Abstract
Marital idealization is defined as an interpersonal mode of self-deception whereby husbands and wives convey an exceedingly positive portrayal of their spouse and relationship (e.g., “My spouse has never made me angry”). For the Marriage and Health Study, we obtained responses from 119 long-wed couples at baseline, 1- and 2-years later (M = 34 years married). We first computed and compared contemporaneous actor-partner interdependence models (APIMs) suggesting that marital satisfaction predicts marital idealization within and between spouses; the reverse APIM was not supported (i.e., marital idealization did not predict marital satisfaction). Yet our analyses suggest the question should be answered with longitudinal data. When reported contemporaneously, husbands’ marital satisfaction predicts marital idealization by their wives. The same cross-over effect is observed for wives—but not concomitantly, only in future. That is, marital satisfaction and idealization reported by wives predicts marital idealization reported by their husbands 2-years later.
Keywords
Across cultures, marital idealization or the propensity to idealize one's spouse and marriage is significantly associated with marital satisfaction (Gana et al., 2016; Liu & Jackson, 2019; O’Rourke & Cappeliez, 2003a). Moreover, marital idealization predicts adaptation to widowhood (O’Rourke, 2004), the psychological well-being of spousal caregivers (O’Rourke et al., 2012; O’Rourke et al.,1996), and the mental and physical health of married and cohabiting couples (de la Rubia, 2008a; Gana et al., 2016; O’Rourke & Cappeliez, 2005). Despite the adaptive value of marital idealization, conceptual uncertainty regarding this phenomenon persists.
First and foremost, is marital idealization a result of marital satisfaction or vice versa? Existing research supports both hypotheses (cf. Liu & Jackson, 2019; O’Rourke & Cappeliez, 2003a) but no research to date directly compares the two. Moreover, all existing research is cross-sectional with responses collected at one point.
Marital idealization appears unrelated to age, education, years married, number of times married, religious denomination or religious service attendance (O’Rourke & Cappeliez, 2003b). Nor does marital idealization differ between men and women (de la Rubia, 2016a, 2016b; O’Rourke, & Cappeliez, 2001). Yet it remains unclear whether the male or female partner in heterosexual relationships is the genesis or locus of marital idealization (from husband to wife vs. wife to husband). Any gender differences may emerge only over time.
For this study, we computed and compared contemporaneous actor-partner interdependence models (APIMs) in which (1) marital satisfaction predicts marital idealization within couples, versus an alternate model in which (2) marital idealization predicts marital satisfaction within and between spouses. The emerging model was extended over time to determine if associations between marital satisfaction and idealization are cross-sectional or longitudinal, and if associations are similar for wives and their husbands.
Marital Idealization
Marital idealization (aka marital aggrandizement, marital conventionalization) has been defined as an interpersonal mode of self-deception by which one conveys an exceedingly positive portrayal of his/her spouse and relationship (Karandashev, 2019; O’Rourke & Cappeliez, 2002). According to O’Rourke and Wenaus (1998), marital idealization … is a propensity to discount negative interpersonal perceptions and experience either by forgetting memories incongruent with existing schemata, or radical reframing of interpersonal experience so that memories are redefined to assume a different meaning, often antithetical, from initial perception.
Marital idealization can be understood within Brandtstädter’s (1999) theory of resilience and adaptation to loss in which personal continuity is maintained using assimilative, accommodative and immunizing coping processes. It is the latter which captures the essence of marital idealization (O’Rourke et al., 2011a, 2011b). That is, immunization serves to protect strongly-held and well-entrenched beliefs and ideals, even in the face of contrary information (Rothermund & Brandtstädter, 2003). When situations cannot be proactively managed, ego-entrenched beliefs are retained contrary to objective reality. Such positive illusions enable subjective well-being to be maintained despite significant losses (Schütz & Baumeister, 2017). The related concept of positive or marital sentiment override (Hawkins et al., 2002; Story et al., 2007) describes a similar positive bias in which the marital bond is thought to ‘override’ perceptions of negative spousal attributes.
Marital idealization was originally seen as a form of intentional deception (e.g., impression management) that inflates marital satisfaction and confounds responses to other couples measures (Cammarano, 2011; Edmonds, 1967; Visschers et al., 2017). Yet other research indicates that marital idealization is not intentional deception as much as protective self-deception, serving various adaptive functions for both married men and women (O’Rourke & Cappeliez, 2005). Marital idealization and other positive illusions are associated with the health of young and older married adults (O’Rourke, 2004; Taylor et al., 2000).
Marital Satisfaction, Idealization and Gender
Research suggests that marital idealization emerges synergistically between happily married wives and their husbands (Fowers & Applegate, 1996; Fowers et al., 1994; O’Rourke et al., 2010). More recent research suggests that spouses who idealize one another are also free of psychological distress (Gana et al., 2016). That is, low anxiety (wives) and few depressive symptoms (husbands) predict marital idealization reported by both husbands and wives (i.e., within-couple effects).
The dyadic hypothesis is supported by initial research from the Marriage and Health Study (O’Rourke et al., 2010). At baseline, participants completed both self and spousal versions of the NEO-FFI personality inventory, 2-weeks apart (e.g., “I am not a worrier” & “My spouse is not a worrier”; Costa & McCrae, 1992). In contrast to trait levels, reported trait discrepancies between spouses (i.e., higher extroversion, openness, conscientiousness, agreeableness and lower neuroticism than reported by one's spouse) predicted marital idealization by both wives and their husbands (O’Rourke et al., 2011b). For example, wives who described their husbands as relatively more agreeable than their husbands described themselves (and husbands who reported their wives are more agreeable than their wives described themselves) report higher levels of marital idealization. Additionally, wives tend to idealize their marriages when they describe their husbands as comparatively more extraverted, and when their husbands describe their wives as comparatively more conscientious. And both spouses idealize their marriages when husbands described their wives as relatively lower in neuroticism than they described themselves.
These findings emerged independent of the individual trait levels for either spouse. In fact, individual trait levels are comparatively insignificant predictors of marital idealization compared to relative reporting differences within couples. For each of the Big Five personality traits, both perceiving and being perceived more positively than one sees him/herself, significantly predicts marital idealization for both husbands and wives (O’Rourke et al., 2011b). However, this study did not examine marital satisfaction in relation to marital idealization.
The question remains, does marital satisfaction predict marital idealization or vice versa, and does this differ between husbands and wives over time? Actor-partner interdependence modeling (APIM; Kenny et al., 2006) allows us to simultaneously examine the relative contribution of within- and between-spouse factors in relation to dyadic constructs. With APIM, cross-lagged paths indicate partner effects over-and-above within-spouse or actor effects simultaneously for both husbands and wives. APIMs computed for this study allow us to identify the relative strength and direction of crossover effects between spouses over time.
Research Questions
This study is composed of cross-sectional and longitudinal components. We first computed and compared separate contemporaneous APIMs in which: (1) marital satisfaction reported by both wives and their husbands predicts marital idealization reported by both spouses; and the reverse, (2) marital idealization reported by wives and their husbands predicts marital satisfaction reported by both. Path coefficients and goodness of fit statistics are compared between APIMs as these are nested models. See Figure 1.

Competing actor-partner interdependence model, marital satisfaction and marital idealization.
We extend the resulting model over 2-years in order to identify both cross-sectional and longitudinal effects between marital satisfaction and marital idealization. This allowed us to determine if associations between marital satisfaction and marital idealization are similar between long-wed husbands and wives over time.
Methods
Participants
We recruited 125 wives and their husbands 50+ years of age, married 20+ years for the Marriage and Health Study (King et al., 2018). We purposefully recruited long-wed couples, as idealization within dating (Wu et al., 2020) and newlywed couples (Murray et al., 2011) may be wishful thinking, not self-deception in accord with theory (O’Rourke & Cappeliez, 2003a).
Six couples provided data at T1 only (4.8%), while the remaining 119 couples also provided responses 1- and 2-years later. The six who dropped out did not differ significantly in number of years married, t(124) = −.33, p = .74; nor did men/husbands or women/wives differ in educational attainment, t(122) = 1.47, p = .15 & t(122) = .90, p = .37, respectively. In other words, those lost to attrition did not differ from those who provided the longitudinal data examined for this study.
Procedures and Measures
At recruitment (T1), participants were mailed a package including a consent form, self-report measures of marital idealization, marital satisfaction, and a demographic questionnaire (i.e., age, sex, years married, number of previous marriages, ethnicity, education, employment status, work history). Questionnaires were mailed separately to husbands and wives along with separate postage-paid return envelopes; participants were asked to complete questionnaires independently (Claxton et al., 2012). This procedure was repeated two years thereafter. The Marriage and Health Study was undertaken with the approval of the Research Ethics Board at Simon Fraser University, Burnaby (BC), Canada.
Marital Aggrandizement Scale
The Marital Aggrandizement Scale (MAS; O’Rourke & Cappeliez, 2002) was developed as a measure of idealization of one's spouse and marriage. MAS items were written in extreme terms such that they cannot be endorsed without conveying an inordinately positive depiction of marriage (e.g., “If my spouse has any faults, I am not aware of them”; “My spouse has never made me angry”). Respondents indicate the extent to which they agree with each of 18 statements along a 7-point Likert-type scale. The MAS was first developed in English and French (O’Rourke & Cappeliez, 2003b) and has since been translated into Spanish, Mandarin, Urdu, Farsi and Hebrew (e.g., Arshad & Fatima, 2014).
Across cultures, internal consistency of MAS responses ranges between.79 < α < .88 (de la Rubia, 2017; Liu & Jackson, 2019). Responses are statistically indistinguishable between older men (M = 67.83, SD = 16.72) and women (M = 68.83, SD = 16.83; O’Rourke et al., 2010). Nor do age, sex, education, years married, number of times married, religion, or religious service attendance appear related to MAS response levels (de la Rubia, 2016a, 2016b, 2017; O’Rourke & Cappeliez, 2002).
Dyadic Adjustment Scale
Marital satisfaction was measured using the 32-item Dyadic Adjustment Scale (DAS; Spanier, 1976). The DAS is a well-known index of marital satisfaction (e.g., “Do you confide in your spouse?”; “How often do you and your spouse quarrel?”). Responses are recorded along a series of 5- and 6-point Likert-type scales, two yes/no questions, and one final question to which persons are asked to endorse the statement which best reflects their expectation of relationship continuity (six response alternatives provided). Total scores 100–119 suggest marital stability and overall satisfaction (Stuart, 1992) whereas DAS scale responses 120+ suggest inordinate self-sacrifice or relationship idealization (Kazak et al., 1988). The DAS was administered at baseline only.
Analytic Procedures
APIMs for this study were computed using AMOS 26.0 and maximum likelihood parameter estimation (Byrne, 2010). Simultaneous estimation of actor and partner effects allowed us to identify crossover effects between spouses while correcting or covarying for within-actor consistency over time. Such multivariate analyses reduce the likelihood of capitalization on chance. With 119 couples and six variables, path models had sufficient statistical power to detect medium to large effect sizes (d = .80, α = .05; O’Rourke & Hatcher, 2013).
In accord with convention, we report three goodness-of-fit-indices: an incremental, an absolute, and a parsimonious fit index. The Comparative Fit Index (CFI) is an incremental index that represents the extent to which the hypothesized model is a better fit to the data than the null model. The Standardized Root Mean Square Residual (SRMR) is an absolute index that represents the standardized difference between observed and predicted correlations within the hypothesized model. Finally, the Root Mean Square Error of Approximation (RMSEA) is a parsimony index that represents the extent to which the hypothesized model fits the data relative to the general population. Values greater than .94 for the CFI, and values less than .055 for the SRMR and the RMSEA, indicate good fit between models and data (O’Rourke & Hatcher, 2013).
Results
As reported above, 119 couples provided responses at baseline and 2-years later. They had been married 33.8 years on average (SD = 7.9, range 20–56 years); 17% of wives and 19% of husbands reported at least one previous marriage. We neglected to ask participants to report numbers of children and grandchildren.
Consistent with generational norms (Bair, 2007), husbands (M = 60.0 years; SD = 6.6) were slightly older than their wives (M = 58.5 years, SD = 6.4). On average, they had completed 14.8 years of formal education (SD = 4.7). Currently or prior to retirement, most participants reported working in clerical/administrative (44.8%) or professional/ managerial positions (37.2%). Close to 40% had retired from the paid workforce. See Table 1.
Descriptive Statistics and Internal Consistency (Cronbach's α) for Marital Satisfaction and Marital Idealization Over 2 Years.
Note. Marital Satisfaction = Dyadic Adjustment Scale, Marital Idealization = Marital Aggrandizement Scale.
Table 2 reports correlation coefficients between marital satisfaction and marital idealization at baseline and Time 2, between husbands and wives. Like marital satisfaction (r = .47, p < .01), marital idealization is moderately but not strongly correlated between spouses (r = .38, p < .01) yet this appears to increase for marital idealization at Time 2 (r = .53, p < .01). See Table 2.
Correlation Coefficients Between Marital Satisfaction and Marital Idealization Between Spouses.
Note. SAT = marital satisfaction (Dyadic Adjustment Scale), IDL = marital idealization (Marital Aggrandizement Scale). H = husbands, W = wives. p < .01 for all coefficients.
Cross-sectional APIM
As described above, we first computed and compared cross-sectional APIMs in which: (1) marital satisfaction predicts marital idealization within and between spouses; versus (2) marital idealization predicts marital satisfaction within and between spouses. Goodness of fit analyses support the former whereas the latter model was not viable, χ2 (df = 1) = 11.43, p < .01. More precisely, goodness of fit statistics are each below acceptable thresholds (Byrne, 2010) for the model in which marital idealization predicted marital satisfaction as measured by the Comparative Fit Index (i.e., CFI = .91), Standardized Root Mean Residual (SRMR = .064), and the Root Mean Square Estimate of Association (RMSEA = .297).
However, the reverse model in which marital satisfaction predicts marital idealization exhibits good model fit, χ2 (df = 1) = 2.32, p = .13. Fit statistics are in ideal parameters for both the CFI (.99) and SRMR (.030), but not the RMSEA (.100). Each of the three is superior to the first model. The chi-square statistic is also significantly smaller for the model in which marital satisfaction predicts marital idealization, Δχ2 (Δdf = 0) = 9.11, p < .01 (i.e., nested model comparison).
In this baseline model, marital satisfaction is moderately correlated between husbands and wives. Marital satisfaction reported by husbands predicts their marital satisfaction; the same is true for wives. And marital satisfaction reported by husbands predicts marital idealization reported by their wives. The reverse association did not emerge, however. At baseline, marital satisfaction reported by wives does not predict marital idealization reported by their husbands. See Figure 2.

Baseline actor-partner interdependence model, marital satisfaction predicts idealization. Note. Parameters expressed as maximum likelihood estimates (standardized solution). Parenthetical numbers indicate significance levels (CR > 1.96, p < .05; CR > 2.58, p < .01). Dotted line indicates predicted but statistically insignificant path.
Longitudinal Analyses
In accord with previous research (O’Rourke & Cappeliez, 2003a), Study One findings suggest that marital satisfaction predicts marital idealization within- and between spouses, but not vice versa (Gana et al., 2016; Liu & Jackson, 2019). This finding allows us to extend the APIM model over time. Within long-wed couples, we investigated whether marital satisfaction predicted marital idealization at baseline and two years later. See Figure 3.

Longitudinal actor-partner interdependence model predicting marital idealization. Note. Parameters expressed as maximum likelihood estimates (standardized solution). Parenthetical numbers indicate significance levels (CR > 1.96, p < .05; CR > 2.58, p < .01). Thin arrows indicate cross-sectional associations; thick arrows indicate longitudinal effects.
This longitudinal APIM also emerged as viable, χ2 (df = 6) = 8.50, p = .20. The CFI (.99) and SRMR (.039) are ideal. And for this model, the RMSEA (.040) is also in ideal parameters. We again see that husbands’ marital satisfaction predicts marital idealization reported by their wives at baseline, but not vice versa. Instead, marital idealization for both husbands and wives is predicted by their spouses’ marital idealization reported two years earlier. And marital idealization for both husbands and wives is predicted by marital idealization reported by their spouses 2-years earlier. Goodness of fit indices for this longitudinal APIM are superior to both cross-sectional models.
Note also that marital idealization reported by husbands is predicted by marital satisfaction reported earlier by their wives. This finding is notable as we found no such cross-sectional effect. Marital satisfaction reported by wives predicts marital idealization reported by their husbands in future but not contemporaneously. This crossover effect from wives to their husbands is only longitudinal. This finding is noteworthy as it suggests that gender differences in marital idealization emerge over time.
Discussion
As noted by O’Rourke and Cappeliez (2003a), marital satisfaction can exist without idealization but marital idealization does not exist without satisfaction. This is in accord with our finding that, when measured contemporaneously, marital satisfaction predicts idealization; the reverse model was not supported. This result would seem to be at odds with other recent APIM studies. With couples in France, Gana et al. (2016) report that marital idealization predicts marital satisfaction, but marital idealization is also (inversely) predicted by avoidant attachment style. In Mainland China, Liu and Jackson (2019) indicate that marital idealization predicts marital satisfaction, and marital idealization is also predicted by (the absence of) psychological stress. That is, these APIM studies do not examine just marital idealization and marital satisfaction, but in association with psychological variables. In multivariate models, it is problematic to examine specific elements in isolation when models are not nested (e.g., include unrelated variables). By directly comparing alternate models, our results support the reverse association.
However, longitudinal findings suggest that cross-sectional analyses provide an incomplete picture. Marital satisfaction reported by husbands contemporaneously predicts both his and his wife's propensity to idealize him and their marriage. Yet this within-couple, contemporaneous effect of marital satisfaction from wives to their husbands is observed in neither cross-sectional nor longitudinal models; marital satisfaction reported by wives only later predicts marital idealization by their husbands. That is, cross-sectional analyses provide an partial picture of associations between marital satisfaction and marital idealization.
Limitations and Future Research
A growing body of research suggests that marital idealization exists across cultures, with research conducted with couples in Canada, Pakistan, Mexico, France, Israel and China (e.g., Arshad & Fatima, 2014; de la Rubia, 2016a, 2016b). And several APIM studies have now been conducted examining marital idealization in distinct cultures (Gana et al., 2016; Liu & Jackson, 2019; O’Rourke & Cappeliez, 2003a). We can conclude marital idealization is not only an Anglo-Saxon phenomenon; that is, the propensity to idealize one's spouse and marriage exists across continents, cultures, languages, and religions. Yet generational and demographic differences between studies limit our ability to draw broad conclusions as study samples include cohabiting or unwed couples as well as young and long-wed couples (de la Rubia, 2016a, 2016b; Gana et al., 2016; Liu & Jackson, 2019; O’Rourke & Cappeliez, 2003a). Longitudinal research should be conducted comparing levels and determinants of marital idealization across cultures (e.g., aboriginal, First Nations). Ideally this research will include same-sex couples and polyamorous relationships.
Couples recruited for the Marriage and Health study were recruited from one Canadian city and almost all participants were Caucasian. Though sample attrition was low, ethnic homogeneity limits our ability to generalize findings widely. Research is needed measuring marital satisfaction and idealization before significant health diagnoses, and measured at multiple points before and after death. This would allow us to determine if marital idealization emerges or increases in response changes in health and life losses over time.
Summary
Does marital satisfaction predict marital idealization or vice versa? APIM research with cross-sectional data appears to support the latter (Gana et al., 2016; Liu & Jackson, 2019). Yet the results of this study suggest the question should be answered with longitudinal data. When reported at the same time, husbands’ marital satisfaction predicts marital idealization by their wives. The same cross-over effect is observed for wives, but not concomitantly, instead over time. That is, marital satisfaction and idealization reported by wives predicts marital idealization reported by their husbands 2-years later. Future research examining both marital satisfaction and marital idealization over multiple points is needed over extended periods.
Footnotes
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 work was supported by the Social Sciences and Humanities Research Council of Canada (grant number 410-2005-2328, 127915, 136727).
