Marital Status,Relationship Distress,and Self-rated Health

Abstract

This paper analyzes data from a nationally representative survey of adults in the United Kingdom (Understanding Society, N = 37,253) to explore the marital status/health nexus (using categories that include a measure of relationship distress) and to assess the role that sleep problems play as a potential mediator. Findings indicate how it is not just the “form” marital status takes but also the absence or presence of relationship distress that is essential to self-rated health. We demonstrate two further findings that: (1) sleep problems act as a mediator of the link between marital status/relationship distress and self-rated health, most notably for those in cohabiting relationships with medium/high distress or who have a history of relationship loss, and (2) the mediating role of sleep problems differs for divorced men and women.

Keywords

health marital status mediation relationship distress sleep problems

A significant body of literature now points “to the importance of social relationships in affecting [health and] health behaviour” (Umberson, Crosnoe, and Reczek 2010:139). Marriage has been central to this discussion. Consistently, and across populations, evidence has shown that married people live longer, happier, and healthier lives than their unmarried counterparts. Marriage has been found to be associated with better outcomes for health conditions, such as heart disease, hypertension, and arthritis (Zhang and Hayward 2006), and men and women who get and stay married are less likely to experience depression (Koball et al. 2010).

These findings tend to be situated in terms of selection, protection, or more recently, crisis. The selection hypothesis denies a causal role for marriage, suggesting instead that better health leads to marriage (Koball et al. 2010). The protection thesis argues that marriage protects people against poor health outcomes while also promoting good health outcomes, although not all married people are “protected” to the same extent. From their analysis of (U.S.) large-scale panel survey data (linked to administrative data), Rendall et al. (2011) conclude that unmarried American men and women have significantly higher mortality than their married counterparts. They find a marriage mortality advantage both among all adults over age 25 and in a subsample ages 65 and older and also report a gender dynamic, with married men showing an additional survival premium.

Numerous conceptual and explanatory frameworks have been put forward to explain the protective link between being married and health. Umberson and Montez (2010) highlight three ways in which social ties may influence health: behavioral, psychosocial, and physiological. Behavioral explanations highlight how a spouse may “monitor, inhibit, regulate, or facilitate health behaviors in ways that promote a partner’s health” (Umberson and Montez 2010:3). Psychosocial explanations include discussion of the symbolic meaning surrounding marriage and how meanings attached to marriage and relationships with children may promote a greater sense of responsibility to “stay healthy.” Physiological explanations emphasize how supportive interactions can benefit such things as cardiovascular and immune functions.

In essence, marriage is considered protective, first, due to socioeconomic reasons; second, because it offers protection from social isolation; and third, because spouses may monitor and attempt to change their partner’s health behaviors (Umberson 1992). Gender differences, for example, are explained through suggestions that “marriage is associated with receipt of substantially more efforts to control health for men than women” and “those who attempt to control the health of others are more likely to be female than male” (Umberson 1992:907). Although these conceptual frameworks primarily focus on potential explanations of the marital status/health relationship, attention is also given to precursors. The major precursors of marital union are related to age, race, place of birth, and education (Ross and Mirowsky 2013).

The final thesis—the crisis model—suggests that marital status differences in health exist because of the short-term stresses and strains of marital/partnership dissolution rather than the protective effects of marriage per se. Although this model is the least supported empirically, some research and discussion has pointed toward its applicability. For Williams and Umberson (2004), if the protection thesis is correct, those in all other marital status categories would be expected to report poorer health than those who are married, but they found that this was not the case. Rather, continuity of marital status was most important, with those continually divorced across three waves of U.S. data (1986 to 1994) and the “never married” being similar to married individuals with respect to health. However, the authors also modify the crisis model by reporting that transitions out of marriage may, in some cases, improve health, as they may bring an end to marital stress and crisis.

Background

The Link between Marital Status and Health: Reconceptualizing Marital Status

There are several reasons for reexploring the suggested link between marital status and health. Many societies today can be considered as societies “in transition” because of changing family structures. The social contexts of the household have grown increasingly complex: for example, the proportion of nonmarried men and women ages 60 and younger who were cohabiting in Britain rose from 11% to 24% between 1986 and 2005, and in 2005, one in four children was born to cohabiting parents (Social Exclusion Task Force 2007). As well as increasing complexity in household formation, commentators suggest that we are living in a “lonely society” (Mental Health Foundation 2010) with little “belonging support.” Our modern way of life is said to have changed the way that people connect, to have isolated us from others, and with adverse health consequences (Mental Health Foundation 2010). Using the terminology of Durkheim, we are witnessing shifts in both “conjugal society” and “family society” (Besnard 2001).

Hughes and Waite (2002) offer a framework through which to consider these changing family formations and the possible impacts they have on health. The authors theorize how within-household relationships can be categorized by a series of tangible and intangible relations, which are characterized as “resources” and “demands.” It is “an individual’s perception of the balance of resources and demands that is consequential for his or her health” (Hughes and Waite 2002:4; emphasis in original). They also state that consideration needs to be given to the fact that “the qualities of relationships within households of the same structure certainly differ” (Hughes and Waite 2002:15). This echoes others, including Franklin (2009), who have called for further research on the quality of family relationships within households or who have suggested that inconsistent findings in the literature on marital status and health may be linked to confounding marital status and marital quality (Gallo et al. 2003). There is therefore a need to reconsider the relationship between legal marital status, conjugal relationships, and relationship satisfaction. This requires separate consideration of legally married and cohabiting and incorporation of measures of relationship quality.

There is also a need to “deconstruct” the unmarried group. Research on marital status and health often compares married to unmarried—a comparison that groups together considerable diversity within the category unmarried. This grouping may mask the “temporality” of relationships deemed key in crisis perspectives noted above, as some relationship forms may be “transitional” categories (for example, movement from an unhappy relationship to separation and through to divorce), while others may be more “permanent” categories (for example, never married).

The Link between Marital Status/Quality and Health: Sleep as a Potential Mediator

Sleep has been largely ignored as a potential mediator of the link between marital status/relationship quality and health. This is perhaps surprising given that the association between sleep and health is now widely recognized (Zee and Turek 2006). Quantity and quality of sleep are predictors of Type 2 Diabetes, and a decrease (or increase) in sleep duration affects all-cause mortality (Cappuccio and Miller 2010). Sleep disturbance is also embedded within discussions of depression, with the former included in diagnostic criteria for the latter (Weich 2010). Given its status as a cause of ill health, a growing body of research has focused on what, in turn, causes poor sleep. Identified factors include psychosocial work stress (Knudsen, Ducharme, and Roman 2007) and worries and pain (Dregan, Lallukka, and Armstrong 2013). Lonely individuals have also been found to “evince” poorer sleep efficiency (Cacioppo et al. 2002).

While the relationship between sleep and health is now widely acknowledged, the relationship between marital status and sleep remains underexplored. Where it has been considered, a complex picture arises. On the one hand, studies have consistently shown that married individuals report better-quality sleep. Arber, Bote, and Meadows (2009), for example, report that married individuals have higher odds of good sleep, compared to the divorced/separated. On the other hand, the shared bed has been described as a “battle-ground” (Hislop and Arber 2003b)—suggesting that being married/partnered may have a negative effect on sleep quality and quantity.

Resolving this tension, Troxel and colleagues have suggested that being married per se might not be beneficial to sleep. Although their initial analysis (Troxel, Buysse, Matthews, et al. 2010) showed that partnered women had better sleep quality than unpartnered women, this effect became nonsignificant when covariates were added to models. Of key importance was stability of the attachment relationship. Maritally happy women reported fewer sleep disturbances (Troxel et al. 2009), and those with a stable relationship history had better sleep compared with those who had lost or gained a partner over the same period (Troxel, Buysse, Hall, et al. 2010). Somewhat similarly, Cartwright and Wood (1991) noted a reduction in delta sleep in those undergoing divorce versus those who had been divorced for some time.

These bodies of literature suggest that sleep may act as a mediator between marital status/relationship quality and health. When framed within Baron and Kenny’s (1986) four steps to establishing mediation, this literature indicates that at least two steps are met: that the causal variable is correlated with the mediator (marital status to sleep) and that the mediator affects the outcome variable (sleep to health). The literature also hints at the complex role sleep may play as a mediator. First, it is known that factors such as stress, worries, anxiety, feelings of insecurity, and vulnerability can all negatively impact sleep (Arber et al. 2007, 2009), and sleep can therefore be situated as a physiological response through which pathways of crisis move from marital status to health. Second, marital status/relationship quality may influence sleep through symbolic meanings and the (shifting) contexts of the sleep environment. For example, bed sharing is associated with intimacy, security, and a sense of stability, closeness, and reassurance (Rosenblatt 2006).

The mediating effect of sleep may also be moderated by gender. Research has indicated that as well as there being gender differences in the link between marital status and health, the relationship between marital status and sleep may be gendered. For example, Troxel et al. (2009) suggest that women show greater levels of relational interdependence, and women’s sleep may also show greater sensitivity to the dissolution of marriage (which in turn impacts their health). Venn et al. (2008) found that physical and emotional care for young children at night was largely provided by women. From this, they conclude that women undertake a “fourth shift,” where their daytime involvement with physical caring and sentient activities continue into the night, disturbing their own sleep.

Building on the foregoing discussion, this paper examines the following research questions and subquestions:

How do marital status and relationship quality link to self-rated health?

Do the links between relationship quality and self-rated health differ between those who are married and those cohabiting?

Do groups within the umbrella term unmarried differ in their associations with self-rated health?

To what extent do sleep problems mediate any link between marital status/relationship quality and self-rated health?

What are the associations between marital status/relationship quality and sleep problems?

When considering the marital status/health nexus, does the mediating role of sleep problems differ by gender?

Data and Methods

This paper analyses cross-sectional data from the Understanding Society (Wave 1) survey to answer these research questions. Wave 1 of Understanding Society includes a range of questions on marital status, marital history, and marital quality as well as questions on sleep and socioeconomic circumstances. Data were collected in 2009 and 2010 from a representative sample of households in Britain (McFall and Garrington 2011). All adults over 16 within sampled households completed a face-to-face interview with additional attitude and sleep questions asked in a self-completion questionnaire. The sample for the current analysis is 37,253 respondents who completed the self-completion questionnaire without the need for a proxy.

Self-rated Health

We analyzed a subjective measure of health captured using a 5-point Likert scale (“In general, would you say your health is . . .,” with 1 being “excellent” and 5 being “poor”). This was reverse coded so that higher values equated to more positive evaluations of health. As Jylhä (2009) notes, while there is some ambiguity as to what self-rated health measures, there is a strong and consistent association with mortality. Self-rated measures have been widely used within research on marital status and health. For example, Ren (1997) used data from the U.S. National Survey of Families and Households (N = 12,274) to show that self-rated health is associated with both marital status and the quality of marriage and cohabitation.

Marital Status/Relationship Distress Measure

Our marital status/quality variable combined marital status (single, married, cohabiting, separated but legally married, divorced, and widowed) with a quality-of-relationship variable. Understanding Society contained nine questions pertaining to relationship quality that were asked of those married or living with a partner (see Appendix 1 at http://jhsb.sagepub.com/supplemental). Factor analysis of these nine items identified three factors. Four of these relationship quality items loaded onto the first factor (“consider divorce/leaving,” “regret getting married/living together,” “quarrel,” “get on each other’s nerves”). An unweighted factor-based scale was created by summing the scores from these four items. Analysis showed that there was no significant difference in the health or sleep quality of those in the highest and middle tertiles of this summated scale. Therefore, the relationship scale was recoded into two categories: low tertile (score less than 20, which relates to more negative values on the questions) versus high/middle tertile (score equal to or greater than 20, which equates to more positive values on the questions). As the items all measured “negative” aspects of relationship quality, this provided a measure of the absence or presence of distress. Our resulting marital status/quality variable thus had the following eight categories: married with little/no relationship distress; married with medium/high relationship distress; cohabiting with little/no relationship distress; cohabiting with medium/high relationship distress; separated but legally married; divorced; widowed; and single.

Sleep Measure

The Understanding Society survey included seven self-completion questions on sleep quality/quantity. These sleep questions, and the response categories, mirrored some aspects of the widely used Pittsburgh Sleep Quality Index (Buysse et al. 1989) and were also similar to the Jenkins Sleep Questionnaire (Jenkins et al. 1988), which asked whether individuals have experienced trouble falling asleep, trouble staying awake, waking up at night, and waking up feeling tired.

Sleep quality was measured by asking, “During the past month, how would you rate your sleep quality overall?” with four response categories: very good; fairly good; fairly bad; and very bad. To measure sleep latency, respondents were asked how often they “cannot get to sleep within 30 minutes” in the past month, with five response categories from “not during the past month” to “more than once most nights.” The item on sleep maintenance asked how often respondents reported they “wake up in the middle of the night or early in the morning” and used the same response categories as sleep latency. Kenny (2013) advocated the use of multiple indicators to remove any biasing effect of measurement error in the mediator. Factor analysis (principle component analysis) showed that these three items loaded onto a single factor, and reliability analysis gave a Cronbach’s alpha of .715, which would be weakened if any of the three items was removed. We therefore constructed a “sleep problems” scale using these three items, which was constrained to those who completed all three sleep questions. As sleep latency and sleep maintenance include more response categories (five) than the sleep quality item (four), each item was standardized prior to combining them.

Other Analyzed Variables

Age—as both a linear variable and as age squared—are controlled for in all models. Other covariates included factors that might be considered “precursors” to marriage, namely, gender, ethnicity, and education. Ethnicity is coded following Nandi and Platt’s (2012) analysis of self-reported ethnicity in Britain (see Appendix 2 at http://jhsb.sagepub.com/supplemental). Education is coded as: “degree or above”; “nursing, professional, and A level”; “GCSE or lower”; and “none of the above.”

Analysis

The conceptual representation in Figure 1 illustrates how the association between marital status/relationship distress and self-rated health was proposed as influenced by sleep problems. Statistically, Figure 1 represents a series of linear equations that captures the direct effects of marital status/relationship distress on self-rated health as well as the extent to which marital status/relationship distress is associated with self-rated health indirectly through sleep problems. (1) The total effect of marital status/relationship distress (X) on self-rated health (Y)—indicated as c—is the unstandardized slope of the regression of Y on X (Rucker et al. 2011). This quantifies “how much two cases differing by one unit on X are estimated to differ on Y independent of the effect of M on Y” (Hayes 2012:6). (2) The effect of marital status/relationship distress (X) on sleep problems (M) is represented by a. This is the unstandardized slope of the regression of X on M. (3) The effect of sleep problems (M) on self-rated health (Y), while controlling for marital status/relationship distress (X), is represented by b. (4) The direct effect of marital status/relationship distress (X) on self-rated health (Y), while controlling for sleep problems (M), is represented by c′. Whether sleep problems mediate the relationship, namely, the indirect effect of marital status/relationship distress on self-rated health via sleep problems, is calculated as a product of the regression coefficients a × b, which is usually equivalent to c – c′ (see Rucker et al. 2011).

First, we ran simple mediation analysis, which included age, age squared, ethnicity, and education as covariates that were assumed precursors of marital union. This model explored the total effect of marital status/relationship distress on self-rated health (controlling for precursors of marital union) and the role of sleep problems as a potential mediator. Second, we ran moderated mediation analysis. This moderated mediation analysis sought to examine whether the mediating effects of sleep problems differed for men and women, and therefore expanded Figure 1 to include three a pathways (the effect of marital status/relationship distress on sleep problems [a₁], the effect of gender on sleep problems [a₂], and the effect of the interaction between gender and marital status/relationship distress on sleep problems [a₃]). The b pathway remained the same (the effect of sleep problems on self-rated health while controlling for marital status/relationship distress). Three c′ pathways were calculated: the direct effect of marital status/relationship distress on self-rated health, controlling for sleep problems (c′₁); the direct effect of gender on self-rated health, controlling for sleep problems (c′₂); and the direct effect of the gender/marital interaction, controlling for sleep problems (c′₃). Here, the conditional indirect effect of marital status/relationship distress on self-rated health via sleep problems was calculated as a product of the regression coefficients (a₁ + a₃W)b₁, where W is the moderating variable (Hayes 2012). For both simple mediation and moderated mediation, the “process” SPSS macro (Hayes 2012) was used to calculate the regression coefficients, and the more robust method of reporting bootstrap standard errors and confidence intervals was used to determine significance.

As the independent variable (X, marital status/relationship distress) was categorical within all models, we represented this with x – 1 dummy-coded variables. All indirect, direct, and total effects were therefore relative to the group not coded (married with little/no relationship distress) (Hayes and Preacher 2013). All variables in the analysis are quantified in the same metric (MacKinnon 2000) and were checked for multicollinearity.

As Understanding Society collects data from all adults in sample households, there is the potential for data to be clustered within households and nonindependent. Initial analysis was undertaken to explore whether the data should be treated as single level or two level. The regression coefficients described above were obtained (1) using simple linear regression techniques and (2) using multilevel models (Mlwin) with individuals nested within households. Results from both a single-level mediation analysis and a multilevel mediation analysis were comparable (see Appendix 3 for actual numbers at http://jhsb.sagepub.com/supplemental). We therefore analyze the data as single level throughout the paper.

Results

Table 1 presents descriptive statistics (with further descriptive statistics found in Appendix 2 at http://jhsb.sagepub.com/supplemental). These tables illustrate how those who are married with little/no relationship distress report the best health and the best sleep.

Table 1.

Marital Status/Relationship Distress and Mean Values on Key Variables, Understanding Society, Wave 1, 2009–2010 (N = 34,474).

Marital Category	N in Marital Category Group	Mean Age (SE)	Self-rated Health (Estimated Marginal Means at Mean Age)	Relationship Distress Score (SE)	Standardized Sleep Problems Score (Estimated Marginal Means at Mean Age)
Married with little/no relationship distress	12,651	50.53 (.13)	3.63 (.01)	21.38 (.02)	−.411 (.02)
Cohabiting with little/no relationship distress	2,766	36.97 (.25)***	3.47 (.02)***	21.50 (.03)*	.211 (.05)**
Married with medium/high relationship distress	4,503	47.81 (.20)***	3.43 (.02)***	17.20 (.03)***	.257 (.04)***
Cohabiting with medium/high relationship distress	1,816	36.50 (.29)***	3.27 (.03)***	16.87 (.04)***	.534 (.06)***
Single	8,287	28.94 (.15)***	3.36 (.01)***		.027 (.03)***
Separated/legally married	614	46.52 (.49)***	3.33 (.04)***		.599 (.10)***
Divorced	2,158	53.18 (.25)***	3.24 (.02)***		.454 (.05)***
Widowed	1,679	72.56 (.29)***	3.38 (.03)***		.268 (.07)***
Total/overall mean	34,474	44.31 (.10)		2.15 (.02)

Note: All figures are calculated where sleep problems score not missing. Significance calculated using pairwise comparisons (GLM), but only significance in comparison to married with little/no relationship distress is shown.

p < .05, **p < .01, ***p < .001.

Our discussion of findings from the initial mediation analysis follows Baron and Kenny’s (1986) four-step approach to identifying mediation: (1) establish that marital status/relationship distress is related to self-rated health; (2) establish that marital status/relationship distress is associated with sleep problems; (3) establish that sleep problems are associated with self-rated health; and (4) establish whether sleep problems fully mediate the link between marital status/relationship distress and self-rated health (see Figure 1). Following these steps enables us to engage with Research Questions 1, 1a, 1b, 2, and 2a simultaneously. We then move to address Question 2b, and whether the mediating role of sleep differs for men and women, through the moderated mediation analysis.

Establishing That Marital Status/Relationship Distress Is Related to Self-rated Health

The values in Step 1 (Table 2, c path) represent the total effect of marital status/relationship distress on self-rated health relative to those married with little/no relationship distress (adjusting for precursors). All other marital categories report significantly poorer health than those married with little/no relationship distress. Those who are separated but legally married (–.263), divorced (–.336), and cohabiting with medium/high relationship distress (–.340) have the greatest unit decrease in self-reported health (thus report the poorest health). Step 1 also highlights how the relationship between marital status and self-rated health is more complex than a married-versus-unmarried comparison, demonstrating that both the form and quality of relationships are important. It is notable that individuals who are cohabiting with little/no relationship distress (–.144) report slightly better health than those who are married with medium/high relationship distress (–.193).

Table 2.

Four Steps in Exploring Mediation (All Models Include Precursors: Age, Age Squared, Ethnicity and Education), Understanding Society, Wave 1, 2009–2010 (N = 34,424).

Marital Status/Quality Category	Step 1: The Total Effect of Marital Status on Self-rated Health (Testing Path c)	Step 2: The Effect of Marital Status on Sleep Problems (Testing Path a)	Step 3: The Effect of Sleep Problems on Self-rated Health (While Controlling for Marital Status; Testing Path b)	Indirect Effect via Sleep Problems (ab)	Step 4: Direct Effect (Path c′) of Marital Status on Self-rated Health
Married with little/no relationship distress	Reference	Reference	Reference	Reference	Reference
Cohabiting with little/no relationship distress	−.144 (.02)	.173 (.03)	−.135 (.002)	−.023 (.01)	−.121 (.02)
Married with medium/high relationship distress	−.193 (.02)	.666 (.02)	−.135 (.002)	−.090 (.01)	−.103 (.02)
Cohabiting with medium/high relationship distress	−.340 (.03)	.910 (.05)	−.135 (.002)	−.123 (.01)	−.217 (.03)
Single	−.238 (.02)	.509 (.04)	−.135 (.002)	−.069 (.01)	−.169 (.02)
Separated/legally married	−.263 (.04)	.955 (.10)	−.135 (.002)	−.129 (.02)	−.134 (.04)
Divorced	−.336 (.03)	.785 (.01)	−.135 (.002)	−.106 (.01)	−.230 (.02)
Widowed	−.207 (.03)	.797 (.07)	−.135 (.002)	−.108 (.01)	−.099 (.03)

Note: All models run to ensure the base number of respondents stayed equivalent (so Step 1 run on those not missing sleep problems score, Step 2 on those not missing health score). The figures in Step 1 represent the total effect of marital status on health relative to those married with little/no relationship distress. Therefore, those who are single are on average .238 units lower in self-reported health (that is, worse) than those married with little/no relationship distress. Step 2 shows that cases that differ by one unit on marital status differ by x units on sleep problems. Step 3 holds condition constant. Two people within the same marital status group but who differ by one unit on sleep will differ by -.135 units on health. Multiplying the coefficients in Step 2 and Step 3 yields the indirect effect.

Establishing That Marital Status/Relationship Distress Is Associated with Sleep Problems

Step 2 (Table 2) illustrates the a coefficient. For example, those who are widowed are on average .797 units higher in their sleep problems score than are married people with little/no relationship distress. Marriage per se does not appear to protect from poor sleep, since those married with medium/high relationship distress report poorer sleep (.666) than both those who are single (.509) and those who are cohabiting with little/no relationship distress (.173). However, there appears to be an interplay between the role of marriage and the role of relationship distress, with legal marriage providing some protection against the association between relationship distress and sleep. Those cohabiting with medium/high relationship distress (.910) report poorer sleep than do married people with medium/high relationship distress (.666).

Establishing That Sleep Problems Affect Self-rated Health

Step 3 (Table 2) reports the effect of sleep problems on self-rated health while controlling for marital status/relationship distress and all precursors of union. The coefficient of –.135 confirms the previously reported significant relationship between sleep and health. Holding condition constant, those with more sleep problems report poorer health.

Sleep as a Partial Mediator of the Link between Marital Status/Quality and Health

Multiplying the coefficients of Step 2 and Step 3 together gives the mediating role of sleep problems (ab). As shown in Table 2, sleep is a partial mediator of the link between marital status/relationship distress and self-rated health, accounting for some, but not all, of the total effect of marital status on health. Relative to those married with little/no relationship distress, those who are separated are .129 units worse in self-rated health as a result of being separated negatively impacting sleep and these sleep problems, in turn, negatively influencing self-rated health. When the indirect effect of sleep problems (ab) is considered as a percentage of the total effect (Step 1, Path c) (Figure 2), the relative indirect effect of sleep problems is most notable for those who are divorced (32%), cohabiting with medium/high relationship distress (36%), married with medium/high relationship distress (47%), separated (49%), and widowed (52%). This suggests that sleep problems play a role in the marital status–to–self-rated health pathway, particularly when in a troubled cohabiting/married relationship or after the loss of a relationship.

Gender Differences in Sleep as a Mediator of the Link between Marital Status/Quality and Health

Table 3 presents the results from moderated mediation analysis exploring gender differences in total and direct effects of marital status/relationship distress on self-rated health as well as gender differences in the indirect effect of sleep problems. In some cases, there are comparable total effects of marital status/relationship distress on self-rated health for men and women. For example, among those who are divorced and cohabiting with medium/high relationship distress, both men and women evaluate themselves over .3 units less healthy on average (than those who are married with little/no relationship distress). In contrast, being single or separated is linked to poorer health for women, while being widowed or in a marriage with medium/high relationship distress is associated with poorer health for men.

Table 3.

Moderated Mediation Analysis: Exploring How Gender Moderates the Relationship between Marital Status/Relationship Distress, Self-rated Health, and Sleep Problems (While Controlling for Precursors: Age, Age Squared, Ethnicity and Education), Understanding Society, Wave 1, 2009–2010 (N = 34,424).

	Men					Women
Marital Category	N in Each Category	Sleep Problems Score (Estimated Marginal Means at Mean Age) (SE)	Total Effect of Marital Status/ Relationship Distress on Self-rated Health	Direct Effect of Marital Status/ Relationship Distress on Self-rated Health (SE)	Indirect Effect via Sleep Problems (Bootstrap SE)	N in Each Category	Sleep Problems Score (Estimated Marginal Means at Mean Age) (SE)	Total Effect of Marital Status/ Relationship Distress on Self-rated Health	Direct Effect of Marital Status/ Relationship Distress on Self-rated Health (SE)	Indirect Effect via Sleep Problems (Bootstrap SE)
Married with little/no relationship distress	6,126	−.648 (.03)	Reference	Reference	Reference	6,525	−.198 (.03)	Reference	Reference	Reference
Cohabiting with little/no relationship distress	1,318	−.435 (.06)	−.134	−.107 (.03)	−.027 (.01)	1,448	.017 (.07)	−.156	−.131 (.03)	−.025 (.01)
Married with medium/high relationship distress	1,874	.009 (.03)	−.248	−.155 (.03)	−.093 (.01)	2,629	.426 (.05)	−.150	−.068 (.02)	−.082 (.01)
Cohabiting with medium/high relationship distress	779	.278 (.08)	−.320	−.195 (.04)	−.125 (.01)	1,037	.749 (.08)	−.355	−.233 (.03)	−.122 (.01)
Single	3,842	−.249 (.04)	−.143	−.072 (.02)	−.071 (.01)	4,445	.307 (.04)	−.322	−.249 (.02)	−.073 (.01)
Separated/legally married	165	−.098 (.18)	−.159	−.083 (.08)	−.076 (.03)	449	.852 (.12)	−.293	−.161 (.05)	−.132 (.02)
Divorced^a	663	−.069 (.09)	−.363	−.296 (.04)	−.066* (.01)	1,495	.659 (.07)	−.317	−.208 (.03)	−.108* (.01)
Widowed	420	−.069 (.12)	−.241	−.153 (.05)	−.089 (.02)	1,259	.292 (.08)	−.182	−.095 (.03)	−.087 (.01)

The indirect effect of the highest-order interaction is significant only for those who are divorced, which can be interpreted as an indication that the indirect effect of marital status on health through sleep differs by gender.

As found in other research, women report greater sleep problems than men (Dzaja et al. 2005), and women’s sleep appears more susceptible to relationship loss (Troxel et al. 2009). The group with the “best” sleep is men who are married with little/no relationship distress with an age-controlled, standardized sleep problems score of –.648, while comparable women (married with little/no relationship distress) have a score of –.198. Despite these differences, for both men and women, those who are married and those cohabiting with medium/high relationship distress have high sleep problems scores.

Table 3 also illustrates how sleep acts as a partial mediator of the link between marital status/relationship quality and health for both men and women. Men who are married with medium/high relationship distress are .09 units worse in self-rated health associated with poor sleep, while comparable women (married with medium/high relationship distress) are .08 units worse in self-rated health through the association with poor sleep. The indirect effect of the highest-order interaction is significant only for the divorced. While the total effect of divorce on health is similar for both men and women, sleep is a much stronger mediator of the relationship between marital status/relationship distress and self-rated health for divorced women than for divorced men.

These findings regarding gender differences in the mediating role of sleep are represented in Figure 3, which shows the percentage of the total effect of marital status/relationship distress on health that is attributed to the indirect effect of sleep problems separately for men and women. Dividing the indirect effect of sleep (from Table 3) by the total effect illustrates that for divorced men, sleep accounts for only 18% of the total effect of marital status on health, whereas for divorced women, the figure is 34%.

Discussion and Conclusion

Marital Status and Health: Protection/Crisis

By examining marital status differentials in self-rated health, the paper has shown that those in marriages with little/no relationship distress have better self-rated health than all other groups. While those who are married with medium/high distress have comparable self-rated health to those cohabiting with little/no relationship distress, those who are divorced and those cohabiting with medium/high relationship distress report the poorest health. The findings situate largely within the “protection idea,” as it is not simply living with someone or being in a happy relationship that protects. Rather, (legal) marriage is protective when relationship distress is absent and (legal) marriage diminishes the negative impact of relationship distress on self-rated health.

Marriage appears to “protect” differentially for men and women. It has been suggested that women benefit more from the material well-being marriage offers, whereas married men benefit more from social/emotional support that marriage brings (Liu and Umberson 2003:249). Current findings offer support for this. We would suggest that married men with relationship distress have reduced social/emotional support, and it is for this reason that they report poorer health compared to single men and cohabiting men with little/no relationship distress.

Marital Status/Relationship Distress and Sleep Problems

When considering marriage/cohabitation and sleep problems, it is relevant to consider the meaning of co-sleeping. Some studies indicate that co-sleeping is associated with better sleep outcomes, while others suggest that the shared bed may be a “battle-ground” (Hislop and Arber 2003b). For Troxel (2010:5), “the divergent findings concerning the effects of co-sleeping on sleep . . . suggest that the psychological need for closeness and security, particularly at night, trumps the equally important need for good quality sleep.” Troxel (2010) suggests that the interplay between closeness/security and good-quality sleep is likely to be moderated by qualitative dimensions of the relationship. Our findings add to these ideas by demonstrating that both the form and quality of marital relationships are associated with sleep quality. With respect to form, legal marriage appears important; those who are married with little/no relationship distress report the best sleep (Table 2), and among those with medium/high relationship distress, the cohabiting have poorer sleep than the married. However, the importance of relationship quality can be gauged by noting that those in a relationship with medium/high distress—whether based on marriage or cohabitation—report poorer sleep than those who are single.

The Mediating Role of Sleep Problems

This paper has demonstrated how sleep is part of the process by which social relations translate into health outcomes (cf. Hale 2010). Sleep problems act as a partial mediator of the effect marital status/distress has on self-rated health, with the indirect effect of sleep problems most notable for those in relationships with medium/high relationship distress or those who have left a marital relationship (divorced/separated). The current analysis does not enable us to understand fully why sleep is acting as a partial mediator. However, it is possible to situate sleep within the known pathways through which social ties influence health—namely, behavioral, psychosocial, and physiological pathways (Umberson and Montez 2010). For example, financial stresses and worries can lead to intrusive thoughts/rumination and physiological changes in cortisol levels, both of which negatively impact sleep and can have adverse consequences for health (Troxel 2010). Supportive social ties may offer increased socioeconomic security and trigger physiological sequelae—such as reduced blood pressure and stress hormones (Umberson and Montez 2010)—which may protect against these stress-related sleep problems.

Sleep is also bound up with the symbolic meaning of particular social ties and therefore links to pathways of emotional support. As noted above, research has suggested a possible interplay/conflict between these ideals of bed sharing and sleep quality. Pankhurst and Horne (1994:315) found “a general preference among [their] subjects for sleeping with their partner rather than without, despite the objective evidence that showed that sleep was ‘poorer’ when they did so.” For those who are in relationships with medium/high relationship distress, the symbolic meaning of the shared bed, and the associated emotional support, may negatively shift. For those who have left a marital relationship (divorced/separated), part of the mediating role of sleep on health may be explained by the loss of the symbolic shared sleep environment.

Consideration of these possibilities enables us to begin to explain the observed gender difference. While the association of divorce with poor self-rated health is comparable for men and women, a greater proportion of the total effect of divorce on women’s self-rated health is mediated via sleep problems. For women, the adverse effects of divorce, such as material deprivation and changed relationships with family members (Arber, Davidson, and Ginn 2003), could be operating partly through sleep problems. Alongside this, the loss of the shared sleeping environment could be impacting sleep problems (and, in turn, self-rated health). As Hislop and Arber (2003a) note, divorce can bring a sense of freedom to some women as well as the opportunity to improve sleep patterns. However, the empty double bed can also serve as a constant reminder of the “couple relationship that was” (Williams 2005).

Further longitudinal work is needed here. One of the recognized limitations of the current study is that data are cross-sectional. Following the methodological literature on mediation analysis, we have hypothesized a causal relationship and used the language of causality throughout. However, as data are cross-sectional, in reality we can talk only about associations. Further to this, our postulated causal model describes pathways from marital status/distress to self-rated health via sleep problems. It could be suggested that sleep quality may work “backwards,” influencing marital status/relationship quality. As Rosenblatt (2006) notes, there is evidence that couples with different preferences for sleep timing have more arguments and marital discord. Therefore, longitudinal studies are necessary to fully unravel the mediating role of sleep problems. Longitudinal studies could explore how the relationship history of the couple is associated with sleep quality, taking into account how relationships are formed and ended, and both mental and physical health at different stages in the relationship history. These longitudinal models would also be able to explore more adequately the mechanisms through which sleep is operating.

Our study has added to literature on marital status, sleep, and health by extending studies that have shown that “positive and negative aspects of relationship functioning are linked with sleep via their influence on psychological, behavioural, chronobiological, and physiological mechanisms” (Troxel et al. 2007:398). It has contributed to completing the triangle by suggesting that sleep problems provide a link between marital status/distress and health. It has also highlighted the importance of exploring gender differences in the mediating role of sleep. Women’s sleep may show greater sensitivity to the dissolution of marriage (which in turn impacts their health).

The paper has also highlighted the importance of considering both the form and quality of relationships when exploring the link between marital status and health. Where previous research into relationships and sleep has included measures of relationship quality and stability, it has failed to consider how the form of relationship (marriage vs. cohabitation) is associated with health. Through analyzing various categories among the unmarried, we have also highlighted diversity within the unmarried group, for example, noting differences in the mediating role of sleep for those who are divorced and those separated but legally married.

Footnotes

Acknowledgements

We would like to thank the reviewers and editor for their useful and constructive comments throughout the submission process. We acknowledge both the UK Data Archive and the Institute for Social and Economic Research for making the Understanding Society data available for use. We would also like to thank Professor Andrew Hayes for making the “process” macro available to the research community.

Supplemental Material

The online appendices are available at .

Author Biographies

Robert Meadows is a senior lecturer in the Department of Sociology, University of Surrey. He has considerable experience of researching the social aspects of sleep. He is currently working on two British Academy funding studies, one of which is using longitudinal dyadic data analysis techniques to explore sleep loss over time within households.

Sara Arber is a professor of sociology and codirector of the Centre for Research on Ageing and Gender (CRAG) at University of Surrey, Guildford, United Kingdom. She was president of the International Sociological Association Research Committee on Sociology of Aging (RC11) (2006–2010). Sara has published over 300 articles on gender and ageing, gender and class inequalities in health, and the sociology of sleep. She is a fellow of the British Academy and the Gerontological Society of America (GSA).

References

Arber

Sara

Bote

Marcos

Meadows

Robert

. 2009. “Gender and Socio-economic Patterning of Self-reported Sleep Problems in Britain.” Social Science & Medicine 68(2):281–89.

Arber

Sara

Davidson

Kate

Ginn

Jay

. 2003. Gender and Ageing: Changing Roles and Relationships. Berkshire, UK: McGraw-Hill International.

Arber

Sara

Hislop

Jenny

Bote

Marcos

Meadows

Robert

. 2007. “Gender Roles and Women’s Sleep in Mid and Later Life: A Quantitative Approach.” Sociological Research Online 12(5):3.

Baron

Reuben M.

Kenny

David A.

1986. “The Moderator–Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations.” Journal of Personality and Social Psychology 51(6):1173–82.

Besnard

Philippe

. 2001. “Marriage and Suicide: Testing the Durkheimian Theory of Marital Regulation a Century Later.” Pp. 133–55 in Emile Durkheim: Critical Assessments of Leading Sociologists, vol. 4, edited by Frederick

William Stuart

. London: Pickering, Taylor & Francis.

Buysse

Daniel J.

Reynolds

Charles F.

III Monk

Timothy H.

Berman

Susan R.

Kupfer

David J.

1989. “The Pittsburgh Sleep Quality Index: A New Instrument for Psychiatric Practice and Research.” Psychiatry Research 28(2):193–213.

Cacioppo

John T.

Hawkley

Louise C.

Berntson

Gary G.

Ernst

John M.

Gibbs

Amber C.

Stickgold

Robert

Hobson

J. Allan

. 2002. “Do Lonely Days Invade the Nights? Potential Social Modulation of Sleep Efficiency.” Psychological Science 13(4):384–87.

Cappuccio

Francesco

Miller

Michelle A.

2010. “The Epidemiology of Sleep and Cardiovascular Risk and Disease.” Pp. 83–110 in Sleep, Health and Society: From Aetiology to Public Health, edited by Cappuccio

Francesco P.

Miller

Michelle A.

Lockley

Steven W.

Oxford, UK: Oxford University Press.

Cartwright

Rosalind D.

Wood

Ellen

. 1991. “Adjustment Disorders of Sleep: The Sleep Effects of a Major Stressful Event and Its Resolution.” Psychiatry Research 39(3):199–209.

10.

Dregan

Alex

Lallukka

Tea

Armstrong

David

. 2013. “Potential Pathways from Biopsychosocial Risk Factors to Sleep Loss Due to Worry: A Population-based Investigation.” Journal of Public Mental Health 12(1):43–50.

11.

Dzaja

Andrea

Arber

Sara

Hislop

Jenny

Kerkhofs

Myriam

Kopp

Caroline

Pollmächer

Thomas

Polo-Kantola

Päivi

Skene

Debra

Stenuit

Patricia

Tobler

Irene

Porkka-Heiskanen

Tarja

. 2005. “Women’s Sleep in Health and Disease.” Journal of Psychiatric Research 39(1):55–76.

12.

Franklin

Adrian S.

2009. “On Loneliness.” Geografiska Annaler: Series B, Human Geography 91(4):343–54.

13.

Gallo

Linda C.

Matthews

Karen A.

Kuller

Lewis H.

Sutton-Tyrrell

Kim

Edmundowicz

Daniel

. 2001. “Educational Attainment and Coronary and Aortic Calcification in Postmenopausal Women.” Psychosomatic Medicine 63(6):925–35.

14.

Hale

Lauren

. 2010. “Sleep as a Mechanism through Which Social Relationships Affect Health.” Sleep 33(7):862–63.

15.

Hayes

Andrew F.

2012. “PROCESS: A Versatile Computational Tool for Observed Variable Mediation, Moderation, and Conditional Process Modelling” [White paper]. Retrieved April 16, 2014 (http://www.afhayes.com/).

16.

Hayes

Andrew F.

2013. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-based Approach. New York: Guilford Press.

17.

Hayes

Andrew F.

Preacher

Kristopher J.

2013. “Statistical Mediation Analysis with a Multicategorical Independent Variable.” British Journal of Mathematical and Statistical Psychology. Advance online publication.

18.

Hislop

Jenny

Arber

Sara

2003a. “Sleep, Gender and Aging: Temporal Perspectives in the Mid-to-later Life Transition.” Pp. 225–46 in Age Matters: Realigning Feminist Thinking, edited by Calasanti

Toni M.

Slevin

Kathleen F.

New York: Routledge.

19.

Hislop

Jenny

Arber

Sara

. 2003b. “Sleepers Wake! The Gendered Nature of Sleep Disruption Among Mid-life Women.” Sociology 37(4):695–711.

20.

Hughes

Mary Elizabeth

Waite

Linda J.

2002. “Health in Household Context: Living Arrangements and Health in Late Middle Age.” Journal of Health and Social Behavior 43(1):1–21.

21.

Jenkins

C. David

Stanton

Babette-Ann

Niemcryk

Steve J.

Rose

Robert M.

1988. “A Scale for the Estimation of Sleep Problems in Clinical Research.” Journal of Clinical Epidemiology 41(4):313–21.

22.

Jylhä

Marja

. 2009. “What Is Self-rated health and Why Does It Predict Mortality? Towards a Unified Conceptual Model.” Social Science & Medicine 69(3):307–16.

23.

Kenny

David A.

2013. “Mediation.” Retrieved October 2, 2014 (http://davidakenny.net/cm/mediate.htm).

24.

Knudsen

Hannah K.

Ducharme

Lori J.

Roman

Paul M.

2007. “Job Stress and Poor Sleep Quality: Data from an American Sample of Full-time Workers.” Social Science & Medicine 64(10):1997–2007.

25.

Koball

Heather L

Moiduddin

Emily

Henderson

Jamila

Goesling

Brian

Besculides

Melanie

. 2010. “What Do We Know about the Link between Marriage and Health?” Journal of Family Issues 31(8):1019–1104.

26.

Liu

Hui

Umberson

Debra J.

2008. “The Times They Are a Changin’: Marital Status and Health Differentials from 1972 to 2003.” Journal of Health and Social Behavior 49(3):239–53.

27.

MacKinnon

David P.

2000. “Contrasts in Multiple Mediator Models.” Pp. 141–60 in Multivariate Applications in Substance Use Research, edited by Rose

Jennifer S.

Chassin

Laurie

Presson

Clark C.

Sherman

Steven J.

Mahwah, NJ: Lawrence Erlbaum.

28.

McFall

Stephanie L.

Garrington

Chris

. 2011. Understanding Society: Early Findings from the First Wave of the UK’s Household Survey. Colchester, UK: Institute for Social Research.

29.

Mental Health Foundation. 2010. The Lonely Society. London: Author.

30.

Nandi

Alita

Platt

Lucinda

. 2012. “How Diverse Is the UK?” Pp. 13–14 in Understanding Society: Findings 2012, edited by McFall

Stephanie

Garrington

Chris

. Colchester, UK: Institute for Social and Economic Research, University of Essex.

31.

Pankhurst

Francesca P.

Horne

James A.

1994. “The Influence of Bed Partners on Movement During Sleep.” Sleep 17(4):308–15.

32.

Ren

Xinhua Steve

. 1997. “Marital Status and Quality of Relationships: The Impact on Health Perception.” Social Science & Medicine 44(2):241–49.

33.

Rendall

Michael S.

Weden

Margaret M.

Favreault

Melissa M.

Waldron

Hilary

. 2011. “The Protective Effect of Marriage for Survival: A Review and Update.” Demography 48(2):481–506.

34.

Rosenblatt

Paul C.

2006. Two in a Bed: The Social System of Couple Bed Sharing. Albany: State University of New York Press.

35.

Ross

Catherine E.

Mirowsky

John

. 2013. “Theory and Modeling in the Study of Intimate Relationships and Health.” Journal of Health and Social Behavior 54(1):67–71.

36.

Rucker

Derek D.

Preacher

Kristopher J.

Tormala

Zakary L.

Petty

Richard E.

2011. “Mediation Analysis in Social Psychology: Current Practices and New Recommendations.” Social and Personality Psychology Compass 5(6):359–71.

37.

Social Exclusion Task Force. 2007. Think Child, Think Parent, Think Family. London: Cabinet Office.

38.

Troxel

Wendy M.

2010. “It’s More Than Sex: Exploring the Dyadic Nature of Sleep and Implications for Health.” Psychosomatic Medicine 72(6):578–86.

39.

Troxel

Wendy M.

Buysse

Daniel J.

Hall

Martica

Kamarck

Thomas W.

Strollo

Patrick J.

Owens

Jane F.

Reis

Steven E.

Matthews

Karen A.

2010. “Social Integration, Social Contacts, and Blood Pressure Dipping in African-Americans and Whites.” Journal of Hypertension 28(2):265–71.

40.

Troxel

Wendy M.

Buysse

Daniel J.

Hall

Martica

Matthews

Karen A.

2009. “Marital Happiness and Sleep Disturbances in a Multi-ethnic Sample of Middle-aged Women.” Behavioral Sleep Medicine 7(1):2–19.

41.

Troxel

Wendy M.

Buysse

Daniel J.

Matthews

Karen A.

Kravitz

Howard M.

Bromberger

Joyce T.

Sowers

MaryFran

Hall

Martica H.

2010. “Marital/Cohabitation Status and History in Relation to Sleep in Midlife Women.” Sleep 33(7):973–81.

42.

Troxel

Wendy M.

Robles

Theodore F.

Hall

Martica

Buysse

Daniel J.

2007. “Marital Quality and the Marital Bed: Examining the Covariation between Relationship Quality and Sleep.” Sleep Medicine Reviews 11(5):389–404.

43.

Umberson

Debra

. 1992. “Gender, Marital Status and the Social Control of Health Behavior.” Social Science & Medicine 34(8):907–17.

44.

Umberson

Debra

Crosnoe

Robert

Reczek

Corinne

. 2010. “Social Relationships and Health Behavior across the Life Course.” Annual Review of Sociology 36:139–57.

45.

Umberson

Debra

Montez

Jennifer Karas

. 2010. “Social Relationships and Health: A Flashpoint for Health Policy.” Journal of Health and Social Behavior 51(1 Suppl.):S54–S66.

46.

Venn

Susan

Arber

Sara

Meadows

Robert

Hislop

Jenny

. 2008. “The Fourth Shift: Exploring the Gendered Nature of Sleep Disruption among Couples with Children.” British Journal of Sociology 59(1):79–97.

47.

Weich

Scott

. 2010 “The Epidemiology of Sleep and Depression.” Pp. 178–90 in Sleep, Health and Society: From Aetiology to Public Health, edited by Cappuccio

Francesco P.

Miller

Michelle A.

Lockley

Steven W.

Oxford, UK: Oxford University Press.

48.

Williams

Kristi

Umberson

Debra

. 2004. “Marital Status, Marital Transitions, and Health: A Gendered Life Course Perspective.” Journal of Health and Social Behavior 45(1):81–98.

49.

Williams

Simon J.

2005. Sleep and Society: Sociological Ventures into the (Un)Known. London: Routledge.

50.

Zee

Phyllis C.

Turek

Fred W.

2006. “Sleep and Health: Everywhere and in Both Directions.” Archives of Internal Medicine 166(16):1686–88.

51.

Zhang

Zhenmei

Hayward

Mark D.

2006. “Gender, the Marital Life Course, and Cardiovascular Disease in Late Midlife.” Journal of Marriage and Family 68(3):639–57.

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

0.55 MB