Direct and Indirect Transmission of Relationship Functioning Across Generations

Abstract

Relationship functioning is assumed to propagate across subsequent generations, but most studies have lacked appropriate methodologies to test this assumption prospectively. In a randomly selected sample of youth (N = 821) followed prospectively for over 25 years across multiple generations, we examined the association of romantic engagement (i.e. emotional involvement and closeness) between parents with offspring romantic relationship quality. We tested two developmental pathways linking parents’ romantic engagement with offspring adult romantic relationship quality, the first operating via parenting practices, and the second operating via adolescent depression. Parents’ romantic engagement predicted offspring romantic relationship quality a mean of 17 years later, net age and socioeconomic status. Results supported a developmental pathway from parents’ romantic engagement at offspring mean age 14, to parenting at offspring mean age 16, to offspring socio-emotional functioning at mean age 22, and offspring romantic relationship quality at mean age 33. However, the influence of parents’ romantic engagement on offspring’s adult romantic relationship quality does not appear to operate via a pathway of adolescent depression. Implications for prevention are discussed.

There is an implicit assumption among family scholars that the effects of poor marital relationships and divorce are propagated across subsequent generations as they age (Wallerstein, Lewis, & Blakeslee, 2000; Whitton et al., 2008). In fact, empirical research supports the association of childhood experiences of marital discord and separation on problems in romantic relationship development and psychological well-being (Hayashi & Strickland, 1998; Kwong, Bartholomew, Henderson, & Trinke, 2003; Story, Karney, Lawrence, & Bradbury, 2004).

On the other hand, evidence from longitudinal studies suggests that marital withdrawal may be just as deleterious to emotional development as is open conflict (Sturge-Apple, Davies, & Cummings, 2006). Some have suggested that marital withdrawal is a representation of overall family disengagement, and may generalize to a lack of emotional availability for affective parent-child relationships (Cox, Paley, & Harter, 2001) and for consistent childrearing practices (Crockenberg & Covey, 1992). As such, children may experience marital withdrawal as a threat to their security (Cox, Paley, Payne, & Burchinal, 1999). Further, given the clear association of maternal depression with social withdrawal (Pettit & Joiner, 2006), a focus on the intergenerational influence of emotional disengagement in marriage has strong theoretical importance. Yet, research on the influence of withdrawal and disengagement in marriage is sparser than that on the impact of conflict.

Several theoretical models suggest that marital withdrawal would have important implications for children’s interpersonal development. First, parents’ marital disengagement may either signal or contribute to parents’ overall emotional unavailability and social withdrawal (Cox et al., 1999). Adolescents with emotionally disengaged parents may be forced to navigate their formative romantic relationships without parental support. Interpersonal stress models of depression and relational functioning suggest that adolescent depression may compromise the ability to develop adequate coping skills in peer and romantic relations (Davila et al., 2009). Stress generation models propose a bidirectional association of depression and romantic discord, suggesting that depressed individuals behave in ways that lead to interpersonal stress, and this stress in turn contributes to the maintenance or worsening course of depressive symptoms (Davila, Hammen, Burge, Paley, & Daley, 1995; Hammen, 2009). In this model, marital withdrawal may influence offspring relationship functioning by increasing the risk for adolescent depression. Social learning theory suggests that children learn patterns of social interaction via parental modeling, including both hostile and avoidant emotional regulation strategies in interpersonal relationships (Kim, Pears, Capaldi, & Owen, 2009). Finally, attachment theories propose that warm, responsive parenting practices lead to secure attachment in offspring, and these patterns of attachment predict subsequent warm, responsive interpersonal relationships with others in adulthood, whereas the absence of such practices predicts poor relational functioning (Hazan & Shaver, 1987). Avoidant attachment may predict later avoidance in intimate relationships.

This study aims to directly address the influence of parents’ withdrawal/disengagement on offspring functioning, in view of the limited research on this marital process. We review presently four facets of the family environment with theoretically relevant mechanisms for the transmission of romantic disengagement and withdrawal across generations - parental depression, parenting practices, offspring depression, and socio-emotional functioning.

Parental Depression

Substantial evidence documents the influence of maternal depression on child development, including poor self regulation, insecure attachment, low social competence, and behavioral problems (Cummings & Davies, 1994; Gotlib & Goodman, 1999). Later research has turned its focus to the mechanisms by which such risk is conferred, including heritability, neuroregulatory deficits, exposure to problematic maternal cognition and behavior, and stress, including marital discord (Goodman & Gotlib, 1999). Parental depression prospectively increases the risk in particular in girls for depression, interpersonal incompetence with peers and romantic partners, and early childbearing (Hammen, Brennan, & Le Brocque, 2011). Moreover, marital disengagement may be expected to differentially influence mood symptoms in adolescent girls, as a function of their greater sensitivity to interpersonal stress (Rudolph, 2002).

Parenting Practices

Parenting styles are often considered the “vehicle” of socialization across childhood and adolescence (Gray & Steinberg, 1999), and thus may account for intergenerational transfer of intimate functioning. Parents with harmonious marriages tend to engage in positive parenting styles characterized by warmth, affection, involvement, structure, and empathy (Barber, Bolitho, & Bertrand, 2001; Gray & Steinberg, 1999). On the other hand, research consistently demonstrates that stressful family conditions, particularly marital turmoil, affect children via the mediating corrosion of parental disciplinary responses to child misbehavior and positive encouragement for prosocial behavior (Erel & Burman, 1995; Fauber, Forehand, Thomas, & Wierson, 1990; Patterson, Reid, & Dishion, 1992).

Parenting may be affected by both parental depression and marital disengagement. Depressed mothers are less consistent, less warm, and more hostile in their parenting interactions (Lovejoy, Graczyk, O’Hare, & Neuman, 2000). In a longitudinal study of a community sample, marital withdrawal was a consistent predictor of less sensitive, responsive parenting, net the effects of parental education, parental depression, and child negativity (Cox et al., 1999).

Offspring Depression

The development of depression merits special consideration in studies of intergenerational transmission of romantic relationship quality. Depression is associated with interpersonal avoidance that contributes to the maintenance of depressive symptoms and affects the development of adequate social networks (Pettit & Joiner, 2006). In turn, social withdrawal may impact functioning in interpersonal relationships (Allen et al., 2006). Specifically, depressed adolescents are at risk for compromised romantic relationship functioning and early sexual activity (Davila, 2008), early marriage and marital dissatisfaction (Gotlib, Lewinsohn, & Seeley, 1998). Recent evidence further suggests that depression may exacerbate the negative influence of adverse family environment to increase the risk for poor romantic outcomes (Davila et al., 2009).

A profusion of research has concluded that children who live in homes with elevated marital conflict are consistently at risk for numerous behavioral and emotional problems including internalizing and externalizing symptoms (e.g., Katz & Gottman, 1993; Sternberg, Lamb, Guterman, & Abbott, 2006). Few studies have linked marital disengagement/withdrawal with subsequent offspring psychopathology. One exception found evidence that interparental withdrawal was associated with offspring symptomatology including internalizing and externalizing behaviors, as well as school adjustment, independent of the effects of marital hostility (Sturge-Apple et al., 2006). From the point of view of security, parents’ romantic engagement with one another may reflect an overall commitment to the relationship that may be protective and may have long-term implications for engagement in offspring’s peer and romantic relationships. That is, offspring may in turn have difficulty connecting deeply with others.

Socio-Emotional Functioning

There is ample evidence that marital conflict predicts poor social adjustment in children, including poorer quality of relationships with peers (Eisenberg & Fabes, 1992; Lindsey, Colwell, Frabutt, & MacKinnon-Lewis, 2006; Schulz, Waldinger, Hauser, & Allen, 2005). Some have proposed that marital conflict may interfere with the process of parent-child emotional reciprocity, wherein the child’s perception that the parent accurately responds to his or her emotional signals results in a sense of trust in the caregiver and generalizes that trust to other relationships (Lindsey, Caldera, & Tankersley, 2009). In contrast, positive parenting styles promote in the offspring a healthy capacity for relation to others and intimacy (Gray & Steinberg, 1999), and social competence (Baumrind, 1991; Collins, Maccoby, Steinberg, Hetherington, & Bornstein, 2000). As children learn patterns of emotional intimacy from their parents, a lack of marital engagement between parents may also reinforce patterns of low engagement with peers and dating partners.

Marital disengagement may lead to poor offspring socio-emotional functioning via its effects on depression as well. Research suggests that depression may directly impact interpersonal problems such as rejection by peers, lack of intimacy and support in friendships, and lack of warmth and support in the family (Rudolph, Flynn, & Abaied, 2008). In particular, because of its relation to social withdrawal (Goodyer & Cooper, 1993), adolescent depression may set in motion a pattern of disengagement in romantic relationships.

Current Study and Hypotheses

Taken together, prior findings suggest that parents’ romantic disengagement/withdrawal may have long term influence on romantic functioning in offspring. We tested these relationships in a community sample of youth and their parents followed prospectively for over 25 years. We expected parents’ romantic engagement to predict offspring romantic relationship quality by middle adulthood (M age 33). Literature linking parents’ romantic engagement with parent-child processes, depression, and social adjustment suggested two developmental pathways by which intergenerational transmission might develop. In the first pathway, emotional security theory lead us to hypothesize that low romantic engagement between parents would predict less close, less consistent and harsher parenting in middle adolescence, in turn leading to poor socio-emotional functioning in emerging adulthood, and reduced romantic relationship quality in adulthood. In the second, interpersonal stress theory lead us to hypothesize that low romantic engagement between parents would predict depression in adolescence (net the influence of maternal depression), which in turn would set in motion a trajectory of impaired socio-emotional functioning by emerging adulthood, ultimately reducing offspring’s quality of romantic relationships in adulthood.

Method

Participants and Procedure

The Children in the Community (CIC) cohort derives from a randomly selected sample residing in two upstate New York counties in 1975 (Kogan, Smith & Jenkins, 1977). The area sampled for this study was selected to be generally representative of the United States on socioeconomic status and the majority of demographic variables, but reflected the sampled region with regard to high proportions of Catholic (54%) and Caucasian (91%) participants.

Maternal interviews were conducted on a range of health, behavioral and environmental factors. Of the original 975, 767 youth and their mothers were retained at subsequent follow-up, and 54 new families were recruited from urban poverty areas to replace families lost to urban renewal. This sample as constituted in 1983 (N = 821) was demographically representative of the sampled regions, and retention of study members has been 95% since that time. The youths and their mothers were assessed in four follow-up interviews (1983, 1985-86, 1991-93, and 2001-2004) regarding demographic, psychiatric, and other psychosocial factors. Where a family had more than one eligible child, one child was randomly selected. Interviews were conducted in the home by intensively trained and supervised lay personnel. In the most recent assessment, interviews carried out in a computerized questionnaire by visiting layperson interviewers, web-based responses, and telephone interviews, were followed by two highly trained and experienced clinicians who evaluated responses to follow-up questions to validate diagnoses. Additionally, research with this sample has shown that diagnoses made by laypersons and trained clinicians have moderate yet significant concordance over a time span of 3-4 months (Cohen, O’Connor, Lewis, & Malachowski, 1987). Detailed descriptions of the sample characteristics, procedures, psychometric support for using lay interviewers, and follow-up are available in earlier reports (Cohen & Cohen, 1996). Study procedures conformed to institutional guidelines.

In 1983 (Wave 2) the mean age of offspring was 13.8 years (SD = 2.6, range = 9 – 19). In 1985-1986 (Wave 3), the mean age was 16.2 (SD = 2.8, range = 11 – 22). In 1991-1993 (Wave 4), the mean age was 22.1 (SD = 2.7, range = 17 – 28), and in 2001-2004 (Wave 6), the mean age was 33.2 (SD = 2.90, range = 28 – 40).

Measures

Parent socioeconomic status (M age 16)

Parental socioeconomic status (SES) was assessed in Wave 1 (1975), Wave 2 (1983), and Wave 3 (1985-86), as a standardized sum of: (a) maternal and paternal years of education, (b) maternal and paternal occupational status, and (c) family income category (Hollingshead & Reidlich, 1958). In the present study, we employed the Wave 3 score (1985-86), when offspring were a mean age of 16 years old, as it had more complete SES data.

Parent romantic engagement (M age 14)

Mothers were asked questions concerning their relationship with a spouse or romantic partner during the Wave 2 (1983) assessment when offspring were a mean age of 14. Single mothers (18% of the sample) were also included in the study if they were in a current relationship. The Romantic Engagement Scale (α = .91) consists of 6 items measuring affective relations between spouses. Items include “We engage in outside interests together,” “We are very affectionate with each other,” “We comfort and help each other when we have troubles,” “We can talk with each other about everything,” “How much do you admire the child’s father in his role as a companion to you?” and “How much do you admire the child’s father in his role as someone who helps you out?.” Low scores on the Romantic Engagement Scale equate to low romantic engagement and vice versa. Of mothers who scored 1 SD lower than the mean (n = 88, 15%), less than 50% (n = 42) reported high conflict (Argue Scale, Cronbach’s alpha = .54) concurrently in their relationship, suggesting that romantic engagement is measuring a separate set of behaviors from conflict.

Parent depression symptoms (M ages 14 and 16)

Mothers completed a 9-item index of depression based on the Symptom Checklist-90 Depression Scale (Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974) to assess severity of somatic (e.g., low energy, appetite, and sleep problems), cognitive (e.g., difficulty concentrating), and melancholic (e.g., sadness, loss of interest in things) symptoms of depression during the first three assessments (Cronbach’s α = .85, .87, and .88, respectively). Participants rated items on a 5-point scale, from “not at all bothered by in the last year” to “extremely bothered by.” Our prior work indicates the scale is associated with marital discord, parenting, and offspring reports of internalizing symptoms (Kasen, Cohen, Chen, & Castille, 2003). As the measure does not require a definable episode of depression, it more closely assesses the chronic depressive affect found in dysthymia (Cohen et al., 2008; Kasen et al., 2003).

Offspring depression symptoms (M ages 14, 16, and 33)

The parent and youth versions of the Diagnostic Interview Schedule for Children (Costello, Edelbrock, Duncan, & Kalas, 1984) were administered in all waves for a variety of Axis I and II disorders. We combined mother and youth reports, such that symptoms were considered present if endorsed by either the parent or the child. To improve on the specificity of the resulting diagnoses, we created a scale for each syndrome based on all the relevant items, including associated impairment. This approach generates better construct validity and prevalences consistent with clinical practice (Cohen, Velez, Kohn, Schwab-Stone, & Johnson, 1987). Here, we used scaled scores for Wave 2 (M age 14), Wave 3 (M age 16), and Wave 6 (M age 33). Internal consistency reliability was .86 in earlier waves and increased to .91 by Wave 6 (Crawford, Cohen, Midlarsky, & Brook, 2001).

Parenting practices (M age 16)

Parenting practices were assessed by a combination of youth and parent report at Wave 3 (1985-86), when offspring were a mean age of 16. Participants who reported having at least one child were asked 33 items assessing parental behaviors, that when high (e.g., inconsistent parental discipline) or when low (e.g., parental affection), have been found to be indicative of problematic parenting (Johnson, Cohen, Kasen, Smailes, & Brook, 2001). These items were adapted from the Disorganizing Poverty Interview and other validated measures of child-rearing behavior (Avgar, Bronfenbrenner, & Henderson, 1977; Schaefer, 1965) which are reliable and valid (Cohen & Cohen, 1996; Johnson et al., 2001; Kogan et al., 1977), and predict offspring risk for psychiatric symptoms and disorders (Johnson et al., 2001; Johnson, Cohen, Chen, Kasen, & Brook, 2006). For the present study, we employed scaled scores of inconsistency, closeness (reverse coded), and harsh punishment.

Offspring socio-emotional functioning (M age 22)

The construct of socio-emotional functioning was indexed by three separate measures assessing different aspects of one’s overall social functioning during the Wave 4 youth assessment (M age 22). Items in these three scales may assess social functioning within family, peer, and romantic relationships, but given the lack of specificity in the wording of questions, these scales represent overall social functioning. The Trust/Suspiciousness Scale (α = .66) was indexed by 5 items on suspiciousness of others, social alienation, and lack of trust in others. This scale assesses the person’s ability to trust and engage with other people within their social sphere. Sample items include “I often wonder if the people I know can really be trusted,” and “I often go out of my way to protect myself from others.” The Interpersonal Difficulty Scale (α = .79) was indexed by 6 items asking the respondent how much they were bothered by anger and aggression, and their ability to interact with others. This scale assesses the person’s ability to emotionally regulate within social contexts, and to connect with and feel close to others in a general sense. Example items include “Feeling anxious, nervous, or tense inside,” and “Difficulty in feeling close to other people.” The Social Support Scale (α = .56) was indexed by 13 items covering number and duration of friendships, presence of a confidante, the availability of help, and closeness with others. This scale specifically addresses social support received from friends and relatives. Sample items include “How often have you felt lonely or cut off from people in the last month?” and “In general, do you have opportunities to meet new people that you might enjoy as friends.” We reverse coded the Trust/Suspiciousness and Interpersonal Difficulty scales to match the direction of the Social Support Scale.

Offspring romantic relationship quality (M age 33)

Offspring romantic relationship quality was assessed by self-report at Wave 6 (2001-2004) when the offspring were a mean age of 33. The Relationship Quality Scale (α = .89) consists of 9 questions designed to measure various aspects of personal relationships including affection, communication, and loyalty with the person whom the respondent was closest (e.g., spouse or best friend). Questions were asked on a Likert scale from 1 (never) to 5 (always), including items such as “We engage in outside interests together,” “We really talk with one another,” and “We give each other information, guidance, and suggestions.” Lower scores on this scale are indicative of low romantic relationship quality.

As the current study was designed to measure romantic relationship functioning specifically, only participants who described a relationship between them and a spouse or romantic partner were included in the study. Of the 821 participants, 581 (70.8%) had reported on their romantic relationship (66.8% spouse; 33.2% romantic partner) and were therefore included in the study. Fifty-two percent of the subsample was female, and 92.6% were Caucasian. Participants who did not report on a romantic relationship (n = 95) did not differ from those who reported about a romantic relationship in terms of age, sex, race, or socioeconomic status.

Data Analysis

Variables were initially analyzed for normality and transformed if necessary. Missing data was handled using the Expectation Maximization (EM) method in the SPSS Missing Values module to impute missing data, resulting in a complete dataset (N = 581). Expectation maximization is an iterative procedure comprising two steps, the first computing the expected value of the complete data log likelihood, while the second step substitutes the expected values for the missing data obtained in the prior step to obtain new parameter estimates (Dempster, Laird, & Rubin, 1977). The EM method performs better than listwise and pairwise deletion, and stochastic regression imputation in terms of parameter estimates and standard errors, and performs similarly to multiple imputation in some cases (Newman, 2003).

The relationship between parents’ romantic engagement and offspring romantic relationship quality was tested via hierarchical linear regression, controlling for age, sex, and socioeconomic status. The developmental pathways were tested using the structural equation modeling (SEM) program AMOS (Version 16; Arbuckle, 2007). To test the significance of the total indirect (i.e., mediating) effect, we used Preacher and Hayes’ (2008) bootstrapping method to obtain 2,000 random samples to derive 95% confidence intervals and estimates for the p values.

Model fit was assessed with several commonly used indices in addition to the traditional chi-square test of model fit, which is fairly sensitive to sample size and nonnormality (Bentler & Bonett, 1980). We reported the Normed Fit Index (NFI; Bentler & Bonett, 1980), the Tucker-Lewis Index (TLI; Tucker & Lewis, 1973) and the Root Mean Square Error of Approximation (RMSEA; Browne & Cudeck, 1993). Generally speaking, for the TLI and CFI, values of .90 to .95 indicate acceptable fit while values greater than .95 indicate good fit (Bentler & Bonett, 1980; Hu & Bentler, 1999), and RMSEA values of less than .05 are indicative of good fit, while values between .05 and .08 are considered acceptable (Browne & Cudeck, 1993).

Standardized variables were utilized in linear regression models, while non-standardized variables were utilized in the SEM analysis. In path diagrams, standardized estimates and multiple squared correlations are reported. Additionally, although each wave of data collection corresponds to a mean age of the offspring, for simplicity, we will refer to these as age 14, age 16, etc. in the remainder of the manuscript.

Results

Correlations

Table 1 depicts the correlations among study variables. Parents’ romantic engagement was significantly associated with offspring depression, parent depression, all parenting variables, interpersonal difficulty, and offspring romantic relationship quality. Punishment was significantly associated with trust/suspiciousness; closeness was significantly associated with social support, trust/suspiciousness, and interpersonal difficulty, and inconsistency was significantly associated with trust/suspiciousness and interpersonal difficulty. Parent depression at offspring age 14 was significantly associated with all parenting variables, offspring depression, and interpersonal difficulty. Offspring depression at age 16 was significantly associated with all parenting variables, trust/suspiciousness, and interpersonal difficulty, but not with social support. Offspring romantic relationship quality was significantly associated with all variables, except punishment at age 16. Although most correlations between variables were significant, the effect sizes were in the small to medium range.

Table 1.

Partial Correlations^a among Study Variables (N = 581)

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13
1.G1 Romantic Engagement (14)	--
2. G1 Depression (14)	-.37**	--
3. G1 Depression (16)	-.25**	.65**	--
4. G2 Depression (14)	-.16**	.34**	.27**	--
5. G2 Depression (16)	-.12**	.25**	.30**	.48**	--
6. G2 Depression (33)	-.14**	.12**	.12**	.26**	.23**	--
7. G1 Punishment (16)	-.08*	.25**	.24**	.18**	.18**	.17**	--
8. G1 Inconsistency (16)	-.16**	.26**	.16**	.23**	.21**	.09*	.14**	--
9. G1 Closeness† (16)	-.16**	.15**	.16**	.24**	.13**	.20**	.19**	.11**	--
10. G2 Social Support (22)	.07+	-.01	-.07+	-.13**	-.07+	-.18**	-.04	-.07+	-.18**	--
11. G2 Trust/Suspiciousness† (22)	.05	-.07+	-.14**	-.16**	-.19**	-.28**	-.18**	-.13**	-.17**	.20**	--
12. G2 Interpersonal Difficulty† (22)	.10*	-.13**	-.18**	-.24**	-.22**	-.28**	-.08+	-.11**	-.14**	.29**	.39**	--
13. G2 Relationship Quality (33)	.12*	-.09*	-.10**	-.11**	-.10*	-.22**	-.07+	-.13**	-.19**	.21**	.15**	.16**	--

^aControlling for age, gender, and socioeconomic status,

G1 = Generation 1 (Parents); G2 = Generation 2 (Offspring). (#) = Age of G2 offspring at time of measure.

^*p <= .05.

^**p <= .01.

⁺p <= .10.

^†Reverse coded variable.

Does Parents’ Romantic Engagement Predict Offspring Romantic Relationship Quality?

We tested the relationship between parents’ romantic engagement at offspring age 14 and subsequent offspring romantic relationship quality at age 33, via hierarchical linear regression analyses, controlling for age, sex, and socioeconomic status. The model was significant (B = .12, SE(B) = .04, Beta = .12, t = 2.89, p < .01), with demographic variables explaining a larger percentage of the variance than parents’ romantic engagement (3% compared to 1%). We then split the file by sex and re-ran the analyses controlling for age and socioeconomic status. For females (N = 302), parents’ romantic engagement predicted offspring romantic relationship quality a mean of 17 years later (B = .18, SE(B) = .06, Beta = .18, t = 3.19, p < .01) and explained 3.2% of the variance. For males (N = 279), parents’ romantic engagement did not predict offspring romantic relationship quality (B = .03, SE(B) = .06, Beta = .03, t = .53, p = .60), and therefore, males were excluded from additional analyses.

Developmental Pathways

Prior to model testing, we added offspring and parent depression at age 14 to the model to control for the autoregressive effects of depression, in addition to covariances between error terms of items measured at the same time to control for potential within-wave measurement error. Additionally, we added offspring depression at age 33 to control for its effect on concurrent romantic relationship quality. Initial testing indicated the model was a good fit for the data, χ² (46, N = 302) = 66.4, p = .03, CFI = .97, TLI = .95, RMSEA = .04 (95% CI = .01 - .06, p = .82); however, there were numerous insignificant paths. Parent depression at offspring age 14 did not predict offspring depression or parenting at age 16. Offspring and parent depression at ages 14 and 16 did not predict socio-emotional functioning or romantic relationship quality. Likewise, parents’ romantic engagement at offspring age 14 did not predict parent or offspring depression at age 16 although it was correlated concurrently with offspring and parent depression at age 14.

After trimming the non-significant paths, the final model remained a good fit for the data, χ²(59, N = 302) = 83.1, p = .02, CFI = .96, TLI = .95, RMSEA = .04 (95% CI = .02 - .05, p = .89), with all paths significant at the p ≤ .05 level. Figure 1 depicts the trimmed model with standardized regression weights and squared multiple correlations. Consistent with the parenting model, parents’ romantic engagement at offspring age 14 predicted parenting at offspring age 16 (estimate = -.26, SE = .08, CR = -3.07, p < .01), parenting at offspring age 16 predicted offspring socio-emotional functioning at age 22 (estimate = -.62, SE = .09, CR = -3.82, p < .001), and offspring socio-emotional functioning at age 22 predicted offspring romantic relationship quality at age 33 (estimate = .33, SE = .09, CR = 3.76, p < .001). Offspring socio-emotional functioning at age 22 (estimate = -.41, SE = .20, CR = -4.34, p < .001) and parents’ romantic engagement at offspring age 14 predicted offspring depression at age 33 (estimate = -.14, SE = .06, CR = -2.60, p = .01). Offspring depression at age 14 predicted parenting at offspring age 16 (estimate = .59, SE = .08, CR = 5.16, p < .001). There was also a direct relationship between parents’ romantic engagement and offspring romantic relationship quality, albeit smaller in magnitude (estimate = .12, SE = .08, CR = 2.13, p = .03). The total indirect effect of parents’ romantic engagement on offspring romantic relationship quality was significant (estimate = .05, SE = .02, 95% CI = .02 -.12, p = .001). The final model explained 14% of the variance in offspring romantic relationship quality.

Figure 1.

Trimmed model with standardized estimates – G1 romantic engagement, G1 depression, G1 parenting, G2 depression, and G2 socio-emotional functioning predicting G2 romantic relationship quality for females.

G1 = Generation 1 (Parents); G2 = Generation 2 (Offspring).

Note: Subscripted numbers represent squared multiple correlations.

All paths are significant at the p <= .02 level.

* Reverse Coded.

In summary, results suggest that parents’ romantic engagement at offspring age 14 directly and indirectly affects offspring romantic relationship quality at age 33. In the final model, offspring depression at age 16 did not predict offspring socio-emotional functioning at age 22, rather, it appears that the concurrent association of parents’ romantic engagement and offspring depression at age 14 predicts parenting at offspring age 16, which in turn predicts offspring socio-emotional functioning at age 22. Thus, when controlling for prior offspring depression at age 14, offspring depression at age 16 is not directly associated with offspring socio-emotional functioning at age 22. Parenting at offspring age 16 was associated with parents’ romantic engagement and offspring depression at age 14 and offspring socio-emotional functioning at age 22. Additionally, parents’ romantic engagement at offspring age 14 did not predict parent or offspring depression at age 16.

Discussion

This study investigated the continuity of relationship functioning across generations in a community sample of youth and their parents followed prospectively for over 25 years. We expected that parents’ romantic engagement (emotional closeness and involvement) in one generation would subsequently predict romantic relationship quality in the next generation. Further, we tested two possible developmental pathways to explain this link. The first pathway would operate via an increased risk for maladaptive parenting, which again would influence poor socio-emotional functioning, and in turn, poor romantic relationship quality. The second pathway would operate via an increased risk for offspring depression, in turn leading to compromised socio-emotional functioning and ultimately, to poor romantic relationship quality.

Results partially confirm our hypotheses. Parents’ romantic engagement in one generation directly predicted romantic relationship quality in the next generation, but only for females. This adds to evidence that females not only fare worse than males after exposure to marital stress (Crawford et al., 2001; Turner & Barrett, 1998), but may be more positively influenced than males when exposed to strong affective engagement between their parents. As the first generation of parents consisted solely of mothers, it is also possible that female children are more attuned to their mothers’ reported romantic engagement than male children, and thus more likely to benefit from the positive qualities of their mother’s relationships.

The evidence from the path models seems to support one of the hypothesized developmental pathways. Parents’ romantic engagement in early adolescence predicted parenting in middle adolescence, which then predicted offspring’s socio-emotional functioning in emerging adulthood and romantic relationship quality in adulthood. Prior research has found that marital hostility predicts maladaptive parenting (e.g., Kim et al., 2009; Patterson et al., 1992). The results of the present study suggest a lack of parental romantic engagement, that is, withdrawal and emotional disengagement from one’s romantic partner, is also a risk for maladaptive parenting, specifically inconsistency, low closeness, and harsh punishment. Children exposed to marital conflict and maladaptive parenting are at risk for emotional dysregulation, leading to difficulties in peer relationships and social interactions (e.g., Eisenberg & Fabes, 1992; Schulz et al., 2005). Similarly, our findings suggest that parenting is a significant mechanism for transmitting poor emotional engagement in subsequent close relationships. Specifically, inconsistent, detached, and harsh parenting in adolescence predicts interpersonal difficulty, mistrust and suspiciousness of others and less receipt of social support from others in peer relationships in emerging adulthood. The finding that parenting and offspring socio-emotional functioning predict romantic relationship quality up to a decade later supports prior evidence that children with impaired emotion regulation and psychosocial development are likely to extend these impairments to other relationships (Eisenberg & Fabes, 1992; Kim et al., 2009; Schulz et al., 2005; Whitton et al., 2008).

Our results did not support the second proposed developmental pathway, whereby the influence of parents’ romantic engagement in early adolescence on risk for offspring’s adult romantic relationship quality would operate via higher risk for offspring depression in middle adolescence, subsequently influencing offspring socio-emotional functioning. Although parents’ romantic disengagement was correlated with adolescent depression at age 16, after controlling for prior adolescent depression at age 14 and parent depression at ages 14 and 16, parents’ romantic disengagement no longer predicted adolescent depression at age 16. The present study suggests that affective disengagement between parents does influence the child and parent’s concurrent depressive symptoms, but that the development of such symptoms does not directly predict one’s socio-emotional functioning years later. Rather, it appears the influence of adolescent depression on socio-emotional functioning may be indirect via its concurrent association with parents’ disengagement.

Our findings have theoretical implications, as they are consistent with emotional security models linking parental withdrawal with parents’ reduced emotional availability, and the modeling of avoidant regulation strategies in intimate relationships (Cox et al., 1999; Sturge-Apple et al., 2006). The support of a pathway that operates to increase risk for low adult romantic relationship quality via parenting may also have important clinical implications. From the point of view of prevention, our results suggest that programs targeting marital discord with the goal of improving child outcomes would benefit from a consideration not only of open conflict, but of emotional disengagement. Such a pattern of withdrawal appears to weaken parents’ emotional availability for offspring, and may later reduce their ability to effectively navigate the critical development task of forming trusting, supportive relationships with peers and intimate partners. Similarly, our findings did not support interpersonal stress models emphasizing the association of parental relationship functioning with adolescent depression. It appears that parental conflict has important implications for later offspring depression, but in the context of romantic disengagement, the influence on parenting is greater. Parents’ romantic disengagement has critical implications for parenting and in turn for children’s own capacity for intimacy across the lifespan.

Despite the relevant findings of the present study, there are a number of limitations. First, parents’ romantic engagement and offspring romantic relationship quality were based solely on self-report and did not incorporate the spouse’s experience within their romantic relationships. On the other hand, we were able to draw on reports of romantic relationship quality by both male and female offspring participants, an unusual feature in most studies linking parenting with marriage. Second, although the majority of correlations between variables were statistically significant, the majority were in the small to medium range. However, given the span of time between measurement of many of these variables, it is not surprising that the effect sizes are relatively small. Third, although the sample was chosen to be representative of the region, the ethnic diversity of the sample is quite low, thereby affecting the generalizability of the findings. Additionally, because our final model only pertains to females, the results of the present study do not inform us as to mechanisms by which parents’ romantic engagement may or may not influence subsequent peer relationships and romantic relationships for males. Similarly, as we only examined outcomes for females with a romantic relationship at age 33, it is unclear how parents’ romantic disengagement and withdrawal may have impacted females who were not in a relationship. Fourth, the study could have been strengthened by inclusion of observational measures of peer interactions and marital interactions, as a combination of reporting methods is generally desirable. However, the Children in the Community Study was designed to study a very wide range of environmental risks for mental health disorders. As such, the inclusion of observational measures of parenting and parents’ romantic engagement would not have been practical, especially as this sample began to drift geographically in later years. Fifth, as the study was designed originally to study the development of mental health disorders broadly, we lack measures of children’s perceptions of their parents’ romantic withdrawal and depression. Such perceptions influence children’s adjustment to parents’ depression (Goodman, Tully, Connell, Hartman, & Huh, 2011) and marital conflict (Grych, Seid, & Fincham, 1992). Sixth, since our assessments took place in the early 1980s, we did not have strong measures of attachment that would have reflected current trends in measurement. However, our findings on parents’ romantic withdrawal and offspring’s later mistrust and lack of support with peers do hint at the possible operation of attachment security in the transfer of romantic relationship quality, and this merits further attention in future studies. Lastly, we only looked at a single time point for offspring romantic relationship quality in adulthood. The extent to which these associations extend to other developmental periods or persist over time remains unknown.

Yet, the current study has numerous strengths. First and foremost, the prospective design spanned two generations followed across six time points. Further, the sample was representative of the sampled region, with a wide range of ages for parents and children, socioeconomic levels from very poor to very wealthy, and a mixture of urban, suburban, and rural populations. Perhaps most importantly, the present study examined a theoretically and clinically meaningful, alternate aspect of romantic relationship functioning, rather than focusing on open conflict, thereby adding to the existing literature on the continuity of romantic relationship quality over time.

This study provides evidence suggestive of a developmental pathway for the direct and indirect transmission of relationship functioning across generations. Clinicians working with couples with children should include an assessment of engagement between parents in addition to open conflict. Such affective withdrawal may reflect a general pattern of familial emotional withdrawal, with important implications for offspring’s social development in multiple contexts.