Tempting Fate or Inviting Happiness? Unrealistic idealization prevents the decline of marital satisfaction

Abstract

The authors examine whether unrealistically viewing a romantic partner as the image of one’s ideal partner accelerates or slows declines in marital satisfaction among newlyweds. A longitudinal study linked unrealistic idealization at the point of marriage to changes in satisfaction over the first three years of marriage. Overall, satisfaction declined markedly, consistent with past research. However, seeing a less-than-ideal partner as a reflection of one’s ideals predicted a certain level of immunity to the corrosive effects of time: People who initially idealized their partner highly experienced no declines in satisfaction. The obtained benefits of idealization remained in analyses that separately controlled for the positivity of partner perceptions and the possibility that better adjusted people might be in better relationships.

The excesses of hope must be expiated by pain; and expectations improperly indulged, must end in disappointment.

Samuel Johnson

What we need most is not so much to realize the ideal as to idealize the real.

Francis Herbert Hedge

Conventional wisdom sides with Johnson in cautioning romantic partners that being too idealistic is the road to disillusion – and even divorce (Brehm, 1988; Brickman, 1987; Huston, Caughlin, Houts, Smith, & George, 2001). However, empirical research suggests there is a case to be made for encouraging people to maintain positive, even unrealistic, perceptions and hence “idealize the real” (Fletcher & Kerr, 2010; Miller, Niehuis, & Huston, 2006; Murray, Holmes & Griffin, 1996a; 1996b; Rusbult, Van Lange, Wildschut, Yovetich, Verette, 2000). In this spirit, this paper focuses on the specific role that idealizing a partner plays in buffering against the forces that might lead to declines in marital satisfaction.

When is the Honeymoon Truly Over?

Sustaining a satisfying marital relationship is crucial for physical and psychological health (Baumeister & Leary, 1995). Yet time does not treat marriages well. Over the newlywed years, partners shift their activities from fun and leisure to chores and drudgery (Huston, McHale & Crouter, 1986). They also relax efforts to be considerate and behave less responsively (Huston et al., 2001). Conflicts increase. Satisfaction declines, often precipitously (Huston et al., 1986; Karney & Bradbury, 1997). Such declines increase the likelihood of divorce (Kurdek, 1999; 2002). Because marital instability puts physical and psychological health at risk, identifying factors that might slow declines in satisfaction is crucial (Karney & Bradbury, 1995).

This paper examines whether unrealistically viewing one’s partner as a mirror of one’s ideals sustains satisfaction over time, or alternately, sets newlyweds on the path to discord. Some existing research suggests that idealistic biases might hasten distress. For example, people who blame their partner for problems, yet naively expect their partner to behave perfectly in the future, experience steeper declines in satisfaction over the newlywed years (McNulty & Karney, 2004). Newlyweds who make charitable attributions for serious problems also experience steeper declines in satisfaction (McNulty, O’Mara & Karney, 2008). In this cautionary light, people who believe their partner mirrors their ideals might only be disappointed when time later reveals how their partner falls short of these lofty standards.

Nonetheless, the literature also points to an upside to idealization. Even though objectively harsh realities can contrast with one’s expectations, people find ways to assimilate ambiguous evidence to desired beliefs about themselves and others (see Gagne & Lydon, 2003; Kunda, 1990; Neff & Karney, 2004; Neff & Karney, 2009; Martz, Verette, Arriaga, Slovik, Cox, & Rusbult, 1998; Murray & Holmes, 1993; 1997; Taylor & Brown, 1988). In fact, research on positive illusions in relationships points to the benefits of seeing one’s partner generously. For instance, people in satisfying marital relationships see their own relationship as superior to others’ relationships (Rusbult et al., 2000). They also see virtues in their partners that are not obvious to anyone else (Murray et al., 1996b; Murray, Holmes, Dolderman & Griffin, 2000). People in stable dating relationships even redefine what qualities they want in an ideal partner to match the qualities they perceive in their own partner (Fletcher, Simpson & Thomas, 2000; Murray et al., 1996b).

In this charitable light, seeing a partner as a mirror of one’s ideal partner might function as a generous filter that affords the optimism needed to effectively cope with the challenges that come with time. For instance, as interdependence increases, partners transgress more (Kelley, 1979). People who see their partner as a better match to their ideals might perceive such behaviors as more forgivable (Arriaga, Slaughterbeck, Capezza, & Hmurovic, 2007; Miller et al., 2006; Murray et al., 1996b). Such charitable perceptions might motivate them to take more constructive remedial action (Murray, Holmes, Aloni, Derrick & Leder, 2009; Rusbult, Verette, Whitney, Slovik & Lipkus, 1991). As the rush of infatuation fades, people might also find more of their attention drawn to alternative partners (Fisher, 1998). Seeing one’s partner as a match to one’s ideals might help motivate people to derogate such temptations, further sustaining satisfaction (Johnson & Rusbult, 1989; Lydon, Menzies-Toman, Burton, & Bell, 2008). As time passes, being unduly anxious about a partner’s rejection takes a greater toll on satisfaction (Murray, Holmes & Collins, 2006). However, people who see insecure partners generously might subtly communicate such idealized perceptions, and thereby, ameliorate their partner’s rejection anxieties (Drigotas, Rusbult, Wieselquist, & Whitton, 1999; Murray et al., 1996).

This paper is the first longitudinal study to examine how the illusion of perceiving one’s partner as ideal affects the fate of new marriages. Existing research points to some long-term benefits of seeing one’s partner positively in marriage. Specifically, perceiving a spouse more generously than their behavior warrants predicts marginal increases in love (Miller et al., 2006); seeing one’s own relationship as superior to others also predicts increased commitment (Rusbult et al., 2000). However, no existing research has examined how unrealistically seeing a partner as a match to one’s ideals affects the trajectory that satisfaction takes in new marriages.

In this study, newlyweds described themselves, their partner, and their hopes for an ideal partner on 20 interpersonal qualities every six months over three years. They also reported satisfaction at each of these 7 time points. Building on prior research on the benefits of positive illusions (Miller et al., 2006; Murray et al., 1996a; 1996b), we expected unrealistic idealization to temper, not accelerate, declines in satisfaction.

Separating Unrealistic and Realistic Idealization

Models of social perception suggest that perceptions of one’s partner are shaped by one’s hopes, yet constrained by reality (Kunda, 1990). Such models imply that Ron’s perception of his partner Gayle may reflects both Gayle’s actual qualities (realism) and Ron’s hoped-for ideal qualities (idealization bias). Empirically identifying such models requires a proxy for reality because there is typically no objective way to determine what qualities Gayle actually possesses (Funder, 2003). We utilize Gayle’s self-perceptions as a benchmark for reality because self-perceptions tend to be grounded in reality (John & Robins, 1994), yet inflated on average (John & Robins, 1994; Taylor & Brown, 1988). This bias in self-perceptions makes our “reality” benchmark conservative for most people because Ron’s idealization of Gayle would only qualify as an illusion or bias if it transcends any tendency she has to see herself positively.

We conceptualize idealization as a motivated process in which Ron comes to believe that the specific qualities he hopes for in an ideal partner (e.g., someone who is warm rather than demanding and lazy rather than athletic) match the qualities Gayle actually possesses. Thus, idealization does not simply involve seeing a partner positively; rather it involves seeing a match between the particular characteristics that one’s own partner and one’s ideal partner possess. We indexed idealization through the “perceived-ideal” correlation, an intraclass correlation between the actor’s ratings of his/her partner and the actor’s ratings of his/her ideal partner across 20 qualities. More positive “perceived-ideal” correlations capture a higher degree of match between the actor’s perception of the partner and the actor’s ideals. Correspondingly, we indexed realism through the “real-ideal” correlation, an intraclass correlation between the partner’s 20 self-ratings and the actor’s ratings of the ideal partner. More positive “real-ideal” correlations capture a high degree of match between the partner’s self-perceptions and the actor’s ideal partner, indicating the actor’s more justified perception of the partner as ideal. Calculating these idiographic correlations involved treating the 20 qualities (e.g., warm, demanding) as rows and the target of perception (i.e., ideal, perceived partner, actual partner) as columns within the data set.

Our key analysis strategy identified unrealistic idealization or bias. In our analyses, we statistically isolate bias by controlling for the actual match between the actor’s ideals and the partner’s self-perceptions (i.e., the “ideal-real” correlation) when we examine the effects of the actor’s perceptions (i.e., the “perceived-ideal” correlation).

Hypotheses

We expected that actors who initially idealized those they married would start off more satisfied. But what would happen over time? We expected that actors who perceived their partner as being similar to their ideals – holding the actual fit constant – would continue to benefit. That is, we expected people who idealized their partner more initially to experience smaller declines in satisfaction over the first years of marriage. We also expected the same processes to occur when the actor was the focus of unrealistic idealization from the partner.

Our approach invites two alternative explanations for the hypothesized effects. First, the “ideal-real” correlation between Ron’s ideals and Gaye’s self-perceptions might not capture any meaningful “reality” about Gayle. If that were the case, controlling for this correlation would leave the “realism” component of the “perceived-ideal” correlation largely intact. In such a scenario, any apparent benefits of idealization might reflect the benefits of possessing a more desirable and well-adjusted partner (i.e., someone who is higher in self-esteem, less neurotic, less depressed, or more secure). Second, the “perceived-ideal” correlation between Ron’s ideals and his perception of Gayle might capture nothing more than how positively Ron sees Gayle overall. If this is the case, any apparent benefits of idealization might simply reflect the benefits of seeing one’s partner generously. We first test our basic hypotheses about the longitudinal benefits of unrealistic idealization. We then assess and rule out each of these alternative explanations.

Method

Participants

Two hundred and twenty-two childless couples in first marriages between two and six months in length participated in a 7-wave longitudinal study. The sample consisted of 193 couples providing 3 or more data points. Eleven couples separated or divorced. The sample was predominantly White (89%). At Time 1, participants averaged 27.2 (SD = 4.0) years in age and the median family income ranged from $40,000–$70,000 per year. Participants received escalating payment for each wave.

Procedure

Both members of the couple completed measures of personality and psychological well-being at Time 1. They described themselves, their partner, and their hopes for an ideal partner on the Interpersonal Qualities Scale at 7 bi-annual assessments. They also completed a satisfaction measure. Participants completed these measures in the laboratory at the initial, 12, 24, and 36-month assessments and by mail at 6, 18, and 30-month assessments.

Measures

Interpersonal qualities scale

This 20-item measure, developed from the interpersonal circle (Leary, 1957, Wiggins, 1979) tapped perceptions of an individual’s positive (i.e., kind & affectionate, self-assured, sociable/extraverted, intelligent, open & disclosing, witty & humorous, patient, rational, understanding, warm, responsive, tolerant & accepting) and negative (i.e., critical & judgmental, lazy, thoughtless, controlling & dominant, moody, distant, complaining, immature) interpersonal qualities (Murray et al., 1996a). In counter-balanced order, participants described themselves, their partner, and their own ideal or most preferred partner on these attributes (0 = not at all, 8 = completely characteristic).

Satisfaction

This 4-item measure (α = .84, Murray et al., 1996a; 1996b) tapped global evaluations of the relationship’s quality (e.g., “I am extremely satisfied with my relationship”; “I have a very strong relationship with my partner”).

Global self-esteem

The 10-item Rosenberg (1965) self-esteem scale tapped feelings of self-worth (e.g., “I feel like a person of worth, at least on an equal basis with others).

Depression

The 6-item Center for Epidemiological Studies (CES-D) scale indexed the occurrence of depressive symptoms over the prior month (e.g., “I felt depressed”).

Neuroticism

The 12-item subscale of the NEO-FFI (Costra & Macrae, 1989) indexed neuroticism and negative affectivity (e.g., “I am not a worrier”, reversed).

Attachment

The 18-item Collins and Read (1990) attachment scale tapped general feelings of anxiety (e.g., “I often wonder whether romantic partners really care about me” and avoidance in interpersonal relationships (“I know that people will be there when I need them”).

Results

Our analyses proceed in four steps. First, we computed intraclass idealization (i.e., perceived-ideal) and realism (i.e., real-ideal) correlations for each participant at each time point, transformed using Fisher’s (1921) procedure. Second, we confirm that satisfaction, on average, declined over time, and did so more for some people. Third, we show that unrealistic idealization predicted the slope of this decline. Fourth, we rule out the alternative explanations.

Does Satisfaction Differentially Decline Over Time?

To answer this question, we used MLwiN (Goldstein et al., 1998) to model our data as a three-level nested structure with time at the lowest level, person at the second level, and gender within couple at the highest level. This approach simultaneously estimates one regression equation for women and one for men, controlling for the interdependence between dyad members. It also allows tests of gender differences and pooling of coefficients in their absence.¹

We predicted satisfaction at each time point from: (1) an intercept term, (2) the linear effect of time (scored 0 to 6), a within-person fixed effect that captures the average trajectory of change in actors’ satisfaction, and (3) error terms. The effect of time (pooled across gender) was significant and negative. Satisfaction declined on average, b = −.131, SE = .011, z = 11.91, p < .001. We then allowed the linear effect of time to vary across people. Specifying time as a random effect significantly improved model fit, indicating that some people experienced steeper declines in satisfaction than others.

Does Unrealistic Idealization Slow Satisfaction’s Decline?

Next, we added main effect and interaction terms to the model to try to identify the individuals for whom satisfaction declined least over time. Table 1 lists the variables in this model, itemizing those central to our hypotheses first and controls second. It also lists the coefficients for the equations for men and women, respectively. (We repeat pooled coefficients across columns).² To address the idealization hypotheses, we added (4) actors’ Time 1 idealization index (i.e., perceived-ideal correlation), a between-person effect that captures whether actors are initially more satisfied when they idealize their partner more, (5) partners’ Time 1 idealization index, a between-person effect that captures whether actors are initially more satisfied when their partners idealize them more, and (6) the interaction between time and actors’ Time 1 idealization indices and the interaction between time and partners’ Time 1 idealization indices. These effects capture whether actors’ initial idealization of the partner or partners’ initial idealization of actors predict actors’ satisfaction trajectories, as hypothesized.

Table 1.

Predicting Declines in Satisfaction from Idealization

	Men		Women
Predictor	b (SE)	z	b (SE)	z
Quantifying Satisfaction’s Trajectory
Intercept	6.893 (.041)	–	6.893 (.041)	–
Time	−.097 (.012)	−8.08***	−.097 (.012)	−8.08***
Testing the Hypothesized Effects of Idealization
Actor’s initial idealization	.189 (.077)	2.45*	.189 (.077)	2.45*
Partner’s initial idealization	.142 (.071)	2.00*	.142 (.071)	2.00*
Actor’s initial idealization by time	.122 (.023)	5.30***	.122 (.023)	5.30***
Partner’s initial idealization by time	.055 (.021)	2.62**	.055 (.021)	2.62**
Controls to Isolate the Effect of Bias
Actor’s initial realism	−.123 (.094)	−1.31	−.123 (.094)	−1.31
Partner’s initial realism	−.161 (.107)	−1.50	.078 (.119)	0.66
Actor’s initial realism by time	−.017 (.029)	−0.59	−.017 (.029)	−0.59
Partner’s initial realism by time	.024 (.028)	0.86	.024 (.028)	0.86
Actor’s realism correlation at each measurement point	.062 (.066)	0.94	.062 (.066)	0.94
Partner’s realism correlation at each measurement point	.669 (.091)	7.35***	.135 (.090)	1.50
Controls to Remove Synchronous Associations between Idealization and Satisfaction
Actor’s idealization at each measurement point	.716 (.062)	11.55***	1.074 (.063)	17.05***
Partner’s idealization at each measurement point	.214 (.047)	4.55***	.214 (.047)	4.55**
Actor’s initial satisfaction	.786 (.039)	20.15***	.786 (.039)	20.15***
Time by actor’s initial satisfaction	−.092 (.012)	7.67***	−.092 (.012)	7.67***

^*p < .05,

^**p < .01,

^***p < .001.

We also included two sets of control variables (entered simultaneously with the central variables). To isolate unrealistic idealization, we added: (7) actors’ and partners’ Time 1 realism indices (i.e., real-ideal correlations) and their interactions with time, between-person effects that capture whether declines in satisfaction depend on the match between reality and ideals, and (8) actors’ and partners’ realism indices at each time of measurement, within-person effects that capture whether people are happier at those times when reality better matches their ideals. To remove the synchronous association between idealization and satisfaction from the trajectory, we added: (9) the effects of actors’ and partners’ idealization indices at each time of measurement, within-person effects that capture whether actors are happier (relative to their own mean) at those points when they idealize their partner more and their partner idealizes them more, and (10) actors’ initial satisfaction and its interaction with time, a between-person effect that captures whether declines in satisfaction depend on its initial level.³

As expected, the interaction between actors’ initial level of unrealistic idealization and time was significant (see Table 1). Figure 1 illustrates this interaction. It illustrates the satisfaction trajectories for actors one standard deviation above and below the mean on initial unrealistic idealization (controlling for all other variables in the model, Aiken & West, 1991). Satisfaction declined sharply for actors who initially idealized their partner less than average, b = −.169, SE = .018, z = −9.39, p < .001. However, actors who idealized their partner more than average initially did not experience any significant decline, b = −.026, SE = .018, z = −1.44.

Figure 1.

Declines in Actors’ Satisfaction as a Function of Actors’ Initial Unrealistic Idealization

The interaction between partners’ initial level of unrealistic idealization and time was also significant. Actors’ satisfaction declined more when their partner idealized them less initially (see Figure 2). The synchronous effects of unrealistic idealization on satisfaction at each time point (i.e., the within-person effects) were also significant. Actors reported relatively greater satisfaction at the times when they idealized their partner more and their partner idealized them more (relative to their own mean). The interaction between time and satisfaction revealed that satisfaction declined more when people started off happier, and thus, had further to fall.

Figure 2.

Declines in Actors’ Satisfaction as a Function of Partners’ Initial Unrealistic Idealization

A Better Partner?

The apparent benefits of actors’ unrealistic idealization in fostering continued satisfaction could simply reflect the benefits of possessing a more desirable partner if the “real-ideal” correlation fails to capture something meaningful about the partner’s reality. To assess this possibility, we correlated the partners’ self-esteem, neuroticism, depression, anxiety and avoidance with the index of partners’ degree of match to actors’ ideals (see Table 2). Partners who were a better match to actors’ ideals (i.e., higher real-ideal correlations) were higher in self-esteem, lower in neuroticism and depression, and less anxious and avoidant in attachment style. Furthermore, the benefits of unrealistic idealization could not be explained by partners’ personal adjustment. We added measures of actors’ and partners’ initial adjustment (e.g., self-esteem) and their interactions with time to the equations listed Table 1. In each analysis, the obtained interaction between actors’ unrealistic idealization and time and partners’ unrealistic idealization and time remained significant (see Table 3). Thus, actors reported more stable satisfaction the more they idealized their partner and the more their partner idealized them initially regardless of actors’ and partners’ level of self-esteem, neuroticism, depression, or security. Therefore, the benefits of unrealistic idealization cannot be explained away by the alternative that better-adjusted people are involved in being in better-functioning relationships.

Table 2.

Convergent validity correlations for the “realism” index.

	Partner’s Personal Adjustment
Partner’s Match to Actor’s Ideals	Self-esteem	Depression	Neuroticism	Anxiety	Avoidance
Female partner	.38	−.40	−.38	−.18	−.25
Male partner	.49	−.33	−.40	−.19	−.37

NB: All correlations are significant, p < .01.

Table 3.

The Effects of Unrealistic Idealization on Satisfaction Trajectories Controlling for the Actor’s and the Partner’s Personal Adjustment

	Unrealistic Idealization Coefficient Predicting Actor’s Satisfaction Trajectory
	Actor’s Idealization		Partner’s Idealization
Personal Adjustment Control Variable	b (SE)	z	b (SE)	z
Self-esteem	.121 (.023)	5.26***	.054 (.021)	2.57*
Depression	.111 (.023)	4.83***	.042 (.021)	2.00*
Neuroticism	.120 (.023)	5.22***	.053 (.022)	2.41*
Anxiety	.119 (.023)	5.17***	.049 (.021)	2.33*
Avoidance	.120 (.023)	5.22***	.052 (.021)	2.48*

^*p < .05,

^**p < .01,

^***p < .001.

Simple Positivity?

The apparent benefits of actors’ unrealistic idealization could still capture the benefits of seeing one’s partner more generously in general. To distinguish idealization – defined as the fit between the ideal and perceived partner – from general positivity, we averaged each participant’s ratings of the partner’s traits across the 20 interpersonal qualities. (We reverse-scored the negative traits.) We then added the actors’ and the partners’ general positivity indices and their interactions with time to the equations listed in Table 1. The actors’ unrealistic idealization by time interaction remained significant, b = .106, SE = .033, z = 3.21, p < .01. Actors reported less steep declines in satisfaction when they idealized their partner more at the point of marriage even controlling for how generously they perceived their partner overall. Moreover, actors’ general positivity did not interact with time in predicting actors’ satisfaction trajectory, b = .017, SE = .022, z = .77. Thus, it is the specific capacity to see one’s partner as a match to one’s ideals that sustained people’s satisfaction over time.⁴

Discussion

Seeing an imperfect partner as a match to one’s ideals predicts considerable resilience in newlywed marriages. Looking at a point in time, unrealistically idealizing one’s partner predicted greater satisfaction. That is, people reported relatively greater satisfaction at those time points when they idealized their partner more (relative to their own means). Looking over time, unrealistically idealizing one’s partner predicted continued satisfaction. Remarkably, actors who were initially high on unrealistic idealization experienced no decline in satisfaction. In contrast, actors who idealized their partner the least initially experienced precipitous declines. Moreover, the same benefits were evident when the actor was the target of the partner’s unrealistic idealization.

There is one notable limitation to these data, one that is inherent to studying how real-life relationships change over time. The data are correlational. Examining change over time provides our best means of studying causation naturalistically, but it does not offer definitive proof. Thus, we cannot rule out the possibility that a third variable might account for our effects. However, we did assess (and rule out) the two most obvious third variables.

First, actors who idealize their partners might have better partners to begin with (and controlling for the real-ideal correlation might do little to capture such an effect). This was not the case. Partners who were a better match to actors’ ideals (and, thus, scored highly on the reality correlation) were higher in self-esteem, lower in neuroticism, lower in depression, and less anxious and avoidant in attachment style, all of which are desirable qualities in a spouse (Murray et al., 2006). This suggests that the “real-ideal” correlation provides a reasonable proxy for reality. Furthermore, unrealistic idealization still retained its protective power in analyses that explicitly modeled the effects of possessing a more desirable partner on satisfaction’s decline. Second, the apparent benefits of seeing a partner as a match to one’s ideals might simply capture the benefits of seeing one’s partner more positively. This was not the case either. Actors reported less steep declines in satisfaction when they idealized their partner more initially even controlling for their tendency to see their partner more generously.

The findings are also noteworthy in several respects. To start, the protective effects of unrealistic idealization emerged despite the fact that those who were initially most ebullient generally had further to fall. People who were the most satisfied initially experienced steeper declines in satisfaction. Further analyses revealed that people who initially idealized their partner the most also experienced steeper declines in the perception that their partner met their ideals. Despite such evident risks of disappointment, initial idealization still predicted sustained satisfaction over the course of marriage. Also, the protective effect of idealization emerged using an indirect measure – the tendency to ascribe the same specific traits to one’s own partner and one’s ideal partner. Finally, the apparent benefits of unrealistic idealization emerged in analyses that controlled for the degree to which the actors actually did find the qualities he/she hoped to find in the partner (at least according to the partner). The findings thus speak to the prevalence and power of positive perceptual biases in relationships (Fletcher & Kerr, 2010).

Idealizing a partner might have protective effects because people have the power to shape their romantic fates through their behavior. Indeed, the behaviors that sustain relationships (e.g., being supportive) and the behaviors that undermine relationships (e.g., being critical) are controllable ones. Therefore, believing a partner reflects one’s hopes might predict continued satisfaction because it fosters the optimism that is needed to behave well and cope admirably with the costs and challenges that come with interdependence (Brickman, 1987; Murray & Holmes, in press). The power that idealization might have in fostering such resilience likely rests in the flexibility of the construal process itself (Griffin & Ross, 1991). Over time, people in dating relationships redefine their ideals to match the qualities they perceive in their partners (Fletcher et al., 2000; Murray et al., 1996b). The same processes may be at work here. Namely, as time and greater interdependence reveal exactly how a partner disappoints, people who flexibly adjust their ideals to match the qualities they now perceive in their partners might stay satisfied nonetheless (Kunda, 1990).

Conclusion

Rather than setting couples up for disappointment, unrealistic idealization predicted resilience against the corrosive effects of time. Idealizing one’s partner and being idealized by one’s partner predicted sustained relationship satisfaction. Rather than tempting fate, seeing one’s partner as a closer reflection of one’s ideals seems to invite happiness.