Relationship quality among cohabitors and marrieds in older adulthood

Abstract

The rapid growth in cohabitation in recent decades has coincided with a burgeoning literature on the topic. Yet despite a sustained increase in cohabitation among middle-aged and older adults, this group has received little research attention. Close relationships are integral to well-being and the quality of these relationships has consequences for health, especially among older adults. We use data from the National Social Life, Health, and Aging Project (NSHAP), a nationally representative sample of 3005 people ages 57–85 to compare the relationship quality of older cohabitors versus marrieds. The two groups are remarkably similar. Cohabitors and marrieds do not significantly differ in their reports of emotional satisfaction, pleasure, openness, time spent together, criticism, and demands. Cohabitors are less likely than marrieds to report that their relationship is very happy. There is some evidence of gender by union type differences. Cohabiting unions among older adults tend to be of relatively long duration. Overall, these results indicate that cohabitation may operate as an alternative to marriage for older adults.

Cohabitation in the U.S. has increased dramatically in recent decades, climbing from 500,000 couples in 1970 to nearly 6.8 million couples today (U.S. Census Bureau, 2009). More than one-half of people in their twenties and thirties have cohabited. And, cohabitation is now the modal path of entry into marriage (Smock, 2000). It is also the relationship context for about one in five births (Mincieli et al., 2007). Nonetheless, the rise in cohabitation is not confined to younger adults; cohabitation is also growing among middle-aged and older adults. In less than a decade, cohabitation levels among those 50 and older have nearly doubled, rising from 1.2 million persons in 2000–2.2 million persons in 2008 (U.S. Census Bureau, 2009).

Despite a burgeoning literature on cohabitation among younger adults, comparatively little is known about this phenomenon among older adults (although see Brown et al. (2005, 2006), Chevan (1996), de Jong Gierveld (2004), Hatch (1995), King and Scott (2005)). Several scholars have noted the importance of investigating new family forms, including cohabitation, in older adulthood (Allen et al., 2000; Cooney and Dunne, 2001). A declining share of older adults is married, meaning that a larger proportion is eligible to cohabit. Moreover, as the U.S. population ages and baby boomers move into older adulthood, the growth in cohabitation is likely to accelerate.

Close relationships are integral to adult well-being (Lui and Umberson, 2008; Waite, 1995). It is well-established that marriage, in particular, provides unique benefits for older adults, who typically enjoy higher levels of physical and mental health as well as greater financial resources and social support than unmarrieds (Calasanti and Kiecolt, 2007; Connidis, 2001). High quality relationships enhance these benefits as marital quality is linked to both individual health and subjective well-being (Connidis, 2001; Kiecolt-Glaser and Newton, 2001) and this association is most pronounced among older adults (Umberson et al., 2006). Whether cohabitation confers similar benefits to older adults is unclear.

The purpose of this study is to investigate the extent to which cohabiting and married relationships are similar. Do older cohabiting adults report relationship quality comparable to that of marrieds? We use data from the 2005 to 2006 National Social Life, Health, and Aging Project (NSHAP), a nationally representative sample (N = 3005) of the noninstitutionalized U.S. population ages 57–85, to compare older cohabitors and marrieds across multiple domains of relationship quality. Our study extends prior research on relationship quality among cohabitors (e.g., Brown, 2004, 2003, 2000b; Brown and Booth, 1996; King and Scott, 2005; Nock, 1995; Skinner et al., 2002) by considering older adults. It also informs gerontological research on marital status and well-being, which largely has been confined to spousal loss (e.g., Carr, 1996; Carr et al., 2000; Lee et al., 2001, 1998) and, to a lesser extent, remarriage (e.g., Bulcroft et al., 1989; Burch, 1990), by incorporating cohabitation.

1. Background

To formulate our expectations about how the relationship quality of cohabitors and marrieds differs in older adulthood, we begin by reviewing the literature on this topic for adults in general. Next, we draw on the few studies to consider cohabitation among older adults to refine our hypotheses for this distinct age group. Finally, we propose and test a series of hypotheses about the linkages between union type (i.e., cohabitation versus marriage) and relationship quality among older adults.

1.1. Relationship quality among cohabitors and marrieds

The first national study of relationship quality among cohabitors versus marrieds was conducted by Nock (1995). Using data from the 1987 to 1988 National Survey of Families and Households (NSFH), he found that cohabitors tend to report poorer relationship quality than their married counterparts. Specifically, cohabitors perceive fewer costs to exiting their unions and report less happiness in their relationships. This pattern obtained regardless of whether marrieds had premaritally cohabited. Nock concluded that cohabitation, similar to remarriage, is an incomplete institution (cf. Cherlin, 1978) in which the norms and roles for cohabiting partners are unclear, setting the stage for relationship disharmony.

Brown and Booth’s (1996) analysis of the NSFH yielded similar findings. Cohabitors are less happy and report more frequent disagreement and conflict than marrieds, on average. Whereas Nock (1995) examined variation among marrieds by distinguishing between those who had and had not cohabited premaritally, Brown and Booth explored variation among cohabitors according to whether they reported plans to marry. Roughly 75% of cohabitors report plans to marry their partners, and these cohabitors do not differ from marrieds in terms of relationship quality. The overall differential is an artifact of very low relationship quality among a minority of cohabitors—those with no eventual or definite plans to marry their partners. Perceptions of relationship instability are no exception. In general, cohabitors report less certainty that their relationship will remain intact, but cohabitors with plans to marry are no less (or more) certain than marrieds.

This finding is notable because cohabiting unions remain highly unstable, enduring for less than two years, on average, before couples either tie the knot or sever their relationship. During the 1980s and early 1990s, a majority of cohabitations were formalized through marriage. Over time, this pattern has shifted such that today, more cohabiting couples end their relationships than marry (Bumpass and Lu, 2000). Less than one-half of couples formalize their unions through marriage, although a majority of marriages are preceded by cohabitation. Relationship instability is positively associated with separation among cohabitors, however, high levels of relationship quality are not predictive of the transition from cohabitation to marriage (Brown, 2000b). Additionally, relationship instability is consequential for other domains of well-being. For instance, the poorer psychological well-being of cohabitors versus marrieds is largely attributable to the higher levels of relationship instability characterizing cohabiting unions (Brown, 2000a).

The short duration of most cohabiting relationships raises the question of whether relationship quality evolves in distinct ways for cohabitors and marrieds over time. For both groups, relationship duration is negatively associated with positive dimensions of relationship quality, such as happiness and interaction, over the first decade of the union (Brown, 2003). For negative dimensions, such as relationship instability, the role of duration depends on union type. Relationship instability does not significantly vary by relationship duration among marrieds. Among cohabitors, relationship duration is positively associated with relationship instability. Further investigation reveals that controlling for marriage plans reduces the positive effect of duration to nonsignificance. Additionally, plans to marry interact with relationship duration such that during the first few years of a cohabiting relationship, plans to marry are linked to lower levels of relationship instability. At longer durations, this association reverses, suggesting that the failure to realize marriage plans can be detrimental to relationships (even though the relationship remains intact). Cohabiting unions that are not formalized quickly through marriage are characterized not only by higher levels of instability, but also lower levels of happiness and interaction (Brown, 2003).

Both plans to marry and relationship duration may operate in unique ways among older adults. In their comparison of older (i.e., 50 and older) and younger cohabitors using the NSFH, King and Scott (2005) find that older cohabitors report higher levels of relationship quality. Specifically, older cohabitors report fewer disagreements and conflict as well as more interaction with their partners than do younger cohabitors. Relationship instability levels are also lowest for older cohabitors. At the same time, older cohabitors are in unions of longer average duration (nearly 5 years versus about 2 years) and are less likely to report marriage plans than their younger counterparts. Consistent with prior research (Brown and Booth, 1996), age is negatively associated with plans to marry. King and Scott also document an interaction between age and marriage plans such that the positive effect of plans to marry on relationship quality diminishes with age. This distinctive pattern indicates that cohabitation may serve a different purpose among older adults, operating as more of a long-term alternative to marriage. For younger adults, cohabitation often appears to be a prelude or stepping stone to marriage (King and Scott, 2005). Whether this is a true age effect or actually reflects cohort variation is unclear and the authors acknowledge that they cannot test these competing possibilities.

1.2. Cohabitation in older adulthood

The motivations for cohabitation are likely to differ among older and younger adults as each group is at a different life stage characterized by distinct opportunities, constraints, and cultures. Older adults are more likely to have been previously married. Only about 10% of cohabitors over age 50 are never-married (Brown et al., 2006). Prior marital experience is positively associated with cohabitation (Bumpass and Lu, 2000). Previously married cohabitors are less likely to report marriage plans and tend to cohabit for longer durations, perhaps because they are less sanguine about remarriage (Brown and Booth, 1996; King and Scott, 2005). They also tend to report lower levels of relationship quality. Indeed, many older unmarried adults, especially women, are relatively uninterested in marriage, yet they do desire companionship (Bulcroft and Bulcroft, 1991; Talbott, 1998). Cohabitation may offer older adults most of the benefits of marriage without imposing the institutional norms of marriage that stipulate a traditional gender-based exchange (which could translate into heavy caregiving responsibilities for women). Moreover, it allows couples to share living quarters and day-to-day expenses while maintaining financial autonomy, protecting their wealth for transfer to their offspring and ensuring that partners are not responsible for the potential financial burdens entailed by old age (e.g., long-term care). In short, cohabitation offers many of the advantages of marriage without the legal, economic, and normative constraints.

Alternatively, the absence of such constraints may not be entirely beneficial to the extent that the incomplete institutionalization of cohabitation among older adults leads to uncertainty about relationship norms, roles, and expectations. In fact, the relative rarity of cohabitation among older adults suggests it is more incompletely institutionalized for this population. As Nock (1995) posited, the incomplete institutionalization of cohabitation may operate as a relationship stressor and ultimately undermine relationship quality.

Moreover, cohabitation does not require the same level of commitment as marriage. Older adults may be willing to cohabit with someone they would not marry. Stated differently, cohabitation may be selective of those who are poor marriage material (Booth and Johnson, 1988; Nock, 1995). Older cohabitors may be less desirable partners to the extent that they are in poorer health, report more depressive symptoms, are more likely to consume alcohol, and have fewer social ties than marrieds (Brown et al., 2006, 2005; Chevan, 1996).

For older adults who came of age before the rise in cohabitation during the 1970s and 1980s, cohabitation may retain some of its stigma as a less desirable living arrangement, reflecting an era during which cohabitation was pejoratively referred to as “living in sin” or “shacking up.” Although cohabitation has gained widespread acceptance, the attitudes of older adults towards cohabitation remain less favorable (Hansen et al., 2007). Despite this potentially lingering stigma, anecdotal evidence suggests that adult children may strongly encourage their unmarried parent to cohabit rather than remarry to protect family assets and financial autonomy (Hatch, 1995). The potential influence of older adult peers is unknown.

To our knowledge, only one empirical study attempts to examine the reasons why older adults might cohabit. King and Scott (2005) explore various reasons for cohabiting, including compatibility, independence, sexual satisfaction, and commitment, but find no significant variation by age. They acknowledge that this set of reasons is limited in scope, reflecting motivations typically driving decisions to cohabit among younger adults and ignoring factors that may be unique to older adults (e.g., pressure from adult children, a desire to retain financial autonomy), and therefore we should not conclude that there are no differences in the factors motivating younger and older people to cohabit. To the contrary, they maintain that older cohabitors are distinct from their younger counterparts: “prior research on cohabitation cannot be generalized to older adults” (King and Scott, 2005, p. 283).

2. The current study

Using data from a nationally representative sample of people ages 57–85, the present study is designed to examine the association between union type and relationship quality among older adults, a largely neglected but growing share of cohabitors. This study is important for several reasons. First, the U.S. population is aging, and older adults are less likely to be married now and in the future than in the past (Cooney and Dunne, 2001), meaning that more are eligible to cohabit. Second, cohabitation is accelerating among older adults. As a larger share experiences this living arrangement, it is essential that we identify its consequences for individual well-being. Third, cohabitation appears to operate differently among older versus younger adults in terms of relationship quality (King and Scott, 2005), but how the unions of older cohabitors compare to those of marrieds is unclear.

Indeed, a review of the prior literature yields competing hypotheses. On the one hand, older cohabitors may report lower levels of relationship quality than marrieds, whether because of the poorer relationship quality characterizing cohabitors vis a vis marrieds in the general population (Brown and Booth, 1996; Nock, 1995), the incomplete institutionalization of cohabitation in later life (cf. Nock, 1995; King and Scott, 2005), selection factors that result in poorer matches through cohabitation (Chevan, 1996), or the stigma associated with unmarried partnerships (Hansen et al., 2007). On the other hand, older cohabitors seemingly enjoy many of the benefits of marriage, their unions tend to persist over time, and cohabitation may serve as an alternative to marriage (King and Scott, 2005), leading to the hypothesis of no difference in relationship quality between older cohabitors and marrieds.

In addition to establishing the bivariate association between union type and relationship quality, we also consider the roles of factors related to either union type or relationship quality, including demographic characteristics, economic resources, health, and social support (Brown et al., 2006; Brown and Booth, 1996; Chevan, 1996; King and Scott, 2005; Nock, 1995). Salient demographic factors include gender, age, race, and union duration. Older men are more likely to cohabit (or be married) than older women and cohabitors are younger than marrieds, on average (Brown et al., 2006). White marrieds tend to report higher levels of relationship quality than Black marrieds (Bulanda and Brown, 2008), although there do not appear to be race differences in relationship quality among cohabitors (Brown, 2003). Union duration is longer, on average, for marrieds than cohabitors (Brown and Booth, 1996; Nock, 1995), and the association between union duration and relationship quality differs for cohabitors and marrieds (Brown, 2003).

Economic factors, including education, employment, and wealth, are associated with union type and relationship quality. Education and employment are negatively related to cohabitation and positively related to marriage (Smock and Manning, 1997), although among older adults, cohabitors are more likely to be working than marrieds (Brown et al., 2006), perhaps because they are younger, on average. Given our focus on older adults, many of whom are not working, a measure of assets rather than income seems more appropriate. Prior research on older adults reveals no statistical difference in household income between cohabitors and marrieds (Brown et al., 2006). Whether assets (an indicator of wealth) operate similarly remains an empirical question. Income is positively associated with relationship quality (Rogers and DeBoer, 2001).

Health measures of interest include having private health insurance, activities of daily living (ADL) limitations, and alcohol consumption. Older cohabitors are less likely than older marrieds to have private health insurance and are more likely to consume alcohol (Brown et al., 2006). The two groups do not differ in terms of ADLs (Brown et al.), but physical health is linked to relationship quality (Umberson et al., 2006).

Social support, including religiosity and having living children, is weaker among older cohabitors than marrieds (Brown et al., 2006). As an incomplete institution, cohabitation tends to be characterized by less social support, which in turn can undermine relationship quality (Eggebeen, 2005; Ross, 1995).

3. Method

Data come from the 2005 to 2006 National Social Life, Health, and Aging Project (NSHAP), a nationally representative sample of 3005 community dwelling persons ages 57–85 (i.e., persons born 1920–1947). Fielded by the National Opinion Research Center and the University of Chicago, the sample design was developed by the Health and Retirement Study (HRS), using their household screening process. Of the 4400 persons selected from screened households, 92% were eligible for inclusion in the NSHAP and of those, 76% completed the NSHAP interview. The NSHAP includes an in-person interview, a self-administered questionnaire, and a biomeasures collection. The topics covered by the NSHAP are wide ranging: demographic characteristics, sexual and union histories, social networks, physical and mental health, well-being and illness, and social and cultural activities.

These data are particularly well-suited to our research questions for several reasons. First, the NSHAP distinguishes cohabitors from other unmarrieds in a recently fielded sample, permitting us to draw conclusions about the current population of older cohabitors in the U.S. Second, there are detailed marital and cohabiting histories for all respondents that include the start and end dates of unions, allowing us to calculate current relationship duration. Third, several indicators of both positive and negative relationship quality are captured. Positive dimensions include relationship happiness, emotional satisfaction, openness, time spent together, and physical pleasure. Negative dimensions measure partner demands as well as criticism by partner. Although the HRS is a larger sample with more cohabitors, it does not have measures of relationship quality and thus is not appropriate for the current study.

Nearly all older cohabitors are previously married, but a majority of older marrieds are in long-term first marriages (69% of marrieds in the NSHAP are in first marriages). Prior marital experience is related to union type (Bumpass and Lu, 2000) and there is mixed evidence about its association with relationship quality (Coleman et al., 2000). For ease of comparison, we restrict our analyses to previously married cohabitors (N = 54) and remarrieds (N = 558). This restriction only omits six cohabitors who are never-married. It also excludes 1243 respondents in first marriages. The modest size sample of cohabitors is in line with what would be expected based on population level estimates. Nearly 2% of persons over age 60 are cohabiting (Brown et al., 2005), which would yield about 60 cohabitors from a sample of 3005 respondents. Still, some of our independent variables are measured as dummies rather than ordinal or interval level variables because of small cell sizes.

3.1. Measures

3.1.1. Dependent variables

The NSHAP includes several measures of relationship quality. There are five measures of positive relationship quality. Happiness is a dummy variable that derived from a seven point scale ranging from 1 = very unhappy to 7 = very happy such that those who are very happy (i.e., a 6 or 7 on the scale) are coded 1 and all others are coded 0¹. Emotional satisfaction is a dummy variable that distinguishes between those who are 1 = extremely or very emotionally satisfied with their relationship versus those who are 0 = moderately, slightly, or not at all satisfied. Pleasure, which measures the extent to which respondents view their relationship as physically pleasurable differentiates those responding extremely or very (=1) from those who report their relationship is moderately, slightly, or not at all pleasurable (=0). Openness gauges how often respondents can open up to their partner about their worries: 1 = often versus 0 = some of the time or hardly ever (or never). Time together distinguishes those respondents who report liking to spend free time: 1 = together versus 0 = some together, some apart or separate. There are two measures of negative relationship quality. Criticizes captures how often respondents believe their partner criticizes them: 1 = often versus 0 = some of the time or hardly ever (or never). Similarly, demands gauge how often the partner makes too many demands of respondents: 1 = often versus 0 = some of the time or hardly ever (or never).

3.1.2. Independent variable

The focal independent variable, union type, distinguishes between 1 = previously married cohabitors and 0 = remarrieds.

3.1.3. Control variables

Control variables include a series of demographic characteristics, economic factors, health measures, and social support. Demographic characteristics are gender (1 = man, 0 = woman), age (coded in years), race (1 = white, 0 = nonwhite), and union duration (measured in years). Economic factors include education (1 = some college or more, 0 = high school degree or less), working (1 = yes, 0 = no), and assets (dollar value). Health measures are private health insurance (1 = yes, 0 = no), any activities of daily living (ADL) limitations² (1 = yes, 0 = no) and alcohol consumption (1 = yes, 0 = no). Social support variables are respondent has one or more living children (1 = yes, 0 = no) and regular (i.e., at least weekly) attendance at religious services (1 = yes, 0 = no).

3.2. Analytic strategy

Descriptive statistics (means or percentages, as appropriate) are presented for all variables, separately for cohabitors and remarrieds (hereafter referred to simply as marrieds), to establish bivariate comparisons. Then, since all dependent variables are binary, logistic regression models are estimated in which relationship quality is regressed on union type and the controls. The table includes both the coefficients and the odds ratios for each covariate. Odds ratios of less than one indicate a negative association with the dependent variable whereas odds ratios greater than one reflect a positive relationship. Supplemental models explore whether union type interacts with union duration (cf. Brown, 2003). All estimates are generated in Stata using the svy commands to account for the complex sampling design of the NSHAP, which is cluster-based.

4. Results

4.1. Descriptive results

Table 1 shows the descriptive statistics for all variables by union type. For the most part, cohabitors and marrieds report similar levels of relationship quality. On four out of five dimensions (emotional satisfaction, pleasure, openness, and time together) of positive relationship quality, the two groups do not significantly differ. Cohabitors are less likely to report being very happy in their relationship though. Roughly 47% of cohabitors versus 63% of marrieds report a very happy relationship. There is no evidence that cohabitors experience higher levels of negative relationship quality. The two groups are similarly likely (p < 0.10) to report being frequently criticized by their partner (or spouse). And, cohabitors are actually less likely to report that their partner often makes unreasonable demands of them than marrieds. Whereas 38% of marrieds report that their spouse often makes unreasonable demands of them, only 23% of cohabitors report this frequency.

Table 1.

Descriptive statistics (means or percentages) for all variables used in the analyses, by union type.

Variable	Union type
	Cohabiting	Remarried
Dependent variables
Happiness	46.8%	63.2%*
Emotional satisfaction	76.5%	78.6%
Pleasure	72.2%	77.2%
Openness	76.1%	75.1%
Time together	51.7%	54.6%
Criticizes	31.5%	42.8%+
Demands	23.2%	37.9%**
Demographic characteristics
Gender (1 = man)	44.0%	62.0%*
Age (in years)	65.6	66.3
Race (1 = white)	84.9%	88.6%
Union duration (in years)	9.3	20.6
Economic resources
Education (1 → H.S.)	57.8%	60.8%
Employment (1 = working)	46.6%	43.5%
Assets ($1,000s)	283.7	555.5**
Health
Private insurance	61.9%	59.6%
ADLs	50.3%	45.7%
Alcohol consumption	64.2%	64.7%
Social support
Living children	85.9%	88.4%
Regular religious attendance	7.2%	36.6%***
N	54	558

⁺p < 0.10.

^*p < 0.05.

^**p < 0.01.

^***p < 0.001 (one-tailed tests).

Across the various sets of controls, there are also few differences between cohabitors and marrieds. The two groups are similar in terms of age and race. Cohabitors are less likely than marrieds to be men. Whereas 62% of marrieds are men, just 44% of cohabitors are men. Union duration is much shorter among cohabitors, who average about 9 years, versus marrieds at roughly 20 years. Similar proportions of cohabitors and marrieds have more education than a high school degree and are working. Cohabitors have significantly fewer assets than marrieds. The former group averages $283,000 whereas the average among the latter group is $555,000. Cohabitors and marrieds are similarly likely to have private insurance, to report having an ADL, and to consume alcohol. Although the two groups do not differ in the proportion with living children, a higher percentage of marrieds (37%) than cohabitors (7%) attend religious services regularly.

4.2. Multivariate results

The full models predicting each of the seven dimensions of relationship quality are shown in Table 2. Notably, the inclusion of the control variables does not substantively alter the pattern of the bivariate association between union type and relationship happiness as shown in the first panel. The odds that cohabitors report a very happy relationship is only about one-half that of marrieds. This finding is consistent with our initial hypothesis drawn from prior research that has shown cohabitors are typically less happy than marrieds. Men and whites tend to be more likely to report happy relationships than women and nonwhites, respectively. Union duration is negatively associated with relationship happiness. The odds that those with at least some college are very happy are about 28% lower than the odds for those with no more than a high school diploma. Regular religious attendance is associated with higher odds of being very happy in the relationship.

Table 2.

Coefficients and odds ratios from logistic regression models predicting relationship quality for cohabitors versus marrieds, net of controls (N = 612).

Relationship quality	Happiness		Emotional satisf.		Pleasure		Openness		Time together		Criticizes		Demands
	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR
Union type
Cohabiting	−0.65 **	0.52	−0.22	0.80	−0.51	0.60	0.06	1.06	−0.41	0.66	−0.15	0.86	−0.53	0.59
Demographic characteristics
Gender (1 = man)	0.53 **	1.70	0.84 ***	2.32	0.51 ***	1.67	0.64 **	1.90	−0.12	0.89	0.87 ***	2.39	0.55 ***	1.73
Age (in years)	0.00	1.00	0.02	1.02	0.02	1.02	−0.02 +	0.98	0.02	1.02	−0.01	0.99	−0.02 **	0.98
Race (1 = white)	0.66 **	1.93	0.20	1.22	0.50	1.65	0.43	1.54	0.07	1.07	−0.35	0.70	0.08	1.08
Union duration (in years)	−0.02 ***	0.98	−0.03 ***	0.97	−0.04 ***	0.96	−0.03 ***	0.97	−0.03 ***	0.97	0.02 ***	1.02	0.01	1.01
Economic resources
Education (1 → H.S.)	−0.33 *	0.72	0.28	1.32	−0.02	0.98	0.29 +	1.34	−0.14	0.87	−0.31	0.73	0.20	1.22
Employment (1 = working)	−0.07	0.93	0.29	1.34	0.01	1.01	−0.16	0.85	0.09	1.09	0.11	1.12	0.10	1.11
Assets ($1,000s)	0.00	1.00	0.00	1.00	0.00	1.00	0.00	1.00	0.00	1.00	0.00 **	1.00	0.00	1.00
Health
Private insurance	−0.04	0.96	0.01	1.01	0.03	1.03	0.08	1.08	0.10	1.11	0.25	1.28	−0.43 **	0.65
ADLs	−0.12	0.89	−0.08	0.92	0.22	1.25	0.14	1.15	−0.14	0.87	−0.12	0.89	0.16	1.17
Alcohol consumption	−0.24	0.79	0.10	1.11	−0.03	0.97	0.16	1.17	0.02	1.02	0.12	1.13	0.44 *	1.55
Social support
Living children	0.29	1.34	0.32	1.38	0.34	1.40	0.41	1.51	0.29	1.34	−1.01 ***	0.36	−0.45	0.64
Regular religious attendance	0.50 **	1.65	0.35	1.42	0.36	1.43	0.91 ***	2.48	0.13	1.14	−0.22	0.80	−0.08	0.92
F (13, 38)	1.92 *		5.42 ***		2.50 **		3.70 ***		2.55 ***		4.63 ***		2.68 **

⁺p < 0.10.

^*p < 0.05.

^**p < 0.01.

^***p < 0.001 (one-tailed tests).

For the other dimensions of positive relationship quality, there are no significant differences between cohabitors and marrieds, which support our competing hypothesis that among older adults, the benefits from cohabitation appear comparable to marriage. Net of controls, cohabitors report similar levels of emotional satisfaction, pleasure, openness, and time spent together. There are some notable linkages between the various controls and relationship quality. The second and third columns show the models for emotional satisfaction and pleasure, respectively. For both dimensions, men are more likely than women to report high quality. And, union duration is negatively related to both dimensions of relationship quality. Although cohabitors and marrieds do not significantly differ in their likelihoods of reporting that they can open up to their partners about their worries, men and those with at least some post-high school education are more likely to report high openness. Age and union duration are negatively associated with relationship openness. Similarly, as union duration increases, the odds of spending time together often decrease.

Turning now to the two negative dimensions of relationship quality, criticism and demands by the partner, union type differences documented in the bivariate diminish to nonsignificance with the inclusion of controls, again supporting the competing hypothesis of no differences between cohabitors and marrieds. The odds that men report their partner is critical of them is 2.39 times that of women. A longer union duration is associated with an increase in the odds of frequent criticism. Assets exhibit a similar pattern. Having living children is associated with reduced odds of reporting a critical partner. Men are more likely than women to report having a demanding partner. With age, the likelihood of reporting a partner that is often demanding appears to decline. Private insurance is negatively associated with a demanding partner, whereas alcohol consumption is positively related to a demanding partner.

The modest number of cohabitors precludes estimating models separately for men and women. We tested gender by union type interactions (results not shown) to determine whether the linkage between cohabitation (or marriage) and relationship quality differs by gender. Two significant interactions emerged. First, there is a negative interaction between gender and union type (coef. = −1.26, p < 0.01) on pleasure such that cohabiting men are much less likely than married men to report a very pleasurable relationship. There is no union type effect among women. Married women are also less likely than married men to report a very pleasurable relationship, but among cohabitors, there is no gender difference. Second, gender and union type interact negatively (coef = −1.56, p < 0.01) in their effects on criticism. Married men are especially likely to perceive their spouses as highly critical versus married women. There is no gender effect among cohabitors. Married men are more likely than cohabiting men to report a highly critical spouse, but there is no union type difference among women.

Supplemental analyses were conducted to determine whether union type and union duration interact in their effects on relationship quality as documented by prior research using NSFH (Brown, 2003). Among older adults in the NSHAP, none of these interactions attains statistical significance (results not shown), suggesting that union duration’s negative association with relationship quality follows a similar linear trajectory for both cohabitors and marrieds.

5. Discussion

Cohabitation is increasingly common among adults of all ages, including older adults. Although only about 2% of the population over age 50 is currently cohabiting (6% of unmarrieds over age 50 cohabit), this figure represents over 2 million persons, and prevailing trends portend an acceleration of cohabitation among this age group (Brown et al., 2006; Cooney and Dunne, 2001). The share of older adults who are unmarried continues to climb, and baby boomers—the first cohort to cohabit in large numbers during young adulthood—are moving into older adulthood. Still, very little is known about older cohabitors. Prior research offers a demographic portrait of this group (Brown et al., 2005; Chevan, 1996), but does not advance our knowledge of how cohabitation operates as a union context among older adults. A study by King and Scott (2005) comparing the relationship quality of older and younger cohabitors coupled with the literature on living arrangements and partnering in later life (Blieszner, 2007; Cooney and Dunne, 2001) suggest that cohabitation may be distinctive in older adulthood, perhaps operating as a long-term alternative to marriage.

To evaluate the role of cohabitation in older adulthood, we compare the relationship quality of cohabiting and married adults age 57–85 using data from the 2005 to 2006 NSHAP. Early research on relationship quality in cohabiting versus marital unions reveals that cohabitors are in unions that are relatively short-lived, characterized by greater relationship instability and lower levels of relationship quality (Brown, 2003; Brown and Booth, 1996; Nock, 1995). But compared to younger cohabitors, older cohabitors are in relationships of higher quality and longer duration, on average (King and Scott, 2005), leading us to anticipate that the variation between older cohabitors and marrieds may not be especially large.

Indeed, our results show very few differences in relationship quality between the two groups. Although cohabitors are less likely to be very happy in their relationships than are marrieds, the two groups are similarly likely to report high levels of emotional satisfaction, time together, openness, and physical pleasure. They also are about as likely to report that their partner (or spouse) is often critical of them (this association is marginally significant, p < 0.10). And, cohabitors are no more likely to report their partner often makes too many demands of them; in fact, the findings support the opposite conclusion. Marrieds are more likely to be highly demanding than cohabitors. Controlling for demographic characteristics, economic resources, health, and social support, which prior research (Brown et al., 2006; Chevan, 1996) shows to be related to union type among older adults, does not substantively change the bivariate associations, although the union type differential on demands reduces to nonsignificance.

Taken together, it seems cohabitors and marrieds enjoy similar levels of relationship quality in older adulthood. A key advantage of this study is that it examines multiple indicators of both positive and negative relationship quality. On six of the seven dimensions, cohabitors fare no worse than their married counterparts. Notably, cohabitors are less likely than marrieds to report that they are very happy in their relationships, a finding that is in line with prior research on the general population using NSFH data (Brown and Booth, 1996; Nock, 1995). This global measure of relationship quality may capture underlying variation in general life happiness between the two groups, and the literature shows cohabitors tend to exhibit more depressive symptoms, on average, than do marrieds (Brown, 2000b; Brown et al., 2005).

Our study provides preliminary evidence that the linkages between union type and relationship quality may differ for men and women. Despite the modest number of cohabitors in the sample, we document both a union type effect among men (but not women) and a gender effect among marrieds (but not cohabitors) for two of the dimensions of relationship quality, pleasure and criticism. The salience of gender in older adult unions merits greater attention in future research.

This research shows that union duration is negatively associated with all dimensions of relationship quality, except high demands. Over time, relationship quality among older cohabitors and marrieds follows a similar linear, downward path; union type and union duration do not significantly interact in their effects on relationship quality. This finding differs from that established by Brown (2003) using NSFH data to examine cohabitors and marrieds in the first ten years of the relationship, and provides additional evidence that cohabitation in later life is distinct from cohabitation at earlier life course stages. Moreover, the relatively long duration of cohabiting unions among older adults is in line with the results obtained by King and Scott (2005) and supports their assertion that cohabitation operates as a long-term alternative to marriage in later life.

This study has some limitations. First, the number of cohabitors in the NSHAP is small (although the proportion of the sample that is cohabiting is consistent with the proportion in the population), reducing the statistical power of the analyses. The models test directional hypotheses, so one-tailed tests for statistical significance were used. Still, the occurrence of a type II error (i.e., nonsignificant coefficients are actually significant) is not unlikely. Ancillary power analyses (not shown) suggest that the sample provides adequate power to detect differences of the magnitude obtained for relationship happiness, but more modest differences in relationship quality between cohabitors and marrieds are less likely to be detected as statistically significant because of the small number of cohabitors in the sample. Second, the small sample of cohabitors precludes gender-specific analyses, although we were able to establish that gender and union type sometimes interact in their effects on relationship quality. Third, we rely on a rather basic set of controls to maximize the degrees of freedom. Some controls are simplified into dummies because several nominal categories (e.g., race, education, and health insurance variables) are not tenable due to small cell sizes. Fourth, the NSHAP does not include measures of either plans to marry or relationship instability. Plans to marry may be related to relationship quality (Brown and Booth, 1996), although this linkage could be rather weak among older adults (King and Scott, 2005). Relationship instability would inform conclusions about the meaning of cohabitation in later life (e.g., levels similar to those observed for marrieds would be in line with the notion that cohabitation is an alternative to marriage for older adults). Finally, the analyses are cross-sectional, meaning that we cannot establish causality. Nor can we easily evaluate the possibility of selection. Cross-sectional analyses disproportionately represent longer-term unions as short-term (and perhaps poorer quality) unions are selected out through break up. This is not a big concern unless cohabitors and marrieds have different thresholds of relationship quality for terminating their relationships (e.g., the lower commitment levels associated with cohabitation may translate into a higher likelihood of break up at a given level of relationship quality).

Nonetheless, this study makes important contributions to both the family and gerontology literatures. It demonstrates that unlike the pattern documented among younger adults, for whom cohabitation is typically associated with lower levels of relationship quality than is marriage (Brown, 2003; Brown and Booth, 1996; Nock, 1995), in older adulthood, cohabitors and marrieds enjoy similar levels of relationship quality. Although union duration is related to union type (cohabiting unions are more recently formed), average duration among both groups far exceeds the overall average among cohabitors (about 2 years, according to Bumpass and Lu (2000)), suggesting both union types provide stability and tend to persist over time. Indeed, the linkage between union duration and relationship quality follows the same trajectory regardless of union type, unlike the distinctive patterns documented among younger adults (Brown, 2003). This study provides new evidence that cohabitation is a long-term alternative to marriage among older adults, which is consistent with prior research (King and Scott, 2005). Cohabitation appears to offer many of the benefits of marriage in older adulthood, namely comparable relationship quality and an enduring union. The extent to which comparable benefits prevail in other domains of well-being, including physical health and mortality, await future research.