Conjoint trajectories of couples’ marital and parental conflictual behaviors and later-life mental, physical, and relational health

Abstract

Conflictual marital and parental relationships mutually reinforce each other generating family-level stress. The cumulative experiences of stressful family conflictual circumstances (a family-level construct of family conflictual circumstances (FCC), based on marital and parental conflictual behaviors) affect a couple’s well-being. The present study, utilizing longitudinal data of 370 couples in enduring marriages and a person-centered approach, examined: a) the existence of heterogeneous groups of couples with FCC trajectory patterns, b) whether individual and contextual factors are associated with FCC trajectory patterns, and c) differential later-life health and relational consequences of these groups. We identified four heterogeneous groups of couples with distinct FCC trajectory patterns in the early middle years (from 1990 to 1994; approximately age 40 for both husbands and wives). Personal (neurotic vulnerability) and contextual factors (family financial hardship) influenced the development of the FCC trajectories, and FCC trajectory patterns were consequential for spouses’ later mental, physical, and relational health (2001). Two features of the longitudinal synchrony in FCC trajectory patterns (severity and synchrony) were utilized to explain the differential impacts of the trajectory patterns on spouses’ later health and relational outcomes.

Previous studies have found that marital and parental conflictual behaviors influence one another (Cox, et al., 2001; Erel & Burman, 1995; Gao et al., 2019). Previous meta-analytic studies support a “spillover” process, whereby the emotion, affect, and mood created by marital interactions transfers to parent–child relationships and vice versa (Anderson et al., 2010; Doss et al., 2009; Erel & Burman, 1995; Krishnakumar & Buehler, 2000). Also, a daily diary study supported the spillover process, showing lower marital quality predicted lower levels of parent–child relationship quality on the same day (Kouros et al., 2014). However, recent research has argued that significant intrafamilial processes (experiencing conflicts in one family subsystem can lead to conflicts in another family subsystem) were explained by stable differences between families, rather than changes occurring within a family (Mastrotheodoros et al., 2019). Specifically, families with higher interparental conflict compared to other families tended to have higher parent–child conflicts, especially for mothers. However, within families, mothers’ experiences of intense interparental conflicts were unrelated to intense mother-adolescent conflicts (Mastrotheodoros et al., 2020), and interparental conflicts did not induce changes in the quality of parent–adolescent relationship (Mastrotheodoros et al., 2019). Findings about family associations have not been consistent in the literature. Also, previous cross-sectional studies and meta-analytic reviews have focused on the associations between marital and parenting behaviors during specific life-stages (e.g., toddlerhood, see Chester & Blandon, 2016; or adolescence, see Conger et al., 1994), the developmental consequences for children/adolescents (Conger et al., 2010; Taylor et al., 2010) or parent–adolescent relationship quality (Kouros et al., 2014; Mastrotheodoros et al., 2019). Less is known about co-occurring trajectories in marital and parental conflictual behaviors over an extended period, and how they combine to influence parents’ well-being over the life course. In addition, conflictual marital relationships lead to less responsive parenting or escalating negativity in parent–child interactions (Krishnakumar & Buehler, 2000; Ponnet et al., 2013), which form a conflictual family context. This stressful family circumstance (either acutely or chronically) can have various psychological and physical health consequences for husbands and wives (Bryant et al., 2017; Quittner et al., 1990; Wickrama, O’Neal, & Klopack, 2020). Thus, the present study expands existing research by investigating how the co-occurrence of trajectories of marital and parenting conflictual behaviors over the middle years combines to influence couples’ mental, physical, and relational health over the life course.

Middle age is commonly defined as the time between ages 45 and 65, and we specifically focus on the influence of couples’ conflictual behaviors during their early middle years (for 4 years; parents were approximately age 40) on their later health over the following 11 years. We expect there is heterogeneity in co-occurring trajectories of marital and parental conflictual behaviors over the early middle years (conjoint trajectory patterns), as marital and parental conflict behaviors may influence one another over time, and individual resources of couples are various. We posit that these different conjoint trajectory patterns of marital and parental conflictual behaviors may predict later physical and relational health outcomes differently. Taken together, we will investigate (a) the existence of groups of couples with conjoint trajectory patterns of marital and parental conflictual behaviors over the early middle years (1990–1994), (b) the influences of individual and contextual factors on the membership of these groups of couples, and (c) differential later-life health and relational outcomes of these groups of couples over the life course (2001).

Co-occurrence of marital and parental conflictual behaviors trajectories

Family systems theory conceptualizes the family as an emotional unit, describing emotional and behavioral interconnectedness in the family (Cox & Paley, 1997; Papero, 1990). Family members are highly interrelated with interactions in one family subsystem (e.g., marital subsystem) affecting other family subsystems (e.g., parent–child subsystem) (Papero, 1990). Life experiences, spouses’ behaviors, emotions, and incidents occur within the family context, and reciprocal influences exist between them, explaining concurrent and longitudinal associations between partners (Wickrama, O’Neal, & Klopack, 2020; Wickrama, O’Neal, & Lee, 2020). Negative interactions in the marital subunit can affect partners’ behaviors in the parental subunit, and the subunits can mutually reinforce each other creating overall stressful family conflictual circumstances. Past research has found that marital hostility influences overall family interactions (e.g., reduced warmth) and decreased responsiveness toward children (especially for fathers, Stroud et al., 2011). Specifically, past research has highlighted associations between marital discord and parent–child difficulties (Erel & Burman, 1995; Krishnakumar & Buehler, 2000; Sturge-Apple et al., 2006). That is, parents’ consistent experiences of marital discord undermine their ability to effectively parent their children (Cowan et al., 2007; Erel & Burman, 1995) (e.g., being less attentive and decreased positive parenting behaviors, Grych, 2002; Sturge-Apple et al., 2006). Fathers’ destructive conflict behaviors in the marital relationship led to mothers’ negative reactions toward children’s negative emotions, and mothers’ marital conflictual behaviors were associated with their unsupportive parenting behaviors (Gao et al., 2019). For families with more destructive interparental conflicts, their adolescents tend to experience low parental support, to feel sadness and anger, and to have higher parent–adolescent negative interactions (Cui & Conger, 2008; Mastrotheodoros et al., 2019).

Consistent with the notion of circular causality in relationships, the distress from parental difficulties can also transfer to marital quality, influencing marital functioning (Lavee et al., 1996; Fincham & Hall, 2005). For both husbands and wives, parenting stress negatively affects their perceived marital quality (Lavee et al., 1996). These consistent processes between family subsystems can generate family-level stress, affecting family well-being. In the present study, we expect that couples’ marital and parental conflictual behavior trajectories are mutually linked in a longitudinal context, resulting in conjoint trajectory patterns. These trajectory patterns reflect family conflictual circumstances (FCC) in which couples experience family-level stress across multiple relationships.

Heterogeneity in family conflictual circumstances trajectories

According to family systems theory, families strive for a sense of balance when facing life challenges with different patterns emerging (Cox & Paley, 1997; Papero, 1990). Families adjust to stressful situations differently depending on available resources and various contextual factors, which include diverse families with different patterns of stressful family interactions for a short or long time. Specifically, some families may return to more positive marital and family interactions in a short term, while others maintain negative interactions or continue to decline. Also, some couples that have highly conflictual marital relationships may have little conflict with their children and vice versa. For some couples, as levels of parenting stress increase, they may experience more intense marital conflicts at home. However, others may experience high levels of parent–children conflict initially but develop constructive ways to handle conflict over time. Kamp Dush and Taylor (2012) also found that frequency of marital conflict is not stably low or high over time, and they identified three distinct trajectory groups for marital conflict (high, middle, and low) over 20 years. All three groups were overall stable over the first 12 years and then tapered slightly in the final 8 years. Specifically, the high marital conflict group (23% of the sample) had slight upside-down U-shape meaning that conflict gradually increased at the initial 8 years and decreased across the final 12 years. The low marital conflict group (17%) had stable low conflict, and for the middle marital conflict group, (61%, right above the mean for marital conflict), the levels of marital conflict were largely stable with a slight decline over the later years of the study.

Married couples’ behavioral trajectories are deeply affected by various factors, including personal characteristics and life course experiences (Bryant et al., 2017; Karney & Bradbury, 1995). Anderson and colleagues (2010) identified five different marital happiness trajectory groups of couples with distinct trajectory shapes (i.e., high and stable levels of happiness, continuously low happiness, declining happiness, and a curvilinear pattern of increased happiness, decline, and recovery) over the study period. These distinct trajectory groups were differentially associated with individuals’ economic hardship. Similarly, among low-income young mothers, researchers identified groups of mothers with distinct parenting stress trajectories (Chang & Fine, 2007). Mothers’ personal resources (individual factor) and their low-income family circumstances (contextual factor) were related to their experiences of chronically high or decreasing parenting stress. Thus, we posit that the existence of heterogeneity in couples’ FCC trajectories.

Individual and contextual factors and multiple family relationships

Karney and Bradbury (1995) argued that individuals bring their own personal resources to marriage, creating variation in marital adaptation in terms of behavioral interactions and relational outcomes. Neurotic tendencies are relatively stable tendencies to respond to threat or frustration with negative emotions. Neuroticism has multiple facets that are correlated with one another including anger, anxiety, vulnerability, and hostility (Costa & McCrae, 1992). Individuals who are high on neuroticism have negative emotional responses to challenges frequently (McCrae & Costa, 2003) and tend to be self-critical and sensitive to other’s criticisms (Watson, Clark, & Harkness, 1994). Also, trait neuroticism has been associated with individuals’ emotional instability (e.g., being easily upset and agitated by stressful situations, Costa & McCrae, 1992), which can foster disruptive or hostile behaviors in family relationships. Neurotic tendencies in one spouse can trigger or amplify similar trends in their partner, and this reciprocal couple process can form patterns of hostile marital interactions over time. As this neurotic vulnerability seems to be closely related to ways how individuals respond to stressful situations, especially emotional and relational stress responses, the present study focuses on neuroticism as an individual factor. In addition, previous research has shown that parents’ hostility in the home (in the context of marital relationships) is associated with more aggressive or hostile behaviors displayed by their children (Stover et al., 2016). A child’s aggressive and disobedient behaviors may erode their parents’ emotional capabilities, and parents may negatively perceive their child’s aggressive behaviors generating stressful FCC (Bryant et al., 2017). Thus, we expect that husbands’ and wives’ neurotic vulnerabilities (individual factors) are associated with the memberships of distinct groups of couples with FCC trajectories over the middle years.

From a life course perspective, specific family or life experiences, such as economic hardship (contextual factor), may have persistent impacts on marital interactions and family well-being (Conger et al., 2010; Wickrama, O’Neal, & Lee, 2020). Stressful family economic conditions often provoke anxiety and increase tension between spouses, which decreases spouses’ abilities to regulate emotional reactions and solve relationship difficulties (e.g., difficulty communicating in couples and discordant interactions between parents and children). This ultimately results in destructive family interactions (Buck & Neff, 2012). Thus, resource scarcity (e.g., personality and neuroticism) and stressful family experience may be associated with distinct FCC trajectories.

Characteristics of family conflictual circumstances trajectories

We posit that there are two essential features of these family trajectory patterns: the severity (the average level) and longitudinal synchrony. The average level reflects the severity of co-occurring conflictual behaviors. Longitudinal synchrony reflects how these marital and parental conflictual behaviors are closely moving together in a similar manner (shape) over the study period (i.e., the degree of FCC resemblance of marital and parental conflictual behaviors trajectories over time). This longitudinal resemblance reflects how parallel or similar couples’ marital and parental conflictual behavior trajectories are. We propose that the group-specific longitudinal synchrony in FCC can be assessed by the similarity (or discrepancy) in mean slope (i.e., rate of change) between marital and parental conflictual behavior trajectories and by the similarity (or difference) in the levels of trajectories (i.e., couples have similar severity and couples change together).

We expect that these features of conjoint conflictual trajectories may have implications for health and relationship outcomes in later years. Specifically, we argued that high longitudinal synchrony may amplify the effect of severity for better or worse, affecting physical and relational outcomes in later years.

Couple family conflictual circumstances trajectory patterns and spouses’ health and well-being outcomes

Chronic stressful marital and parental experiences cause psychological, behavioral, and physiological arousal, which can wear down individuals’ regulatory systems leading to increased vulnerability to later physical health problems (Lee et al., 2020; McEwen & Gianaros, 2010). Chronic stressful marital and parent–child relationships have implications for subsequent depressive symptoms of both husbands and wives in later years especially for couples with low marital integration (Bryant et al., 2017). Also, marital and parent–child relationship stress (e.g., spousal and parental roles) may influence each other, proliferating across the life course, which could be consequential for couples’ later psychological distress (Wickrama, O’Neal, & Klopack, 2020).

Family stress can activate the hypothalamic–pituitary–adrenal (HPA) axis in stressful circumstances, leading to increased cardiovascular reactivity and decreased immune functioning (McEwen & Gianaros, 2010; Kiecolt-Glaser et al., 2002, 2005). With repeated exposure to stressors, this activation becomes sustained over time, which results in the dysregulation of multiple body systems. Cumulative physiological dysregulation is associated with low energy expenditure, elevated BMI, and premature aging (McEwen & Gianaros, 2010). Past research has also emphasized the salient role of psychological distress (e.g., depressive symptoms, Lee et al., 2021 in linking sustained stressful marital interactions and physical health outcomes. In addition to being a severe mental health condition, depressive symptoms have been linked to subsequent health risks, including obesity, diabetes, inflammation, and cardiovascular and autoimmune diseases (Goldston & Baillie, 2008; Kiecolt-Glaser & Glaser, 2002; Kiecolt-Glaser et al., 2015). We posit that chronic stressful family circumstances in middle years (i.e., FCC) have detrimental health impacts and negative influences on both spouses’ marital satisfaction over the years. Further, this influence may be moderated by the longitudinal synchrony of FCC for better or worse.

The current study

As previously discussed, group-specific patterns of FCC can be distinct based on the two features: severity and longitudinal synchrony, leading to differential health and relational outcomes. We expect that relative to other groups, couples with strong longitudinal synchrony and high severity of FCC will have the most detrimental health and relational outcomes as high FCC severity may be amplified by the strong and close linkage between marital and parental conflictual behaviors. This amplified FCC reflects a stress proliferation process in multiple family relationships, leading to detrimental health and relational outcomes. Conversely, couples with strong longitudinal synchrony and low severity of FCC likely experience the best health outcomes compared to other groups of couples because the benefit of low level (severity) of FCC is intensified by the strong synchrony of the behaviors trajectories, potentially reflecting their shared constructive conflict management abilities. Other potential FCC groups with various features may emerge (e.g., moderate level of severity and strong synchrony and high level of severity and weak synchrony), and these groups may have varying health and relational outcomes in later years.

Thus, in the current study, we hypothesize:

1. Heterogeneity exists in conjoint trajectories of couples’ marital and parental conflictual behaviors (FCC—Family Conflictual Circumstance) over the middle years (1990–1994).

2. Individual (i.e., neurotic vulnerability) and family contextual (i.e., family economic problems) factors are associated with conjoint trajectory classes of FCC (1990–1994).

3. These FCC classes are associated with husbands’ and wives’ mental and physical health and relationship outcomes (marital satisfaction) in the later years (2001). The unique features of distinct groups of FCC (severity and longitudinal synchrony) may create variance in detrimental health influences of FCC as high or low severity can be amplified by strong longitudinal synchrony of FCC.

Method

Participants and procedures

To evaluate these hypotheses, we used longitudinal data from 370 heterosexual married couples who initially participated in the Iowa Youth and Family Project (IYFP) between 1989 and 1994 and continued to participate in the Iowa Midlife Transition Project (MTP) in 2001. The 370 married couples in this study (80% of the initial 459 couples in IYFP) were those who participated in multiple waves of data collection (in 1990, 1991, 1992, 1994, and 2001) and remained married throughout the study period. Comparisons between our final sample (n = 370) and those excluded (n = 89; some families were no longer able to participate due to relocation, divorce, or discontinued participation in the IYFP/MTP) revealed no significant differences in respondent characteristics at initial data collection time (e.g., years of education and annual family income). On average, the couples had been married for 17 years in 1990 and had at least one child who was a seventh grader in 1989 (Conger & Elder, 1994). The couples were in their early middle years in 1990; the ages ranged from 31 to 68 years for husbands (M = 41) and from 29 to 53 years for wives (M = 39). Most of the couples had entered late midlife in 2001, when the average ages of the husbands and the wives were 52 (range 43–80) and 50 (range 41–65) years, respectively. These are all White, heterosexual couples, and mostly living in rural communities in Iowa. Specifically, in 1989, 34% of the families lived on a farm, 54% of the families lived in small towns (approximately 5000 people or less), and 12% of the families resided in rural areas but not on a farm. In 1989, the median annual family income was $33,240, and 78% of the women were employed (clerical, service worker, and officials), and 96% of the men were employed (farmers, craftsmen, professional, and manager). Husbands and wives had a median of 13 years of education.

Measures

Neurotic vulnerability (NV): The NEO-PI personality inventory was utilized to measure neurotic vulnerability (Costa & McCrae, 1992), one facet of the Neuroticism domain. In 1990, husbands and wives responded to eight items with responses ranging from 1 (strongly agree) to 7 (strongly disagree), including “It’s often hard for me to make up my mind,” “I feel I am capable of coping with most of my problems.” At each time point, husbands’ and wives’ responses were reversed where appropriate and averaged with higher scores representing a higher level of neurotic vulnerability (Cronbach’s αs were .76 and .77 for husbands and wives, respectively.

Family financial hardship (FFH) was assessed using a list of economic problems adapted from Dohrenwend et al. (1978). In 1990, husbands and wives were asked to respond to 22 items based on the question, “In the last 12 months, has your family made any of the following adjustments because of financial need?” The list of financial difficulties included items such as “used savings to meet daily living expenses,” “changed residence to save money,” and “fallen behind in paying bills.” Husbands’ and wives’ “yes” responses (1 = yes, 0 = no) were summed to assess family-level financial strain, with higher scores indicating more family financial hardship.

Couples’ marital conflictual behaviors (MCB): In 1990, 1991, 1992, and 1994, husbands and wives responded to eight items (Conger, 1988) on a 7-point Likert scale from 1 (always) to 7 (never), indicating how frequently their spouses exhibit specific behaviors when they have a problem to solve (i.e., destructive problem-solving behaviors). Sample items include “criticize you or your ideas for solving the problem?” and “ignore the problem.” At each time point, husbands’ and wives’ responses were reversed where necessary and averaged within a spouse, with higher scores representing a higher level of marital conflictual behaviors. Cronbach’s αs ranged from .89 to .94 for husbands and wives across the years. Then, husbands’ and wives’ marital conflictual behaviors were averaged together to create couples’ marital conflictual behaviors at each time point.

Couples’ parental conflictual behaviors (PCB): In 1990, 1991, 1992, and 1994, husbands and wives responded to eight items (Conger, 1988) on a 7-point Likert scale from 1 (always) to 7 (never), indicating how frequently their child (who was in the seventh grade when the study began in 1989) exhibited specific behaviors when each parent needed to solve a problem with the child (e.g., destructive problem-solving behaviors). Sample items included “refuse, even after discussion, to work out a solution to the problem?” As with MCB, responses were reversed where necessary and averaged within a spouse, with higher scores representing a higher level of parent–child conflict that husbands and wives’ perceived. Cronbach’s αs ranged from .87 to .91 for husbands and wives across the years. Then, husbands’ and wives’ parental conflictual behaviors were averaged to create couples’ parental conflictual behaviors at each time point.

Marital relationship satisfaction: In 2001, husbands and wives indicated a 5-point Likert scale ranging from 1 (completely satisfied) to 5 (not at all satisfied); how satisfied are you, all things considered, with your relationship. Then, husbands’ and wives’ responses were reversed, with higher scores representing a higher level of marital satisfaction.

Depressive symptoms: Thirteen items from the Symptom Checklist (SCL-90-R) (Derogatis, 1996) were utilized to measure self-reported ratings of depressive symptoms for husbands and wives in 2001. Sample items included “feeling down,” “feeling hopeless about the future,” “crying easily,” and “feeling of worthlessness.” Participants were asked their distress level on a 5-point Likert scale ranging from 1 (not at all) to 5 (extremely). Responses were averaged, with higher scores indicating more depressive symptoms. Cronbach’s α was .88 both for husbands and wives.

Physical impairment: In 2001, the severity of physical impairment was measured by a list of 10 items adapted from the Rand Health Science Program in Health Survey 1.0 (Hays et al., 1993). Husbands and wives were asked to indicate on a 3-point scale from 1 (not limited at all) to 3 (yes, limited a lot) how much their health condition or a memory problem interferes with their daily activities (e.g., dressing and showering). This scale captures physical function impairment, especially for vigorous (e.g., lifting heavy objects) and moderate (e.g., pushing a vacuum) daily activities, with higher scores representing a higher level of physical impairment. Cronbach’s α was .88 and .91 for husbands and wives, respectively.

Statistical analyses

Using Mplus software (version 8; Muthén & Muthén, 2017), we first investigated the stability and change in couples’ marital and parental conflictual behaviors (MCB and PCB) by estimating latent growth curves (LGCs) over 4 years (from 1990 to 1994). The comparative fit index (CFI; acceptable fit > .90; Little, 2013) and root mean square error of approximation (RMSEA; acceptable fit <.08; Little, 2013) were utilized to assess model fit. Full information maximum likelihood estimation (FIML estimation) was used to handle missing values as this approach produces deterministic results and improves the accuracy of the analysis (Enders, 2010).

Second, utilizing growth mixture modeling (Muthén et al., 2003; Wickrama et al., 2016), latent classes of family conflictual context (FCC) trajectory patterns were identified based on growth parameters of couples’ marital and parental conflictual behaviors. With this analytic approach, trajectory class membership is inferred from the data empirically based on posterior class membership probability. Four growth parameters, the initial levels and slopes for couples’ marital and parental conflictual behaviors trajectories, are simultaneously used to identify latent classes of family-level conflictual circumstances. These latent classes of FCC trajectory patterns capture couples’ severity and synchrony of MCB and PCB. The Bayesian Information Criterion (BIC; preferably lower values), adjusted Lo–Mendell–Rubin likelihood ratio test (LMR–LRT), and entropy values were used to assess the model fit (acceptable fit > .80; Wickrama et al., 2016). For the LMR–LRT, a significant p value indicates that class K model has better fit compared to a class K-1 model, and entropy is a standardized index of model-based classification accuracy reflecting the accuracy of class membership assignment (higher values generally indicating the precise assignment of individuals to latent class membership) (Wickrama et al., 2016).

The FCC trajectory patterns were characterized by the two dimensions, severity and longitudinal synchrony. We calculated severity by taking the average level of couple marital and parental conflictual behaviors at the middle time point. Two indicators of longitudinal synchrony (long-term similarity) were calculated: (1) the degree of closeness by first calculating of the difference in the levels of marital and parental conflictual behaviors at the mid-point (DL) and then taking the reciprocal (1/DL) (i.e., 1/ the difference in the level between marital and parental conflictual behaviors) and (2) the degree of shape similarity by first calculating the difference in slopes (DS), then taking the reciprocal (1/DS) (i.e., 1/ the difference in slopes between marital and parental conflictual behaviors). To assess overall longitudinal synchrony, the product term of closeness and shape/slope similarity was used because we expect the influence of closeness may be amplified by the degree of shape synchrony (i.e., interaction). Higher scores of the product term represent overall high synchrony and smaller scores represent low overall synchrony in couples’ FCC trajectories over the years.

Third, to examine the differential effects of individual (neurotic vulnerability) and contextual (family financial hardship) factors on FCC trajectories, logistic regression analyses were performed by specifying the neurotic vulnerability of husbands and wives (1990) and family financial hardship (1990) as independent variables with the derived class membership (obtained in the step 2) as dependent variables.

Last, we investigated the association between latent classes of FCC and husbands’ and wives’ subsequent mental/ physical health and relationship outcomes in 2001 (i.e., depression, physical impairment, and marital relationship satisfaction) controlling for the autoregressive effect of depression, global health, and marital satisfaction. To consider the interdependence between husbands and wives into consideration, we utilized the dyadic structural equation modeling. Health and relationship outcomes in 2001 were regressed on FCC group dichotomies using the lowest FCC risk group (lowest FCC level and strong synchrony) as the reference group. This approach provides information about the unique influence of the FCC trajectory group on the mental and physical health and relationship outcomes in 2001, compared to the reference group.

Results

Estimating couple marital and parental conflictual behavior trajectories

The descriptive statistics and correlations for couples’ marital and parental conflictual behaviors, neurotic vulnerability, family financial hardship, mental and physical health indicators, and marital satisfaction are shown in Table 1. With four repeated measures (1990, 1991, 1992, and 1994), we also checked if there were quadratic patterns of couples’ marital and parental conflictual behaviors trajectories (CFI/RMSEA = 1/.00). We found that no significant means and variances of the latent quadratic growth factors in both marital and parental conflictual behaviors, reflecting that the changes in these behaviors of couples did not follow quadratic patterns of the trajectories. Therefore, the linear growth trajectories for couples’ marital and parental conflictual behaviors were selected as the optimal trajectory model. As shown in Table 2, models estimating linear growth had an acceptable fit with the data for couples’ marital and parental conflictual behaviors (CFI/RMSEA for marital and parental conflictual behaviors growth models = 1/.05 and 1/.00, respectively). The initial levels of couples’ marital and parental conflictual behaviors were 2.27, p < .001 and 2.66, p < .001, respectively. The mean rate of change in conflictual marital behaviors was significant (.04, p < .001), whereas the mean rate of change in parental conflict behaviors was not statistically significant. The variances of all four growth parameters were significant (p < .01).

Table 1.

Correlation matrix and descriptive statistics for study variables (N = 370 married couples).

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
Marital conflictual behaviors
1. 1990	—
2. 1991	.77**	—
3. 1992	.75**	.79**	—
4. 1994	.63**	.58**	.56**	—
Parental conflictual behaviors
5. 1990	.35**	.33**	.31**	.28**	—
6. 1991	.37**	.33**	.34**	.32**	.74**	—
7. 1992	.36**	.35**	.39**	.35**	.71**	.81**	—
8. 1994	.29**	.26**	.31**	.39**	.57**	.67**	.68**
Neurotic vulnerability
9. Husband 1990	.12**	.10	.08	.04	.08	.02	−.03	−.05	—
10. Wife 1990	.27**	.23**	.19**	.27**	.18**	.16**	.25**	.20**	.00	—
Family financial hardship
11. 1990	.19**	.17**	.17**	.06	.06	.08	.04	.06	.14*	.11*	—
Depressive symptoms
12. Husband 2001	.20**	.21**	.16**	.17**	.19**	.17**	.12*	.11*	.23**	.15**	.23**	—
13. Wife 2001	.12*	.16**	.14**	.13*	.13*	.14**	.19**	.17**	.01	.25**	.15**	.17**	—
Physical impairment
14. Husband 2001	.06	.03	.04	.07	.08	.03	.02	−.01	.13*	.01	.12*	.25**	.03	—
15. Wife 2001	.02	.06	.03	.05	.05	.04	.14**	.13*	−.07	.05	.10	−.01	.16**	.13*	—
Marital satisfaction
16. Husband 2001	−.34**	−.32**	−.24**	−.29**	−.16**	−.13*	−.11*	−.05	−.06	−.08	−.11	−.18**	−.17**	−.02	.05	—
17. Wife 2001	−.24**	−.23**	−.22**	−.35**	−.06	−.10	−.16**	−.14*	−.09	−.16**	−.10	−.10	−.25**	−.00	−.08	.35**	—
M (SD)	2.27 (.73)	2.35 (.85)	2.39 (.91)	2.40 (.76)	2.67 (.68)	2.68 (.73)	2.65 (.73)	2.65 (.79)	2.28a (.53)	2.45a (.46)	9.10 (7.62)	1.43a (.44)	l.53a (.50)	1.24a (.33)	l.30a (.39)	4.19a (.77)	4.02a (.76)

Note: Means (M) and standard deviations (SD) are presented at the bottom of the table. Letter superscripts on mean values denote significant gender differences.

^*p < .05.

^**p < .01.

Table 2.

Results from univariate growth curves of couples’ marital and parental conflictual behaviors (n = 370 married couples).

	Initial level		Rate of change		RMSEA/CFI
	Mean	Variance	Mean	Variance
Couple
Marital conflictual behaviors (1990, 91, 92, 94)	2.27 ***	.61 ***	.04 ***	.02 **	.05/1.00
Parental conflictual behaviors (1990, 91, 92, 94)	2.66 ***	.39 ***	−.00	.02 ***	.00/1.00

Note: RMSEA = root mean square error of approximation; CFI = comparative fit index. Significant coefficients shown in bold.

^**p < .01.

^***p < .001.

Examining distinct groups of couples with similar family conflictual context (FCC) trajectory patterns

Utilizing Growth Mixture Modeling (GMM), we estimated FCC trajectory patterns using couples’ marital and parental conflictual behaviors growth factors to identify models with two to six classes. The fit indices are presented in Table 3. The 4-class estimation produced the best fitting model, as indicated by Adj. LMR–LRT (value = 111.73, p = .21), an acceptable entropy value (value = .83), and group size. To identify valid groups in the study population, group sizes should be appropriately large for each group (at least 5% of the sample; Wickrama et al., 2016).

Table 3.

Model fit indices for family conflictual context latent class trajectories (n = 370 married couples).

Fit statistics	Family conflictual context class trajectory classes				6 classes
	2 classes	3 classes	4 classes	5 classes
# of free parameters	23	28	33	38	43
BIC	4815.940	4731.056	4644.507	4582.237	4547.517
SSABIC	4742.983	4642.239	4539.830	4461.700	4411.119
Entropy	0.823	0.805	0.828	0.826	0.850
Adj. LMR-LRT (p value)	372.730 (.003)	110.115 (0.11)	111.726 (0.21)	88.255 (0.27)	121.383 (0.11)
Group size (n, %)
Class 1	121 (36.4%)	75 (22.6%)	80 (24.1%)	48 (14.5%)	24 (7.5%)
Class 2	211 (63.6%)	182 (54.8%)	158 (47.6%)	31 (9.3%)	127 (38.3%)
Class 3		75 (22.6%)	58 (17.5%)	74 (22.3%)	56 (16.9%)
Class 4			36 (10.8%)	128 (38.6%)	48 (14.5%)
Class 5				51 (15.4%)	3 (0.9%)
Class 6					73 (22.0%)

Note: BIC = Bayesian information criteria. SSABIC = sample size adjusted BIC. Adj. LMR–LRT = adjusted Lo–Mendell–Rubin log-likelihood ratio test. Bold and italic values represent the optimal class model.

Figure 1 depicts FCC trajectory patterns for each of the four groups. Severity and overall longitudinal synchrony scores across groups are presented in Table 4. Group A (see Panel a in Figure 1) was comprised of 49 couples (13.3% of the sample). Couples in this group had the lowest average level of FCC (severity, 1.67) compared to other groups and had a high degree of stability in marital/ parental conflictual behaviors. The FCC severity was computed using the middle time point (in 1992), consistent with our emphasis on longitudinal context. The synchrony indicators, closeness and shape/slope similarity, for this group were 1.79 (1/DL [.56] = 1.79) and 25 (1/DC [.04] = 25), respectively. The product term, as shown in Table 4, indicating overall synchrony was 44.75 (i.e., 1.79 × 25). This group showed the lowest severity (average level) and high longitudinal synchrony among the four groups, and we labeled this group “lowest FCC severity and high synchrony.”

Figure 1.

Estimated mean trajectories of marital and parental conflictual behaviors for couples over 4 years (n = 370 married couples). Note. b= unstandardized coefficients. SE= standard error. * p < .05. * p < .01. *** p < .001.

Table 4.

Severity and overall longitudinal synchrony scores across groups (n = 370 married couples).

	FCC Group A	FCC Group B	FCC Group C	FCC Group D
	(49 couples, 13.3%)	(165 couples, 44.6%)	(95 couples, 25.7%)	(61 couples, 16.5%)
	Lowest FCC severity	Moderate FCC severity	Moderate–high FCC severity	Highest FCC severity
	High synchrony	Moderate synchrony	Moderate synchrony	Highest synchrony
FCC severity	1.67	2.26	2.91	3.28
Closeness synchrony	1.79	2.44	1.37	9.09
Shape/slope synchrony	25	14.29	25	14.29
Overall synchrony	44.75	34.87	34.25	129.90
(Closeness synchrony × shape synchrony)

Note: FCC = family conflictual circumstances.

The second group, Group B (see Panel b in Figure 1), was comprised of 165 couples (44.6% of the sample) with a moderate level of FCC (2.26). This group had a moderate increase in marital conflictual behaviors. In this group, the degree of closeness was 2.44 (1/DL [.41] = 2.44) and the shape/slope similarity was 14.29 (1/DC [.07] = 14.29). The product term indicating overall synchrony was 34.87 (i.e., 2.44 × 14.29). The overall synchrony value was smaller than Group A, and the severity (average level) was higher than Group A, but smaller than those of Groups C and D. Thus, we labeled this group “moderate FCC severity and moderate synchrony.”

A third group (Group C, see Panel c in Figure 1) was characterized by couples who had a high average level of FCC over the study period (2.91 at the middle time point; 95 couples, 25.7% of the sample) and a high increase in marital conflictual behaviors. In this group, the synchrony indicators were 1.37 (1/DL [.73] = 1.37) and 25 (1/DC [.04] = 25) for the closeness and shape similarity, respectively. As shown in Table 4, the product term indicating overall synchrony was 34.25 (1.37 × 25), similar to Group B. The severity of this group was higher than Groups A and B, thus, we labeled this group “moderate-high FCC severity and moderate synchrony.”

Group D (see Panel d in Figure 1) was characterized by the highest FCC severity over the study period (3.28 at the middle time point; 61 couples, 16.5% of the sample) and stability in marital/ parental conflictual behaviors. In this group, the synchrony indicators for the closeness and shape similarity were 9.09 (1/DL [.11] = 9.09) and 14.29 (1/DC [.07] = 14.29), respectively. As shown in Table 4, overall synchrony was 129.90 (i.e., 9.09 × 14.29). This group had the highest overall synchrony among the groups. Thus, we labeled this group “highest FCC severity and highest synchrony.”

Husbands’ and wives’ neurotic vulnerability and family financial hardship influences

We investigated the differential influences of husbands’ and wives’ neurotic vulnerability and family financial hardship on group membership by utilizing logistic regression analyses. The reference group was Group A with the lowest FCC severity and high synchrony, which we expected to be the most protected group. Wives with higher levels of neurotic vulnerability were more likely to be in Groups B, C, and D, compared to the Group A over the middle years. However, there are no significant differential effects in husbands’ neurotic vulnerability across the groups. For family financial hardship, couples experienced greater family financial hardship were more likely to be in Group D with highest severity and high synchrony in conjoint FCC trajectories, compared to those with less family financial hardship (see corresponding ORs in Table 5).

Table 5.

Association between neurotic vulnerability of husbands and wives and family financial hardship and FCC class membership (n = 370 married couples).

	FCC Group B		FCC Group C		FCC Group D
	Moderate FCC severity		Moderate–high FCC severity		Highest FCC severity
	Moderate synchrony		Moderate synchrony		Highest synchrony
	OR	95% CI	OR	95% CI	OR	95% CI
Husbands’ neurotic vulnerability	1.63	.69, 3.82	1.81	.73, 4.45	2.51	1.00, 6.29
Wives’ neurotic vulnerability	5.02 ***	2.15, 11.70	5.84 ***	2.37, 14.41	9.20 ***	3.47, 24.41
Family financial hardship	.99	.95, 1.04	1.02	.97, 1.07	1.06 *	1.00, 1.12

Note: OR = odd ratio. FCC = family conflictual circumstances. Group A (reference group) = lowest FCC severity and high synchrony. Significant coefficients shown in bold.

^*p < .05.

^**p < .01.

^***p < .001.

Differential mental, physical, and relational outcomes of family conflictual circumstances trajectory groups

Figure 2 presents the SEM results using couple FCC trajectory group to predict husbands’ and wives’ depressive symptoms, physical impairment, and marital satisfaction in 2001. The reference group was Group A (lowest FCC severity and high synchrony). Membership in Group B (moderate FCC severity and moderate synchrony), compared to Group A, was associated with higher levels of physical impairment for wives and lower levels of marital satisfaction for both wives and husbands in 2001. Membership in Group C (moderate-high FCC severity and moderate synchrony), compared to Group A, was associated with a higher level of depressive symptoms for wives and husbands, higher levels of physical impairment only for wives, and lower levels of marital satisfaction for wives in 2001. Membership in Group D (highest FCC severity and highest synchrony), compared to Group A, showed the association with more depressive symptoms for husbands and lower levels of marital satisfaction for both husbands and wives in 2001, suggesting the detrimental health influence of high severity is amplified by strong synchrony.

Figure 2.

Effects of family conflictual circumstances trajectory patterns on health and relationship outcomes (n = 370 married couples). Significant paths (standardized coefficients) are shown in the figure. Group A = lowest FCC severity and high synchrony, Group B = moderate FCC severity and moderate synchrony, Group C = moderate-high FCC severity and moderate synchrony, and Group D = highest FCC severity and highest synchrony. X² (df) = 69.29 (17), p = .0001, comparative fit index (CFI) = .90, RMSEA = .06. H. = husband, W = Wives, H. Ph. Impairment = husbands physical impairment. W. Ph. Impairment = wives physical impairment. FCC = family conflictual circumstances. * p < .05, ** p <.01, *** p <.001.

In summary, Group A showed the most favorable health consequences compared to Groups B, C, and D, suggesting a protective effect of lowest severity was amplified by strong synchrony. It seems this group has the least mental and physical health risk as well as most positive relationship outcomes in later years. Thus, the results provided evidence that synchrony moderates the health and relational influence of FCC severity for better or worse, especially in the longitudinal context. Also, it seems that moderating influence of synchrony on the health effect of severity is more profound at low level of severity.

Discussion

Past research has highlighted how conflict in the marital subunit and the parental subunit of the overall family unit can mutually reinforce each other, leading to detrimental physical, mental, and relational health outcomes. However, much of this research has been cross-sectional, has tended to focus on either one of them separately, or has examined child, rather than parental, outcomes (Conger et al., 2010; Gao et al., 2019; Krishnakumar & Buehler, 2000). The current study expands on past research by utilizing longitudinal data from husbands and wives in long-term marriages with sufficient follow-up to assess health and relational consequences. We utilized GMM to assess heterogeneous groups of parents who cluster around different family conflict trajectories unobserved in population-level data. Additionally, we examined personal (neuroticism) and contextual (family financial hardship) factors that were hypothesized to affect the family FCC.

Consistent with family systems theory (Papero, 1990), the results supported the existence of a family-level construct of couple FCC trajectory patterns and indicated that MCB and PCB form highly variable latent trajectories of FCC for couples. Respondents tended to cluster into four classes based on these latent trajectories. These classes generally displayed a high degree of synchrony between MCB and PCB, suggesting that conflict in these family subunits is mutually reinforcing. This synchrony is consistent with past research that showed the overlap between parental and marital conflicts (Krishnakumar & Buehler, 2000; Ponnet et al., 2013). Our finding suggests that researchers should consider the whole family unit rather than focusing on a single subunit, as these subunits are highly interconnected and affect one another. Additionally, this finding suggests that practitioners should take conflict in multiple domains into account. Interventions designed to reduce conflicts and improve physical, mental, and relationship health need to address the conflict across family subunits.

Neurotic vulnerability was a strong predictor of class membership for wives, but not husbands. Consistent with previous research (Lynn & Martin, 1997), wives are higher in neurotic vulnerability than husbands in our study sample. Previous research showed that masculinity is negatively associated with neuroticism (Marusic & Bratko, 1998). Masculinity represents traditional gendered social expectations associated with strength, aggression, and dominance. Masculine qualities were strongly expected for men in the rural communities in the 1980s (Campbell et al., 2006), which possibly influenced on the husbands’ overall lower neurotic vulnerability and lack of significant differences in husbands’ neurotic vulnerability across groups. However, for women, it seems that neurotic vulnerability can be an important risk factor for conflictual family environments. That is, women with high neurotic vulnerability may experience more distress from conflictual relationships with spouses and children. It is possible that neurotic tendencies provoke intense emotional reactivity to negative interactions in multiple family relationships, which may lead to subsequent conflictual behaviors and form conflictual family context. Individuals’ neurotic vulnerability seemed to not only influence interactions in the multiple family relationships but also intensify emotional stress reactivity, affecting wives’ feelings of distress. This is consistent with previous research and provides evidence for the essential role of individuals’ personal resources in establishing couples’ behaviors in marital (Karney & Bradbury, 1995; Wickrama et al., 2018) and parent–child relationships. Cumulative experiences of stressful FCC led to more depressive symptoms, more severe physical impairment, and lower marital satisfaction. Wives in Group D had the highest level of neurotic vulnerability, indicating these wives were the most vulnerable for mental, physical, and relational health consequences over the years.

For husbands, neurotic vulnerability did not seem to contribute to membership in FCC classes. This might be explained by our study samples’ specific characteristics, such as a shared similar regional, social, and economic background. Previous research found that couple-level neurotic vulnerability (the combination of husbands’ and wives’ neurotic vulnerability) was also associated with marital hostility, emphasizing couple-level vulnerability may create a context that influences marital interactions (e.g., increased marital hostility) and development of depressive symptoms for both husbands and wives (Wickrama et al., 2018). Future research needs to extend further what other potential individual and couple-level resources (reflecting couples’ characteristics), or vulnerabilities may affect couples’ marital and parental behaviors, using more diverse and extensive population. Similarly, family financial hardship was a risk factor for membership in the high severity and high synchrony class (Group D). This result is consistent with past research that has found that financial stress is a significant risk factor for family conflict (Wickrama & O’Neal, 2019). These findings highlight the importance of personal and family-level contextual factors in the development of the FCC. Interventions seeking to protect families against conflict might target financial stress and neuroticism.

We hypothesized that the identified FCC groups have different trajectory patterns that refer to longitudinal synchrony characterized by two features: severity (the average FCC level at the mid-point) and synchrony (the similarity between marital and parental conflictual behavior trajectories). We assessed the degree of longitudinal synchrony presented in each FCC group using these two indicators, focusing on (a) how close the mean levels of trajectories were and (b) how parallel their MCB and PCB rates of change were over time. All four identified groups showed considerable levels of longitudinal synchrony, supporting the notions of spillover and circular causality in family relationships (e.g., MCB and PCB are reciprocally related). However, the four identified groups had FCC trajectory patterns with different levels (at the middle time point). The groups with the lowest and highest FCC severity also had high overall synchrony, suggesting that the effects of severity of FCC were amplified by the degree of synchrony. As hypothesized, Group A (lowest FCC severity and high synchrony) had the best health and relational outcomes. In contrast, membership in Group D was associated with substantial and consistent negative health and well-being effects. The moderating influence of synchrony on the health effect of severity seems to be more profound at low level of severity, as seen in Group A.

Aside from the vulnerable and protected groups, the other two identified groups (one characterized by moderate FCC severity and moderate synchrony, and the other by moderate–high FCC severity and moderate synchrony) both had a moderate increase in MCB over the study period. Memberships in these groups were associated with lower marital satisfaction for husbands and wives. Group C membership had negative influences on both spouses’ depressive symptoms. The effects of different levels of FCC severity combined with the moderate synchrony seemed to affect spouses’ depressive symptoms differently.

Husbands’ later physical impairment was not associated with membership in any of these groups, and there were minimal differences in physical impairment among these groups. In contrast, wives’ physical health was differentially associated with memberships in each group. These results revealed a differential impact of conflictual family relationships for women and men on developing physical health consequences. This trend is somewhat consistent with previous research emphasizing the greater health benefits of marriage for men compared to women, especially in heterosexual marriages (Kiecolt-Glaser & Newton, 2001; Wanic & Kulik, 2011). Specifically, women often experience worse health outcomes as they seem to be more negatively impacted by marital conflicts. Laboratory studies have shown that women, compared to men, experience more physiological distress during negative interactions with their spouse, such as elevated endocrine response (Kiecolt-Glaser et al., 1996), cardiovascular reactivity (Smith et al., 2004), and immunological dysregulation (Kiecolt-Glaser et al., 2002). Some researchers argue that this is because of women’s tendency to be relationally interdependent and sensitive to social relationships (Kiecolt-Glaser & Newton, 2001), which may lead to their strong physiological stress responses to conflictual family interactions. Another explanation is that husbands generally receive more significant social support from their wives than vice versa (Umberson et al., 1996), specifically in terms of wives’ healthy initiative to maintain health-promoting behaviors and preventive practice benefiting their spouse (Umberson, 1987). Our findings may reflect that these gender-related factors played a role in the associations between conflictual family circumstances and physical health outcomes for husbands. Future research needs to explore other mechanisms in these associations, focusing on longitudinal context and multiple family relationships.

There are important limitations to consider in the current study. First, we utilize an all-White US sample, and replication with more diverse samples is needed. Because we were interested in marital and parent–child relationships in the longitudinal context, we only use data from consistently married parents. Also, family dynamics may vary between different types of families such as stepfamilies and divorced-parent families. Future research should try to replicate whether similar patterns of our findings would emerge in these different types of family structures, as well as families with different racial, social, and cultural backgrounds. Second, we use self-reports for all of our measures, which are vulnerable to the issues with all self-report instruments. In this study, we conceptualized parent–child conflictual behaviors by how husbands and wives perceived their children’s destructive problem-solving behaviors toward them. These reports reflect discordant interactions between parents and a child at home. But, as mentioned above, these self-reports are potentially biased, and observations would add strength to this study. Third, we created a dyad-level variable by averaged dyad members’ scores. We understand the benefit of utilizing a common fate approach in estimating couple-level constructs (Ledermann & Kenny 2012), which allows researchers to estimate accurate couple-level constructs. However, in the context of the longitudinal growth models, a common fate growth model (Ledermann & Macho, 2014) typically has to estimate more parameters (due to the estimation of the couple-level latent variables), leading to increased model complexity compared to the growth model including couple-average variables applied in the current study. Given the model complexity and the limited sample size, we argue the couple-average approach is applicable and estimates reasonable parameters of couple-level constructs in our model. Future studies need to consider using common fate approach to estimate more accurate couple-level constructs. Last, it is important to note that our sample was comprised of families who lived in rural communities during a specific economic downturn—1980s Farm Crisis. Even though living in a different historical and social context, in many ways, these families are similar to other families who experience significant financial hardship in these times. Life challenges and various acute/ chronic stressors caused by financial hardship are not unique and uncommon. Thus, this study can provide valuable insight into how significant financial stress influences families in terms of relationships, mental, and physical health.

Despite these limitations, this research expands past literature on the effects of family conflict on health and well-being and offers important insights for practitioners and further reinforces calls to focus on multiple family subunits. Additionally, practitioners may want to focus interventions on families with high severity and high synchrony in conflict overall family unit as these families may be at the most significant risk due to this multiplicative effect.