Just the Two of Us? How Parents Influence Adult Children’s Marital Quality

Abstract

We work from a life-course perspective to explore how relationships with parents affect adult children’s marital quality. We further ask whether the effects of parents on adult children’s marital quality depend on the adult child’s gender, age, marital duration, and childhood family experiences. Growth-curve analysis of national, longitudinal data (Americans’ Changing Lives) indicated that relationships with fathers (n = 336) and mothers (n = 520) differentially affected the marital quality of adult children over time. Findings suggest that the effects of the parent – adult child relationship on adult children’s marriages depend on age, marital duration, and levels of family stress experienced in childhood.

Theorists have long emphasized that married couples do not exist in social isolation. Rather, social contexts and relationships outside the marital dyad influence marital quality (Lewis, 1973; Milardo & Lewis, 1985). Popular culture and psychiatric literature have pointed to relationships with parents as having a strong influence on marital dynamics, yet there is little research that directly addresses how ongoing relationships with parents affect the quality of adults’ marital relationships. Given the salience of intergenerational relationships for adult children (Lye, 1996; Umberson, 1992), one would expect parents to influence their children’s marital relationships (Bryant, Conger, & Meehad, 2001). On the one hand, external stressors undermine marital quality (Karney & Bradbury, 1995), and if relationships with parents are strained, this might impose stress on the marital relationship. On the other hand, supportive ties with parents might be a resource for adult children in ways that promote marital quality. We work from a life-course perspective to address three specific issues. First, we analyze longitudinal data to consider how relationships with parents shape trajectories of change in adult children’s marital quality over time. Second, we ask whether these effects vary by the gender, age, and marital duration of the adult child. Third, we consider whether childhood family stress confounds or moderates the effect of adult child – parent ties on adult children’s marital quality.

Parent – Adult Child Relationships and Adult Children’s Marital Quality

A life-course perspective emphasizes that parents and children have linked lives that travel in interlocked trajectories throughout the life course (Elder, Johnson, & Crosnoe, 2003). In this view, parents continue to influence children’s well-being into adulthood (Knoester, 2003). Most adult children have supportive relationships with parents and report that they feel loved and cared for (Cooney & Uhlenberg, 1992; Lye, 1996). Yet recent studies have shown that conflict and stress in relationships with parents are also important aspects of intergenerational relationships (Bengtson, Giarrusso, Mabry, & Silverstein, 2002) that influence adults’ well-being (Umberson, 1992). Because parent – adult child ties may be typified as both supportive and strained at various points over time, life-course scholars recommend that both positive and negative aspects of intergenerational ties be analyzed (Bengtson et al., 2002).

Marital-quality researchers have also demonstrated that multiple dimensions of marital quality—some positive and some negative—change over the life course (Bradbury, Fincham, Beach, 2000; Umberson, Williams, Powers, Chen, & Campbell, 2005). Periods of marital strain may be typified by conflict or by feeling bothered and upset about the relationship, whereas feelings of satisfaction and feeling loved characterize marital closeness (Bradbury et al., 2000). Marital closeness and strain are conceptually and empirically distinct aspects of marital quality (Bradbury et al., 2000; Fincham, Beach, & Kemp-Fincham, 1997); analyzing both dimensions allows for the detailed description of change in each. Moreover, supportive and strained aspects of a parent – adult child relationship may differentially affect closeness and strain in adult children’s marriages over the life course. We use the term marital quality to refer to both closeness and strain in marital ties, and parent-child relationship quality to refer to support and strain in relationships with parents.

Relationships with parents may influence marital quality in several ways. Parents may make children feel loved and cared for, thus fostering psychological well-being in the adult child (Umberson, 1992); that sense of well-being may spill over, fostering marital closeness for adult children. Adults may rely on parents as sources of support during times of stress (e.g., work, child care), lessening the burden on spouses to provide continuous support, thus enhancing marital quality. In contrast, strained intergenerational ties may be a considerable source of stress for adult children (Umberson, 1992), and several studies have shown that external stressors undermine marital quality (Neff & Karney, 2004; Umberson, Williams, Powers, Chen et al., 2005).

Although there are theoretical reasons to suggest that intergenerational ties affect adult children’s marital quality, we were unable to locate any longitudinal studies that examine this issue. Some studies have examined the link between social network involvement and marital quality. These studies have not differentiated the influence of parents apart from social networks generally; however, results suggest that social ties with others can influence marital quality. A cross-sectional analysis of 166 Australian families revealed a positive association between support from network members and marital satisfaction for men (Cotton, Conningham, & Antill, 1993). But a longitudinal study of long-term marriages showed that emotional support from social networks was not a significant predictor of marital satisfaction for either men or women (Bryant & Conger, 1999). In a later study, Bryant et al. (2001) found that discord with in-laws was detrimental to adult children’s marital satisfaction. Given this overall body of evidence, we hypothesize that supportive relationships with parents foster marital quality and that strained relationships with parents undermine marital quality.

Age and Marital Duration

A life-course approach suggests that the longer couples are married, the more they come to rely on one other and the less they rely on others for support (Bryant & Conger, 1999). This may make longer term marriages less susceptible to the outside effects of relationships with parents. Moreover, social norms associated with the development of children point to adult children’s differentiation from parents (Lye, 1996). In accordance with these norms, parents are most likely to provide support to children in early adulthood (Cooney & Uhlenberg, 1992), and even if levels of support from parents remain constant, parents may have less influence over their children as children age. We hypothesize that supportive relationships with parents enhance marital quality and that strained relationships with parents undermine marital quality, and that these effects decrease with increasing age and marital duration.

Gender

Gender is a central theme in the literature on intergenerational relationships (Kaufman & Uhlenberg, 1998), yet we know of no longitudinal studies that examine gender differences in the effect of relationships with parents on marital quality. Intergenerational ties may be more salient to the lives of women than men, and mothers and fathers play different roles in the lives of their adult children (Amato, 1994; Videon, 2005). Daughters visit and talk with their parents more often than sons do (Umberson, 1992; Ward & Spitze, 1998), and mothers are more likely to provide emotional support to daughters than to sons (Suitor, Pillemer, & Sechrist, 2006). Compared to fathers, mothers remain more involved with children throughout the life course and report higher levels of emotional attachment to their adult children (Lye, 1996; Silverstein & Bengtson, 1997). We hypothesize that relationships with mothers have stronger effects on the marital quality of adult children and that the effects of the parent – adult child tie are greater for daughters.

Childhood Family Stress

A growing body of evidence demonstrates that stressful childhood experiences have enduring negative effects on children’s health and well-being throughout life, a process known as cumulative disadvantage (Dannefer, 2003). We build on this perspective to suggest that stressful experiences with parents in childhood are likely to influence adult children’s marital quality in two ways. First, studies have suggested that family experiences in childhood are related to both adult marital quality (Amato & Booth, 2001; Umberson, Williams, Powers, Liu, & Needham, 2005) and adult child – parent relationships (Booth & Amato, 1994). Therefore, it is likely that these early relationships, in part, confound the impact of parent – adult child relationships on marital quality. Second, childhood family stress may moderate the link between intergenerational ties and marital quality. Adults who experienced high levels of childhood family stress may be more vulnerable to stressors that occur in adulthood (Umberson, Williams, Powers, Liu et al., 2005). We hypothesize that adult children with higher levels of childhood family stress are more strongly affected by strained intergenerational relationships (in adulthood), and this increased strain, in turn, undermines marital quality.

Taken together, past research suggests that the quality of relationships with parents may affect children’s marital quality. We address three specific research questions: (a) How do relationships with parents affect trajectories of change in adult children’s marital quality over time? (b) Does the effect of relationships with parents on adult children’s marital quality depend on the child’s age, marital duration, or gender? and (c) Do levels of childhood family stress confound and/or moderate the relationship between parent – adult child relationship quality and adult children’s marital quality?

Method

Data

We analyze data from the first three waves of the Americans’ Changing Lives (ACL) panel study (House, 1986). Multistage stratified area probability sampling was used to obtain the original sample of individuals in the contiguous United States (aged 24 – 96). Face-to-face structured interviews lasting approximately 90 minutes each were conducted with individuals in 1986 (N = 3,617), 1989 (N = 2,867), and 1994 (N = 2,398). In the original sample (1986), 1,904 married individuals (not married to one another) were interviewed. Seventy-one percent (n = 1,352) of the original 1,904 married individuals were interviewed in all three waves of data collection. Of those who are not accounted for, 17.6% (n = 335) did not respond to one or both of the follow-up surveys, and the remaining 11.4% (n = 217) had died by 1994. Of the 1,352 individuals interviewed in all three waves, 78.3% (n = 1,059) remained married to the same spouse over the 8-year period, 12.1% (n = 164) were widowed, 8.4 % (n = 113) divorced, and 1.2% (n = 16) were separated.

In this study, we draw two subsamples to examine how relationships with mothers and fathers affect adult children’s marital quality. The first sample includes 520 individuals who remained married to the same spouse over the study period and had a living mother who was mentally and physically capable of providing help or advice at Waves 1 and 2. The second sample includes 336 individuals who remained married to the same spouse over the study period and had a living father who was mentally and physically capable of providing help or advice at Waves 1 and 2. We used parental support/strain at Wave 1, and changes in parental support/strain between Wave 1 and 2, to predict marital quality trajectories from Wave 1 to Wave 3. We use parental support/strain only at Waves 1 and 2 because of measurement limitations due to survey changes in Wave 3. We excluded from the analysis observations with missing information on marital quality due to being lost to follow-up (less than 3%). Respondents lost to follow-up are more likely to be from lower-socioeconomic-status (SES) groups and have lower marital quality (discussed below).

Selection Bias

Selection bias based on the possibility of divorce is a potential problem for an analysis of marital quality over time. Because we examine only those marriages that continue over the study period, marriages most likely to be negatively affected by relationships with parents may be the most likely to end in divorce and drop out of the study. Therefore, estimates of the effects of relationships with parents on marital quality may be biased because only higher quality marriages among individuals of older ages are in the sample. Such nonrandom selectivity may lead to biased estimates of the effects of independent variables on marital quality. Although we are unable to completely remedy this type of selection bias in the present study, we address this issue in two ways. First, we include a covariate for previous divorce in an effort to partially control for one’s propensity for divorce (1 = previously divorced; 0 = not previously divorced). Second, we use a Heckman-type correction approach to address this selection issue. This approach involves first modeling the hazard that a respondent may divorce over the 8-year study period (using the entire sample who were married at Wave 1 with those who divorced later included), conditional on a set of predictors (age at marriage, number of previous marriages, marital duration, high financial strain, employment of wife, stepchild in home, recent thoughts of divorce, years of education, African American) (Heckman, 1976; White, 1990). Next, using a sample of only those who are married, we modeled marital quality as a function of a set of independent variables, including the estimated risk of getting divorced from the first stage. Following this correction, covariate estimates on marital quality should be interpreted as adjusted for observed and unobserved variables that may affect marital quality and tendency for divorce. We note that those with better quality marriages may experience fewer fluctuations in marital quality, and this selection effect may lead to more conservative estimation of the effects of parent – adult child relationships on marital quality.

Measures

Marital quality

Life course and marital quality researchers suggest that periods of marital strain and closeness are conceptually and empirically distinct with different correlates and should be analyzed separately (Bradbury et al., 2000). We consider two dimensions of marital quality—marital strain and marital closeness—and adopt the marital-quality measures Umberson, Williams, Powers, Liu et al. (2005) used. Marital closeness is a latent variable composed of four items based on four questions: (a) “How satisfied are you with your marriage” (responses from 1 to 5, not at all satisfied to completely satisfied); (b) “How much does your husband/wife make you feel loved and cared for?” (response from 1 – 5, not at all to a great deal); (c) “How much is he/she willing to listen when you need to talk about your worries or problems?” (responses from 1 – 5, not at all to a great deal); and (d) “Is your husband/wife a person with whom you can share private feelings and concerns” (0 = no; 1 = yes). Marital strain is a latent variable composed of two items based on two questions: (a) “How often do you feel bothered or upset by your marriage?” and (b) “How often would you say the two of you typically have unpleasant disagreements or conflicts?” (responses range from 1 – 7, never to almost daily). Higher values in the scale indicate higher levels of the intended construct. To ensure that change over time reflects growth rather than change in the measurement scale, the scales for marital closeness and marital strain are further standardized at each wave of measurement based on the mean and standard deviation at Wave 1 (Bryk & Raudenbush, 1987).

Parent support and strain

We used four variables to measure support and strain from parents based on constructed indices in the ACL data: mother support, father support, mother strain, and father strain. Mother support and father support are two separate index variables, each based on responses to the following questions: (a) “How much does your mother/father make you feel loved and cared for?” and (b) “How much is she/he willing to listen when you need to talk about your worries or problems?” Mother and father strain are two separate index variables based on responses to the following questions: (a) “How much do you feel your mother/father makes too many demands on you?” and (b) “How much is she/he critical of you or what you do?” Responses for both sets of questions range from 1 – 5, not at all to a great deal, where a greater value indicated a higher level of support/strain from mothers and fathers. We constructed the mother and father support/strain indices by taking the mean of the two items and then further standardized at each wave of measurement based on the mean and standard deviation at Wave 1. One missing value on mother strain at Wave 2 was imputed by regression in the ACL data set.

Childhood family stress

The childhood family stress measure is based on reports of seven family stressors that occurred when the respondent was age 16 or younger: at least one parent died, parents divorced, parents had marital problems, at least one parent had a mental health problem, at least one parent had an alcohol problem, never knew father, and family economic hardship. Family economic hardship is coded 1 for respondents who report their family was somewhat worse off or a lot worse off than the average family in their community. All other childhood stressors were coded 0 if the experience did not occur and 1 if it did occur. All scores on all dichotomous variables were summed to create an indicator of the number of childhood stressors. Reports of childhood stress may be subject to bias associated with retrospective reports. Although we note this limitation, previous studies provide some support for the validity and reliability of these reports, including for retrospective reports of death or divorce of a parent (McLeod, 1991) and depressive symptoms (Schraedley, Turner, & Gotlib, 2002).

Life course and sociodemographic variables

We included measures for age (in years) of the respondent and marital duration (in years, at Wave 1). We also controlled for other sociodemographic covariates that are relevant to both marital quality and parental support/strain. They include gender (0 = adult daughter; 1 = adult son), race (0 = other; 1 = African American), education measured in number of years completed at Wave 1, total family income (in $1,000s) measured as a time varying covariate, and divorce hazard. Because parental status of adult children influences the adult child’s marital quality (Umberson, Williams, Powers, Chen et al., 2005) and may be related to the adult child – parent relationship, we controlled for respondents’ parental status measured as a time varying covariate in the analyses. Parental status include three categories: (a) no children (reference group), (b) minor child (younger than 18 years old), and (c) at least one adult child (18 or older). Table 1 presents means and standard deviations for all variables in the analysis for the two subsamples. Age, marital duration, education, family income, divorce hazard, and childhood family stress are centered at means of each subsample in the analyses. Significance tests are conducted to compare the difference between the two subsamples. Results in Table 1 shows that those in the mother subsample tend to be older, have longer marital duration, have lower risk for divorce, are more likely to have only one living parent, are less likely to have all of their children under 16, and are more likely to have at least one adult child than are those in the father subsample.

Table 1.

Weighted Descriptive Statistics of Variables Analyzed

	Subsample for Mother Relationship (n = 520)		Subsample for Father Relationship (n = 336)
	Mean	SD	Mean	SD
Marital closeness T1	0.009	0.978	0.026	0.958
Marital closeness T2	−0.115	1.029	−0.134	1.056
Marital closeness T3	−0.157	1.083	−0.218	1.098
Marital strain T1	0.046	1.000	0.033	1.001
Marital strain T2	0.223	1.195	0.183	1.188
Marital strain T3	0.375	1.198	0.371	1.195
Parent support T1	0.011	0.939	−0.058	1.018
Parent strain T1	−0.075	0.913	−0.093	0.981
Change in parent support T1 – T2	−0.057	0.893	0.063	0.858
Change in parent strain T1 – T2	−0.030	0.810	0.020	0.839
Childhood family stress	0.680	1.028	0.585	0.982
Age (T1, in years)	38.072	9.504	35.725	7.661*
Gender (1 = adult son)	0.501	0.500	0.531	0.500
Race (1 = African American)	0.086	0.280	0.068	0.252
Years of education (T1)	13.511	2.406	13.810	2.253
Family income (time varying, $1,000)	37.904	19.184	39.460	17.879
Marital duration (T1, in years)	14.578	9.937	12.135	8.110*
Previously divorced (T1)	0.159	0.366	0.164	0.371
Divorce hazard	−0.219	0.160	−0.247	0.172*
Only one living parent	0.326	0.469	0.115	0.319*
Parental status (Time varying)
No kids	7.80%		7.64%
Children all under 16	44.05%		52.43%*
At least one adult child	48.15%		39.93%*

^*p < .05 (two-tailed T test for comparing means and two-tailed Z test for comparing proportions of two subsamples).

Analytical Design

We use latent linear growth-curve models (i.e., random effects models) to consider the effects of independent variables on both baseline levels and change in marital closeness and strain over time. In using growth-curve techniques, we are able to consider that individuals start the study period with different levels of marital strain and closeness and that each individual may experience different rates of change in marital strain and closeness as a function of support and strain in relationships with parents. A major advantage of growth-curve models over traditional regression models is their ability to distinguish the two levels (i.e., within-individual and between-individual) of heterogeneity in estimating population average differences. The model we used for this study can be specified as follows:

$$\begin{array}{l}{Y}_{ij}={\beta }_{0i}+{\beta }_{1i}{T}_{ij}+\sum {\alpha }_k{\mathit{TVX}}_{\mathit{kij}}+\sum {\omega }_k{\mathit{TVX}}_{\mathit{kij}}\ast T_{ij}+{\epsilon }_{ij}\\ {\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}SP{1}_i+{\gamma }_{02}SN{1}_i+{\gamma }_{03}(SP{2}_i-SP{1}_i)+{\gamma }_{04}(SN{2}_i-SN{1}_i)+\sum {\pi }_{0P}{X}_{Pi}+{\zeta }_{0i},\\ {\beta }_{1i}={\gamma }_{10}+{\gamma }_{11}SP{1}_i+{\gamma }_{12}SN{1}_i+{\gamma }_{13}(SP{2}_i-SP{1}_i)+{\gamma }_{14}(SN{2}_i-SN{1}_i)+\sum {\pi }_{1P}{X}_{Pi}+{\zeta }_{1i}[1pt]\end{array}$$ (1)

where Y_ij represents the outcome variable (e.g. marital closeness/strain). β_0i, the latent intercept, and β_1i, the latent slope of marital quality trajectories are predicted by Wave 1 parental support (indexed by SP1_i), Wave 1 parental strain (indexed by SN1_i), and changes in parental support (SP2_i – SP1_i) and strain (SN2_i – SN 1_i) between Wave 1 and Wave 2. All time varying covariates (indexed by TVX _kij, including income and parental status) are included in Level 1 of the model, and all other time invariant covariates (indexed by X_ki) are included in Level 2 of the model. The residual terms are ε_ij, ζ_0i, and ζ_1i. We estimated all models using STATAXTMIXED (StataCorp, 2007). The focus of inference in this study is on estimating the effects of parental support/strain on the baseline level (i.e., latent intercept) and rate of change (i.e., latent slope) in marital closeness/strain trajectories over time. We ran the models separately for mother support/strain and father support/strain. In the models examining gender, age, marital duration, and childhood stress variation in the relationships between marital strain/closeness and parental support/strain, we include interaction terms for parent support/strain by gender, age, marital duration, and childhood stress. We compare results from the models with and without controlling for childhood stress to determine whether family stress confounds the association between parental support/strain and marital quality.

Results

Consistent with past research (Umberson, Williams, Powers, Chen et al., 2005; Umberson, Williams, Power, Liu et al., 2005), results from unconditional growth models with no covariates (not shown) indicate that marital closeness decreased (b = −.013; p < .001) and marital strain increased (b = .016; p < .001) over the study period. Tables 2 – 5 show results from the growth-curve analysis and present the estimated effects of the parental support and strain measures and other covariates on the baseline level (latent intercept) and the rate of change (latent slope) in marital strain and closeness over time.

Table 2.

Estimated Effects of Mother and Father Support/Strain on Marital Closeness/Strain From Growth-Curve Models

	A. Mother (n = 520)				B. Father (n = 336)
	Marital Closeness		Marital Strain		Marital Closeness		Marital Strain
	Latent Intercept	Latent Slope	Latent Intercept	Latent Slope	Latent Intercept	Latent Slope	Latent Intercept	Latent Slope
Parent support T1	0.234***	−0.003	−0.108*	0.009	0.174***	0.016**	−0.052	0.001
Parent strain T1	0.006	−0.000	0.117*	−0.000	0.025	0.001	0.167**	0.006
Change in parent support T1 – T2		0.010		0.006		0.018**		−0.009
Change in parent strain T1 – T2		−0.001		0.002		−0.004		0.007
Age (T1)	−0.014	−0.002**	0.000	0.002*	−0.034*	−0.002	0.015	0.002
Gender (1 = adult son)	0.370***	0.009	−0.018	−0.008	0.444***	0.013	0.044	−0.005
Race (1 = African American)	0.455***	−0.000	0.389***	−0.008	−0.254	−0.007	0.242	0.004
Years of education (T1)	0.009	0.003	−0.007	0.001	−0.034	0.003	−0.003	0.002
Family income (time varying, $1000)	0.001	−0.000	0.000	−0.000	0.001	−0.000	−0.001	−0.000
Marital duration (T1, in years)	0.015*	0.003***	−0.014	−0.001	0.011	0.003*	−0.018	−0.000
Previously divorced (T1)	−0.094	−0.000	0.003	0.025*	0.073	−0.020	−0.288	0.027
Divorce hazard	0.160**	−0.004	−0.070	−0.003	1.027***	−0.026	−0.567*	0.004
Parental status (time varying)
Children all under 16	−0.199*	0.012	0.069	0.013	−0.220*	0.003	0.162	0.033
At least one adult child	−0.297**	0.007	0.066	0.003	−.373**	0.008	0.203	0.020
Only one living parent	0.153	0.018*	−0.102	−0.007	0.193	−0.004	−0.125	0.006
Mean	1.305**	−0.063	−0.611	0.015	3.055***	−0.109	−1.830	0.021
Random effects
Level 2 variance	0.743***	0.002***	0.966***	0.003***	0.741***	0.002***	0.971***	0.003***
Level 1 variance	0.139***		0.103***		0.142***		0.087***
Log likelihood	−1,541.154		−1,525.589		−993.100		−938.390

^*p < .05.

^**p < .005.

^***p < .001 (two-tailed test).

Table 5.

Estimated Effects of Mother and Father Support/Strain and Childhood Family Stress on Marital Closeness/Strain From Growth Curve Models

	A. Mother (n = 520)				B. Father (n = 336)
	Model 1		Model 2		Model 1		Model 2
	Intercept	Slope	Intercept	Slope	Intercept	Slope	Intercept	Slope
Marital Closeness
Parent support T1	0.221***	−0.003	0.243***	−0.003	0.186***	0.016**	0.215***	0.013*
Parent strain T1	0.004	0.000	0.011	−0.000	0.049	0.001	0.060	−0.005
Change in parent support T1 – T2		0.010*		0.010		0.019**		0.015*
Change in parent strain T1 – T2		−0.001		−0.001		−0.004		−0.004
Childhood stress	−0.056	0.000	−0.051	0.000			−0.088	−0.006
Parent support T1 × childhood stress			−0.054	0.001			−0.089*	0.011*
Parent strain T1 × childhood stress			−0.126**	0.009*			−0.049	−0.004
Change in parent support T1 – T2 × Childhood stress				0.003				−0.002
Change in parent strain T1 – T2 × Childhood stress				0.013**				0.022***
Mean	1.357**	−0.063	1.372**	−0.070	3.348***	−0.107	3.409***	−0.094
Random effects
Level 2 variance	0.730***	0.002***	0.713***	0.002***	0.703***	0.002***	0.691***	0.002***
Level 1 variance	0.139***		0.139***		0.142***		0.142***
Log likelihood	−1,536.832		−1,527.363		−984.636		−972.318
Marital stain
Parent support T1	−0.103*	0.009	−0.106*	0.010	−0.053	0.002	−0.065	0.001
Parent strain T1	0.113*	0.000	0.112*	0.000	0.167**	0.006	0.173**	0.008
Change in parent support T1 – T2		0.006		0.006		−0.008		−0.008
Change in parent strain T1 – T2		0.002		0.002		0.006		0.005
Childhood stress	−0.011	0.003	−0.019	0.003	−0.000	0.006
Parent support T1 × Childhood stress			0.001	−0.001			0.019	0.002
Parent strain T1 × Childhood stress			0.074	−0.005			0.085	0.010
Change in parent support T1 – T2 × Childhood stress				−0.001				0.004
Change in parent strain T1 – T2 × Childhood stress				−0.003				−0.007
Mean	−0.722	0.018	−0.759	0.021	−1.830*	0.024	−1.875*	0.014
Random effects
Level 2 variance	0.950***	0.004***	0.945***	0.004***	0.971***	0.003***	0.965	0.003
Level 1 variance	0.102***		0.102***		0.087***		0.088
Log likelihood	−1,521.105		−1,519.312		−937.63095		−933.347

Note: Age, gender, race, education, family income, marital duration, previously divorced, divorce hazard, parental status and only one living parent are all controlled in all models. All significant age and marital duration interaction effects discussed in text are also controlled in models.

^*p < .05.

^**p < .005.

^***p < .001 (two-tailed test).

Marital Closeness and Marital Strain

We first examine results from growth-curve models that estimated the effects of support and strain with parents on marital closeness and strain trajectories. Table 2 shows the results for support/strain with mothers (Panel A) and fathers (Panel B), controlling for all sociodemographic covariates (models without controls for sociodemographic covariates revealed similar results). We begin with the results on marital closeness. Results indicated that baseline support from mothers (Panel A) and from fathers (Panel B) was positively associated with baseline marital closeness. Results showed no significant effects of baseline mother support/strain on the rate of change (the latent slope) of marital closeness (Panel A). The significant positive coefficient for baseline support from fathers (Panel B) on the rate of change of marital closeness (b = .016, p < 0.01) indicated that every one-unit increase in baseline support from fathers slowed the rate of decline in marital closeness over time by .016 units. Because levels of support and strain may change over time, we also assessed the effects of the changing rate of parent support/strain on marital closeness. There were no significant effects of change in mother support/strain (Panel A) on the rate of change of marital closeness. As Panel B shows, increased support from fathers (T1 to T2) slowed the decline in marital closeness over time. Decreased support from fathers (T1 to T2) fostered a more rapid decline in marital closeness over time.

Next, we report results on marital strain. Results revealed that baseline strain with mothers and fathers was positively associated with baseline marital strain and that baseline support from mothers was negatively associated with baseline marital strain. Mother and father support/strain were not significantly associated with the rate of change in marital strain trajectories, and there were no significant effects of changing levels of support/strain with parents (T1 to T2) on marital strain.

Age, marital duration, and gender interactions

We considered whether the effect of relationships with parents on adult children’s marital quality depended on the child’s age, marital duration, and gender. We first present results for age and marital duration interactions (Table 3) beginning with results on marital closeness (upper half of Table 3). There were significant Age × Marital duration interaction effects on baseline levels but not on the changing rate of marital closeness trajectories. The significant interaction effect of baseline support from fathers and age on baseline marital closeness (b = −0.014, p < 0.05) indicated that baseline support from fathers was positively related to baseline marital closeness for individuals 49 (i.e., 0.187/0.014 + 35.725) years old and younger, and negatively related to baseline marital closeness for individuals older than 49. The significant interaction effect of baseline strain from fathers and age on baseline marital closeness (b = 0.016, p < 0.05) indicated that baseline strain with fathers was positively related to baseline marital closeness for individuals 36 years old and older, and negatively related to baseline marital closeness for individuals younger than 36. The significant interaction effect of support from mothers (b = −0.010, p < 0.05) and fathers (b = −0.013, p < 0.05) and marital duration on baseline marital closeness indicated that the relationship between baseline support from parents and baseline marital closeness was positive for those with marital durations of 37 years or less for relationships with mothers and 27 years or less for relationships with fathers, and negative for those with longer marital durations.

Table 3.

Estimated Interaction Effects of Mother and Father Support/Strain by Age/Marital Duration on Marital Closeness/Strain From Growth-Curve Models

	A. Mother (n = 520)				B. Father (n = 336)
	Age Interaction Model		Marital Duration Interaction Model		Age Interaction Model		Marital Duration Interaction Model
	Intercept	Slope	Intercept	Slope	Intercept	Slope	Intercept	Slope
Marital closeness
Parent support T1	0.235***	−0.003	0.228***	−0.003	0.187***	0.016**	0.187***	0.017**
Parent strain T1	0.005	−0.000	0.004	0.000	0.047	−0.001	0.044	0.000
Change in parent support T1 – T2		0.010*		0.010*		0.020**		0.020**
Change in parent strain T1 – T2		−0.001		−0.001		−0.006		−0.005
Age/marital duration	−0.014	−0.002**	0.017*	0.003***	−0.035**	−0.002	0.015	0.003*
Parent support T1 × Age/marital duration	0.000	0.000	−0.010*	0.000	−0.014*	0.000	−0.013*	−0.001
Parent strain T1 × Age/marital duration	0.001	0.000	0.002	0.000	0.016*	−0.001	0.006	−0.001
Change in parent support T1 – T2 × Age/marital duration		0.000		0.000		0.000		0.000
Change in parent strain T1 – T2 × Age/marital duration		0.000		0.000		−0.001		−0.001
Mean	1.305**	−0.062	1.417**	−0.061	3.393***	−0.111	3.235***	−0.106
Random effects
Level 2 variance	0.743***	0.002***	0.733***	0.002***	0.710***	0.002***	0.722***	0.002***
Level 1 variance	0.139***		0.139***		0.142***		0.142
Log likelihood	−1,540.669		−1,536.216		−985.233		−987.170
Marital strain
Parent support T1	−0.119*	0.009	−0.110*	0.009	−0.058	0.001	−0.054	0.001
Parent strain T1	0.113*	0.000	0.115*	0.000	0.153*	0.006	0.153*	0.006
Change in parent support T1 – T2		0.006		0.005		−0.009		−0.008
Change in parent strain T1 – T2		0.002		0.001		0.006		0.007
Age/marital duration	−0.001	0.002*	−0.015	−0.001	0.016	0.002	−0.020	−0.001
Parent support T1 × Age/marital duration	−0.005	0.000	0.000	0.000	−0.002	0.000	−0.001	0.001
Parent strain T1 × Age/marital duration	−0.010*	0.001	−0.012*	0.000	−0.009	0.000	−0.007	0.000
Change in parent support T1 – T2 × Age/marital duration		0.000		0.000		0.000		0.001
Change in parent strain T1 – T2 × Age/marital duration		−0.000		0.000		0.000		0.000
Mean	−0.612	0.014	−0.697	0.014	−1.935*	0.022	−1.903*	0.015
Random effects
Level 2 variance	0.957***	0.004***	0.951***	0.004***	0.967***	0.003***	0.968***	0.003***
Level 1 variance	0.102***		0.102***		0.087***		0.087***
Log likelihood	−1,522.4932		−1,521.315		−937.489		−936.822

Note: Age, gender, race, education, family income, marital duration, previously divorced, divorce hazard, parental status, and only one living parent are controlled in all models.

^*p < .05.

^**p < .005.

^***p < .001 (two-tailed test).

Next, we report results for age and marital duration interactions on marital strain (lower half of Table 3). There were significant age and marital duration interaction effects for strain from mothers on baseline marital strain but not on the changing rate of trajectories. Results revealed a positive relationship between baseline mother strain and baseline marital strain for individuals 49 years old and younger and those married for 24 or fewer years, but this effect became negative for individuals older than 49 and those married more than 24 years. We did not find significant gender interaction effects on baseline level or change in marital closeness or strain (Table 4).

Table 4.

Estimated Interaction Effects of Mother and Father Support/Strain by Gender on Marital Closeness/Strain From Growth-Curve Models

	A. Mother (n = 520)		B. Father (n = 336)
	Intercept	Slope	Intercept	Slope
Marital closeness
Parent support T1	0.267***	−0.010	0.156*	0.015*
Parent strain T1	0.022	−0.005	0.030	−0.005
Change in parent support T1 – T2		0.008		0.015
Change in parent strain T1 – T2		0.001		−0.001
Gender (1 = adult son)	0.369***	0.009	0.445***	0.013
Parent support T1 × Adult son	−0.065	0.013	0.036	0.002
Parent strain T1 × Adult son	−0.025	0.008	−0.007	0.009
Change in parent support T1 – T2 × Adult son		0.002		0.006
Change in parent strain T1 – T2 × Adult son		−0.004		−0.003
Mean	1.300**	−0.060	3.058***	−0.108
Random effects
Level 2 variance	0.742***	0.002***	0.740***	0.002***
Level 1 variance	0.139***		0.142***
Log likelihood	−1,539.577		−992.298
Marital strain
Parent support T1	−0.090	0.011	−0.055	0.001
Parent strain T1	0.149*	−0.001	0.206*	0.011
Change in parent support T1 – T2		0.001		−0.004
Change in parent strain T1 – T2		0.008		0.010
Gender (1 = adult son)	−0.018	−0.008	0.043	−0.004
Parent support T1 × Adult son	−0.028	−0.006	0.006	0.000
Parent strain T1 × Adult son	−0.065	0.002	−0.066	−0.009
Change in parent support T1 – T2 × Adult son		0.009		−0.007
Change in parent strain T1 – T2 × Adult son		−0.015		−0.006
Mean	−0.631	0.018	−1.844*	0.020
Random effects
Level 2 variance	0.965***	0.004***	0.970***	0.003***
Level 1 variance	0.103***		0.088***
Log likelihood	−1,522.159		−937.625

Note: Age, gender, race, education, family income, marital duration, previously divorced, divorce hazard, parental status and only one living parent are controlled in all models.

^*p < .05.

^**p < .005.

^***p < .001 (two-tailed test).

Childhood family stress

We examined whether childhood family stress confounded the effect of support/strain from parents on children’s marital quality. We accomplished this by adding the childhood stress variable into models predicting marital closeness and strain in Model 1 of Table 5. Comparing results in Model 1 of Table 5 with Table 2 indicated little difference in the estimated effects of strain/support with mother and/or father on trajectories of marital closeness and strain. Additional analyses (not shown) indicated that higher childhood stress was associated with lower levels of parental support and strain, but we did not find evidence that childhood stress was a significant predictor of marital closeness or strain. This indicates that stressful family experiences in childhood did not play a major role in explaining the association between support/strain from parents in adulthood and respondents’ marital quality.

Next, we considered whether childhood family stress moderated the relationship between support/strain from parents and children’s marital quality. We added interaction terms for childhood family stress with parent support and strain in Model 2 of Table 5 to predict growth trajectories of marital closeness and strain. We first discuss the results on marital closeness (Model 2, upper half of Table 5). Results in Model 2, Panel A, of Table 5 indicated that, when childhood family stress was high (above the mean level), those with higher baseline levels of strain from mothers had lower levels of marital closeness at baseline, but they experienced a less rapid decline in marital closeness over time. In contrast, when childhood family stress was low (below the mean level), those with higher baseline levels of strain from mother had higher levels of marital closeness at baseline but experienced a more rapid decline in marital closeness over time. Results in Model 2, Panel B, of Table 5 indicated that, when childhood family stress was low, those with higher baseline levels of support from fathers had higher levels of baseline martial closeness, but they experienced a more rapid decline in marital closeness over time. When childhood family stress was high, those with higher baseline levels of support from fathers had lower levels of baseline closeness, but they experienced a slower decline in martial closeness over time. Moreover, an increase in strain from both mothers (Panel A) and fathers (Panel B) (T1 to T2) slowed the decline in marital closeness over time for those with higher childhood family stress but fostered decline in marital closeness for those with lower childhood family stress. There were no significant interactions between support/strain with mothers and/or fathers and childhood stress in predicting adult children’s marital strain levels (Model 2, lower half of Table 5).

Discussion

Marriage and family scholars emphasize that relationships with people outside the marital dyad have significant effects on the marital quality of the dyad (Bryant et al., 2001). Life-course scholars emphasize that adult children tend to remain closely involved with their parents over the life course (Lye, 1996) and that relationships with parents, especially in childhood, have been shown to influence adults’ marital interactions and marital quality (Amato & Booth, 2001). Considering these literatures together, we hypothesized that the quality of relationships with parents would affect the marital quality of adult children. Our results suggest that relationships with parents influence the marital quality of adult children, and that the nature of these effects depend on the gender of the parent, the age, and marital duration of the adult child, and how much family stress the adult child experienced during childhood.

In regard to gender, supportive and strained relationships with both mothers and fathers were positively associated with adult children’s marital closeness and strain (respectively) at the baseline level. Yet our longitudinal findings suggest differences in the ways mothers and fathers influence the marital quality of adult children over time. Although we hypothesized that mothers would have more influence on the marital quality of their adult children, only support from fathers, as well as an increase in support from fathers over time, slowed the decline in marital closeness that tends to occur over time. Research suggests that mothers and fathers play different roles in the lives of their children and that it is normative for mothers to be more involved in their children’s lives throughout the life course (Amato, 1994; Videon, 2005). Because mothers are culturally expected—and more likely—to provide support to adult children (Lye, 1996), receiving anticipated support from mothers may have little effect on adult children’s marital quality. In contrast, because of cultural expectations of fatherhood, support from fathers may be unanticipated (Doherty, Kouneski, & Erickson, 1998; LaRossa, Jaret, Gadgil, & Wynn, 2000). Receiving unanticipated support may foster feelings of positive affect and facilitate closeness between spouses. Alternatively, the type of support measured in this analysis (i.e., availability to talk about problems) may connote different symbolic meanings when received from fathers versus mothers, thus influencing adult children’s marital quality in different ways. We note that, contrary to expectations, relationships with parents affected the marital quality of adult sons and daughters in similar ways. This finding is consistent with research on adult child – parent relationships (Amato & Booth, 2001).

Previous studies have suggested that the marital quality of older individuals or those in longer term marriages may be less reactive to relationships with parents than those who are younger or in the early phases of marriage (Bryant & Conger, 1999). In contrast, our study revealed that the parent – adult child relationship shapes marital quality in different—but not less significant—ways depending on age and marital duration. Support from mothers and fathers was positively associated with baseline levels of marital closeness for young to midlife adult children and for those with shorter marital durations but negatively associated for mid-to late-life adult children and for those with longer marital durations. Upon marriage, marital partners typically become a primary source of support, while the quality of the parent – adult child bond becomes less salient (Sarkisian & Gerstel, 2008). Support from parents as adult children’s age and marital duration increases may reflect problematic circumstances in the adult child’s life that independently influence marital quality. Similarly, it may be the case that parents provide nonnormative support during this life-course stage because of the adult child’s marital problems. Alternatively, nonnormative support from parents to those married longer and those in mid- to late life may be a source of alternative companionship, thus creating conflict between spouses (Felmlee, 2001). Our results in regard to strain with parents suggest that those who are older or married longer may be better able to manage strain from parents in ways that bolster marital closeness and protect against increasing marital strain, whereas younger individuals and those with shorter marital durations have not developed such mechanisms.

Finally, we drew on recent life-course work on cumulative disadvantage to consider whether childhood family stress moderates the effects of adult child – parent relationships on adult children’s marital quality. Previous studies have found that people with more childhood stress were more reactive to stress in adulthood (Dannerfer, 2003) and that marital quality diminished more rapidly in response to stress for adults who experienced high levels of childhood family stress (Umberson, Williams, Powers, Liu et al., 2005). Consistent with these studies, our baseline results revealed that strained ties with mothers were negatively associated with marital closeness only for those with higher levels of childhood stress, whereas supportive ties with fathers were positively associated with marital closeness only for those with lower levels of childhood stress. In contrast, however, our longitudinal findings suggest that adult children who reported higher levels of childhood stress had levels of marital closeness that were less adversely affected by strain with mothers, less adversely affected by increasing levels of strain with mothers and fathers, and more positively affected by support from fathers. Stress researchers suggest that individuals learn to cope with high levels of stress over time (Lazarus & Folkman, 1984). It may be that adult children with a history of stress in relationships with parents have learned how to cope with strain in ways that protect against marital strain. Similarly, this legacy may make those who grew up with high levels of childhood family stress more likely to react positively to what may be unexpected support from fathers. Alternatively, those with higher levels of childhood stress may have different types of relationships with parents. Strained relationships with parents may be less salient—and supportive relationships more salient—to adult children who experienced greater childhood family stress. Adult levels of marital closeness, in turn, may be affected.

Limitations and Considerations for Future Research

We are unable to address several issues in this article that warrant further attention. There are theoretical reasons to expect that other types of support (e.g., instrumental) as well as other aspects of the parent-child relationship (e.g., parent disapproval of a spouse) may affect adult children’s marital quality (Bryant & Conger, 1999). Because of measurement constraints in the ACL data, we were unable to consider other aspects of the parent-child tie in our analysis. Further, although we used growth-curve analysis of longitudinal data, the analysis remained susceptible to some issues of causality. For example, because the same time points were used to partially determine the slope of marital quality, it is possible that the marital relationship partially affected relationships with parents or that other external factors jointly shaped marital relationships and parent – adult child ties. Even with limitations, the current study extends past research by using national, longitudinal data that allow for a consideration of the consequences of the parent – adult child tie on trajectories of change in marital quality over an 8-year period. Our multidimensional approach to measuring parent – adult child relationship quality and marital quality allowed us to develop a more complete picture of the effects of relationships with parents on children’s marital relationships—for better or worse—over time. It is notable that our results revealed more significant findings on marital closeness than on marital strain, and research should further investigate the ways marital closeness and strain have different predictors across the life course. For example, studies have shown that childhood stress, such as a parent divorce, causes problems in later marital relationships (Amato & Booth, 2001). Despite this evidence, few studies have examined how supportive aspects of relationships with parents in childhood might influence positive dimensions of marital quality later in life. Moreover, our attention to the how levels of childhood stress shaped the effect of relationships with parents on marital quality adds to a burgeoning area of research on the complex ways that childhood experiences buffer against, or contribute to, the effects of later life stressors (Dannefer, 2003). Given this evidence, future research should further investigate the ways early childhood and adult relationships with parents remain important in the lives of adult children.