W.

(2017). The psychology of close relationships: Fourteen core principles. Annual Review of Psychology, 68, 383–411. https://doi.org/10.1146/annurev-psych-010416-044038

20.

Finkel

E. J.

Slotter

E. B.

Luchies

L. B.

Walton

G. M.

Gross

(2013). A brief intervention to promote conflict reappraisal preserves marital quality over time. Psychological Science, 24(8), 1595–1601. https://doi.org/10.1177/0956797612474938

21.

Fletcher

G. J.

Simpson

J. A.

Thomas

(2000). The measurement of perceived relationship quality components: A confirmatory factor analytic approach. Personality and Social Psychology Bulletin, 26(3), 340–354. https://doi.org/10.1177/0146167200265007

22.

Ford

Kamerow

(1989). Epidemiologic study of sleep disturbances and psychiatric disorders. An opportunity for prevention? JAMA, 262(11), 1479–1484. https://doi.org/10.1001/jama.262.11.1479

23.

Gordon

A. M.

Carrillo

Barnes

C. M.

(2021). Sleep and social relationships in healthy populations: A systematic review. Sleep Medicine Reviews, 57, 101428. https://doi.org/10.1016/j.smrv.2021.101428

24.

Gordon

A. M.

Chen

(2014). The role of sleep in interpersonal conflict: Do sleepless nights mean worse fights? Social Psychological and Personality Science, 5(2), 168–175. https://doi.org/10.1177/1948550613488952

25.

Gordon

A. M.

Chen

(2014). The role of sleep in interpersonal conflict: do sleepless nights mean worse fights? Social Psychological and Personality Science, 5(2), 168–175.

26.

Gottman

J. M.

(2014). What predicts divorce? The relationship between marital processes and marital outcomes. Psychology Press.

27.

Grandner

M. A.

(2019). Epidemiology of insufficient sleep and poor sleep quality. Sleep and health (pp. 11–20). Elsevier.

28.

Gross

Jazaieri

(2014). Emotion, emotion regulation, and psychopathology: An affective science perspective. Clinical Psychological Science, 2(4), 387–401. https://doi.org/10.1177/2167702614536164

29.

Gunn

H. E.

Troxel

W. M.

Hall

M. H.

Buysse

D. J.

(2014). Interpersonal distress is associated with sleep and arousal in insomnia and good sleepers. Journal of Psychosomatic Research, 76(3), 242–248. https://doi.org/10.1016/j.jpsychores.2013.11.010

30.

Hasler

Troxel

(2010). Couples’ nighttime sleep efficiency and concordance: Evidence for bidirectional associations with daytime relationship functioning. Psychosomatic Medicine, 72(8), 794–801. https://doi.org/10.1097/PSY.0b013e3181ecd08a

31.

Hisler

G. C.

Krizan

(2017). Anger tendencies and sleep: Poor anger control is associated with objectively measured sleep disruption. Journal of Research in Personality, 71, 17–26. https://doi.org/10.1016/j.jrp.2017.08.009

32.

Janocha

Shpakou

Sobieszchzanska

Kalka

Klimatckaia

Wozniak

Pilecka-Radziszewska

Malyszczak

(2018). Emotions and their cognitive and adaptive functions. Progress in Health Science, 8(1), 182–188. https://doi.org/10.5604/01.3001.0012.1149

33.

Joel

Eastwick

P. W.

Allison

C. J.

Arriaga

X. B.

Baker

Z. G.

Bar-Kalifa

Bergeron

Birnbaum

G. E.

Brock

R. L.

Brumbaugh

C. C.

Carmichael

C. L.

Chen

Clarke

Cobb

R. J.

Coolsen

M. K.

Davis

de Jong

D. C.

Debrot

DeHaas

E. C.

Wolf

(2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences of the United States of America, 117(32), 19061–19071. https://doi.org/10.1073/pnas.1917036117

34.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2020). Dyadic data analysis. Guilford press.

35.

Loving

T. J.

Slatcher

(2013). Romantic relationships and health. In Simpson

J. A.

Campbell

(Eds.), The oxford handbook of close relationships (pp. 617–737). Oxford University Press.

36.

Mollayeva

Thurairajah

Burton

Mollayeva

Shapiro

C. M.

Colantonio

(2016). The Pittsburgh sleep quality index as a screening tool for sleep dysfunction in clinical and non-clinical samples: A systematic review and meta-analysis. Sleep Medicine Reviews, 25, 52–73. https://doi.org/10.1016/j.smrv.2015.01.009

37.

Motomura

Katsunuma

Yoshimura

Mishima

(2017). Two days’ sleep debt causes mood decline during resting state via diminished amygdala-prefrontal connectivity. Sleep, 40(10), 1–8. https://doi.org/10.1093/sleep/zsx133

38.

Neff

Karney

(2007). Stress crossover in newlywed marriage: A longitudinal and dyadic perspective. Journal of Marriage and Family, 69(3), 594–607. https://doi.org/10.1111/j.1741-3737.2007.00394.x

39.

O’Leary

Bylsma

L. M.

Rottenberg

(2017). Why might poor sleep quality lead to depression? A role for emotion regulation. Cognition and Emotion, 31(8), 1698–1706. https://doi.org/10.1080/02699931.2016.1247035

40.

Palmer

Alfano

(2016). Sleep and emotion regulation: An organizing, integrative review. Sleep Medicine Reviews, 31, 6–16. https://doi.org/10.1016/j.smrv.2015.12.006

41.

Polo-Kantola

(2007). Sleep and menopause. Women's Health, 3(1), 99–106. https://doi.org/10.2217/17455057.3.1.99

42.

Reis

H. T.

(2007). Steps toward the ripening of relationship science. Personal Relationships, 14(1), 1–23. https://doi.org/10.1111/j.1475-6811.2006.00139.x

43.

Renshaw

Blais

Smith

(2010). Components of negative affectivity and marital satisfaction: The importance of actor and partner anger. Journal of Research in Personality, 44(3), 328–334. https://doi.org/10.1016/j.jrp.2010.03.005

44.

Richman

DeWall

Wolff

(2015). Avoiding affection, avoiding altruism: Why is avoidant attachment related to less helping? Personality and Individual Differences, 76, 193–197. https://doi.org/10.1016/j.paid.2014.12.018

45.

Richter

Adam

Geiss

Peter

Niklewski

(2016). Two in a bed: The influence of couple sleeping and chronotypes on relationship and sleep. An overview. Chronobiology International, 33(10), 1464–1472. https://doi.org/10.1080/07420528.2016.1220388

46.

Roberson

Shorter

Woods

Priest

(2018). How health behaviors link romantic relationship dysfunction and physical health across 20 years for middle-aged and older adults. Social Science and Medicine, 201, 18–26. https://doi.org/10.1016/j.socscimed.2018.01.037

47.

Rockwood

N. J.

Hayes

A. F.

(2017, May). MLmed: An SPSS macro for multilevel mediation and conditional process analysis. Poster presented at the annual meeting of the Association of Psychological Science (APS).

48.

Saghir

Syeda

Muhammad

Abdalla

(2018). The amygdala, sleep dept, sleep deprivation, and the emotion of anger: A possible connection? Cureus, 10(7), 1–5. https://doi.org/10.7759/cureus.2912

49.

Selig

J. P.

Preacher

K. J.

(2008). Monte Carlo method for assessing mediation: An interactive tool for creating confidence intervals for indirect effects [Computer software].

50.

Simor

Krietsch

K. N.

Köteles

McCrae

C. S.

(2015). Day-to-day variation of subjective sleep quality and emotional states among healthy university students—a 1-week prospective study. International Journal of Behavioral Medicine, 22(5), 625–634. https://doi.org/10.1007/s12529-015-9464-4

51.

Slep

A. M.

Heyman

R. E.

Lorber

M. F.

Tiberio

S. S.

Casillas

K. L.

(2021). Couples’ anger dynamics during conflict: Interpersonal anger regulation, relationship satisfaction, and intimate partner violence. Family Process, 60(4), 1280–1294. https://doi.org/10.1111/famp.12634

52.

Slotter

E. B.

Emery

L. F.

Luchies

L. B.

(2014). Me after you: Partner influence and individual effort predict rejection of self-aspects and self-concept clarity after relationship dissolution. Personality and Social Psychology Bulletin, 40(7), 831–844. https://doi.org/10.1177/0146167214528992

53.

Slotter

E. B.

Luchies

L. B.

(2013). Relationship quality promotes the desire fore closeness among distressed avoidantly attached individuals. Personal Relationships, 1–11. https://doi.org/10.1111/PERE.12015

54.

Snijders

Bosker

(2012). Multilevel analysis: An Introduction to Basic and advanced multilevel modeling. Sage Publishers.

55.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

(1999). Manual for the personal and relationships profile (PRP). University of New Hampshire, Family Research Laboratory. Available in: http://pubpages.unh.edu/∼mas2

56.

Sun Han

Kim

Shim

(2012). Stress and sleep disorder. Experimental Neurobiology, 21(4), 141–150. https://doi.org/10.5607/en.2012.21.4.141

57.

Suni

Vyas

(2023). Improve your child’s school performance with a good night’s sleep. Sleep Foundation. https://www.sleepfoundation.org/children-and-sleep/sleep-and-school-performance

58.

Triantafillou

Saeb

Lattie

E. G.

Mohr

D. C.

Kording

K. P.

(2019). Relationship between sleep quality and mood: Ecological momentary assessment study. JMIR Mental Health, 6(3), Article e12613. https://doi.org/10.2196/12613

59.

Troxel

Robles

Hall

Buysse

(2007). Marital quality and the marital bed: Examining the covariation between relationship quality and sleep. Sleep Medicine Reviews, 11(5), 389–404. https://doi.org/10.1016/j.smrv.2007.05.002

60.

Troxel

W. M.

(2010). It’s more than sex: Exploring the dyadic nature of sleep and implications for health. Psychosomatic Medicine, 72(6), 578–586. https://doi.org/10.1097/PSY.0b013e3181de7ff8

61.

Tsuno

Besset

Ritchie

(2005). Sleep and depression. The Journal of Clinical Psychiatry, 66(10), 1254–1269. https://doi.org/10.4088/jcp.v66n1008

62.

Van Der Helm

Gujar

Walker

M. P.

(2010). Sleep deprivation impairs the accurate recognition of human emotions. Sleep, 33(3), 335–342.

63.

Van Veen

M. M.

Lancel

Beijer

Remmelzwaal

Rutters

(2021). The association of sleep quality and aggression: A systematic review and meta-analysis of observational studies. Sleep Medicine Reviews, 59, 101500. https://doi.org/10.1016/j.smrv.2021.101500

64.

Velotti

Balzarotti

Tagliabue

English

Zavattini

Gross

(2016). Emotional suppression in early marriage: Actor, partner, and similarity effects on marital quality. Journal of Social and Personal Relationships, 33(3), 277–302. https://doi.org/10.1177/0265407515574466

65.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS Scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. https://doi.org/10.1037//0022-3514.54.6.1063

66.

Wolpert

(2008). Depression in an evolutionary context. Philosophy, Ethics, and Humanities in Medicine, 3(1), 8. https://doi.org/10.1186/1747-5341-3-8

67.

Yoo

Gujar

Jolesz

Walker

(2007). The human emotional brain without sleep – a prefrontal amygdala disconnect. Current Biology: CB, 17(20), 877–878. https://doi.org/10.1016/j.cub.2007.08.007

68.

Zhang

Zyphur

M. J.

Preacher

K. J.

(2009). Testing multilevel mediation using hierarchical linear models: Problems and solutions. Organizational Research Methods, 12(4), 695–719. https://doi.org/10.1177/1094428108327450

69.

Zoppolat

Visserman

Righetti

(2020). A nice surprise: Sacrifice expectations and partner appreciation in romantic relationships. Journal of Social and Personal Relationships, 37(2), 450–466. https://doi.org/10.1177/0265407519867145

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.92 MB