Abstract
Using a national sample of married (N = 752) and cohabiting (N = 323) couples, we examined the association between disillusionment and self-perceived breakup likelihood. Because disillusionment had not previously been studied in cohabiting couples, its extent and consequences for them were not known. We found considerable disillusionment in cohabiters, their mean level exceeding that of married couples. Based on a conceptual model of relationship change, we tested further whether disillusionment would predict self-perceived breakup likelihood, controlling for relationship satisfaction, commitment, and length. Furthermore, based on assumptions about barriers to leaving different types of relationships, we examined whether disillusionment’s association with breakup likelihood would be stronger in cohabiting than married couples. Results supported disillusionment’s ability to predict perceived breakup likelihood, even with rigorous controls, and the greater strength of this association in cohabiters. In addition, we found a significantly positive partner effect: Male partners’ disillusionment predicted female partners’ breakup likelihood.
Sociologist Willard Waller (1938) argued that during courtship, partners engage in impression management and idealize each other. Once married, however, spouses are less motivated to impress their partner and have difficulty sustaining idealized images of them. Thus, idealized images give way to more realistic ones and the intense romance begins to weaken, eventually leading to disillusionment and divorce in some couples. Disillusionment is a temporal construct reflecting a change for the worse in various relationship qualities (e.g., love, affection, and ambivalence) over time. The cause for this change is presumed to be the contrast between how partners experienced their premarital relationship, and the reality of spouses’ day-to-day married life. In the past, it has been measured either by actual changes in such relationship qualities when longitudinal panel data were available, or by a questionnaire measure of perceived change in cross-sectional studies (Niehuis, 2007; Niehuis & Bartell, 2006). Recent research has supported the deleterious effects of disillusionment on marital stability (e.g., Birditt, Hope, Brown, & Orbuch, 2012; Huston, Caughlin, Houts, Smith, & George, 2001). No study has examined the effect of disillusionment in cohabiting relationships, however, even though their number has steadily increased in the United States (Manning, 2010) and elsewhere. Using a national cross-sectional data set of 752 married and 323 cohabiting couples, the present study explores whether disillusionment also occurs in cohabiting relationships, and, if it does, whether its potential negative association with perceived relationship stability is similar to that found in married couples.
Disillusionment: Theoretical and Empirical Background
According to the “disillusionment” model, most couples enter marriage happily, but then experience a decline in positive feelings over time (Huston et al., 2001). This phenomenon has also been referred to as the “honeymoon is over” or the “honeymoon followed by blandness” effect (Aron, Normon, Aron, & Lewankowski, 2002; Kurdek, 1998). One explanation for this drop in marital quality emerged originally from Waller (1938), who asserted that idealization of the partner before marriage may hold the danger of leading to disillusionment in marriage. According to Waller, idealization is an important characteristic of courtship. During initial dating, partners see each other fairly realistically, he maintained. However, as the relationship develops and partners’ feelings for each other deepen, they increasingly feel a strong sense of enchantment and seek to display only some of their personality characteristics in order to live up to the images they believe that the other partner may have of them (Crosby, 1985; Waller, 1938). This process of idealization becomes increasingly stronger as the premarital relationship develops, until—according to the theory—one of the partners
builds up an almost completely unreal picture of a person which he calls by the same name as a real person and vainly imagines to be like that person, but in fact the only authentic thing in the picture is the emotion which one feels toward it. (Waller, 1938, p. 200)
Outstanding issues in the study of idealization, disillusionment, and their possible joint operation are examined in a theoretical/review article by Niehuis, Lee, Reifman, Swenson, and Hunsaker (2011).
Although empirical support for this notion of increasing idealization during courtship is lacking (e.g., Pollis, 1969), other writers, such as Halford, Kelly, and Markman (1997) and Nichols (1987), agree with Waller’s contention that individuals who have fallen in love lose their ability to accurately assess the character and personality of the object of their affection. According to Waller (1938), dating partners tend to discount information that might undermine their romantic feelings for, or commitment to, each other, and therefore, perceive both the partner and the courtship in an idealized fashion (Halford et al., 1997; Huston et al., 2001; Kayser, 1993; Swann, De La Ronde, & Hixon, 1994). One line of research (e.g., Hall & Taylor, 1976; Martz et al., 1998; Murray & Holmes, 1997; Murray, Holmes, & Griffin, 1996a, 1996b; Van Lange & Rusbult, 1995) has suggested that early idealization may serve the function of maintaining high expectations and deep levels of love and affection for the partner. For example, Murray and Holmes (1997) and Murray et al. (1996a) have found that people’s ability to idealize a flawed partner and an imperfect relationship predicts greater satisfaction, love, trust, and relationship stability, and less conflict and ambivalence. Furthermore, Murray and colleagues’ 1-year follow-up studies revealed a positive association between strong relationship illusions and subsequent increases in satisfaction.
However, other researchers (e.g., C. R. Berger & Roloff, 1982; Crosby, 1985; Hall & Taylor, 1976; Huston et al., 2001) caution that “illusionment” in dating relationships may hold the danger of disillusionment during the first few years of marriage. At that point, partners settle down to the daily tasks of married life, become increasingly more interdependent, are less concerned with impression management, get to know each other better, and compare their expectations from courtship with the reality of their marriage. According to Huston et al. (2001), disillusionment is bound to happen, particularly for those couples whose illusions are less a form of embellishment and more a form of fantasy. As spouses’ illusions about their relationship and each other vanish, problems that were latent during courtship surface in marriage, partners begin to devalue the person they once idealized (Nichols, 1987; Waller & Hill, 1968), and the strong romantic climate of their marriage disappears, creating feelings of disappointment (Karney & Bradbury, 1997; Kayser, 1993; Waller, 1938). The disillusionment model suggests that the decline in marital quality predicts subsequent likelihood of divorce (Huston et al., 2001).
Four studies demonstrate the significance of marital disillusionment for subsequent divorce. Kayser (1993) wrote about the disaffection process, and defined it as the “deterioration of emotional attachment in marriage” (p. 257). On the basis of disaffected spouses’ qualitative characterizations of their feelings, thoughts, and behaviors, she concluded that disillusionment was part of the disaffection process. The disillusioned husbands and wives in her sample often unfavorably compared their spouse’s behavior after marriage with that prior.
Huston et al. (2001) examined the early marital antecedents of conjugal distress and divorce using the disillusionment model as one paradigm to frame their research. They found that disillusionment during the first 2 years of marriage, as reflected in (a) loss of love, (b) declines in affection, (c) decreases in one’s perception of the partner as a responsive person, and (d) increases in feelings of ambivalence about the relationship, distinguished couples headed for divorce from those who stayed married. Furthermore, these authors have been able to show that the timing of marital dissolution was a function of the newlywed partners’ initial feelings for each other, their behavior toward each other, and their perceptions of each other, as well as the extent to which they experienced disillusionment during the following 2 years of their marriage. That is, couples who divorced after more than 7 years of marriage generally entered matrimony on a more positive note before they subsequently became disillusioned, whereas couples who divorced before their seventh wedding anniversary started out on a less positive note before they experienced disillusionment. The latter findings suggest that idealization may involve not only embellishing a partner’s good qualities but also discounting or dismissing their faults or limitations.
In a longitudinal study of newlywed marriages, Lavner and Bradbury (2010) examined whether changes in marital satisfaction over the first 4 years of marriage could be categorized into types of trajectories and whether such types were associated with different rates of divorce at the 4- and 10-year marks. They identified five different trajectories. Whereas three of the five groups were characterized by small, if any, declines in marital satisfaction over time, the remaining two showed linear declines that were more pronounced. Moreover, “rates of divorce corresponded closely with levels of marital satisfaction within the groups” (Lavner & Bradbury, 2010, p. 1182). Specifically, at both the 4- and 10-year points, spouses in the three most satisfied groups had far lower rates of ending their marriage than did those in the two least satisfied groups. Lavner and Bradbury also explored various factors that might differentiate among the different types and found that those with the most substantial decline in marital satisfaction also had the most problematic personalities (neuroticism, anger, low self-esteem), the most chronic stress during the first 6 months of marriage, and the most aggression (verbal and physical violence), negative affect and the least positive affect during problem-solving conversations as newlyweds. These latter findings provide support for the idea that some couples may not have been ignorant of their partner’s flaws prior to marriage but discounted or minimized them. Lavner and Bradbury did not find any differences between the trajectory types based on demographic characteristics.
Finally, Birditt et al. (2012) examined changes in marital happiness in a six-phase longitudinal study that followed newlywed White (N = 174) and Black (N = 199) couples over a period of 16 years. Using group-based trajectory modeling to identify patterns of marital happiness among husbands and wives, these researchers found (a) that qualitatively different trajectory groups of marital happiness were present in their data (confirming similar findings by Anderson, Van Ryzin, & Doherty, 2010; Kamp Dush, Taylor, & Kroeger, 2008; and Lavner & Bradbury, 2010) and (b) that these trajectory groups were differentially associated with divorce rates. The couples with the most rapidly declining marital happiness had the highest divorce rates. This finding, thus, supported the disillusionment model.
Cohabitation
During Waller’s (1938) time, few couples cohabited before, or as an alternative to, marriage. However, as noted above, cohabitation has increased in recent decades, judged by metrics such as the percentage of U.S. women in different age-groups who have ever cohabited or have cohabited prior to first marriage (Manning, 2010). Thus, the question arises whether disillusionment would also occur in cohabiting couples and, if so, to what extent. It could be argued that entering into cohabitation ought to be associated with as much disillusionment as entering into marriage without prior cohabitation, all other factors such as dating length being equal. Cohabiting with a partner ought to make prolonged impression management, idealization of the partner, and unrealistic expectations of the partner and of the relationship unsustainable.
A study by Rhoades, Stanley, and Markman (2012) provides some evidence for this idea. Using a nationally representative sample of unmarried individuals in heterosexual relationships, Rhoades and colleagues found that cohabiting relationships, as opposed to noncohabiting (i.e., dating) relationships, were characterized by more commitment but lower relationship satisfaction and other indices of relationship quality. Importantly, longitudinal follow-ups for a smaller portion of the sample showed that once individuals began to cohabit, commitment, relationship satisfaction and other indices of relationship quality decreased. Although cohabiters’ amount of commitment may vary depending on the presence or absence of marital intentions (e.g., Poortman & Mills, 2012), they generally seem to invest less in their relationships than married individuals and couples, especially with regard to joint investments, such as having children or purchasing a house (e.g., Heimdal & Houseknecht, 2003; Kiernan, 2001). Hence, cohabiters will typically have fewer barriers than married couples to ending their relationships. Furthermore, Rhoades, Stanley, and Markman (2009) showed with a large national data set that individuals who cohabited prior to their engagement reported lower levels of satisfaction (and other relationship-quality indices) and greater self-perceived odds for divorce than individuals who cohabited after they had gotten engaged or those who did not cohabit at all prior to marriage. Similar findings were reported by Kline et al. (2004) with a convenience sample of 136 couples.
Poortman and Mills (2012) recently developed and tested a typology of unions that vary in interpersonal commitment within marriage and cohabitation. Their typology “ranges across four situations: (a) cohabiting without marriage intentions or uncertain plans, (b) cohabiting with marriage plans, (c) married after a period of cohabitation, and (d) direct marriage” (Poortman & Mills, 2012, p. 359). The researchers then related this typology to the level of joint investments, such as having children together and/or purchasing a home. Using panel data from the first two waves of the Netherlands Kinship Panel Study (N = 2,362), the authors found a positive relationship between interpersonal commitment and joint investments. Cohabiters without marriage plans invested the least and couples who directly married without prior cohabitation invested the most.
Relationship Dissolution
The present study focuses on the relationship between disillusionment and perceived breakup likelihood. Because the study is cross-sectional, following couples longitudinally to observe whether actual breakups occurred was not possible. However, prior research (e.g., Brown, 2000) provides evidence linking perceived breakup likelihood to actual relationship termination. The present study carries the potential for theory development in the area of relationship dissolution. Conceptual models of the relationship dissolution process (Rollie & Duck, 2006; VanderDrift, Agnew, & Wilson, 2009) differ in their details, but they generally incorporate an internal cognitive stage in which one or both partners privately consider the state of the relationship and whether he or she wants to end it. Either during or after this contemplation, the couple member leaning toward termination informs the partner (and possibly others in one’s social network) of this intention, leading to a series of discussions and decisions (e.g., whether to end the relationship or work toward reconciliation). The construct of disillusionment appears highly compatible with the proposed internal cognitive phase of the relationship dissolution process, as disillusionment perceptions (e.g., beginning to see the relationship in a more negative light) likely either trigger or intensify thoughts about ending the relationship.
Hypotheses and Research Questions in the Present Study
The present study sought to extend disillusionment research by testing hypotheses (firm directional predictions) and research questions (exploratory inquiries where we feel the literature does not currently allow firm predictions) derived from the disillusionment model, cohabitation research, and relationship dissolution literature in a national sample of married and cohabiting couples.
Research Question 1 examined whether mean levels of disillusionment would differ between married and cohabiting participants. Because disillusionment has not previously been studied in cohabiting couples, there is little basis to assert a firm prediction regarding differences between cohabiters and married persons. In marriage, considerable opportunity exists (in terms of relationship length) for idealization during courtship to transform into postwedding disillusionment. However, as we noted above, perpetual idealization may also be unsustainable in day-to-day living once couples begin (nonmarital) cohabitation.
Hypothesis 1, based on the disillusionment model, predicted a positive association between disillusionment and self-perceived breakup likelihood. To overcome the potential alternative explanation that disillusionment is simply redundant with other relationship constructs such as satisfaction and commitment, these other variables were controlled statistically. Because disillusionment represents a perceived change for the worse in relationships, whereas satisfaction and commitment may be more absolute perceptions in the moment, disillusionment is expected to have unique predictive power.
Hypothesis 2, based on recent cohabitation research, predicted that the association between disillusionment and perceived breakup likelihood would be stronger in cohabiting than in married participants. Rhoades et al.’s (2012) findings of declining commitment, satisfaction, and other relationship-quality indices on cohabitation suggest the potency of processes akin to disillusionment among cohabiting couples. Furthermore, one can assume that marriage carries more barriers to ending a relationship than does cohabitation, and that such barriers would weaken the connection between disillusionment and perceived breakup likelihood. These arguments, thus, predict a stronger association between disillusionment and relationship breakup likelihood in cohabiting than in married couples.
Hypothesis 3 stems from the dyadic nature of our data. Whereas virtually all studies test what are known as “actor effects” (i.e., how independent and dependent variables are related within the same persons), dyadic data further allow the testing of “partner effects” (Kenny, Kashy, & Cook, 2006). The latter represent possible cross-partner predictive relationships (e.g., from one partner’s disillusionment to the other’s perceived breakup likelihood). Assuming that individuals’ private thoughts and feelings of disillusionment may sometimes be expressed in nonverbal or verbal behavior (e.g., unflattering statements to their partners), their partners may sense that there are difficulties in the relationship. This impression could then provide a pathway from Partner A’s sense of disillusionment to Partner B’s increased estimation of breakup likelihood. Accordingly, we predicted partner effects showing positive relations from each couple member’s disillusionment to the other’s perceived breakup likelihood.
Method
Project Overview
This study was part of a larger project by the National Center for Family & Marriage Research (NCFMR), in conjunction with the survey research firm Knowledge Networks (KN), to survey members of married and cohabiting couples. The data set included core items pertaining to close relationships (e.g., satisfaction, social support), plus specialized items submitted by seven teams of researchers who were selected in a competitive process to have content of interest to them included in the survey. Publications emanating from the project thus far include a study of advance-care planning for medical treatment (Carr, 2012) and work–family conflict (Nomaguchi & Milkie, 2011).
Sample
From July 26 to October 13, 2010, KN conducted an Internet-based survey on a national sample of heterosexual couples (married and cohabiting) aged 18 to 64 years in the United States. The firm maintains a national panel of approximately 50,000 persons, originating from address-based sampling, which “involves probability-based sampling of addresses from the U.S. Postal Service’s Delivery Sequence File” (KN, 1998-2011). For any given project, a random sample is drawn from the larger panel. In the present study, both members of each couple participated. For the married subgroup, both spouses in all couples were active members of the KN panel. Men (assumed to be reluctant participants) were contacted and screened before women in an attempt to obtain both partners’ participation most efficiently. The company
assigned the survey to 1,500 [married] men of whom 1,060 completed [it]. The wives of the men who completed were assigned their survey and 752 completed [it]. 1504 survey[s] representing 752 [married] couples were included in the final data file (personal communication, NCFMR, January 21, 2011)
Of the original 1,500 married men approached, therefore, 50.1% eventuated in complete-couple data.
For the cohabiting subgroup “266 men [of the panel] were assigned to the survey and 159 completed. All 159 female partners of these men were assigned the survey and 108 completed. Thus . . . 108 couples were included in the data file” (personal communication, NCFMR, January 21, 2011). To augment this small number of cohabiting couples from entirely within the panel, other sources were pursued. A second source of cohabiting couples were those in which one member was an active panel member and the other was not. Of 580 panel members (of either sex) who were contacted, “170 completed the survey and provided their partners[’] email address.” These 170 inquiries yielded only 31 completed partner surveys. Finally, an opt-in panel (recruited via online ads) was consulted and yielded both partners from 184 cohabiting couples. The cohabitation subgroup thus consisted of 323 complete couples in total. Of the 846 cohabiting couples who potentially could have been obtained via the panel, only 139 couples participated (16.4%). Comparisons of the panel and off-panel cohabiters by NCFMR revealed that the former were significantly older, higher in income, and more likely to be living in a house (vs. apartment).
Even though intricate systems of sample weights are available for the larger panel and the study-specific sample to correct for under-/overrepresentation on characteristics such as gender, age, race/ethnicity, and education (personal communication, NCFMR, January 21, 2011), they were not used in the present study, except when noted. We found that weighted and unweighted results were very similar in descriptive analyses.
Demographic characteristics of the final unweighted sample (1,075 couples) were as follows. Among men, 79.4% were White, 8.0% Hispanic, 5.8% Black/African American, 4.7% other, and 2.1% multiracial/non-Hispanic. Among women, 81.6% were White, 7.5% Hispanic, 4.2% Black/African American, 5.2% other, and 1.5% multiracial/non-Hispanic. On education, the sample distribution of men was 6.5% with less than a high school diploma, 26.5% with a high school diploma, 32.4% with some college, and 34.6% with a bachelor’s degree or higher. Among women, 4.2% had less than a high school diploma, 20.0% a high school diploma, 40.7% some college, and 35.1% a bachelor’s or higher. Compared with 2010 national estimates (U.S. Census Bureau, 2011), Blacks and Hispanics were underrepresented in the present sample. Furthermore, compared with National Center for Education Statistics (2010) figures, the sample was somewhat more educated than the U.S. population. Because population representativeness was not achieved, we simply use the term national sample to describe the participants.
Procedures
Recruited individuals were asked to complete the survey online. Those with computers and Internet service could use them, whereas a laptop and Internet service were provided to those without. Panel members received various types of prizes and cash awards for survey completion. Information on KN’s national panel is available via the company website (KN, 1998-2011).
Measures
Disillusionment
Disillusionment was measured by 11 items from the Relationship Disillusionment Scale. Niehuis and colleagues (Niehuis, 2007; Niehuis & Bartell, 2006) originally developed the Marital Disillusionment Scale, which contained 16 items, with high internal consistency, convergent validity, and criterion validity (Niehuis, 2007; Niehuis & Bartell, 2006). For the present study, some items were revised to encompass cohabiting couples; existing items’ references to “marriage,” for example, were changed to “marriage/relationship.” Hence, the revised instrument is called the Relationship Disillusionment Scale. The Relationship Disillusionment Scale version used in the present study was shortened to 11 items, because of space limitations in the NCFMR/KN survey. Example items were “I’m beginning to see my relationship in a somewhat more negative light” and “I feel no longer quite as positively about my spouse/partner as I once did.” Items were assessed on a 1 (strongly agree) to 5 (strongly disagree) scale, then reverse-coded and averaged into an index on which higher scores indicated greater disillusionment. Cronbach’s alpha for the main four subgroups (married men, cohabiting men, married women, and cohabiting women) consistently was in the range of .92 to .93.
Relationship Satisfaction
One item, similar to items used in the Changing Lives of Older Couples national survey and the National Survey of Families and Households, asked, “Taking all things together, how satisfied are you with your relationship with your spouse or partner?” and was scored on a response format of very satisfied (1) to very dissatisfied (5). The item was reverse-scored, so that higher values represented greater relationship satisfaction.
Relationship Commitment
Based on Poortman and Mills (2012), a five-level ordinal variable measuring commitment was created from information in the survey on cohabitation, marriage, and marital intent (our extra group came from being able to differentiate, among cohabiters with marriage plans, those who formulated those plans before vs. during cohabitation). Members of cohabiting couples were asked, “Before you were officially living together, had you and your partner already decided to get married in the future?” and “Now that you are living together, have you and your partner agreed to get married in the future?” Married participants were asked, “Did you live with your spouse before you got married?” Answer options for these three questions were “yes” or “no.” The ordinal commitment variable was created as follows. Cohabiters who had not yet agreed to marry at the time of the survey (either prior to or during cohabitation) were assigned a score of 1 (least committed). Individuals who were cohabiting at the time of the survey, but had made a decision midway through their cohabitation to marry, were assigned a score of 2. Cohabiters who had decided to marry even before moving in together, but had not yet entered matrimony, were assigned a 3. Married persons who had previously cohabited with their (future) spouse received a score of 4, and their married counterparts who did not previously cohabit together received a 5. Of the 1,066 couples with commitment scores for both partners, 988 (92.7%) exhibited a perfect match between partners/spouses.
Relationship Length
A measure of overall relationship length (in years) from the time of first dating the current spouse/partner to the present was derived by subtracting the year that dating began from 2010. Men’s values had fewer missing data than women’s; thus, we used the former. Average overall relationship length in married participants was 21.3 years (range = 1-52 years); roughly one third of married couples had been together for 13 or fewer years, another third from 14 to 26 years, and a final third for 27 or more years. For the cohabiters, overall time together averaged 7.8 years (range = 0-41); nearly 75% of these couples had been together 10 or fewer years.
Relationship Breakup Likelihood
To measure self-perceived breakup likelihood, the survey asked the following item, which is similar to one from the NSFH: “What are the chances you and your spouse/partner will break up in the future?” with options: no chance (1), little chance (2), 50-50 chance (3), a pretty good chance (4), and an almost certain chance (5). Measures of perceived breakup or divorce likelihood in cross-sectional studies are common, including in the form of a single item (e.g., Rhoades et al., 2012). To support the validity of a single-item measure of perceived breakup likelihood similar to the present one, Rhoades et al. cite research by Brown (2000) showing that, among cohabiters, high couple scores on perceived likelihood of separation were associated with actual breakup occurrence at a 4-year follow-up.
Results
Preliminary Descriptive Information
The following preliminary analyses were conducted for the two purposes of describing the sample and examining demographic characteristics for possible inclusion in later analyses as control variables.
Descriptive Statistics and Correlations Among Major Study Variables
Table 1 presents means and standard deviations for, and correlations between, the disillusionment, relationship-satisfaction, perceived breakup likelihood, and relationship-length measures. Information is presented for men and women, in married and cohabiting couples. All variables other than relationship length were solidly correlated (absolute r > .50) in the expected directions. Relationship length showed little association with the other variables, except for small but significant negative correlations with perceived breakup likelihood, in both married and cohabiting men. Taking advantage of the couple nature of the data, correlations between male and female partners on disillusionment, satisfaction, and breakup likelihood were computed for both married and cohabiting couples (see matrix diagonals in Table 1). These correlations were consistently in the .50s. Because the commitment item encompasses both married and cohabiting participants, it could not be analyzed separately in the two subgroups. In the full sample, however, ordinal commitment correlated significantly (p < .001) with all the other major variables: disillusionment (−.16 in men/−.17 in women), satisfaction (.19/.19), perceived breakup likelihood (−.32, −.35), and relationship length (.50/.50).
Correlations Between Major Variables in Married (Top) and Cohabiting (Bottom) Groups, Along With Descriptive Statistics, for Men and Women.
Note. Men’s (M) correlations are shown below the diagonal, whereas women’s (W) are shown above it, in the respective sections of the table for married and cohabiting participants. Along the diagonal in parentheses are correlations between male and female partners’ values on the same variables. Relationship length was treated as a couple-level variable, hence means and standard deviations are not separated into men’s and women’s.
p < .05. **p < .01. ***p < .001 (two-tailed).
Demographic Comparisons on Disillusionment
Demographic subgroups were compared on mean disillusionment, using between-group analyses of variance (ANOVAs). Each of these analyses was conducted separately in men and in women. Because the large sample size and multiple comparisons warranted caution, we required both overall F-tests and Tukey follow-up comparisons to be significant before claiming a significant result. By this standard, there were no significant differences in disillusionment by age categories (18-29, 30-44, 45-59, 60-64) or race-ethnicity. Education showed weak negative relations to disillusionment in the ANOVA results. However, in women, specific linear contrasts were significant (p < .05). Women with less than a high school diploma reported the highest disillusionment mean (2.17), whereas those with a bachelor’s degree or higher had the lowest (1.86). Because of these weak findings, the above demographic variables were not included as control variables in the later regression analyses.
Examination of the Research Question and Testing of the Hypotheses
Research Question 1
Differences in disillusionment between men and women (within-couple variable) and married and cohabiting individuals (between-participants variable) were examined in a two-way mixed-model ANOVA. Partial eta-square (η2) was obtained as a measure of effect size (akin to proportion of variance accounted for; Richardson, 2011). Means for the four subgroups are plotted in Figure 1. The ANOVA revealed significant main effects of sex, F(1, 1069) = 19.16, p < .001, partial η2 = .02 (considered a small effect); and relationship type, F(1, 1069) = 25.19, p < .001, partial η2 = .02. As seen in the figure, women consistently reported greater disillusionment than did men, and cohabiters reported greater disillusionment than did their married counterparts. The Sex × Relationship Type interaction nearly reached conventional significance, F(1, 1069) = 3.45, p = .064, partial η2 = .003. The interaction appears to reflect the slightly greater gap between women’s and men’s disillusionment means (2.15 vs. 1.99, respectively) among cohabiters than among the married (1.85 vs. 1.79).
Mean comparisons on the Relationship Disillusionment Scale (maximum = 5).
Hypothesis 1
To provide a very conservative test of the disillusionment index’s ability to predict self-perceived breakup likelihood in line with Hypothesis 1, hierarchical regression analyses (Table 2) were conducted in combined married–cohabitation samples, separately in men and in women. Relationship satisfaction and length (the two most established predictors in the data set) were entered in the first block, the newly formulated commitment variable in the second, and disillusionment (the variable of focal interest) in the third. With very rare exception, all variables in all blocks were statistically significant, for men and women. As expected, greater relationship satisfaction, length, and commitment were all associated with lower self-perceived breakup likelihood. Above and beyond these other significant predictors, disillusionment positively predicted perceived breakup likelihood. In fact, disillusionment had clearly the largest standardized beta coefficient (in absolute value) in the final step of the men’s and women’s equations, and the addition of disillusionment increased the variance accounted for in perceived breakup likelihood (change in R2) by roughly 10 percentage points in both men and women. These findings support Hypothesis 1.
Multiple Regression Analyses Predicting Self-Perceived Breakup Likelihood.
Note. All R2 changes between Blocks 1, 2, and 3 were significant, p < .001.
p < .05. **p < .01. ***p < .001 (two-tailed).
Hypothesis 2
To test whether a stronger link existed between disillusionment and perceived breakup likelihood in cohabiting than in married couples in line with Hypothesis 2, partial correlations were computed between these variables, within subgroups (male married, male cohabiting, female married, female cohabiting). These correlations controlled for relationship satisfaction, commitment, and length. Then, using Preacher’s (2006) method, we compared partial correlations for married and cohabiting groups (separately for men and women). Male cohabiters exhibited a larger partial correlation (r = .46) than their married counterparts (r = .38; significance of difference, p < .01). Female cohabiters also showed a larger partial correlation than their married counterparts (rs = .51 vs. 31; p < .001). These findings support Hypothesis 2.
Hypothesis 3
A path model in AMOS (Arbuckle, 2006) operationalized Kenny et al.’s (2006) actor–partner interdependence model to investigate simultaneously actor and partner effects. Of specific interest was Hypothesis 3’s prediction of positively signed pathways from one partner’s disillusionment to the other’s perceived breakup likelihood. The full (i.e., combined married-cohabiting) sample was used. Because males’ and females’ reports of relationship length were nearly perfectly correlated (.98), and likewise their reports of commitment (.97), only one length variable and only one commitment variable were used. Because men’s versions of the length and commitment variables had fewer missing data than women’s, men’s were used. The model fit well (χ2 = 6.3, df = 4; normed, nonnormed, and comparative fit indices > .99; root mean square error of approximation = .02). Results are shown in Figure 2. The actor–effect results of the path model replicate the earlier regression findings. However, two partner effects (in bold) were also found. Greater male disillusionment was associated with the female partner estimating a higher likelihood of relationship breakup (consistent with Hypothesis 3). Also, the more satisfied the female partner was with the relationship, the less the male partner saw the couple as likely to break up. Correlations between independent variables and between dependent variables are not shown, for ease of viewing. Similar correlations (broken down by married and cohabiting groups) can be gleaned from Table 1. Also not shown are nonsignificant paths from women’s disillusionment to men’s perceived breakup likelihood, and from men’s relationship satisfaction to women’s perceived breakup likelihood.
Path-analysis model of actor (regular line-thickness) and partner (bold line-thickness) effects. Only significant associations are shown. Coefficients are standardized. Correlations between independent variables and between dependent variables are not shown.
Discussion
The present study sought to extend disillusionment research by testing hypotheses derived from the disillusionment model and cohabitation research in a national sample of married and cohabiting couples. Hypothesis 1, based on the disillusionment model, predicted a positive association between disillusionment and self-perceived breakup likelihood. To overcome the potential alternative explanation that disillusionment is simply redundant with current perceptions of relationship quality, we statistically controlled for relationship satisfaction. Moreover, to rule out the possibility that the relationship between disillusionment and perceived breakup likelihood might be a function of commitment (especially among cohabiting individuals), we also controlled for this variable. Our findings were in support of our first hypothesis. Disillusionment, representing a perceived change for the worse in relationships, predicted perceived breakup likelihood above and beyond current perceptions of relationship satisfaction, commitment, and length and, thus, had unique and meaningful predictive power. This result suggests that disillusionment, though closely related to satisfaction and commitment, captures an additional important aspect of relationships (i.e., perceived change) that enhances researchers’ ability to predict relationship outcomes.
In addition, the study examined whether disillusionment, heretofore only studied in the context of married couples, also occurred to a comparable extent in cohabiting relationships (Research Question 1), and whether any positive association between disillusionment and perceived relationship breakup likelihood in cohabiters would be larger than that found in married couples (Hypothesis 2). Mean comparisons showed that cohabiters exhibited somewhat higher disillusionment than married persons, supporting the idea that disillusionment also occurs in cohabiting relationships. However, the overall degree to which both cohabiting and married men and women experienced disillusionment in their relationships was relatively low. Regarding Hypothesis 2, we were able to document that the association between disillusionment and self-perceived breakup likelihood was indeed stronger in cohabiters than in married persons.
The latter finding suggests that breakup likelihood may not only be a matter of disillusionment but also a function of barriers to leaving the relationship. Presumably, barriers to leaving the relationship (Johnson, 1991) are stronger in married than in cohabiting couples, thus weakening the connection between disillusionment and breakup propensity in married couples. Further research into the dynamics of “sliding” and “deciding” to advance one’s relationship from dating to cohabiting to marriage, the disillusionment process, and the role of barriers to leaving the relationship would provide further insight into the general associations found in the present study.
In partial support of Hypothesis 3, one disillusionment-based partner effect occurred in the path model. Specifically, men’s disillusionment was positively related to women’s perceived breakup likelihood. In contrast, women’s relationship satisfaction predicted lower perceived breakup likelihood in men. Speculatively, it is possible that women are attuned to men’s negative affect in assessing how well the relationship is going, whereas men are attuned to women’s positive affect for this purpose. Again, the notion that women may be more sensitive than are men to negative developments in relationships (Lavner & Bradbury, 2010) might explain this finding. However, additional research is needed to confirm this intriguing hypothesis.
Mean differences between demographic groups (as tested in preliminary analyses) were relatively rare. Women exhibited somewhat higher disillusionment than men. Lavner and Bradbury (2010) suggested the possibility that women may be more sensitive than men to disappointment with how their relationships are going, because men may well benefit more than women from the relationship in tangible ways (e.g., more often than not having housework done for them). Such a phenomenon could also manifest through women’s greater reports of disillusionment than men’s.
Limitations and Future Directions
As with virtually all studies, the present one had its limitations. One involved sampling issues, foremost the procedures involved in obtaining the cohabitation sample. As noted, an insufficient number of cohabiting couples existed in the KN panel, necessitating supplementation from other sources (including online opt-in recruitment). Furthermore, although KN uses conventional probability-sampling methods to assemble its panel and samples, decisions of prospective participants to join or not join this particular study may well have been nonrandom, as there was overrepresentation of highly educated persons and underrepresentation of minority-group members. Sample sizes for various groups of minority couples were small, preventing statistically elaborate analyses with them. Another limitation was the cross-sectional design used to examine the relationship between disillusionment and perceived relationship breakup likelihood. Because of the national study’s cross-sectional design, we assessed disillusionment as perceived change in partner and relationship qualities. Based on several studies reviewed above (e.g., Birditt et al., 2012; Huston et al., 2001; Kayser, 1993; Lavner & Bradbury, 2010), the association between disillusionment and marriage termination appears robust, regardless of whether relationship decline was assessed via actual longitudinal change or perceived change. However, it would be helpful to have both types of disillusionment measure included in the same study to verify this apparent robustness, with cohabiting couples also included. Longitudinal studies are, thus, needed that (a) test for convergent validity between the two types of disillusionment assessment (actual change over time in relationship qualities and perceived change, as with the Relationship Disillusionment Scale) and (b) examine the relationship between disillusionment and actual relationship breakup, rather than merely perceived breakup likelihood. Prospects for establishing convergent validity of the two forms of disillusionment assessment (perceived and actual change) seem promising, given research by Sprecher (1999) with related constructs. In a multiwave study with most assessment intervals separated by 1 year, Sprecher found, for a composite measure of love, commitment, and satisfaction, that reports of perceived change over the past year were indeed related to “actual change in the contemporaneous scores” between adjacent waves (p. 50). Also, more process-oriented studies are needed of how disillusionment-based uncoupling occurs in married couples, marriage-committed cohabiters, and non-marriage–committed cohabiters.
As noted above, a potentially promising way to propel the disillusionment construct toward more process-oriented studies would be to integrate it with existing conceptual models of relationship dissolution. VanderDrift et al. (2009) introduced a measure of “dissolution consideration” and found evidence showing that it mediates between dating-couple members’ low commitment and leave behaviors. Sample items on the dissolution consideration measure include “More and more it comes to my mind that I should break up with my partner” and “I have been close to telling my partner that I want to end our romantic relationship.” We propose a somewhat different sequence—that disillusionment would precede consideration of dissolving the relationship. Such a formulation situates disillusionment within research on the breakup process and provides a testable hypothesis, namely that disillusionment at one time-point should positively predict dissolution consideration at a later point.
In conclusion, this study has significantly advanced our understanding of disillusionment in cohabiting and married couples. The study’s numerous strengths (e.g., a large national sample of cohabiting and married dyads; inclusion of a newly formulated, ordinal commitment measure, which helped rule out alternative explanations) far outweigh its limitations. As a result, the present findings pave the way for further research on disillusionment, a construct seemingly capable of uniquely and independently predicting outcomes pertaining to relationship breakup likelihood. How disillusionment contributes to relationship dynamics in married versus cohabiting couples remains an important question for future researchers to address.
Footnotes
Authors’ Note
The opinions and conclusions expressed herein are solely those of the authors and should not be construed as representing the opinions or policy of any agency of the Federal government. NCFMR, in conjunction with Knowledge Networks, conducted a national survey of heterosexual married and cohabiting couples in 2010. The present authors were selected by NCFMR in a competitive process to include a measurement instrument in the survey. Kyung-Hee Lee is now at the Military Family Research Institute at Purdue University, West Lafayette, Indiana.
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 author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by the National Center for Family & Marriage Research, which is funded by a cooperative agreement (5 U01 AE000001-04) between the Assistant Secretary for Planning and Evaluation (ASPE) in the U.S. Department of Health and Human Services (HHS) and Bowling Green State University.