Abstract
Extradyadic sex (EDS) is a major relationship violation, yet it occurs in nearly a quarter of United States cohabiting and marital unions. While many relationships dissolve in the wake of EDS, a majority remain intact. Theories of social stress suggest that substantial psychological distress should result unless EDS is a symptom of stress caused by involvement in a relationship marked by other negative characteristics. This study investigates how one’s own EDS, a partner’s EDS, and mutual EDS are related to internalizing and externalizing behaviors: depressive symptoms and heavy alcohol use, respectively. Analyses of data from the National Longitudinal Study of Adolescent to Adult Health suggest that one’s own EDS is associated with heavy alcohol use among cohabiters and spouses and with depressive symptoms among spouses, while partner EDS has no association with either outcome, net of confounders. We discuss the implications of these findings in the study’s conclusions.
Americans strongly value sexual fidelity within committed relationships. Infidelity topped the list of morally unacceptable activities according to estimates from a recent Gallup poll. Only 9% of those surveyed indicated that it was morally acceptable for married men and women to have an affair, compared to 18% for suicide, 17% for polygamy, 36% for pornography, and 73% for divorce (Jones, 2017). These strong opinions condemning sexual infidelity are in line with academic research indicating that 99% of married persons and 94% of cohabiters expect sexual exclusivity from partners (Treas & Giesen, 2000).
Nonetheless, estimates suggest that in opposite-sex marriages, around 20% to 25% of U.S. men and 11% to 15% of U.S. women engage in extradyadic sex (EDS) (Allen et al., 2005). More recent estimates suggest that EDS occurs in a quarter of all opposite-sex marriages and in a third of all opposite-sex cohabiting unions regardless of whether men or women are reporting this behavior (Frisco et al., 2017). Earlier studies have also confirmed that estimates of EDS among opposite-sex cohabiters are higher than estimates of EDS in marital unions (Forste & Tanfer, 1996; Treas & Giesen, 2000).
While a partner’s EDS increases the odds of union dissolution, one’s own EDS is not significantly associated with increased odds of union dissolution and a large proportion of individuals with partners who engage in EDS remain with them (Frisco et al., 2017). In fact, in our own analysis of the nationally representative data used for the current study, 96% of married men and women and 73% and 68% of cohabiting men and women, respectively, remain in their relationships despite their own or their partner’s EDS.
Given that EDS is stressful (Bird et al., 2007; Fife et al., 2013; Gordon & Baucom, 1999; Gordon et al., 2004) and a major violation of trust and relationship expectations (Treas & Giesen, 2000), we ask whether and how EDS is associated with internalizing and externalizing behavior. Unlike prior research which focuses on either victims (Cano & O’Leary, 2000; Shrout & Weigel, 2018; Sweeney & Horwitz, 2001) or perpetrators (Wiggins & Lederer, 1984; Hall & Fincham, 2009) of EDS, our analysis estimates the distress of both partners. Theory regarding stress and stressful life events suggests that EDS should lead to internalizing and externalizing behavior among individuals who are unfaithful because of either the stress of concealing infidelity, guilt about engaging in EDS, or the stress caused by the relationship strain that occurs when partners discover the transgression (Allen et al., 2005; Beach, Jouriles, & O’Leary, 1985; Glass, 2002; Spanier & Margolis, 1983; Wiggins & Lederer, 1984). This relationship strain should also cause stress and, in turn, distress for those who discover the unfaithfulness of their partners because EDS is a major violation of trust (Glass, 2002; Shrout & Weigel, 2018). Yet, it is also possible that infidelity is a symptom of relationship problems and that the associations between EDS and internalizing and externalizing behavior are spurious.
We investigate these competing ideas in the current study by examining whether individuals’ own reported EDS, their partner’s EDS, and mutual EDS are associated with depressive symptoms, a key indicator of internalizing behavior, and heavy drinking, an important indicator of externalizing behavior. We analyze data from a sample of 24–32 year-old men and women who were currently married or cohabiting when they participated in Wave IV of the National Longitudinal Study of Adolescent to Adult Health (Add Health). A strength of our study is that our data source includes indicators of previous psychological distress, previous heaving drinking, and background characteristics that may select individuals into marital and cohabiting unions where one or both partners participated in EDS. In addition, most prior research uses small or non-representative samples, like couples in marital therapy to estimate how EDS is associated with distress (Beach et al., 1985; Wiggins & Lederer, 1984). Our analysis will confirm whether findings of small clinical samples, which by their nature are selective of individuals who seek therapy, are evident among a national sample of young adults.
Background
The majority of adults ages 25 years and older are in cohabiting and marital relationships, with marriages making up a greater proportion of couples living in the same household (Stepler, 2017). Despite the rising ages for first marriages, even among adults ages 25–34 years, 41% are already married and another 14% are cohabiting (Stepler, 2017). In both marital and cohabiting relationships, there is near universal endorsement for sexual exclusivity (Treas & Giesen, 2000), which makes it unsurprising that sexual infidelity is found to be among the most morally unacceptable behaviors among adults in United States (Jones, 2017). Nonetheless, despite similar expectations for sexual fidelity among spouses and cohabiters, estimates of the prevalence of EDS indicate that cohabiting relationships are more likely to involve EDS than marital relationships (Forste & Tanfer, 1996; Treas & Giesen, 2000).
This higher prevalence of EDS among cohabiters is perhaps unsurprising given the nature of these relationships relative to marital relationships. Marriages involve formal and legal agreements between adults that either imply or explicitly specify expectations for fidelity and relationship permanence. Conversely, cohabiting relationships are more informal and cohabiters tend to have lower commitment, trust, and emotional investment than married individuals (Gerbasi, 2007). In addition, cohabiting relationships are relatively short-lived and are very rarely permanent arrangements. Instead, individuals tend to transition out of these relationships as a result of dissolution or marriage, though the risk of dissolution is higher than the risk of marriage, especially among recent cohorts of young people (Guzzo, 2014).
Regardless of the different nature of cohabiting and marital relationships and the higher prevalence of EDS in the former, discovering a partner’s EDS is likely to be stressful regardless of the relationship context. For example, even among younger daters in college, stress and distress is evident when partners cheat (Shrout & Weigel, 2018). Some psychologists have gone so far as to say that EDS produces symptoms similar to post traumatic stress disorder when it occurs among married men and women, that a partner’s EDS is the most distressing event that couples face in their relationship, and the most difficult to treat in therapy (Gordon et al., 2004).
EDS may also lead to stress for unfaithful partners, regardless of whether or not their EDS has been discovered. If their partner knows of the EDS, then the ensuing conflict is likely to be quite stressful given that faithful partners experience feelings of anger, betrayal, resentment, and/or embarrassment as a result of a partner’s EDS (Bird et al., 2007; Gordon & Baucom, 1999; Gordon et al., 2004; Vaughan, 2003). If an individual’s EDS is not known, they are likely to experience guilt and shame in response to their actions and the stress of concealing their actions. Spanier and Margolis (1983) asked divorced men and women about extramarital sexual relationships and found that 34% of men and 29% of women who engaged in extramarital affairs felt guilt as a result. In a more recent study, Eaves and Robertson-Smith (2007) found that 64% of unfaithful men and 60% of unfaithful women reported feelings of guilt.
Within relationships in which both partners participate in EDS, stress may occur for the reasons just described—both partners in the relationship may experience a combination of guilt, shame, and anxiety due to their own EDS, and anger, resentment, and embarrassment due to their partner’s EDS. However, it is also possible that when mutual EDS occurs, partners feel as though they are “even,” thereby mitigating the stress of EDS.
The association between stress, stressful life events and psychological distress has been well established. Both general stress (e.g. Lin & Ensel, 1989, Thoits, 1995) and stressful life events (Estrada-Martínez et al., 2012; Glass et al., 1997; Hammen, 2005; Mazure, 1998; Paykel, 2003; Tennant, 2002) are associated with internalizing disorders like depressive symptoms. The association between stress (Conger, 1956; Cooper et al., 1992), stressful life events (Fife et al., 2013; Shrout & Weigel, 2018), and externalizing behaviors such as heavy alcohol use is also well established. This is a primary reason that we seek to understand whether and how one’s own EDS, a partner’s EDS, and mutual EDS are related to depressive symptoms and heavy alcohol use within the context of marital and cohabiting unions. We focus on indicators of both internalizing and externalizing behavior because research suggests that men and women respond to stress in different ways, with women more likely to respond via internalizing behaviors and men more likely to respond with externalizing behaviors (e.g., Horwitz & Davies, 1994).
Although the research just presented suggests that both internalizing and externalizing behaviors are likely in the wake of being a victim or perpetrator of EDS, another scenario is possible. This association could be spurious and the result of selection bias (Atkins et al., 2001; Blow & Hartnett, 2005; Thompson, 1983). For example, heavy drinking is associated with men’s EDS (Fincham & May, 2017; Jeanfreau et al., 2014) and among daters, psychological distress has been shown to be a precursor to EDS (Abrahamson et al., 2012; Buunk & Van Driel, 1989; Hall & Fincham, 2009). Similarly, EDS is often described as a symptom of negative relationship characteristics (Glass & Wright, 1977; Treas & Giesen, 2000; Munsch, 2015). Therefore, it could be the characteristics of partners or characteristics of the relationship and not EDS per se that explain any baseline associations between EDS and study outcomes.
Limited research has investigated the association between EDS and psychological distress. In a sample of dating college students, Hall and Fincham (2009) used EDS to predict psychological distress and other indicators of internalizing and found that sexually unfaithful individuals experienced greater psychological distress. This type of distress is by far the most common distress outcome examined in prior research on EDS (Allen et al., 2005). But partners may also respond to EDS with externalized responses, like heavy alcohol use. Most prior research on heavy alcohol use and EDS treats alcohol use as a predictor or precursor (see Abrahamson et al., 2012; Hall et al., 2008; Graham et al., 2016, for examples), but researchers suggest that it may be a consequence as well (Shrout & Weigel, 2018). Furthermore, Sweeney and Horwitz (2003) encourage the use of multiple outcomes to more fully understand responses to stress, and they suggest problem-drinking behavior as one such alternative. Therefore, in the current study we estimate two indicators of distress, one of which is an internalizing reaction and one of which is an externalizing reaction. We also move beyond prior research, which tends to rely on clinical, small, or dating samples by analyzing a large, nationally representative data source that includes both married and cohabiting individuals. This enables us to present a national picture of whether and how EDS produces distress among younger adults in United States who share their households and lives.
Data
Add Health is a school-based, nationally representative, longitudinal study of adolescents who were in grades 7–12 during the 1994–1995 academic year (Wave I). At this time, 20,745 students participated in an in-depth, in-home survey that asked about a range of topics including family and school experiences, relationship and sexual experiences, and health and risk-taking behavior. All respondents interviewed at Wave I, except for those that were high school seniors, were targeted for follow-up interviews at Wave II in 1996 (N = 14,378). All Wave I respondents were then targeted for follow-up at Wave III in 2001–2002 when respondents were between the ages of 18 and 26 (N = 15,197) and again at Wave IV in 2007–2008 when respondents were between the ages of 24 and 32 (N = 15,701). More information about the Add Health sampling design can be found on the study’s website (http://www.cpc.unc.edu/projects/addhealth) (Harris et al., 2009). Wave V data were recently collected during 2016-2018 but these data are not yet available.
The sample for this study is restricted to Wave I respondents who were successfully re-interviewed at both waves III and IV, had a survey weight to construct nationally representative estimates, and reported that they were currently in opposite-sex marriages or cohabiting relationships at Wave IV. Of the 15,701 respondents interviewed at Wave IV, 12,288 had a valid sampling weight (78.3%). Of these, 5,192 were in current marriages and 2,245 were cohabiting, leaving a final sample of 7,437 current opposite-sex unions. Note that the proportion of individuals already married or cohabiting in our sample is largely in line with national estimates (Stepler, 2017).
We do not exclude cases with missing data. Instead, we used multiple imputation procedures available in Stata version 13.0 to impute missing data on analytical variables. The procedure iteratively replaces missing values on variables using predictions based on random draws from the posterior distributions of parameters observed in the sample, creating multiple complete data sets (Allison, 2001). We average empirical results across ten imputation samples and account for random variation across samples to calculate standard errors (Royston, 2005). Missing information due to non-response was largest for personal income (5.1%) and 12.8% of the study sample had missing data on at least one variable.
Measures
Our analysis estimates two indicators of psychological distress, depressive symptoms and heavy alcohol use. The former is a measure of internalizing behavior, while the latter is an indicator of externalizing behavior.
Our indicator for depressive symptoms is derived from a modified nine-item version of the Center for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977) included in Add Health at Wave IV. Modified versions of the CES-D are used commonly by researchers (Crockett et al., 2005; Grzywacz et al., 2006; Frisco et al., 2013), and studies have shown versions using as few as four items to be reliable (Grzywacz et al., 2006). The nine-item CES-D instrument in Wave IV of Add Health asks respondents how often in the last week they were bothered by things not normally bothersome, could not shake the blues, felt like they were just as good as others (reverse-coded), had trouble focusing, felt depressed, felt too tired to do things, enjoyed life (reverse-coded), felt sad, and felt that people disliked them. Response options range from never or rarely (0) to most or all of the time (3).
We sum responses to arrive at a final scale score that ranges from 0 to 27, with higher scores indicating more depressive symptoms (α = .80). The overall mean level of depressive symptoms among respondents in our sample at Wave IV is 5.21. We currently show results estimating discrete CES-D scores at Wave IV, but found substantively similar results using a dichotomized variable that indicates CES-D scores as high or higher than those used in previous research to estimate a clinical indicator of depression (Gotlib et al., 1995; Primack et al., 2009; Shrier et al., 2001).
Our second dependent variable, heavy alcohol use, is derived from three questions answered by respondents at Wave IV asking how many days during the previous 12 months they “drank alcohol,” “drank 5 or more drinks in a row,” and “were drunk or very high on alcohol.” Responses to each of the three questions ranged from none (0) to every day or almost every day (6). Responses were summed and then averaged, with higher scores indicating greater levels of heavy alcohol use (α = .87).
Our primary independent variable of interest is respondents’ EDS experiences in their current opposite-sex marital or cohabiting union. It is based on two questions asked during the Wave IV survey. Respondents were asked whether their current partner had any other sexual partners during their relationship. They were then asked whether they had any other sexual partners during their relationship with their current partner.
We used responses to these two questions to create four dummy variables. 1 The reference category in all analyses is no known EDS in the relationship, which indicates the respondent answered “no” to questions about whether they or their partner had other known sexual partners. Respondent-only EDS indicates that the respondent reported another sexual partner but that their partner did not have other sexual partners. Partner-only EDS indicates that the respondent did not disclose any other sexual partners but indicated that their partner did have other sexual partners. Mutual EDS indicates that the respondent reported that both they and their partner had other sexual partners. Note that this measure of EDS has been used in previous research using the Add Health study (Frisco et al., 2017).
Our analysis also accounts for previous indicators of depressive symptoms and heavy drinking. We control for Wave III indicators, which are both measured in the same manner and using the same items as their Wave IV counterparts (α = .80 for depressive symptoms, α = .87 for heavy drinking). The Add Health study also assesses psychological distress and heavy drinking during adolescence, but we use the Wave III items for comparability to Wave IV. During Wave I and II, depressive symptoms are assessed with a 19-item CES-D instrument. Controlling for the prior indicators of psychological distress effectively turns our regression analyses into lagged dependent variable models.
We recognize that these models may produce biased results (Johnson, 2005). Thus, we also estimated models predicting a change in psychological distress and heavy alcohol use. Results from these models were substantively similar to results obtained using lagged dependent variable models, with few exceptions, which we discuss below. We present estimates from lagged dependent variable models to better show the role of selection in study findings.
We also control for sociodemographic characteristics of respondents. This includes gender of the respondent (1 = male), age of the respondent in years (measured at Wave III), dummy variables indicating race/ethnicity (White = reference, Black, Hispanic, or other race/ethnicity), whether the respondent was a high school dropout at Wave III (1 = yes), personal income at Wave III measured in thousands of dollars, and employment status at Wave III (1 = unemployed). We also control for whether the respondent had any children in the household (1 = yes), indicating whether the respondent reports any sons or daughters as part of their household roster at Wave IV.
Additional control variables include five indicators of risk-taking proclivity. We control for parental separation, coded 1 if the respondent lived in a household without both biological parents at the time of the Wave I interview. Drug use is based on four questions in the Wave III survey asking how often during the previous 30 days respondents used: marijuana; any kind of cocaine; crystal meth; and any other type of drug. We dichotomized each of the items and summed the responses to create an index of drug use ranging from 0 (no drug use) to 4 (usage of all four types). We also control for the number of sexual partners each respondent had vaginal intercourse with by the Wave III interview. We truncated the variable at 21 partners due to the skewed distribution. We include a measure of impulsivity created from four items at Wave I asking respondents whether they agree, using a five-point Likert scale, that: When they have a problem to solve, one of the first things they do is get as many facts about the problem as possible; when they are attempting to find a solution to a problem, they usually try to think of as many different ways to approach it as possible; when making decisions, they generally use a systematic method for judging and comparing alternatives; and after carrying out a solution to a problem, they usually try to analyze what went right and what went wrong (α = .74). Higher values indicate greater impulsivity. We rely on Wave I impulsivity (which indicates adolescent impulsivity) because it was not assessed again in the Add Health study. Additionally, to account for risk-aversion, we control for religious attendance, which indicates how often respondents attend religious services throughout the year. Religious attendance ranges from 0 (never) to 3 (once a week or more).
Finally, we control for characteristics of respondents’ current marital or cohabiting relationships at Wave IV. Relationship happiness is based on an item asking how happy respondents are in their relationship with their partner. Responses range from 0 (not too happy) to 2 (very happy). Relationship commitment is based on an item asking how committed respondents are to their relationship with their partner. Responses range from 0 (not at all committed) to 3 (completely committed). Love from partner is taken from an item asking respondents how much they agree that their partners express love and affection to them, with responses ranging from 0 (strongly disagree) to 4 (strongly agree). Our measure of sexual satisfaction is based on an item asking respondents how much they agree that they are satisfied with their sex life, with responses ranging from 0 (strongly disagree) to 4 (strongly agree). Relationship trust is measured with an item asking respondents how much they agree that they trust their partner to be faithful to them, again with responses ranging from 0 (strongly disagree) to 4 (strongly agree). Finally, relationship duration is measured as the total amount of time the respondent had been involved in a romantic/sexual relationship with their partner at the time of the Wave IV survey, with time measured in months.
Analytic Strategy
We begin our analysis by showing descriptive statistics for all study variables for the married and cohabiting respondent subsamples. These estimates are displayed in Table 1 to provide the context for our multivariate analyses estimating the associations between EDS and our study outcomes.
Descriptive Statistics by Relationship Type.
Source: National Longitudinal Study of Adolescent to Adult Health.
Note: All values are population weighted and based on imputed data.
We then use weighted ordinary least squares regression (OLS) models to estimate how EDS is associated with depressive symptoms and heavy alcohol use. Separate models are presented for currently married and cohabiting young adults. For analyses predicting each study outcome, the first model in each table includes EDS and the Wave III indicator of the study outcome being predicted. The second model adds controls for sociodemographic characteristics and risk-taking proclivity. Finally, the third model adds control variables for relationship characteristics.
Although no known EDS is the reference category in all of our models shown in tables, we estimated supplementary analyses switching out the reference category for EDS to test for other between-category differences. All statistically significant between-category differences observed in these ancillary analyses are noted in tables using superscripts and in the text when we describe estimated results. We also conducted supplementary analyses (available upon request), to estimate an additional model that included an interaction between gender and our indicators of EDS but found no statistically significant gender differences in the way that any forms of EDS were related to either outcome.
Results
EDS in Opposite-sex Marital and Cohabiting Unions
Panel 1 of Table 1 shows the prevalence of EDS in opposite-sex marital and cohabiting unions. Roughly three-fourths of married respondents (78%) and cohabiters (72%) report no known EDS. Respondent-only EDS is reported by 11 % of married respondents and 13% of cohabiters. In essence, among individuals in these romantic unions, roughly 1 in 10 admit they have had at least one other sexual partner.
Smaller percentages of respondents report a partner’s EDS; approximately 6% of spouses and 5% of cohabiters. Note that spouses’ and cohabiters’ reports of partner EDS are half as large as reports of respondents’ own EDS. This likely reflects findings from previous research indicating that partner EDS is both concealed well and also increases the odds of union dissolution when discovered (Frisco et al., 2017).
Finally, estimates in Table 1 suggest that mutual EDS is reported by 6% of married respondents and 10% of cohabiters. It is not surprising that more cohabiters report mutual EDS given that cohabiting unions tend to be less committed relationship arrangements (Stanley et al., 2004).
Depressive Symptoms
Turning to results from multivariate analyses, we first show models indicating how EDS is associated with a measure of internalizing behavior—depressive symptoms—net of prior depressive symptoms. Table 2 shows this association for married young adults. Estimates in Model 1 indicate that all forms of EDS are positively associated with depressive symptoms even when we account for Wave III depressive symptoms. 2 Each form of EDS is associated with more than a 1 point estimated increase in CES-D scale scores. After controlling for respondents’ sociodemographic characteristics and risk-taking behavior in Model 2, the estimated association between each form of EDS and depressive symptoms is reduced, but remains statistically significant. When we control for relationship characteristics in Model 3, the coefficients for partner-only and mutual EDS are rendered to statistical non-significance. The size of the coefficient for respondent-only EDS is also reduced but remains statistically significant. In supplementary analyses, when we switched the reference category for EDS to test for between-category differences, we found that, net of all controls, married young adults reporting respondent-only EDS also report significantly more depressive symptoms than those reporting either partner-only or mutual EDS.
Estimates from Weighted OLS Regression Models Predicting the Association between Extradyadic Sex and Depressive Symptoms among Married Young Adults (N = 5192).
Source: National Longitudinal Study of Adolescent to Adult Health.
Note: *p < 0.05, **p < 0.01, ***p < 0.001; rsignificantly different from respondent-only coefficient; psignificantly different from partner-only coefficient; msignificantly different from mutual coefficient.
Table 3 shows estimates of the association between EDS and depressive symptoms among opposite-sex cohabiting young adults. Net of prior depressive symptoms, only mutual EDS is associated with increased depressive symptoms when compared to no known EDS and the other forms of EDS. Controlling for respondents’ sociodemographic characteristics and risk-taking in Model 2 reduces the size of the coefficient for mutual EDS, but it remains a significant predictor of depressive symptoms relative to the other forms of EDS. In the final model, which includes relationship characteristics, the estimated association between mutual EDS and depressive symptoms is reduced to less than half of its original size and to statistical non-significance relative to all other forms of EDS.
Estimates from Weighted OLS Regression Models Predicting the Association between Extradyadic Sex and Depressive Symptoms among Cohabiting Young Adults (N = 2245).
Source: National Longitudinal Study of Adolescent to Adult Health.
Note: *p < 0.05, **p < 0.01, ***p < 0.001; rsignificantly different from respondent-only coefficient; psignificantly different from partner-only coefficient; msignificantly different from mutual coefficient.
In summary, these results suggest very few associations between EDS and depressive symptoms among respondents in marital and cohabiting unions that are not explained by the characteristics of these relationships. Only respondent-only EDS among spouses is positively associated with depressive symptoms when models are adjusted for relationship characteristics.
Heavy Alcohol Use
We now turn to models estimating how EDS is associated with heavy alcohol use. We first show results for married young adults in Table 4. Model 1 reveals that partner-only EDS is not associated with heavy alcohol use at Wave IV net of Wave III heavy alcohol use. However, respondent-only and mutual EDS are both positively associated with heavy alcohol use relative to no known EDS and partner-only EDS. These associations remain statistically significant even after controlling for respondent’s sociodemographic characteristics and risk-taking proclivity in Model 2 and relationship characteristics in Model 3.
Estimates from Weighted OLS Regression Models Predicting the Association between Extradyadic Sex and Heavy Alcohol Use among Married Young Adults (N = 5192).
Source: National Longitudinal Study of Adolescent to Adult Health.
Note: *p < 0.05, **p < 0.01, ***p < 0.001; rsignificantly different from respondent-only coefficient; psignificantly different from partner-only coefficient; msignificantly different from mutual coefficient.
Results indicating the association between EDS and heavy alcohol use among cohabiting young adults are presented in Table 5. In Model 1, neither partner-only nor mutual EDS are positively associated with heavy alcohol use in comparison to no known EDS. However, respondent-only EDS is positively associated with heavy alcohol use relative to no known EDS 3 . This association remains statistically significant after accounting for confounders in Model 2 and Model 3.
Estimates from Weighted OLS Regression Models Predicting the Association between Extradyadic Sex and Heavy Alcohol Use among Cohabiting Young Adults (N = 2245).
Source: National Longitudinal Study of Adolescent to Adult Health.
Note: *p < 0.05, **p < 0.01, ***p < 0.001; rsignificantly different from respondent-only coefficient; psignificantly different from partner-only coefficient; msignificantly different from mutual coefficient.
To summarize, respondent-only EDS is positively associated with heavy alcohol use relative to no known EDS among both married and cohabiting young adults. These associations are observed even after controlling for factors that may select individuals into heavy alcohol use or EDS. Additionally, among married young adults, mutual EDS is also positively associated with heavy alcohol use.
Discussion
EDS in a marital or cohabiting relationship is a major personal and marital stressor, regardless of whether one is the victim or transgressor (Allen et al., 2005; Bird et al., 2007; Cano & O’Leary, 2000; Fife et al., 2013; Gordon & Baucom, 1999; Gordon et al., 2004; Hall & Fincham, 2009; Shrout & Weigel, 2018; Sweeney & Horwitz, 2001; Wiggins & Lederer, 1984). Yet prior research has not sufficiently unpacked the ways in which individuals exhibit internalizing or externalizing behavior in response to their own or their partner’s EDS. Prior research has generally been limited to examining how internalizing or externalizing behavior results from (or predicts) EDS among small, clinical samples of married individuals in therapy (Beach et al., 1985; Gordon et al., 2004; Wiggins & Lederer, 1984) or among college students who are dating (Hall & Fincham, 2009). Additionally, prior research has not investigated how EDS may be associated with internalizing or externalizing behavior among cohabiters. Finally, analyses tend to focus on either the victim or perpetrator of EDS but not both and certainly not mutual EDS. These limitations of prior research left several important gaps in the literature that our study is able to address due to the features of the Add Health study. We are able to estimate how one’s own EDS, a partner’s EDS, and mutual EDS are associated with internalizing and eternalizing behavior on a national level for young adults in two types of opposite-sex unions with different degrees of legal and personal commitment. Results of our study suggest four primary conclusions.
First, results suggest that spouses who report their own EDS but have partners with no known EDS are at risk of increased depressive symptoms and heavy alcohol use relative to spouses in marriages with no known EDS, and this estimated association cannot be explained by any of the observable personal or relationship characteristics that we account for in statistical models. In essence, engaging in sex with a partner other than one’s spouse leads to both internalizing and externalizing behavior among unfaithful married people. This finding is in line with research that shows that managing the consequences of this non-normative and socially sanctioned behavior produces stress and guilt among those who report their own EDS (Allen et al., 2005; Beach et al., 1985; Glass, 2002; Spanier & Margolis, 1983; Wiggins & Lederer, 1984). For individuals whose spouses have not discovered their other sexual partner(s), guilt stemming from their indiscretions may heighten stress and distress. For those whose partners have discovered their infidelity, increased depressive symptoms and alcohol use may result from the stress of coping with their partner’s response to their behavior.
Consistent with findings for married individuals, respondent-reported EDS was also associated with an increase in heavy alcohol use among cohabiters, even after controlling for personal and relationship characteristics. Unlike spouses, however, respondent-only EDS is not significantly associated with depressive symptoms among cohabiters. While we can only speculate about the reason for this discrepancy between married and cohabiting respondents, the answer may be related to differences in the nature of marriages versus cohabitations. Specifically, the legal barriers to exiting a marriage do not exist for cohabiters. As the sample used here is restricted to couples who remain together after an EDS event, it may be possible that cohabiters who internalize the stress of infidelity as depressive symptoms may have already terminated their relationships. Therefore, they would not appear in our sample. In addition, some cohabiters who engage in EDS may leave their original partners for their EDS partner. These cohabiters would also be omitted from our sample of cohabiters who remain with their original partners. Cohabiters who externalize rather than internalize the stress of their own infidelity, however, may simply turn to heavy alcohol use rather than choosing to leave. This fits with the fact that several characteristics of cohabiters relative to married respondents also predisposes them to heavy drinking (which is more common among cohabiters). Cohabiters’ average frequency of religious service attendance is lower, they are younger, they report more drug use and more lifetime sexual partners, a greater proportion are unemployed, and they are less likely to report children in their households (see Table 1).
A second study conclusion is that partner-only EDS is not associated with depressive symptoms or heavy alcohol use among spouses or cohabiters once we account for relationship characteristics. Findings did reveal a statistically significant bivariate association between partner-only EDS and depressive symptoms among spouses, but the large estimated effect of a 1.52 increase in depressive symptoms was reduced to nearly zero (.08) and statistical non-significance once we accounted for the reported marital characteristics. This suggests that it is the relationship and not the partner EDS that actually predicts increased psychological distress.
It is not entirely surprising that we find very little evidence of internalizing or externalizing behavior in the wake of partner-only EDS. Partner-only EDS increases the odds of union dissolution among spouses and cohabiters (Frisco et al., 2017). Thus, individuals who remain with their partner are a select sample (and may not remain in these relationships in the long run). Additionally, given that these respondents remained in their relationships despite their partners’ infidelity, they may have already coped with their partners’ transgressions and moved on in order to justify staying in the relationship.
A third finding from this study is that mutual EDS is associated with heightened depressive symptoms among married and cohabiting partners, but only in models that do not account for relationship characteristics. This suggests any depressive symptoms experienced by partners as a result of mutual EDS actually are due to the poor quality of relationships rather than EDS. If both partners are seeking sexual partners outside of their relationships, it is not entirely surprising that this is due to the fact that they are in unsatisfying relationships. In supplementary models (available upon request) we confirmed that respondents reporting mutual EDS are less happy in their relationships, less committed to them, and less sexually satisfied than respondents reporting the other forms of EDS.
Finally, a fourth finding is that mutual EDS is also positively related to heavy alcohol use among married respondents even after controlling for relationship characteristics. Among cohabiting respondents, mutual EDS has no association with heavy alcohol use. We conducted supplementary analysis to better understand this difference. First we ensured that this difference is statistically meaningful. We found that it was not significantly different in the model that only accounts for personal characteristics (e.g. drug use, religious attendance, etc.) but it is significantly different when models adjust for relationship characteristics. This suggests that cohabiters who remain in relationships where mutual EDS is reported are a select group and that married individuals in these relationships may face more pressure to stay due to the higher level of commitment involved in marital relationships. Increased externalizing may be a coping mechanism among these spouses. Indeed, a second set of supplementary analyses further reveal that, on average, respondents in the study sample reported decreased heavy alcohol use from Wave III to Wave IV. When changes in heavy alcohol use are examined by relationship type and EDS type, all cohabiting respondents, regardless of their EDS experiences, reported less heavy drinking at Wave IV versus Wave III as do all married respondents with one exception; married respondents who report mutual EDS actually reported an average increase in heavy alcohol between Waves. Therefore, married respondents who experience EDS are a highly select group with respect to drinking behavior, likely because they stay together in spite of their EDS and resort to heavy alcohol use to cope.
While not an explicit purpose of the study, it is notable that we found no statistically significant gender differences in the way that any EDS experiences were associated with depressive symptoms and heavy alcohol use. Our results suggest that at least among a current cohort of young adults in marital and cohabiting unions, infidelity may have similar consequences for men and women. Future research should consider whether modern changes in gender roles and expectations may explain why we find no differences for men and women.
Though this study makes a contribution to our knowledge about EDS and psychological distress, it is not without limitations. First, our measure of EDS is based on respondents’ reports about their own and their current partners’ EDS. However, the questions used to capture EDS only refer to other sexual partners during the course of their current relationship; they do not specify the type of sexual activity involved in the EDS, at what point during the current relationship the EDS occurred, or whether the respondent and their partner had established an expectation of exclusivity prior to the EDS. For this reason, we have not referred to the EDS as “infidelity” or “cheating”. A more ideal way to operationalize infidelity itself would be to ask respondents about expectations of exclusivity in their current relationship, about the timing of any infidelity that occurred, and about the kind of infidelity engaged in. However, while we can only speak to EDS, estimates in the current study are very similar to those of prior studies that have measured infidelity using other national samples (Allen & Atkins, 2012; Munsch, 2015; Treas & Giesen, 2000).
Another limitation of our study is that the couples in our sample are limited to young adults in heterosexual marriages and cohabiting relationships. As such, our findings may not be generalizable to older couples or those involved in same-sex marital or cohabiting unions. Additionally, cohabiters are a diverse group of people, with varying levels of commitment. Future research should consider whether these varying levels of commitment result in differences in response to the stress of infidelity. Finally, our analysis is conducted on a sample of respondents currently in relationships meaning that our findings are not generalizable to the ways in which different forms of EDS may be related to internalized and externalized psychological distress among persons who leave partners in the wake of their own, their partners’ or both partners’ EDS.
Despite limitations, the current study is the first study to our knowledge to show how EDS is related to both married and cohabiting adults’ psychological well-being among a nationally representative sample of currently partnered individuals. Results suggest that cheating may provide new sexual opportunities, but not without a cost. Married and cohabiting respondents who report EDS report significantly more heavy drinking and married respondents additionally face greater depressive symptoms. The fact that this finding remains significant even after accounting for prior drinking and depressive symptoms and the characteristics of relationships, suggests that this finding is not due to selection. Conversely, the consequences of mutual EDS appear to be symptomatic of relationship problems. Study findings suggest that not all EDS experiences significantly impact the well-being of participants who choose to remain in their relationships post-EDS involvement. For many, it is a symptom of other problems in the relationship. For those who do cheat, however, future research is needed to determine strategies to help these individuals with pro-social coping mechanisms.
Footnotes
Acknowledgements
This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (
). No direct support was received from grant P01-HD31921 for this analysis. Finally, we thank Derek Kreager and Brendan Lantz for helpful comments on earlier versions of this manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We acknowledge assistance provided by the Population Research Institute at Penn State University, which is supported by an infrastructure grant by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (5P2CHD041025-17).