Intimacy in Young Adulthood as a Predictor of Divorce in Midlife

Abstract

Using an Eriksonian-based measure (Erikson, 1963), the Inventory of Psychosocial Development (Constantinople, 1969), this longitudinal U.S. study explored the extent to which an individual’s potential for intimacy in young adulthood predicted divorce by midlife. Intimacy was defined as the potential to establish close relationships involving high levels of communication, closeness, and commitment. Marital status 34 years after college graduation was obtained from 167 participants (M age = 55.1 years, 60% male, 30% divorced) originally tested in college in 1966–68 in the United States. Hierarchical logistic regression revealed a significant Gender X Intimacy interaction in predicting marital status at midlife. Women but not men with low intimacy in college had higher risk of divorce in midlife in the sample.

Within the past decade, marriage in the United States has come under increasing scrutiny by governmental and religious institutions. Cohabitation, domestic partnerships, same sex marriages, parenthood outside of wedlock, blended families, and postponement of marriage have received increased public awareness (Hackstaff, 1999). At the same time, the relatively high divorce rate in the United States provides evidence that the once-held view that marital vows were for “better or worse” no longer applies. Throughout adulthood, the percent of ever married individuals who have been divorced rises from approximately 10% of people in their 30s to approximately 40% of current midlife adults (U.S. Bureau of the Census, 2007).

Clearly relevant to understanding the factors that predict divorce is an understanding of marital satisfaction. Surprisingly, although there is extensive literature on marital satisfaction, relatively few studies focus on predictors of divorce, and even fewer employ a prospective design in which individuals or couples are studied prior to marrying. Among the available divorce prediction studies, characteristics of the couple rather than the individual partners serve as the main explanatory variables. Although the dynamics of the couple’s interaction play an important role in predicting the fate of their relationship, there is also value in examining qualities of the individual partners (Whitbourne & Ebmeyer, 1990). For example, using data from three longitudinal samples, Vaillant (1993) found that maturity of defense mechanisms was correlated with long-term marital stability. Although useful, this study did not employ a prospective design. In contrast, the present study used a prospective design to examine the predictive value of the individual’s potential for experiencing intimacy as a predictor of divorce.

The Construct of Intimacy

Intimacy is a complex construct that is notoriously difficult to define and operationalize. Erikson’s (1963) psychosocial theory was one of the first and broadest theories to introduce intimacy not as a quality of a romantic couple but as a potential within the individual. As such, Erikson emphasized three elements of the capacity for intimacy: willingness to make a commitment to another person, ability to share at a deep personal level, and capacity to communicate inner thoughts and feelings. Individuals who favorably resolve the so-called “Intimacy vs. Isolation” psychosocial crisis are, then, high on these three components. Isolation, at the opposite pole of the spectrum, is characterized by an inability to commit, share deep feelings, and communicate.

Although issues related to intimacy can continue throughout adulthood (Whitbourne, Sneed, & Sayer, 2008), Erikson theorized that individuals are maximally vulnerable to the crisis in young adulthood and it is then that intimacy may serve as a predictor for divorce in midlife. Thus, we suggest that individuals who are able to resolve intimacy issues successfully in young adulthood have a greater chance of maintaining a stable marital relationship through midlife. Conversely, we expect that young adults who are isolated are less likely to sustain a relationship, if they even marry.

Of course, intimacy as a quality of the individual should bear a connection to the interactions that take place between married partners. The interpersonal model of intimacy Reis and colleagues (Reis & Patrick, 1996; Reis & Shaver, 1988) developed elaborates the ways in which the ability to experience intimacy, developed and maintained within the context of typical relationship interactions, ultimately influences relationship outcomes over time. The interpersonal model defines intimacy as a reciprocal process in which the behavior and feelings of one partner influence the other and hence, the quality of the relationship. For example, if self-disclosure by one partner is followed up by active listening and responding by the other, the original partner will be even more likely to engage in self-disclosure. Empirical tests of the interpersonal theory of intimacy (Laurenceau, Barrett, & Pietromonaco, 1998; Laurenceau, Barrett, & Rovine, 2005) support the position that intimacy involves both self-disclosure and partner-disclosure with partner responsiveness serving as a mediator.

We propose that intimacy, conceived of in the Eriksonian tradition as a developmental quality, serves as an important link between the interpersonal model of relationship quality and divorce in midlife. Thus, the ability to experience intimacy in young adulthood provides the foundation for the subsequent reciprocal interactions that either sustain and promote or reduce and detract from the quality of the marital relationship.

Gender Differences in Intimacy

Differences between men and women in the expression of intimacy, well documented in the literature (Hyde, 2007), result from early socialization processes that precede young adulthood. Researchers have noted that by the age of three, boys and girls demonstrate differences in how they engage in interactive and intimate play (Cross & Madson, 1997). After childhood, according to this view, women and men continue to interact differently in relationships in ways that affect patterns of romantic relationships and marriage. Philpot, Brooks, Lusterman and Nutt (1997) suggest that men and women develop in different cultures that support different value systems, personality characteristics, communication styles, problem-solving techniques, gender roles, sexuality, and expectations for relationships.

Supporting these theoretical notions, a large meta-analysis of gender differences in self-disclosure (Dindia & Allen, 1992) revealed that, in general, women disclosed slightly more than men (d = .18). Nonetheless, in close relationships, this difference was magnified with women self-disclosing to a significantly higher degree than men. More recently, Leaper and Ayres (2007) have identified gender differences in language use patterns that could have an impact on intimacy; specifically, men are more likely to use assertive speech while women are more likely to use affiliative speech, a fact that may interfere with interactions between partners.

Several empirical investigations reveal that women report higher intimacy levels than men. Heller and Wood (1998) found that wives had higher intimacy levels on a self-report measure of intimacy than their husbands and were more accurate in predicting their partner’s feelings on the same measure. Similarly, Montgomery (2005) reported higher levels of intimacy across age groups for females from middle school through college. Among undergraduates, Hook, Gerstein, Detterich and Gridley (2003) reported that women scored higher on two of the four intimacy factors; namely love and affection, and personal validation. Results from these studies suggest that women are more intimate than men are or at the very least self-report that they are.

Another perspective on gender differences emerges from studies of relationship dissolution. As Duck and Wood (2006) proposed, men and women often deal with conflict in relationships differently. For example, men and women tend to vary in the amount they think about their marital relationship. Women appear to think about their relationships much more than men do and therefore notice even minor fluctuations in the quality of their relationship and communication patterns whereas men are more likely to “brood” rather than openly communicate. This can often lead to an unequal assessment of the quality of the relationship. Based on the literature, it is clear that communication holds a paramount position in relationship quality and stability. Therefore, we believe gender is worthy of investigation when examining the relationship between intimacy and marital outcomes.

Because researchers have identified the potential for intimacy as a crucial factor for maintaining a long-term relationship, we expected that lower intimacy in young adulthood to be associated with higher probability of divorce by midlife. Based on the findings discussed above, we also expected that women would self-report to be more intimate than men would. For exploratory purposes, we also investigated a gender by intimacy interaction such that the effect of intimacy on divorce would be stronger for women than for men. While exploratory, we base this last hypothesis on the gender differences literature (e.g. Duck and Wood, 2006) that suggest that changes in relationship and communication patterns may affect women more than men and thus women carry the majority of the burden in the success of a relationship.

Method

Participants

The current study is part of the Rochester Adult Longitudinal Study (RALS), a larger longitudinal investigation of personality development from college through midlife (Van Manen & Whitbourne, 1997; Whitbourne, Zuschlag, Elliot, & Waterman, 1992). The sample for the current study attended the University of Rochester (Rochester, New York, United States) in the 1960s and was first assessed in 1966. The University of Rochester is an outstanding and selective private university in Upstate New York. The overall sample were a group of well-educated individuals who, by midlife, were predominantly members of the middle and upper-middle socioeconomic classes. The sample initially began with 349 participants spanning the college years from sophomore to senior. In 2000–02, 182 participants completed testing. The final sample used in the current study consisted of the 167 participants who had married at least once; of these, 116 participants were continuously married and 51 had divorced at least once in 2000–02; of these, four had divorced twice and one had divorced three times. At the time of the 2000–02 follow-up, the sample, 60% male, ranged in age from 52 to 59 with an average age of 55 (SD = 1.47).

Measures

Demographic questionnaire

Participants completed a brief open-ended questionnaire in 2000–02 that requested information on their educational, work, and family history since college graduation. They were asked to provide dates and names of educational institutions and employers. Questions on family history asked for dates of major family life events.

Inventory of Psychosocial Development (IPD)

The IPD yields eight stage scores based on Erikson’s psychosocial theory of development with 5 positive items and 5 negative items contributing to each score. Respondents indicated how characteristic or uncharacteristic of them a given item is on a 7-point Likert scale (7=definitely most characteristic of you to 1= definitely most uncharacteristic of you). We obtained difference scores for each stage score by subtracting the summed score of the negative items from the summed scores of the positive items. For each stage, the range of possible scores is thus from −30 to +30. Examples of items on the intimacy subscale are "comfortable in intimate relationships," “very lonely,” “sexually aware,” and “can’t share anything.” We used the 1966 intimacy scores in the current study.

Procedure

The 2000–02 follow-up was initiated in November of 1999, when records were obtained from the most recent version of the University of Rochester Alumni Directory and through a fee-based service that helped to generate possible addresses from the existing information in the University files. The initial mailing began in May 2000 and yielded 105 participants. In subsequent efforts to locate participants from June of 2000 until October of 2002, additional resources were used, including online alumni directories, contact information from the alumni office, and website resources. These efforts produced an additional 77 respondents.

Results

Descriptive Information

We present intimacy scores in 1966 by gender and marital status (see Table 1). College intimacy scores of women who would remain married were significantly higher, t (68) = 2.15, p < .05, than the college intimacy scores of women who would later experience divorce. For men, intimacy scores did not differentiate who would divorce from those who would remain married. Thirty-percent (51/167) of the sample divorced during the time period between 1966 and 2000–02. Out of the 51 participants who divorced, 31 (19%) were men and 20 (11%) were women. Results from a chi-square test revealed a non-significant relationship between gender and divorce status, χ² = .38, df = 1, p > .05. Correlational analyses did not reveal a significant relationship between intimacy and age or length of first marriage.

Table 1.

Means and Standard Deviations

College Intimacy vs. Isolation Stage Scores (1966) and Marital Status as of 2000

	Marital Status
	Married n=116		Divorced n=51		Overall n=167
Variable	M	SD	M	SD	M	SD
Male (n=97)	10.29	6.62	10.97	7.06	10.51	6.73
Female (n=69)	12.14*	6.99	7.84	6.66	10.96	7.12

^*Note: p < .05; difference between married and divorced women’s intimacy scores.

Probability of Divorce

We used hierarchical logistic regression to determine the predictive value of college intimacy scores on midlife marital status among participants who had married at least once. For the purpose of the analyses, we coded participants as divorced regardless of how many times they were divorced (4 were divorced twice, and 1 was divorced three times). We entered the following predictor variables into regressions predicting marital status in midlife in a stepwise fashion: age at first marriage (as a covariate), intimacy, gender, and Intimacy X Gender interaction term.

First we evaluated age and length of first marriage as covariates; although age at first marriage was not significantly associated with marriage status, age at first marriage affected the variance explained by the final regression model so it was retained. Therefore, in the first step of the hierarchical regression, we entered age at first marriage as a predictor of marital status (see Table 2). Next, we entered the variables gender, intimacy and age at first marriage as predictors of marital status. There was no main effect for intimacy or gender in predicting marital status. In the third and final step, we entered the variables gender, intimacy, age at first marriage, and Gender X Intimacy interaction term as predictors of marital status. The exploratory analysis of the Gender X Intimacy interaction yielded a significant prediction of marital outcome in midlife (see Table 2; Wald= 4.19). By breaking down the interaction into simple effects, intimacy was only a significant predictor of divorce for women (p <.05; OR=.917) but not men. Thus, the results indicate that the intimacy effect on divorce varied by gender. For women but not for men, lower intimacy scores predicted greater risk for divorce in midlife. Additionally, the inclusion of age at the time of first marriage in the final model increased the overall fit of the model.

Table 2.

Summary of hierarchical logistic regression analysis for variables predicting marital status (N=167)

Variable	B	SE	Exp (β)	p	R2
Step 1
Intercept	.64	1.00	1.89	.52
Age at 1st marriage	−.04	.04	.96	.27	.01
Step 2
Intercept	.82	1.03	2.26	.43
Age at 1st marriage	−.05	.04	.95	.21
Intimacy	−.03	.03	.97	.23
Gender	−.26	.35	.77	.46	.03
Step 3
Intercept	.30	1.1	1.4	.78
Age at 1st marriage	−.05	.04	.95	.23
Intimacy	.02	.03	1.0	.65
Gender	.85	.65	2.3	.19
Gender x Intimacy v. Isolation	−.11	.053	4.2	.04*	.07

Note: Intimacy v. Isolation is a sum of 10 items on the IPD (Constantinople, 1969).

We coded married as“0” and divorce as “1” and males as “0” and females as “1.” Intimacy data are from 1966.

^*p=<.05

Discussion

The purpose of the present study was to investigate whether the potential of an individual to experience intimacy in young adulthood prospectively predicts divorce by midlife. To the best of our knowledge, the current investigation is the first in the psychological literature to undertake divorce prediction based on intimacy in young adulthood, prior to marriage.

Overall, low intimacy scores in college did not predict divorce. We believe this is the case because predicting marital outcomes from intimacy is complex and dependent on the interaction of intimacy with gender. In addition, men and women did not differ in college intimacy scores. This finding, though not expected, was consistent with the overall pattern of IPD scores in college (Whitbourne et al., 1992), which did not reveal sex differences.

Based on Erikson’s (1963) theory and Reis and Shaver’s (1988) model of interpersonal intimacy, we theorized that intimacy develops in a dynamic interplay through communication. It is likely that the inability of one member of a couple to engage in an intimate interaction such as listening to a partner and responding in an empathic manner may start to pave the way for marital dissatisfaction and ultimately dissolution. This is in line with Duck and Wood’s (2006) theory concerning communication and marital dissolution. Individuals have differing potentials as well as needs for intimacy and expectations of how these needs will be met in the context of marriage or other significant relationships. Therefore, it only takes one person who has a lower need or more difficulty communicating in an intimate way to create potential problems in the relationship.

From our findings of a gender by intimacy interaction, it appears that the burden of this predictive value of intimacy falls on the female partner. It is possible that the expectation that a woman will take the major share of responsibility for the couple’s closeness, including being higher in self-disclosure, may affect partner interactions such that when a woman fails to meet society’s norm of showing a high potential for intimacy, the relationship is more likely to falter. Thus, similar absolute scores on intimacy measures may mean different outcomes for men and women, as was the case in the current investigation.

Yet, there are other plausible explanations which further research could help to disentangle. This type of work could focus on whether or not there are gender differences in the facilitation of intimacy within marital relationships by gathering data from couples and determine if these gender differences have an impact on who initiates the divorce. It is also possible that there are cohort differences and that in future generations, the onus on the female partner may be lessened as socialization patterns move toward greater sexual equality (e.g. (Inglehart & Norris, 2003).

Our results represent an empirical demonstration of how developmental theories can predict that issues confronted at an earlier age influence outcomes that occur in life. This interpretation is consistent with Erikson’s (1963) theory postulating that an impaired ability to form and maintain intimate relationships during late adolescence results in negative consequences later in adulthood. Since communication and openness are crucial to forming intimate relationships, individuals who fail to resolve satisfactorily the Intimacy vs. Isolation psychosocial crisis, particularly women, may put themselves at risk for being unable to maintain a stable marital relationship.

Our findings, though preliminary, have practical implications. First, it can be useful for professionals and couples to identify and reverse destructive processes currently existing in a relationship by examining levels of intimacy within couples, particularly for female partners. Individuals who may be at risk for later marital discord can attempt to enhance preemptively their ability to interact with their spouses in a more intimate fashion. This effort may focus for example on active communication between couples.

There are several limitations to the current study. First, measures of relationship commitment and stability did not include an assessment of satisfaction with the relationship. Without these data, we cannot examine ways in which marital satisfaction may interact or mediate the relationship between intimacy and divorce. The theoretical models we presented suggest that individual intimacy levels may influence marital satisfaction, which in turn, tends to influence divorce outcomes. Second, given that the original sample consisted of volunteers, we cannot generalize the results to a known population. In addition, given that the sample was originally contacted in the 1960’s and has been followed for over thirty years, we cannot determine whether the results of the current study would apply to samples of individuals who may be available for testing on only a single occasion. Of course, we believe that the prospective nature of the data is a strength of this study, but nevertheless generalizability remains an issue. This last point highlights the demographics of the sample, which was primarily homogenous with regard to race and socioeconomic status.

We hope the current investigation continues to give rise to new ways of studying divorce, namely from pre-marital characteristics. This novel area of expanded research can provide relationship researchers additional areas to explore in understanding the many complex factors that can predict relationship dissolution and stability over the years of adulthood.