Empty Nest Status, Marital Closeness, and Perceived Health: Testing Couples’ Direct and Moderated Associations with an Actor-Partner Interdependence Model

Abstract

Building on previous examinations of marital outcomes in the empty nest phase that have been based on surveys of individuals (primarily wives), the current study examined the direct effect of no longer living with children in the home (versus continuing to live with children) on husbands and wives’ ratings of marital closeness as well as their perceived health. The study also tested whether couples’ empty nest status moderated the associations between spouses’ marital closeness and health. Analyses were based on 3,765 mixed-sex couples drawn from the Wisconsin Longitudinal Study. Dyadic multilevel modeling and actor-partner interdependence models (APIM) were used to test hypotheses. Results indicated that, accounting for known covariates, being in the empty nest was directly linked with both husbands and wives reporting higher levels of marital closeness and with wives (only) reporting better health. Moderating findings were less consistent, with the single reliable moderation result indicating that wives’ perceived health was improved at higher levels of their husbands’ marital closeness only among couples who were still living with children in the home. Clinical implications and future research directions are discussed.

A long-standing goal in the psychological study of marriages is to understand whether and why relationship quality changes over time (Bradbury & Karney, 2014; Karney & Bradbury, 1995). Research to date has focused on the early years of marriage, revealing typical declines in quality across the newlywed period (Kurdek, 1998) and increases in strain and conflict following the birth of the first child (Doss, Rhoades, Stanley, & Markman, 2009). Relatively scant research has followed marriages further along the relationship course, that is, until couples re-enter the focus of research in later years on illness, caregiving, and grief (Prokos & Keene, 2005; Williams, 2011). Consequently, major gaps exist in our understanding of intimate relationships during middle adulthood and beyond, leaving highly understudied the question of whether spouses and marriages change as a result of the normative transition of living with no children after having lived with one or more children in the household, namely, the “empty nest” period.

How marriages fare during the empty nest holds critical implications for the well-being and health of spouses, both concurrently and as they transition to older ages (Hawkins & Booth, 2005; Wanic & Kulik, 2011). Prior studies have demonstrated the impact of poorer marital quality (e.g., greater conflict, lower closeness) on health-threatening conditions such as elevated cortisol reactivity, emotional distress, and sleep problems (e.g., Choi & Marks, 2008; Kiecolt-Glaser et al., 1997; Strawbridge, Shema, & Roberts, 2004; Whisman, Uebelacker, & Weinstock, 2004), as well as morbidity following illness (Coyne et al., 2001). Midlife health and well-being reliably predict the developmental transition to old age (Lachman, 2004; Moen & Wethington, 1999). Thus, individual and marital outcomes associated with the empty nest in middle adulthood are likely to set the course for trajectories of various health outcomes later in life.

The few studies conducted to examine spousal and marital outcomes in the empty nest have been based on surveys of individuals and suggest that the empty nest is generally linked to better self-reported marital quality (White & Edwards, 1990) and psychological well-being (Harkins, 1978). For example, Gorchoff, John, and Helson (2008) conducted a longitudinal study of 123 women in middle age (at M ages of 43, 52, and 61). Moving to an empty nest status from one time-point to the next was associated with significant improvements in women’s marital (but not life) satisfaction. This general positive effect of the empty nest on marital quality aligns with research conducted on younger parents showing that the demands associated with child-rearing uniquely strain marital functioning (Crohan, 1996), and that children serve as the most frequent topic of marital conflict in the home (Papp, Cummings, & Goeke-Morey, 2009). Similarly, results from an early study of mothers revealed small but reliable improvements in psychological well-being for women living in the empty nest (Harkins, 1978). Beyond direct effects of empty nest status on (primarily wives’) marital outcomes, investigations of the effects of empty nest status on health outcomes remain scant. Furthermore, no study has examined the potential moderating role of empty nest status on the associations between marriage and health outcomes.

Empty nest status may directly have a positive impact on health outcomes or buffer the impact of marital quality problems on health problems. Acknowledging both the costs and benefits of having children, research has identified extensive demands on health and relationships associated with parenting (e.g., Lavee, Sharlin, & Katz, 1996). According to previous conceptual and empirical work on family resources and strain, it is expected that the empty nest period can be beneficial for parents, in part, because the responsibility of raising children has been lifted. That is, as being in the empty nest period is expected to benefit spouses’ marital quality, empty nest status is likely to directly benefit spouses’ health outcomes since the empty status either may alleviate some strain or allow the couple to experience the positive effect to a greater degree. Furthermore, as the chronic demands and challenges of parenting children diminish for couples who are living in the empty nest, they may have more time, energy, and resources to cope with other stresses and to improve or maintain their relationship quality, particularly compared to couples living with children at home. Therefore, empty nest status will likely play a protective factor by alleviating the impact from marital quality problems on perceived health problems as well.

The Current Study

The current study drew from the original graduate participants and their spouses in the Wisconsin Longitudinal Study to examine 1) the direct effect of the empty nest on spouses’ marital closeness and perceived health and 2) the moderating role of the empty nest on the associations between marital closeness on perceived health. The current study advanced previous literature in several ways. First, a large body of literature has established that marriage affects health (Schilling, Deeg, & Huisman, 2018), but most of the studies used data from only one spouse in the couple dyad. Thus, findings of previous studies might not fully capture the interdependence of romantic relationships. Therefore, the current study builds on previous research by utilizing data from both spouses to examine how one’s marital closeness affects not only his or her own perceived health but also their partner’s perceived health. Second, this study advances the field by examining the moderating role of the empty nest, which is an understudied period of the life course, in the relations between marital closeness and perceived health. Finally, a particular advance over previous work on the effect of empty nest on marital closeness and perceived health is the ability to examine whether findings vary for husbands and wives.

We used dyadic multilevel modeling (MLM) (Raudenbush, Brennan, & Barnett, 1995) to examine the effect of the empty nest on couples’ marital closeness and perceived health. We hypothesized that both husbands’ and wives’ empty nest status would be associated with higher marital closeness for themselves. We hypothesized that both husbands’ and wives’ empty nest status would be associated with higher perceived health for themselves.

We used dyadic MLM as well as the actor-partner interdependence model (APIM; Campbell & Kashy, 2002; Cook & Kenny, 2005; Kenny, Kashy, & Cook, 2006) to examine the moderating role of empty nest status on the associations between marital closeness and perceived health from a dyadic perspective. We expected that empty nest status would play a protective role by strengthening the positive associations between spouses’ marital closeness and their own and their partner’s perceived health.

Methods

Study Design

Participants for this study included a subset of mixed-sex marital dyads in the Wisconsin Longitudinal Study (WLS) who provided cross-sectional data collected in 2004. The WLS began primarily as a study of the transition from high school to college or work and has become a long-term, multidisciplinary study of the life course and aging, encompassing a wide range of well-being domains (e.g., genetic, psychological, physical, financial; Hauser, 2009). Original WLS participants included a random one-third sample of males and females who graduated from Wisconsin high schools in 1957 (N = 10,317). Participants completed informed consent at each round of data collection (Pearce & Parks, 2011). Participants were contacted for research purposes in 1957, 1964, 1975, 1992, 2003–2005, and 2010. Participants were mainly non-Hispanic white men and women who completed at least 12 years of education, representing most Americans at the time the study began. Additionally, WLS obtained data from sub-samples (e.g., siblings and spouses of graduates), permitting investigation of the original participants across relational contexts (Pearce & Parks, 2011). The WLS has maintained exceptionally high retention across the 50+ years spanned by the study. Notably, in 2003-5 (i.e., average age of graduates = 65 years), 7,732 (or 74.9%) of the participants completed phone or mail interviews. Also during this period, 4,076 spouses (or 54.5% of those available) completed phone interviews, providing dyadic data on some of the same constructs collected from the graduates. Our final sample consisted of 3,765 respondents and their spouses who had at least one child and reported on the outcomes of interest. Most participants had children; only 7.6% did not.

Measures

Marital closeness.

Marital closeness was obtained from respondents who completed the marriage interview, which included a single-item rating of how close the respondent currently feels to his or her spouse (1 = not at all close, 2 = not very close, 3 = somewhat close, 4 = very close). Previous analysis of the WLS supports the discriminative validity of this item as a differential predictor of health outcomes of bereaved versus typical parents (Song, Floyd, Seltzer, Greenberg, & Hong, 2010).

Perceived health.

Participants self-rated their health by responding to the question, “How would you rate your health at the present time?” on a scale ranging from 1 (very poor) to 5 (excellent). Self-reported health measures have demonstrated associations with clinician-rated and other objective health outcomes in middle adulthood, including mortality (Fechner-Bates, Coyne, & Schwenk, 1994; Miilunpalo, Vuori, Oja, Pasanen, & Urponen, 1997).

Empty nest status.

The household rosters were used to code empty nest status of participants (1 = not currently living with children, after having lived previously with children in the household, and 0 = currently living with children). Accordingly, the analyses tested the empty nest effect, reflected as not living with children in the household (after having lived with one or more children previously) relative to continuing to live with children in the household, on marital functioning and health. This operationalization follows from previous work showing that partial launching of children from the household was not associated with marital quality (White & Edwards, 1990).

Covariates: Number of children and financial resources.

All models controlled for number of children and household income. Respondents were asked on the household roster, “Including those who are no longer living, do you have any children?” Those who responded positively then entered the number of children, including biological, adopted, step, and foster children, as well as any other children considered to be a part of their family. Total household income also was retained, with higher levels reflecting more financial resources. Consistent with other WLS analyses (Ahern & Hendryx, 2008), we included log of (income + 1) due to the skewed nature of the variable.

Analysis Plan

The present study used IBM SPSS (version 26) for all analyses. Given that marital partners provided reports of the same outcomes, the MLM strategy included dummy codes to generate simultaneous estimates for both respondents, thereby permitting statistical tests to compare relations for husbands versus wives (Laurenceau & Bolger, 2005). MLM for distinguishable dyads framework were conducted (Campbell & Kashy, 2002; Cook & Kenny, 2005). In this study, distinguishability occurs because one member of the couple was the wife and the other was the husband.

To examine the direct associations between empty nest status on marital closeness and perceived health respectively, MLM with a two-intercept approach was used to examine the effect of empty nest status on marital closeness and perceived health:

$${\text{Outcome}}_{\text{ij}}=\left({\text{b}}_1\right) \text{H} \text{intercept}+\left({\text{b}}_2\right) \text{W} \text{intercept} +\left({\text{b}}_3\right) {\text{H}}^{*}{\text{empty nest status}}_{\text{ij}}+\left({\text{b}}_4\right) {\text{W}}^{*}{\text{empty nest status}}_{\text{ij}}+\left({\text{b}}_5\right) \text{number }\text{of}\text{children}+\left({\text{b}}_6\right) \text{log}\_\text{income}+e_{1\text{j}}+e_{2\text{j}},$$

where b₃ and b₄ are the coefficients used for hypothesis testing. Two models were run to accommodate the outcomes.

To test the moderating role of empty nest status on the association between marital closeness on perceived health, APIMs with MLM for distinguishable dyads framework were conducted (Campbell & Kashy, 2002; Cook & Kenny, 2005). APIMs simultaneously estimate effects of wives’ and husbands’ independent variables on their own dependent variables, namely actor effects, and effects of the partner’s independent variables on the person’s dependent variables, namely partner effects. Here, an APIM with a dual-intercept approach was used to simultaneously estimate the interaction effects of empty nest status and a person’s own marital closeness, as well as the interaction of empty nest status and the partner’s marital closeness, on the person’s perceived health, while accounting for main effects:

$$\text{Perceived}{\text{health}}_{\text{ij}}=\left({\text{b}}_1\right) \text{H} \text{intercept}+\left({\text{b}}_2\right) \text{W} \text{intercept}+\left({\text{b}}_3\right) \text{H}{}^{*}\text{Actor}\text{effect}\text{of}\text{marital}{\text{quality}}_{\text{ij}}+\left({\text{b}}_4\right) \text{W}{}^{*}\text{Actor}\text{effect}\text{of}\text{marital}{\text{quality}}_{\text{ij}}+\left({\text{b}}_5\right) \text{H}{}^{*}\text{Partner}\text{effect}\text{of}\text{marital}{\text{quality}}_{\text{ij}}+ ({\text{b}}_6) \text{W}{}^{*}\text{Partner}\text{effect}\text{of}\text{marital}{\text{quality}}_{\text{ij} }++\left({\text{b}}_7\right) \text{H}{}^{*}\text{Actor}\text{effect}\text{of}\text{marital}\text{quality}{}^{*}\text{empty}\text{nest}{\text{ status}}_{\text{ij}}+\left({\text{b}}_8\right) {\text{W}}^{*}\text{Actor effect of marital quality problem}*{\text{empty nest status}}_{\text{ij}}+\left({\text{b}}_9\right) {\text{H}}^{*}\text{Partner} {{\text{effect of marital quality }}^{*}\text{empty nest status}}_{\text{ij}}+\left({\text{b}}_{10}\right){\text{W }}^{*}{\text{Partner effect of marital quality}}^{*}\text{empty}\text{nest}{\text{status}}_{\text{ij}}+\left({\text{b}}_{11}\right) \text{empty}\text{nest}\text{status}+\left({\text{b}}_{12}\right) {\text{log}}_{-}\text{income}+\left({\text{b}}_{13}\right) \text{number}\text{of}\text{children}+e_{1\text{j}}+e_{2\text{j}},$$

where b₇ and b₈ (i.e., actor effects) and b₉ and b₁₀ (i.e., partner effects) are the coefficients examined for hypothesis testing.

Results

Table 1 reports analytic variables drawn from 3,765 mixed-sex couples in the WLS.

Table 1.

Descriptive Statistics for Study Variables

	Husbands		Wives
	M (SD)	Range	M (SD)	Range
Marital closeness	3.84 (0.40)	1–4	3.81 (0.45)	1–4
Perceived health	3.76 (0.97)	1–5	3.81 (0.95)	1–5
	Couple Characteristics
	M (SD) or %		Range
Empty nest status (% empty nest)	91%
Total household income	75022.31 (84878.17)		0–710000.00
Number of children	3.30 (1.53)		1–10

Note. N = 3,765 couples.

Is empty nest status directly associated with marital closeness and perceived health?

In support of our direct hypothesis, empty nest status was associated with higher levels of marital closeness for both wives and husbands. Results are shown in the left-hand column of Table 2. In addition, empty nest status was linked with greater perceived health for wives but not for husbands. Results are shown in the right-hand column of Table 2.

Table 2.

Direct Associations between Empty Nest Status and Spouses’ Marital Closeness and Perceived Health

	Marital closeness		Perceived health
	b (SE)	t	b (SE)	t
Covariates
Income	−0.004 (0.004)	0.99	0.03 (0.01)	3.31**
Number of children	0.005 (0.004)	1.28	−0.02 (0.01)	−3.13**
Wives
Intercept	3.71 (0.02)	152.42**	3.71 (0.05)	71.93**
Empty nest status	0.11 (0.03)	4.27***	0.11 (0.05)	1.98*
Husbands
Intercept	3.78 (0.02)	174.31**	3.68 (0.05)	69.72**
Empty nest status	0.06 (0.02)	2.56*	0.09 (0.05)	1.57

Note. Table reports unstandardized coefficients (b) with standard errors (SE) in parentheses.

^*p < .05.

^**p < .01.

^***p < .001.

Does empty nest status moderate the associations between spouses’ marital closeness and perceived health?

Partial support was obtained for our hypotheses, with one of four associations indicating reliable moderation by empty nest status. In particular, only the link between husbands’ marital closeness and their wives’ perceived health (i.e., partner effect) was moderated by couples’ empty nest status (see Table 3). Probing the significant interaction effect indicated that husbands’ marital closeness was positively associated with their wives’ perceived health among couples who were not empty nesters (b = 0.21, t = 3.82, p < .001), whereas husbands’ marital closeness was not associated with perceived health among those living in the empty nest (b = 0.06, t = 1.00, p = .32). In other words, in the context of still living with children, wives’ perceived health was improved at higher levels of their husbands’ marital closeness while wives in the empty nest tended to maintain good perceived health regardless of their spouse’s marital closeness levels.

Table 3.

The Moderating Role of Empty Nest Status on the Association between Marital Closeness and Perceived Health: APIM Results

	b (SE)	t
Wives
Intercept	3.74 (0.04)	90.97**
Actor effect (Wives->Wives)
Marital closeness	−0.03 (0.09)	−0.32
Marital closeness*EN Status	0.18 (0.10)	1.80
Partner effect (Husbands->Wives)
Marital closeness	0.37 (0.12)	2.98**
Marital closeness*EN Status	−0.26 (0.13)	−1.98*
Husbands
Intercept	3.68 (0.04)	90.97**
Actor effect (Husbands-> Husbands)
Marital closeness	0.31 (0.13)	2.49*
Marital closeness* EN Status	−0.08 (0.13)	−0.62
Partner effect (Wives-> Husbands)
Marital closeness	0.02 (0.10)	0.26
Marital closeness*EN Status	0.04 (0.10)	0.34
Couple Characteristics
Empty nest (EN) status	0.08 (0.04)	2.00*
Income	0.03 (0.01)	3.15**
Number of children	−0.02 (0.01)	−3.24**

Note. Table reports unstandardized coefficients (b) with standard errors (SE) in parentheses.

^*p < .05.

^**p < .01.

Discussion

The current study examined associations between empty nest status and marital closeness and perceived health, respectively. Furthermore, the present study investigated the effects of marital closeness on perceived health and whether empty nest status moderated the relations between marital closeness and perceived health. Finally, we also explored gender differences in the two associations mentioned above. This work heeded calls of Whisman and colleagues (2004) to include perspectives from both partners in order to gain a more complete understanding of the role of empty nest status in marital closeness and perceived health and the relations between marital closeness and perceived health among midlife couples. Dyadic data allowed for the examination of WLS participants along with their partners simultaneously. By identifying whether differences emerge for men and women, the findings can offer potentially tailored prevention and treatment recommendations (Kryspin-Exner, Lamplmayr, & Felnhofer, 2011).

The current study extends the limited body of research on empty nest status in marriage, not only finding the direct effects on marital quality but also the direct effects on perceived health. We had hypothesized that living in the empty nest would yield positive effects on both husbands’ and wives’ marital quality and health outcomes. Our predictions were partially supported in that the empty nest had direct positive effects on both spouses’ marital closeness and on wives’ perceived health.

A significant contribution of the present study was to examine whether empty nest status is a protective or risk factor in the established association between marital functioning and health among midlife and aging couples. Our predictions were partially supported in predicting wives’ perceived health in that empty nest status prevented the negative impact of husbands’ lower marital closeness on wives’ health. Empty nest status did not emerge as a moderator in the model predicting husbands’ marital functioning; indeed, the direct association between husbands’ own marital closeness and their perceived health remained significant and positive (Table 3). All in all, empty nest status may be more important to consider among wives as compared to their husbands in understanding the relationship between marriage and health among midlife and aging adult couples.

The results should be interpreted in the context of some limitations. First, information on marital closeness and perceived health was gathered using self-report measures. Although these measures of marital closeness and perceived health have previously been used with similar populations (e.g., Fechner-Bates et al., 1994; Song et al., 2010), future work should consider using more robust assessments (e.g., observed marital dynamics, objective indicators of health). Additionally, data were collected for the current study at a single time point, providing a snapshot of family members’ functioning and living arrangements. Future studies should expand on these findings with longitudinal designs to account for the dynamic experiences of families, as past research has shown fluctuations in family structure during this time period (i.e., namely as accordion families that grow to accommodate “boomerang children” who return to live at home after leaving; Newman, 2012). This study is also limited by the relatively homogenous sample, which included mainly non-Hispanic white men and women. While the WLS demographics were representative of Midwestern high school graduates in 1957, they lack the cultural and ethnic variation found in contemporary families. As noted above, demographic tests with more heterogeneous samples may yield more fruitful moderating findings.

These findings hold some useful clinical implications. Practitioners who work with midlife to older adults in the empty nest can encourage couples to maximize the benefits as they consider pathways to healthy aging and development for both husbands and wives. Furthermore, understanding the association between husbands’ marital closeness and wives’ perceived health for those who are not empty nesters is important. The results also suggest that midlife couples who are not empty nesters may benefit from interventions to couple-based interventions to improve health by improving marital closeness.

In conclusion, findings suggest that couples anticipating the empty nest transition can be reassured that the period is generally viewed as positive by both husbands and wives, particularly in terms of marital quality and the effects of marital closeness on health. These findings are aligned with previous research on role strain development during early parenthood (Doss et al., 2009) and the role relief that occurs for women following launching of grown children from the home (Gorchoff et al., 2008). Future researchers should build upon these findings by incorporating greater racial, ethnic, and economic diversity in studies of empty nest and boomerang family processes. Moreover, continued dyadic analysis is needed to understand the marital and familial processes as they pertain to both partners’ experiences of individual and relationship functioning during midlife and beyond.