Older married couples’ sexual expression: A dyadic latent class analysis

Abstract

Many older adults continue to engage in sexual behavior, most often in the context of long-term marriages. Prior studies have tended to examine older adults’ sexual behavior, desires and motivations separately. Thus, there remain considerable gaps in our understanding of the multifaceted-nature of older couples’ sexual expression. Using dyadic latent class analysis and data on 953 heterosexual couples in long-term marriages from Wave 2 of the National Social Life, Health, and Aging Project (NSHAP), we describe configurations of couples’ sexual expression. Four latent classes were identified. Within each latent class, couples were largely in agreement about their sexual behaviors, but husbands consistently expressed greater desire and motivation than their wives. In the highest engagement class, couples reported more frequent and wide-ranging activities (vaginal sex, oral sex, intimate touching), more frequent thinking about sex, greater satisfaction with their amount of sexual activity, and placed more importance on sex as part of their lives. In the lowest engagement class, both spouses indicated no sexual activity in the past year, but neither spouse was content with this situation even as both agreed sex was not an important part of life. Demographic, relational, and health factors were associated with membership in different classes. Health factors—especially sexual difficulties—were most predictive, with husbands’ health more associated with latent class membership than wives’ health. Given the complex and multifaceted nature of sexual expression among older married couples, geriatricians and other health practitioners should provide care that recognizes a wider array of activities and the potential for unmet need. Sexual expression is an important component of older adults’ health and well-being.

Sexual activity is an important for adult well-being, and—although activity (Lindau et al., 2007) and quality (Quinn-Nilas, 2020) decrease with age—many older adults continue to engage in sexual behavior. Indeed, more than half of older adults are sexually active (Addis et al., 2006). Prior research indicates that age-related health changes help explain lower sexual activity among partnered older adults (DeLamater, 2012). Nevertheless, even as they recognize that sexual behavior changes with age, older adults believe that a satisfactory sex life is important for maintaining their romantic relationship (Waite et al., 2009).

Sexual behavior in long-term heterosexual marriages is negotiated between spouses (Lodge & Umberson, 2012) and thus a growing number of studies use dyadic data to examine older adults’ sexual expression (e.g., Kwon & Schafer, 2017; Waite et al., 2017). These studies often focus on either overall sexual activity or separate behavioral and attitudinal measures. As a result, there is limited knowledge of how behaviors, desires, and motivations coalesce together at the couple-level to shape sexual expression. Examining multiple components of sexual expression simultaneously is important because although spouses’ reports of partnered sexual behaviors are likely to be similar, the sexual desires and motivations of wives and husbands may or may not be (Lodge & Umberson, 2012). Differences between spouses are especially likely as known gender differences in sexual attitudes (Carpenter, 2010; Lee et al., 2016) may lead to sexual frequency that is less than desired for one or both partners. Moreover, even when spouses’ motivations, desires, and activities are similar, considerable heterogeneity may exist between couples. Thus, there remain considerable gaps in our understanding of the multifaceted-nature of older couples’ sexual expression. We address these gaps using dyadic latent class analysis and data from 953 heterosexual older couples in the National Social Life, Health, and Aging Project (NSHAP).

Background

Older adults’ sexual behavior, motivations, and desires

Sexual expression is not simply the physiological capacity to engage in behavior with a partner, but also involves attitudes developed within a set of culturally-defined beliefs and psychological dispositions. Therefore, building on Waite and colleagues (2017, 2009), we conceptualize older adults’ sexual expression as the combination of sexual activities, motivations, and desires. Although there has been research on each of these sexual expression dimensions separately, little is known about how sexual behaviors, desires, and motivations are simultaneously expressed among couples.

Many adults maintain sexual activities, express motivation for sexual intimacy, and hold attitudes favorable toward sex well into their 70s and 80s (DeLamater, 2012). For example, 58% of U.S. adults over age 45 believe sex is critical to a good relationship, and 42% agree that sex is important to quality of life (Fisher, 2010). More positive attitudes toward sex are associated with higher levels of desire and more frequent sexual activity (DeLamater, 2012; DeLamater & Sill, 2005; Waite et al., 2017). Indeed, more than two-thirds of older adults have had sexual intercourse in the past year (Lindau et al., 2007), with a majority reporting at least twice a month frequency (Addis et al., 2006; Waite et al., 2009). Partnered sexual activity nearly always includes vaginal intercourse, although intimate touching and (to a lesser extent) oral sex are also relatively common (Lindau et al., 2007; Liu et al., 2018; Waite et al., 2009). Solo masturbation is also fairly prevalent (Addis et al., 2006; Herbenick et al., 2010), which signals an underlying desire for sexual intimacy (Waite et al., 2009). Among current cohorts of older adults, sexual behavior occurs largely within marriage as rates of non- and extramarital sex are low (Fisher, 2010; Waite et al., 2017).

Demographic and health correlates of older adults’ sexual expression

Declining sexual frequency and quality with age are well-established (e.g., Forbes et al., 2017; Lindau & Gavrilova, 2010), but other individual and couple characteristics are also correlated with older adults’ sexual behaviors and attitudes. Nonwhites appear to hold more positive attitudes toward sex, place greater importance on sexual activity for well-being (Fisher, 2010), and have more frequent sexual activity (Karraker & DeLamater, 2013; Waite et al., 2017)—although findings depend on sample composition. More educated older adults, especially those with a college degree, have more frequent sexual activity (Allen & Desille, 2017; Stroope et al., 2015), greater levels of desire (DeLamater & Sill, 2005), and fewer sexual difficulties (Laumann et al., 2008). Income is similarly associated with more frequent sexual activity (Addis et al., 2006). Remarried older adults and those whose relationships are of shorter durations also have more frequent sexual activity (Karraker & DeLamater, 2013; Stroope et al., 2015). The quality of the marriage likewise matters as emotional satisfaction and physical pleasure are associated with more frequent sexual activity (Thomas et al., 2015; Waite et al., 2017). Indeed, among long-term married couples, relationship quality and sexual quality tend to change in tandem over time (Quinn-Nilas, 2020).

Health is important for sexual expression. As people age, the probability of physiological changes that present barriers to sexual activity increases. Some of these changes are related to health behaviors, chronic diseases, and other health conditions that interfere with sexual activity (DeLamater & Karraker, 2009; Lee et al., 2016; Syme et al., 2013). For example, physical activity and moderate alcohol use are positively associated with more frequent sexual activity and thoughts about sex (Allen & Desille, 2017), while being non-obese is associated with higher sexual activity (Kwon & Schafer, 2017). Such health behaviors are especially important for men because good cardiovascular health is linked to erectile function (Allen & Desille, 2017). Indeed, older adults with hypertension, heart disease, or diabetes—conditions that affect the vascular system—have less frequent intercourse (Lindau et al., 2007). Having multiple chronic conditions and functional limitations is also negatively related to sexual activity (Kwon & Schafer, 2017; Stroope et al., 2015).

Other physiological changes stem from hormonal shifts that are part of normal aging. Declines in testosterone that affect men’s arousal response and erection strength (DeLamater, 2012) and declines in estrogen that may lead to vaginal thinning and difficulty lubricating among women (Addis et al., 2006; DeLamater & Karraker, 2009; Laumann et al., 2008). Such sexual health problems are associated with reduced odds of sexual activity—especially erectile problems for men (Kwon & Schafer, 2017; Waite et al., 2017) and pain during intercourse for women (Syme et al., 2013).

Older adults’ overall perceptions of their health also are consequential for multiple components of sexual expression. For example, poorer self-rated physical and mental health is associated with less frequent sexual intercourse (Addis et al., 2006; Fisher, 2010; Lindau & Gavrilova, 2010) and oral sex (Herbenick et al., 2010). Poor self-rated health is also associated with thinking about sex less often (Lee et al., 2016) and lower sexual interest (Iveniuk & Waite, 2018).

Gender and older adult couples’ sexual expression

Sexual expression is the product of individuals’ cumulative lifetime experiences within gendered sexual scripts (Carpenter, 2010). Sexual scripts generally refer to how individuals create and understand both their own and other’s sexual behavior. Especially for current cohorts of older adults, these gendered sexual scripts are associated with well-established differences between men and women even as heterosexual spouses both view sex as essential to gendered sexual expression (Lodge & Umberson, 2012). Men are socialized to want, desire, and pursue sex more than women. Thus, it is not surprising that men report higher sexual interest, activity, and quality compared to women (e.g., DeLamater, 2012; Forbes et al., 2017; Waite et al., 2009). For example, Fisher (2010) found that middle-aged and older men are more likely than women to think: about sex at least once each day (45% vs. 8%), that sex is important to their overall quality of life (85% vs. 61%), and key to a good relationship (67% vs. 50%). These attitudinal differences mirror differences in behavior. Older men report more frequent intercourse, more often engaging in oral sex, and more frequent solo masturbation in the past year (Fisher, 2010; Karraker & DeLamater, 2013). Men also indicate higher sexual satisfaction than women (Syme et al., 2013).

Given gendered sexual scripts, most men view engaging in sexual behavior as essential to masculinity and express more worry about the amount of sex they are having, their level of desire, and physiological performance than do women (Lee et al., 2016). Such worries may be more important for older men as age-related health changes mean an increased prevalence of erectile difficulties. For example, although just 16% of men ages 55–59 report erectile difficulties, 88% of men over age 80 do so (Lee et al., 2016).

Sexual expression—and femininity more broadly—for current cohorts of older women is more tied to the spousal role than is true for men. Indeed, older women’s sexual expression is defined almost exclusively within marital unions (Carpenter, 2010). This may be why hormone-related changes in women’s sexual functioning appear less associated with partnered sexual activities (Kwon & Schafer, 2017; Waite et al., 2017) and physical pleasure (Thomas et al., 2015). Thus, unlike among men, there is a less clear connection between specific health conditions and sexual activity and desire among women (Karraker et al., 2011). Nevertheless, women who report lower self-rated health have less pleasurable partnered sex (Laumann et al., 2008) and are less likely to engage in solo masturbation (Lindau et al., 2007).

Current study aims

We pursue three research aims. First, we use dyadic latent class analysis (LCA) to describe the most common configurations of sexual expression among heterosexual older couples. We expect that older adults’ sexual behaviors, motivations, and desires combine in heterogeneous ways. Given that older adults’ sexual activity primarily occurs within the context of long-term marriage and fidelity norms (Waite et al., 2009), we expect agreement between wives’ and husbands’ reports of sexual behavior, but within-couple differences in motivations and desires consistent with gendered sexual scripts. Second, we examine differences in the predictors of latent class membership, hypothesizing that younger age, higher socioeconomic status, better relationship quality, and better health will be associated with couples’ membership in latent sexual expression classes marked by higher levels of engagement and greater agreement among partners. We anticipate health-related deficits (e.g., chronic illness) and sexual difficulties (e.g., pain during intercourse) will predict couples’ membership in classes capturing lower sexual engagement and greater disagreement among spouses on indicators of desire and motivation. Finally, given the extent to which gendered sexual scripts focus on penetrative sexual activities, we expect that husbands’ health-related factors and sexual functioning problems will be more important for predicting dyadic sexual expression latent classes than will wives’ health-related factors.

Methods

We used data from the National Social Life, Health, and Aging Project (NSHAP). NSHAP is a nationally-representative sample of community-dwelling U.S. older adults who were ages 57–85 at Wave 1 in 2005–2006. Wave 2 (n = 3,377) was collected in 2010–2011 and included the original wave 1 respondents (n = 2,261), non-interviewed wave 1 respondents (n = 161), and their partners (n = 955). Data were collected through an in-home interview and a self-administered Leave Behind Questionnaire (LBQ). 87.3% of Wave 2 respondents returned the LBQ. For the current analysis, we used the Wave 2 dyadic sample of 953 heterosexual couples, excluding the two same-sex couples. Nearly all of these couples were married (95.6%).¹

Measures

Sexual expression.

Eight indicators measured three components of sexual expression: sexual behavior, desires, and motivations. We used four indicators of partnered sexual behavior. Frequency of Sex activities with a spouse in the past year was coded 1 = “≤Once a month,” 2 = “2–3 times a month,” and 3 = “≥Once a week.” NSHAP uses a broad definition of sexual behavior, instructing respondents to include “any mutually voluntary activity with another … that involves sexual contact, whether or not intercourse or orgasm occurs.” Respondents were also asked how often in the past year these partnered sexual activities included Vaginal Sex (intercourse), receiving Oral Sex, and Intimate Touch (partner touching genitals with hands); all coded 1 = “never/rarely,” 2 = “sometimes/usually,” and 3 = “always.”

Two indicators measured respondents’ underlying desire for sexual activity (Waite et al., 2009). Solo Masturbation was dummy coded 1 = any masturbation during the past year given high skew. Satisfaction with the amount of partnered sexual activity in the past year (Sex Ok) was coded 1 = “less often than,” 2 = “about as often as,” and 3 = “more often than” desired.

The final two indicators were attitudes reflecting respondents’ motivations for sexual activity (Allen & Desille, 2017). The frequency respondents’ Think about Sex was coded 1 = “≤Once a month,” 2 = “1–3 times a month,” and 3 = “≥Once a week.” Respondents’ assessment of the Importance of Sex in their lives was coded 1 = “not at all/somewhat important,” 2 = “moderately important,” and 3 = “very/extremely important.”

Demographics.

We included seven demographic variables. Individual-level factors included Age (whole years; centered at 62 in the multivariate analyses), a dummy variable for Nonwhite (=1), a dummy variable for at least a College Degree (=1), and a dummy variable for Married Twice or more (=1). The remaining three variables were the couple-level indicators. Household Income was an ordinal measure coded 0 < $25,000 to 3 ≥ $100,000 per year. Relationship Duration was coded in five-year increments from 0 = “<5 years” to 10 = “≥50 years.” Finally, a dummy variable for Currently Cohabiting (=1) controlled for legal status differences.

Relationship quality.

We used three qualitative assessments on the marital relationship. Respondents reported how Physically Pleasurable and how Emotional Satisfying respondents found their relationship (1 = “somewhat,” 2 = “very,” and 3 = “extremely”). Marital Happiness was assessed on a scale of 1 to 7, which due to skew we recoded 1 = “unhappy” (1–4), 2 = “somewhat happy” (5–6), and 3 = “very happy” (7).

Health behaviors.

Four questions gauged respondents’ health behaviors. Vigorous Exercise was a dummy variable (=1) for “3 or more times per week” on average in the past year, where vigorous exercise meant ≥30 min “of things like sports, exercise classes, heavy housework, or a job that involves physical labor.” Heavy Drinking as a dummy variable (=1) for “3 or more drinks” per day, on average, in the last 3 months. Current Smoker was a dummy variable (=1). Finally, respondents were coded Obese (=1) if their body mass index was over 30.

Health status.

Respondent health was measured with four variables. Self-Rated Physical Health and Self-Rated Mental Health were coded from 0 = “poor” to 4 = “excellent.” Chronic Illness was a count (0–7) of seven self-reported physician diagnoses (e.g., hypertension, arthritis, diabetes). Functional Limitations was a count (0–7) of self-reported difficulty with seven functional tasks (e.g., walking several blocks, dressing, eating).

Sexual difficulties.

We included four dummy variables for sexual difficulties lasting “several months or more in the past year.” Wives reported on Problem Lubricating (=1) and physical Pain During Sex (intercourse; =1). Husbands’ had Erectile Problems (=1) if they reported “trouble getting or maintaining an erection.” Finally, Climaxing Too Quickly was coded 1 if either spouse reported they experienced orgasm too quickly.²

Analytic strategy

We used dyadic Latent Class Analysis (LCA) to examine married older couples’ sexual expression. LCA classifies couples into latent sexual expression groups based on each spouses’ response patterns to the observed sexual behavior, desire, and motivation indicators (Geiser, 2012). We used the 3-step estimation (R3STEP) procedure in Mplus, which allows for exploration of how latent sexual expression classes are associated with observed predictors while simultaneously accounting for the inherent uncertainty in assigning cases to their most likely class (see Asparouhov & Muthén, 2014 for full estimation details).

The first step in the 3-step procedure was the estimation of latent classes based on the observed indicators. We examined the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Entropy statistics to assist in determining the appropriate number of latent classes. Smaller AIC and BIC values suggest better model fit. Higher entropy values indicate better class separation. We estimated 2–7 classes. We considered the meanings of the classes, parsimony, and class size, alongside model fit when determining the number of classes. We selected a 4 class model, which had the smallest BIC, after inspecting the meanings and sample sizes of each class. The AIC continued to decrease with more classes as often occurs (Geiser, 2012; see Supplemental Table A.1).

The second step in the 3-step procedure was the probabilistic assignment of each couple to their most likely sexual expression class based on the posterior probabilities estimated in the first step. The third step in the procedure was estimation of the association between covariates and sexual expression latent class membership using multinomial logistic regression while accounting for the measurement error calculated in step two.

All analyses were conducted using Mplus 8.3, weighted to account for the complex sampling design, and multiply imputed. Among the sexual expression indicators, missingness was typically 10–12% for husbands and 7–9% for wives, though nonresponse was higher for questions asked in the LBQ (see Table 1). Non-response on the other covariates was generally low (<5%) with the exception of sexual difficulties (see Supplemental Appendix Table A.2). We imputed 20 data sets, where all study variables—as well as indicators for return of the LBQ, heart disease, and the complex survey design—were included as predictors.

Table 1.

Proportions on sexual expression indicators by gender (n = 953 couples).^a

	Wives		Husbands
	Prop.	Missing	Prop.	Missing
Frequency of Sexual Behavior		10.49%		8.08%
≤Once a month	0.64		0.61
2–3 times a month	0.17		0.20
≥Once a week	0.19		0.19
Vaginal Sex		10.60%		7.35%
Never/rarely	0.51		0.46
Sometimes/usually	0.18		0.18
Always	0.31		0.36
Receiving Oral Sex		11.65%		8.29%
Never/rarely	0.79		0.77
Sometimes/usually	0.17		0.20
Always	0.04		0.03
Intimate Touch		11.33%		8.08%
Never/rarely	0.46		0.44
Sometimes/usually	0.48		0.30
Always	0.06		0.26
Masturbation		10.19%		8.92%
None	0.64		0.49
Any	0.36		0.51
Sex Ok b		18.36%		25.71%
Less often than desired	0.46		0.68
About as often as desired	0.48		0.31
More often than desired	0.06		0.02
Think about Sex		4.93%		4.93%
≤Once a month	0.36		0.14
1–3 times a month	0.36		0.28
≥Once a week	0.28		0.59
Sex Important b		12.91%		14.38%
Not/somewhat	0.50		0.36
Moderately	0.29		0.27
Very/extremely	0.21		0.37

Notes:

^aWeighted and multiply imputed estimates; rate of missingness for each indicator prior to imputation;

^bAsked in the Leave Behind Questionnaire (LBQ).

Results

Descriptive results

We begin by describing the characteristics of the sample (see Supplemental Appendix Table A.2.). As expected, wives were younger than husbands (67.58 vs. 71.09 years; Wald χ² = 4.28, p = .039) and fewer had college degree (25% vs. 35%; Wald χ²= 18.23, p < .001). More than 80% of respondents were White, with non-Whites about equally divided between Black and Hispanic respondents (7% each) and very few of “other” race/ethnicities (<2%; not shown). Couples were in long-term marriages (average duration 35–40 years) and 60% were in first marriages. Wives reported less emotional satisfaction (Wald χ² = 11.58, p < .001) and marital happiness (Wald χ² = 16.23, p < .001) than husbands. Wives were less likely to engage in vigorous exercise (Wald χ² = 3.18, p = .075) or be heavy drinkers (Wald χ² = 24.67, p < .001). Both wives and husbands rated their physical and mental health “good” to “very good,” had 1–2 chronic conditions on average, and very few ADL limitations (>70% reported none; not shown). Sexual functioning problems were common: 47% of husbands indicated erectile problems, 35% of wives reported problems lubricating, and 14% of wives reported pain during intercourse. A quarter of couples (24%) reported climaxing too quickly.

Table 1 presents the distribution of responses on each of the sexual expression indicators. Wives’ and husbands’ accounts of sexual behaviors were largely similar. Majorities of both wives (64%) and husbands (61%) report engaging in sexual activities ≤ once a month; about one in five wives and husbands report engaging in sexual activity ≤ once a week. About half of respondents said their sexual activity at least sometimes included vaginal sex; one-third said it “always” included vaginal sex. Few respondents reported receiving oral sex (79% of wives and 77% of husbands reported “never/rarely”). Similar percentages of wives (46%) and husbands (44%) reported “never/rarely” engaging in intimate touching; among those who reported intimate touching, there was a stark gender difference—just 6% of wives but 26% of husbands reported sexual activity “always” included their partner touching their genitals.

Wives expressed lower desire than husbands did, with lower rates of masturbation and greater satisfaction with their sexual activity frequency. Interestingly, although generally more satisfied with their sexual frequency, 46% of wives also indicated having sex less often than desired (compared to 68% of husbands). Even so, wives indicated less general motivation for sexual activity, thinking about sex much less often and reporting that sex is a less important part of their lives compared to husbands.

Multivariate results

We now turn to the dyadic LCA results. Table 2 presents the item-response probabilities of the sexual expression indicators. These estimates are the probability of a specific response on each indicator for each of the four classes identified and are used to interpret the meaning of the classes.

Table 2.

Item-response probabilities of sexual expression indicators (N = 953 Couples).^a

	Traditionalist		Versatile		Compensatory		Resigned Inactive
	Wives	Husbands	Wives	Husbands	Wives	Husbands	Wives	Husbands
Frequency of Sexual Behavior
≤Once a month	0.68	0.80	0.09	0.04	0.92	0.33	0.99	1.00
2–3 times a month	0.22	0.16	0.37	0.45	0.00	0.24	0.00	0.00
≥Once a week	0.10	0.04	0.53	0.50	0.08	0.43	0.01	0.00
Vaginal Sex
Never/rarely	0.19	0.27	0.10	0.08	1.00	0.15	0.98	0.99
Sometimes/usually	0.28	0.30	0.33	0.20	0.00	0.47	0.01	0.01
Always	0.52	0.44	0.57	0.72	0.00	0.38	0.01	0.00
Receiving Oral Sex
Never/rarely	0.77	0.85	0.50	0.47	1.00	0.47	1.00	1.00
Sometimes/usually	0.19	0.13	0.40	0.44	0.00	0.52	0.00	0.00
Always	0.04	0.02	0.10	0.08	0.00	0.01	0.00	0.00
Intimate Touch
Never/rarely	0.08	0.25	0.05	0.05	1.00	0.14	0.97	0.97
Sometimes/usually	0.50	0.48	0.39	0.42	0.00	0.43	0.02	0.03
Always	0.41	0.26	0.57	0.52	0.00	0.43	0.00	0.00
Masturbation
None	0.62	0.48	0.61	0.44	0.44	0.36	0.74	0.56
Any	0.39	0.52	0.39	0.56	0.56	0.63	0.26	0.44
Sex Ok
Less often than desired	0.38	0.66	0.23	0.43	0.79	0.60	0.76	0.78
About as often as desired	0.56	0.32	0.68	0.50	0.21	0.32	0.23	0.18
More often than desired	0.06	0.02	0.10	0.08	0.00	0.09	0.01	0.04
Think about Sex
≤Once a month	0.26	0.11	0.05	0.00	0.58	0.03	0.63	0.29
1–3 times a month	0.55	0.37	0.39	0.10	0.33	0.34	0.21	0.33
≥Once a week	0.19	0.52	0.56	0.90	0.09	0.63	0.16	0.38
Sex Important
Not/somewhat	0.54	0.34	0.21	0.04	0.73	0.22	0.80	0.60
Moderately	0.34	0.38	0.41	0.19	0.15	0.21	0.12	0.20
Very/extremely	0.12	0.27	0.39	0.77	0.13	0.57	0.08	0.20
Latent Class Prevalence b	26%		29%		9%		36%

Note:

^aWeighted and multiply imputed estimates;

^bBased on the most likely class membership.

We labeled the first class Traditionalist (26%). Couples in this class engage in sexual activity ≤ once a month (although a non-trivial number of couples are having sex more often), and this almost always involves vaginal sex and sometimes intimate touching (especially for wives), but rarely includes receiving oral sex. Wives are content with the amount of sexual activity they are having, whereas two-thirds of husbands in this class desire more frequent sex. Wives think about sex 1–3 times per month and their husbands think about sex a few times per week or more. Consistent with these other modest differences, wives feel sex is only somewhat important to their lives whereas their husbands say that sex is at least moderately important.

Nearly a third of couples (29%) were classified as Versatile—characterized by a high frequency of sexual activity, usually involving vaginal sex and intimate touching, and receiving oral sex at least sometimes, according to both spouses. Thus, not surprisingly, both husbands and wives think about sex frequently and view it as an important part of life. The level and variety of partnered sexual activity in this class are higher than those in any other class. Moreover, the frequency with which wives think about sex and the importance they accord it, although lower than their husbands, is higher than in any other class. Both husbands and wives in this class, compared to the others, are most apt to indicate that they are having sex about as often as desired—although about an equal number of husbands would like to be having sex more often.

We labeled the smallest class (9%) Compensatory. This class is unique in that spouses disagree on their sexual activity. Most wives in this class reported sexual activity as occurring ≤ once a month (91%), but just 33% of husbands did so. Wives report never/rarely engaging in any specific activity even as their husbands say they are “sometimes/usually” having vaginal sex (the lowest of all the sexually engaged classes), are “sometimes/usually” receiving oral sex, and having a fair bit intimate touching. This discrepancy suggests that, even though NSHAP instructed respondents to consider sexual activity broadly (“whether or not intercourse or orgasm occurs”), wives reported limited sexual frequency because they were using vaginal intercourse as a frame (see also Lodge & Umberson, 2012). Thus, couples who appear to have had some difficulty with vaginal intercourse engaged in other “compensatory” activities where husbands are on the receiving end (i.e., oral sex, intimate touching). Two additional item-response probabilities support this interpretation. First, both spouses in this class masturbate more often compared to those in other classes. Second, both spouses wish they were having sex more often. Both spouses appear to share a desire for (vaginal) sexual activity, even as wives rate sex as less important and think about it less often than do their husbands.

The final class, Resigned Inactive, is the largest and captures 36% of couples. This class is characterized by very low rates of partnered sexual behavior as nearly all wives and husbands state they are “never/rarely” engaging in vaginal sex, oral sex, or intimate touching. Masturbation is the lowest in this class compared to the others, with just 26% of wives and 44% of husbands masturbating during the prior year. Neither spouse thinks sex is an important part of life (this is the only class where a majority of husbands hold this view), and a majority of wives do not think about sex very often. Although there is greater variation in the frequency of sexual thoughts among husbands, just 38% report thinking about sex ≥ once a week—the lowest percentage of any class. These couples are not sexually engaged, but interestingly a high proportion of wives (76%) and husbands (78%) state that they are having sex less often than desired.

Table 3 presents these results from the multinomial logistic regression predicting membership in the sexual expression classes. Resigned Inactive was the reference category (other contrasts are discussed in the text). In preliminary analyses, we first added covariates in blocks to determine the unique influence of conceptually similar groups of variables; the substantive pattern of results was comparable to those in the full model (see Appendix Table A.3). Overall, we expected that those who were younger and had higher socioeconomic status, better relationship quality, and better health would have been more likely to be in the classes where spouses are more in agreement and have higher sexual engagement, and this was largely supported.

Table 3.

Predicting sexual expression latent class membership, multinomial regression logit coefficients and standard errors (N = 953 Couples).^a,b

	Traditionalist		Versatile		Compensatory
	B	SE	b	SE	b	SE
Demographics
Age c
Wife	–0.07*	(0.03)	–0.04†	(0.03)	–0.06	(0.04)
Husband	–0.02	(0.03)	–0.07*	(0.03)	–0.07*	(0.03)
Nonwhite
Wife	–0.02	(0.54)	–0.33	(0.68)	0.29	(0.72)
Husband	0.43	(0.64)	1.14†	(0.66)	0.51	(0.73)
College Degree
Wife	–0.09	(0.36)	–0.51	(0.38)	–0.72	(0.71)
Husband	0.16	(0.35)	–0.20	(0.30)	0.92†	(0.55)
Household Income	0.27†	(0.15)	0.23	(0.16)	0.36†	(0.21)
Relationship Duration	–0.02	(0.15)	–0.30**	(0.11)	–0.15	(0.16)
Married Twice
Wife	–0.01	(0.44)	–1.23**	(0.46)	0.07	(0.69)
Husband	–0.54	(0.48)	–0.20	(0.47)	–0.72	(0.73)
Currently Cohabiting	–0.47	(0.88)	0.24	(0.76)	0.85	(0.92)
Relationship Quality
Emotional Satisfaction
Wife	0.01	(0.23)	0.30	(0.27)	–0.40	(0.45)
Husband	0.22	(0.26)	–0.11	(0.30)	–0.30	(0.31)
Physical Pleasure
Wife	0.02	(0.23)	0.59*	(0.24)	0.30	(0.44)
Husband	0.17	(0.22)	1.33***	(0.29)	0.40	(0.42)
Marital Happiness
Wife	0.28	(0.20)	–0.19	(0.26)	0.12	(0.38)
Husband	0.20	(0.27)	–0.20	(0.21)	0.31	(0.48)
Health-Related Behaviors
Vigorous Exercise
Wife	0.27	(0.29)	0.73*	(0.31)	0.77	(0.49)
Husband	0.57*	(0.26)	0.87**	(0.33)	0.87*	(0.42)
Heavy Drinking
Wife	0.31	(0.70)	0.52	(0.67)	–0.12	(0.94)
Husband	0.08	(0.42)	1.29**	(0.41)	1.34*	(0.56)
Smoker
Wife	0.34	(0.40)	0.13	(0.45)	0.27	(0.42)
Husband	–0.10	(0.36)	–0.62	(0.41)	0.28	(0.62)
Obese
Wife	0.07	(0.28)	–0.39	(0.34)	0.33	(0.39)
Husband	0.07	(0.27)	–0.10	(0.31)	–0.49	(0.46)
Health Status
Self-Rated Physical Health
Wife	–0.03	(0.18)	0.10	(0.16)	0.00	(0.26)
Husband	0.20	(0.14)	0.44*	(0.20)	0.20	(0.29)
Self-Rated Mental Health
Wife	0.10	(0.23)	0.02	(0.19)	–0.08	(0.22)
Husband	0.00	(0.16)	0.18	(0.14)	0.45†	(0.24)
Chronic Illness
Wife	0.05	(0.12)	0.17	(0.14)	0.35*	(0.14)
Husband	–0.12	(0.09)	–0.19	(0.13)	0.03	(0.19)
Functional Limitations
Wife	–0.06	(0.11)	0.06	(0.10)	–0.07	(0.13)
Husband	–0.09	(0.12)	–0.07	(0.14)	0.38**	(0.13)
Sexual Difficulties
Problems Lubricating	1.44***	(0.34)	1.65***	(0.38)	0.26	(0.62)
Pain During Sex	0.55	(0.42)	–0.12	(0.47)	–0.64	(1.10)
Erectile Problems	0.04	(0.29)	–1.44***	(0.33)	–0.95†	(0.54)
Climaxing Too Quickly	1.21***	(0.30)	0.77*	(0.35)	0.63	(0.57)

Notes:

^aWeighted and multiply imputed estimates;

^bResigned Inactive is the Reference Category; Intercepts omitted;

^cCentered at age 62.

^†p < 0.10,

^*p < 0.05,

^**p < 0.01,

^***p < 0.001

First looking at the demographic predictors, compared to being in Resigned Inactive older wives were less likely to be in the Traditionalist (b = −0.07, p = .037, Adjusted Odds Ratio [AOR] = 0.94) and, marginally, the Versatile (b = −0.04, p = .088, AOR = 0.96) classes, while older husbands were less likely to be in the Versatile (b = −0.07, p = .012, AOR = 0.94) and the Compensatory (b = −0.07, p = 047, AOR = 0.94) classes. Importantly, there were few age differences in membership across any of the engaged classes (not shown), indicating that—net of other predictors—older wives and husbands are more likely to be classified as Resigned Inactive. Given these are cross-sectional data we cannot say for certain that this reflects aging, per se, although other research finds age-related declines in sexual behavior (see DeLamater, 2012).

Net of other predictors, there were few socioeconomic differences in sexual expression class membership. Couples with college-educated husbands were more likely to be Compensatory compared to Resigned Inactive (b = 0.92, p = .092, AOR = 2.52) and Versatile (b = 1.13, p = .031, AOR = 3.08; not shown). Household income was marginally associated with a greater likelihood of being in the Traditionalist (b = 0.27, p = .075, AOR = 1.31) and Compensatory (b = 0.36, p = .089, AOR = 1.43) classes, but no other comparisons were statistically significant (not shown).

Couples with longer duration relationships were less likely to be in the Versatile class compared to Resigned Inactive (b = −0.30, p = .006, AOR = 0.74). Yet, for sexually engaged couples longer marital durations were associated with an increased likelihood of being in the Traditionalist as opposed to the Versatile class (b = .27, p = .029, AOR = 1.32; not shown). Couples, where wives had been married more than once, were 42% less likely to be in the Versatile class compared to each of the Resigned Inactive, Traditionalist, Compensatory classes (calculations and all comparisons not shown). Husbands’ marital history was not associated with couples’ sexual expression.

Spouses’ assessments of their relationship quality were also associated with class membership—although, net of other covariates, mostly in terms of how physically pleasurable each found it. Wives who reported greater physical pleasure were more likely to be Versatile than either Resigned Inactive (b = 0.59, p = .012, AOR = 1.81) or Traditionalist (b = 0.57, p = .003, AOR = 1.78; not shown). Similarly, husbands who reported greater physical pleasure were more likely to be Versatile than Resigned Inactive (b = 1.33, p < .001, AOR = 3.77), Traditionalist (b = 1.16, p = .001, AOR = 3.19; not shown) or Compensatory (b = 0.93, p = .002, AOR = 2.54; not shown). Note that the effects of physical pleasure for husbands were about twice as large as those of wives. Emotional satisfaction and marital happiness were not associated with membership in any sexual expression class in the full model, regardless of the comparison group (not all results shown).

Health behaviors were somewhat associated with couples’ sexual expression class. When husbands regularly engaged in vigorous exercise couples were more likely to be Traditionalist (b = 0.57, p = .030, AOR = 1.77), Versatile (b = 0.87, p = .008, AOR = 2.39), or Compensatory (b = 0.87, p = .037, AOR = 2.38) than Resigned Inactive (none of engaged classes differed from one another; not shown). Couples where wives engaged in vigorous exercise were more likely to be Versatile than Resigned Inactive (b = 0.73, p = .019, AOR = 2.08).

Couples where husbands reported heavy drinking were more likely to be Versatile (b = 1.29, p = .002, AOR = 3.63) or Compensatory (b = 1.34, p = .016, AOR = 3.83) compared to Resigned Inactive (and the two did not differ from one another; not shown). Husbands’ heavy drinking was also associated with a greater likelihood of couples being in the Versatile rather than Traditionalist class (b = 1.21, p = .015, AOR = 3.34; not shown). Neither wives’ nor husbands’ smoking or obesity—net of other factors—were associated with couples’ sexual expression.

As expected, health status predicted membership across sexual expression classes— principally for husbands. Husbands’ higher self-rated physical health, for example, was associated with higher odds of couples being Versatile compared to Resigned Inactive (b = .44, p = .025; AOR = 1.55). Interestingly, health conditions were mostly associated with an increase in the likelihood of couples being in the Compensatory class. Net of other factors, couples were more likely to be classified as Compensatory when husbands reported functional limitations than either Resigned Inactive (b = 0.358 p = .004, AOR = 1.47) or Versatile (b045, p = .004, AOR = 1.57; not shown). Husbands’ self-rated mental health was also marginally associated with an increased likelihood of couples being in the Compensatory class compared to the Resigned Inactive (b = .45, p = .057, AOR = 1.56) or Traditionalist (b = 0.44, p = .098, AOR = 1.56; not shown) classes. Wives’—but not husbands’—chronic illness increased the likelihood of couples being in the Compensatory class compared to the Resigned Inactive (b = 0.35, p = .011, AOR = 1.42) and Traditionalist (b = 0.30, p = .053, AOR = .1.35; not shown) classes. We note that relative to other predictors, the pattern of associations for health status differed the most between the model were entered separately and the full model. For example, husbands’ chronic illness count was associated with a lower likelihood of couples being in the Versatile and Traditionalist classes (see Appendix Table A.3).

We anticipated that sexual difficulties would predict membership in classes with lower sexual engagement. Surprisingly, couples where wives reported problems lubricating were more likely to be in the Traditionalist (b = 1.44, p < .001; AOR = 4.22) or Versatile (b = 1.65, p < .001; AOR = 5.21) classes compared to the Resigned Inactive class (or the Compensatory class; not shown). Likewise, couples were more likely to be Traditionalist (b = 1.21, p <.001; AOR = 3.36), or Versatile (b = 0.77, p = .029; AOR = 2.16) if either spouse reported problems climaxing too quickly. Of course, it may be that couples were aware of these issues precisely because they were having regular intercourse. As expected, husbands’ erectile difficulty was associated with a lower likelihood of couples being Versatile (b = –1.44, p < .001; AOR = .24) or marginally Compensatory (b = –0.95, p = .077; AOR = .39). Husbands erectile problems, however, increased the odds couples were Traditionalist than either Versatile (b = 1.48, p< .001; AOR = 4.39) or marginally Compensatory (b = 0.99, p= .093, AOR = 2.70) indicating that erectile difficulties were not necessarily associated with more varied (non-penetrative) sexual behaviors (not shown). Wives’ pain during intercourse was not associated with couples’ sexual expression.

Supplemental analyses: Spousal discordance and class membership

In supplemental analyses, we examined the slightly different question as to whether spousal discordance on any given characteristic was associated with couples’ sexual expression. As would be expected given the tendency toward marital homogamy, spouses were overwhelming similar to one another on most demographic, relational, and health behavior characteristics and when they were discordant it was in the expected direction (e.g., 40% of husbands were ≥4 years older, 37% were more emotionally satisfied, and 43% reported higher physical pleasure than their wives). Discordance in health status and sexual difficulties were more evenly split (see Appendix Table A.4 for variable specification and descriptive statistics).

Although direct comparisons between models are difficult, the findings from these supplemental models (see Appendix Table A.5) suggest that characteristics of each spouse—rather than in comparison to one another—are more clearly associated with latent class membership (see also Iveniuk & Waite, 2018) with but a few exceptions. First, there is some suggestion that couples where the husband is ≥4 years older than the wife were more likely to be Traditionalist than Resigned Inactive (b = 0.55, p = .048; AOR = 1.60) or Versatile (b = 0.61, p = .053; AOR = 1.84; not shown). Second, imbalances in marital happiness were associated with an increased likelihood of being Resigned Inactive. Specifically, couples where husbands reported higher marital happiness than their wives were less likely to be Traditionalist (b = –0.60, p = .037; AOR = 0.50) or Versatile (b –0.99, p= .000; AOR = 0.37). Similarly, when wives reported higher marital happiness, couples were less likely to be in the Compensatory class (b = –1.45, p = .045; AOR = 0.23). Finally, the supplemental model showed a consistent positive association between household income and membership in each of the engaged sexual expression classes (which do not differ from each other; not shown) than noted above. This likely reflects the shift in the implied reference group to spouses concordant on all measures. These differences notwithstanding, the substantive conclusions from the supplemental models are similar overall to those of the main findings above.

Discussion

We used nationally representative data from older couples in the U.S. National Social Life, Health and Aging Project (NSHAP) and dyadic latent class analysis to describe sexual expression configurations among couples, and identify how demographic, relational, and health factors were associated with these configurations. We advance our understanding of sexual expression among older couples in two key ways. First, we used LCA to consider not just sexual behaviors but also the underlying sexual desires and motivations that simultaneously define older adults’ sexual expression. Second, our dyadic person-centered approach allowed us to identify heterogeneity in sexual expression between couples.

We identified four distinct sexual expression configurations among older couples: Traditionalist, Versatile, Compensatory, and Resigned Inactive. Consistent with prior studies (e.g., Lindau & Gavrilova, 2010), 64% of older couples reported at least some sexual activity. Spouses were mostly in agreement about their sexual activities, but husbands, on average, reported greater sexual desire and motivation (DeLamater, 2012; Lodge & Umberson, 2012).

The Traditionalist class described a quarter of older couples (26%), who were engaged in sexual activity ≤ once a month, principally vaginal sex, with moderate intimate touching and sometimes oral sex. Wives in these couples had less desire and were more content with their sexual frequency than were their husbands, consistent with expectation. These couples were older, in longer marriages, more likely to have been previously married, to have problems lubricating (wives), or climaxing too quickly during intercourse.

One in three (29%) couples were characterized as Versatile, exhibiting the highest and most varied sexual activities—usually engaging in vaginal sex, oral sex, and imitate touching—compared to other older couples. Both spouses reported high levels of thinking about sex, attributed importance to sex, and were the most likely of all couples to be content with their sexual activity frequency. These couples were somewhat younger, had shorter relationship durations, and wives were less likely to have been previously married. This class was associated with greater reports of physical pleasure by both wives and husbands and, as the supplemental analysis showed, spouses were more apt to have concordant assessments of marital happiness. Husbands’ health and the absence of erectile problems also predicted membership.

The smallest class identified described 9% of couples and differed from the other classes in that wives’ and husbands’ reports of sexual activity were surprisingly discordant. Wives reported a low frequency of sexual activity. Yet, their husbands reported more frequent sexual activity that included vaginal sex, receiving oral sex, or intimate touching at least “sometimes.” Spouses also thought about sex at very different levels and had discordant attributions of its importance. We labeled these couples Compensatory because the discordant reports suggested wives were engaged in other sexual activities where their husbands were the recipient. Indeed, that neither spouse was satisfied with their frequency of sexual activity suggests that both shared a desire for vaginal sexual activity but were engaging in other activities as a means addressing the deficit—at least for husbands. Husbands in this class were more likely to have a college degree; wives were less likely to be remarried. The defining predictor, however, was poor health status. Husbands who reported functional limitations and wives with chronic illnesses were more likely to be in this class. As this was the smallest class, the lack of significance among many predictors may reflect low statistical power. Future research with greater statistical power is needed to increase our understanding of couples in this class, especially their discordant sexual behavior reports.

The final class we identified was not engaged in partnered sexual activity at any meaningful level and comprised the largest class, capturing 36% of couples. We labeled this Resigned Inactive because, even though they each indicated low masturbation or motivation for sex, both wives and husbands said they were having sex less often than desired. Older spouses (including couples where the husband was ≥4 years older than the wife), long-married couples, less educated spouses, lower-income couples, couples where wives had been previously married, and those where spouses had a discordant assessment of marital happiness were more likely to be in this class. Husbands who did not engaged in regular vigorous exercise and those with erectile problems also were more likely to be Resigned Inactive.

Overall, demographic, relational, and health factors were associated with membership in the different sexual expression classes in ways consistent with prior studies (e.g., Allen & Desille, 2017; Karraker & DeLamater, 2013; Stroope et al., 2015; Thomas et al., 2015). We also expected that sexual difficulties would predict membership in less sexually engaged classes as suggested by prior studies (e.g., Kwon & Schafer, 2017; Waite et al., 2017). These measures, however, were associated with sexual expression in some counterintuitive ways—predicting membership in more engaged classes (e.g., wives’ difficulty lubricating) or both limited engagement and inactive classes (e.g., husbands’ erectile difficulties). These findings make sense, however, if we consider that couples not sexually engaged might either not know there was such a problem or, if known, might not view issues like difficulty lubricating as problems. Indeed, sexual problems such as difficulty lubricating can be more easily addressed with over-the-counter products than other performance issues and thus may not present a barrier to continued sexual engagement. Other problems such as erectile difficulty require prescription remedies and are a source of worry because they challenge husbands’ masculinity (Lee et al., 2016; Lodge & Umberson, 2012), and thus, this may explain why husbands reporting this issue were more likely to be Resigned Inactive even as some were sexually engaged at modest levels (i.e., Traditionalist).

The current study is a crucial first step in understanding older adults’ sexual expression from a couple-centered approach. Nevertheless, we measured sexual expression at one point in time, and longitudinal research is needed to address several questions that now arise. First, longitudinal assessment is needed to unpack how relational, health, and sexual functioning factors overtime influence couples’ sexual expression. For example, consider the Versatile class. It is unclear if higher relationship quality leads a couple to engage in more varied activities or if more varied sexual activities lead to greater physical pleasure and more congruent assessments of marital happiness, or if they two are reciprocally related (see Quinn-Nilas, 2020). Second, the longitudinal health consequences of these different sexual expression classes also merit investigation. What are the mental health consequences of being Resigned Inactive, for example, where both spouses express unmet desire for sexual activity? We might speculate that the inability to achieve a widely-held marital expectation may lead to increased distress and loneliness (Warner & Adams, 2016)—as least in late midlife (Lodge & Umberson, 2012). Third, longitudinal research is needed to determine how couples move between the sexual expression classes. For example, do Traditionalist couples typically transition to Resigned Inactive class or do some transition to the Compensatory class. This necessarily then requires understanding what predicts such transitions and what the mental health and relation consequences of such transitions might be.

We also recognize that our analytic sample consists of heterosexual older couples in long-term marriages—a little more than half of which were first marriages for both spouses—from a fairly narrow range of birth cohorts. We might identify different sexual expression configurations in non-married older couples, couples with shorter relationship durations, or sexual minority couples. Given the rise in gray divorce (Brown & Lin, 2012) and the changing norms around gender and sexual behaviors and attitudes (Carpenter, 2010), sexual expression when more recent cohorts enter later-life may be more varied. Moreover, the association between demographic, relational, and health factors might differ among more recent cohorts. For example, we found that couples where wives had been previously married were less likely to be in the high engagement Versatile class. This association might not hold as more recent cohorts—where repartnering is even more common and sexual scripts less gendered—enter later life. Thus, future research on older adults’ sexual expression should include more diverse populations and sample persons from more recent birth cohorts.

Despite these limitations, the current study establishes heterogeneity in older couples’ sexual expression when sexual behavior, desires, and motivations are considered simultaneously. Contrary to the cultural assumption that older couples are disinterested in sexual intimacy, our study, along with others (e.g., Addis et al., 2006; Lindau et al., 2007), highlights that older couples are multifaceted in their sexual expression. Importantly, our study indicates that although spouses’ sexual behavior is similar, the degree to which their desires and motivations differ varies depending on the configuration of behaviors. Moreover, even when both spouses are in agreement that sex is not an important part of life and are not having sex, it does not mean that they are content with this as the Resigned Inactive class shows. Older adults—especially those who are married—view sexual activity as a significant component of a relationship (Lodge & Umberson, 2012; Waite et al., 2009). Couples are not agnostic about such changes despite recognition that sexual frequency declines with age (DeLamater, 2012).

Health practitioners, therapists, and others who work with older adults should, therefore, not assume older couples are content with low sexual activity and recognize that there is not a “one size fits all” approach when it comes to older couples’ sexual expression. This is especially important because there is a reluctance to discuss sexual matters with physicians (Adams, 2014). Recognizing different patterns of sexual expression can lead to better physical and mental therapies for both individuals and couples, as well as, perhaps, adoption of pharmaceutical interventions to maintain desired levels of sexual activity (Lodge & Umberson, 2012).

In conclusion, our study highlights that although a majority of older couples are engaging in sexual behavior, they are diverse in their desires and motivations. Behaviors and attitudes combine in distinct ways that lead to heterogeneous sexual expression. Older couples are not monolithic. Greater attention to the diversity of lived experiences of married older adults and their sexual expression in both research and clinical practice is warranted.

Supplementary Material

Supplemental material for this article is available online.