Introduction

Witnessing destructive interparental conflict negatively impacts children, as evidenced by emotional and physiological measures of distress (El-Sheikh and Erath 2011; Koss et al. 2011). Exposure to interparental conflict is also related to children’s maladjustment, including internalizing and externalizing problems (see Cummings and Davies 2010 for review). Indeed, elementary school-aged children report witnessing interparental conflict as the third most distressing life stressor (Lewis et al. 1984). Further, it is important to note that the effects of interparental conflict are not only evident during childhood; witnessing adult disagreements is distressing even in infancy (Cummings et al. 1981; Du Rocher Schudlichet al. 2011).

Though links between interparental conflict and child outcomes are well-established, not all conflict is equally deleterious for children. Rather, evidence indicates that specific behaviors and emotions expressed during everyday marital conflict differentially impact children (Cummings et al. 2003). The emotional security theory posits that children are most negatively impacted by conflict perceived as threatening to the family system (Cummings and Davies 2010; Davies and Cummings 1994). Children’s emotional security develops in relation to their parents’ marital relationship and is reflected in future emotional responding and effectiveness of coping and emotion regulation skills. Interparental conflict is most damaging to children’s emotional security when it is frequent, aggressive (Davies and Cummings 1994), unresolved (Cummings et al. 1991), centered on child topics (O’Leary and Vidair 2005), or characterized by withdrawal (Du Rocher Schudlich and Cummings 2007). Conversely, conflict that is resolved and dealt with positively may even increase emotional security by reinforcing children’s sense of family stability and providing a constructive model for dealing with difficult emotions (Cummings and Davies 2010; McCoy et al. 2009). A growing body of research has demonstrated the utility of considering marital conflict behavior in three categories: destructive, depressive, and constructive (Cummings and Davies 2002, 2010; Rocher Schudlich 2011; McCoy et al. 2009). Destructive conflict includes behavior that is angry, physically or verbally aggressive, defensive, or contemptuous in nature and is typically accompanied by high levels of conflict. Depressive conflict is characterized by withdrawal and expressions of sadness or anxiety. Constructive conflict behavior, in contrast, is characterized by regulated communication and self-disclosure, demonstrations of support, and attempts at resolution. In general, previous literature indicates that destructive and depressive conflict negatively impact children’s emotional development (Cummings et al. 2002; Du Rocher Schudlich and Cummings 2003; Koss et al. 2011) whereas constructive conflict is positively linked to emotional development (Cummings and Davies 2010; McCoy et al. 2009).

Although the impact of interparental conflict behaviors on older children has been well-investigated, much less is known about how infants respond to interparental conflict styles. Infancy is a time of heightened interparental conflict and young children are highly likely to be exposed to interparental disagreements (Belsky and Rovine 1990), yet infant reactions to interparental conflict have been investigated in very few studies. In 1981, Cummings and colleagues conducted a study in which mothers were trained to observe and record their 10- to 20-month-old infants’ reactions to both simulated and naturally-occurring angry and affectionate adult interactions in the home. Angry interactions typically elicited infants’ distress and other negative emotional reactions, whereas interparental affection elicited infants’ expressions of affection or pleasure. Consistent with the emotional security theory (Davies and Cummings 1994), infants who were repeatedly exposed to naturally-occurring interparental anger were more likely to exhibit negative emotional reactions to these interactions over time. In 2011, Du Rocher Schudlich et al. investigated infants’ responses to live interparental conflict. In the laboratory, parents were videotaped discussing an area of disagreement with their infant present. Infants showed increased discussion attending and negative reactions, such as sadness, frustration, and affect dysregulation, to destructive and depressive conflict, but decreased discussion attending and negative reactions to constructive conflict. Together, these findings demonstrate infants’ sensitivity to interparental conflict behavior and, thus, the utility of better understanding how parents manage conflict in their presence.

However, little is known regarding parents’ management of their conflict when their children are present versus absent, or whether parents attempt to shield their children from their more destructive disagreements. Although empirical evidence regarding this is lacking, family systems theory provides a conceptual framework for understanding parents’ management of conflict in children’s presence versus absence. Systems theorists have emphasized the impact of context on individual and family behavior (Bronfenbrenner 1986, 1999). The family systems perspective holds that family relationships influence one another, so that the quality of the marital relationship is likely to affect parent–child relationships as well as triadic interactions between the child and parents (Cox and Paley 1997, 2003).

As Minuchin (1974) described, in a healthy family system, a boundary exists between the marital and parent–child subsystems, allowing parents to operate efficiently in the separate roles of spouse and caregiver. This boundary may also protect children from spousal behavior that is harmful or distressing, such as conflict. Healthy families are also able to more effectively shift their behavior as differing contexts may demand, thus demonstrating flexibility in their behaviors.

Regarding interparental conflict, family systems theorists may propose that healthy parents will be able to shift the type and intensity of conflict behaviors used based on the context of the interaction, including whether or not their child is present. However, subsystem boundaries may be compromised when interparental conflict is destructive, becoming more diffuse and less protective for children. Alternatively, the presence of the child shifts parents into simultaneous dual roles and subsytems (spousal and parental). Managing the dual roles and relationships in the context of a disagreement may strain parents’ abilities to effectively engage in either subsystem, and in a strained marriage may heighten the marital conflict. Although the theory is ample regarding the ways in which parents’ behavior may be influenced by context, there is a paucity of empirical evidence regarding how and whether parents may shift their conflict behavior in their young children’s presence. Only two studies to date (Deal et al. 1999; Papp et al. 2002) have examined differences in couple conflict when the child is present versus absent; however, the results of these two studies are contradictory.

Deal et al. (1999) examined parents’ marital conflict interactions in dyadic and triadic contexts with their school-aged children. Results indicated that interaction context significantly impacted conflict for both mothers and fathers, decreasing the frequency and intensity of all marital behaviors in triadic interactions. Thus, parents not only decreased their negativity (e.g. hostility, coercion), but also decreased their positivity (e.g. warmth, self-disclosure) in triadic conflicts. This suggests that the introduction of a child lessens the intensity of interparental conflict, perhaps due to parents’ attempts to shield their child from conflict, or because parents are unable to maintain the same high level of dyadic engagement in the triadic context.

Papp et al. (2002) also examined marital conflict when school-aged children were present versus absent by utilizing parents’ own reports of their in-home conflict. In sharp contrast to Deal et al. (1999), Papp et al. found that, in their child’s presence, both parents used fewer positive conflict strategies and more negative strategies, including aggression, hostility, and defensiveness. They also found that conflict was more frequent, longer, and contained more child-related discussions when children were present than absent. This suggests that the child’s presence serves to increase the negativity and intensity of interparental conflict rather than diffuse it.

There are several possible explanations for the disparity in these findings. Papp et al. (2002) utilized an in-home, self-report research design. Although this design has the potential to accurately reflect conflicts that the family experiences on a regular basis, it is also subject to self-reporting biases. Parents may have neglected to report on relatively milder instances of conflict, overlooking them as ordinary discussions, rather than representing constructive conflict. Concern, guilt, or frustration over having fought in front of their children could also lead parents to report their conflict as more negative than it actually was. Alternatively, Deal et al. (1999) examined conflict in a laboratory setting, which may potentially dampen couples’ conflict behavior due to heightened social desirability concerns. Being in an unfamiliar setting may also constrain the intensity of behaviors overall, due to more reservation by the parents. However, findings indicate that conflict observed in the laboratory does not substantially differ from self-reported conflict (author citation). Ultimately, controlled laboratory conditions may ensure greater protocol adherence, thereby maximizing the potential to observe positive behaviors that couples themselves may overlook when reporting on their own conflict.

The studies conducted by Papp et al. (2002) and Deal et al. (1999) focused on school-aged children. However, less is known about how parents of infants may differently manage their conflict. The additional stress of accommodating a new family member may increase conflict and negativity among parents. Furthermore, infants’ less developed language abilities may lead parents to feel less concerned about conflict in their presence, inferring that infants have lowered abilities to discern the meaning of arguments. In fact, research supports that infancy is a time period of heightened parental conflict (Belsky and Rovine 1990) in which increased levels of conflict occur in the infant’s presence (Fantuzzo et al. 1997).

The well-documented effects of destructive conflict on older children’s well-being, combined with emerging evidence of parallel effects in infancy, necessitates a greater understanding of dyadic and triadic patterns of conflict during infancy. Frosch et al. (1998) study is the only one we are aware of that addresses marital behavior differences (although not conflict, specifically) in dyadic versus triadic interactions during infancy. They examined observed differences in marital behavior during a dyadic conflict resolution task and a triadic play interaction and found that infants’ presence altered the way in which parents interacted. Parents’ behavior during the dyadic interactions was more intense and negative than during the triadic interaction, with both parents demonstrating higher engagement, negative affect, and irritation in the dyadic interaction. Frosch and colleagues speculated that the infants’ presence may have diffused the intensity of parents’ interactions by allowing them to focus on the infant, rather than just one another. Thus, these findings parallel those obtained by Deal et al. (1999). However, the triadic interaction involved a play session rather than conflict, which may in part explain the decreased interparental negativity in comparison to the dyadic interaction. Overall, findings are inconclusive across these three studies, with some research indicating overall dampened affect and behaviors (Deal et al. 1999), some indicating heightened negativity and less positivity (Papp et al. 2002), and, finally, other research indicating decreased negativity (Frosch et al. 1998). Additionally, the nature of interparental conflict during infancy remains understudied, as are factors which may contribute to distinctions between dyadic versus triadic interparental conflict.

Relationally distressed couples with a history of heightened destructive conflict become increasingly more sensitized to their own conflict over time, engaging in more frequent conflict (Driver et al. 2012), demonstrating greater emotional arousal and emotion dysregulation (Verhofstadt et al. 2005), developing negative relationship and partner attributions (Fincham 2004), and becoming insecurely attached to their partner (Treboux et al. 2004). Each of these processes erodes the marriage and may impact parents’ abilities to keep their conflict from entering parent–child interactions. The spillover hypothesis from family systems theory, which proposes that disturbances in the marital relationship spillover into parent–child relationships owing to impaired subsystem boundaries, has been well supported empirically (e.g. Davies et al. 2004; Erel and Burman 1995; Grych 2002). Davies and Cummings (1994) reported a lack of such boundaries in families with heightened interparental conflict, such that children become drawn into and attempt to mediate parental conflict. Thus, heightened conflict may impair parents’ ability to safeguard their children from conflict. It is worthy of noting that, although infants are unlikely to interfere in their parents’ conflicts as older children do, they have been observed to exhibit developmentally-salient negative reactions (e.g. crying, yelling, hitting parents or objects) to destructive and depressive parental conflicts, thus indicating that spillover from interparental conflicts is likely to impact children even during infancy (Rocheret al. 2011).

Another important process to consider, given its strong bidirectional links with interparental conflict, is parental depression (Du Rocher Schudlich et al. 2011; Morris et al. 2007). Given the harmful effects of depression on interparental conflict and parenting, depression symptoms are also likely to compromise family subsystems boundaries. Depressed individuals engage in less functional communication and problem solving, demonstrate greater withdrawal and distress, and exhibit a lessened ability to resolve conflict (Du Rocher Schudlich et al. 2011; Heene et al. 2007; Jacob and Johnson 1997). Depression has also been shown to negatively affect parenting (Morris et al. 2007). Mothers with depression are less firm, consistent, warm and nurturing, and employ less positive discipline practices (Letourneau et al. 2010). Furthermore, Jacob and Johnson (1997) observed that this pattern is consistent across family contexts; parents with depression exhibited less positivity and congeniality during problem-solving in both dyadic and triadic interactions. Thus, depression symptoms can weaken both spousal and parental boundaries, resulting in more negative conflict and perhaps a lessened ability to protect children from these negative conflict interactions.

The present study addressed the above gaps in the literature by examining distinctions in interparental conflict behavior during dyadic versus triadic interactions with infants present, utilizing behavioral observations in a laboratory setting. Parental depression symptoms and history of conflict were considered as factors that may contribute to differences in parental behavior in dyadic versus triadic contexts. Based on findings by Frosch et al. (1998) and Deal et al. (1999) who used similar observational laboratory methods for assessing conflict, we hypothesized that children’s presence would dampen the intensity of parents’ conflict behaviors and that parents would engage in less negative and positive conflict behaviors during triadic versus dyadic contexts. Furthermore, based on family systems theory and notions of diffusion of boundaries in poorly functioning couples and individuals, we hypothesized that differences in conflict behavior in the triadic versus dyadic contexts would vary as a function of couples’ history of conflict and depression symptoms. Couples with more destructive conflict histories or with greater levels of depression symptoms were expected to have greater difficulty in shielding their infants from conflict during the triadic conflict interaction. We anticipated that this could result in one of two outcomes: (1) couples with destructive conflict histories or depressive symptoms may display even more negative and less positive conflict expressions during the triadic than the dyadic conflict, which is consistent with previous research documenting more negative conflict and parenting in couples with depression symptoms or high conflict and with Papp et al. (2002) findings; or (2) couples with frequent and intense conflict or depressive symptoms may engage in more similar conflict behaviors during the dyadic and triadic conflict interactions relative to unaffected couples, which would indicate a lack of shielding children from conflict, but not necessarily a worsening of conflict with children’s presence.

Method

Participants

We collected the data during the years 2007–2009. Participants were recruited by contacting families listed in local birth records from a small county in the Pacific Northwest of the United States, as well as couples recommended by previous participants. Families were required to have been living together since the birth of the child, regardless of marital status, in order to ensure that couples described their current relationship circumstances. The study was explained to parents as being concerned with everyday interparental interactions, family relationships, and children. No references were made to couple’s mental health, well-being, or relationship distress.

Participants were 74 nuclear families (mothers’ M age = 29.56 years, SD = 5.54; fathers’ M age = 31.62 years, SD = 5.87) with 33 male and 41 female infants aged 6.20–14.48 months old (M age = 10.07 months, SD = 2.10). Sixty-four of the parent couples (85 %) were married (M length of marriage = 4.83 years, SD = 3.15 years) and couples had been living together for an average of 5.78 years (SD = 3.34). All parents reported being the biological parents of the target child in the study. Parents reported having an average of 1.66 children (SD = .75) and 57 % of infants in the study were first born children. For mothers, 8.2 % completed only high school, 38.3 % attended some college or trade school, 37 % held a bachelor’s degree, and 16.5 % held a master’s degree or higher. For fathers, 1.4 % did not complete high school, 15.1 % completed only high school, 42.5 % attended some college or trade school, 26 % held a bachelor’s degree, and 15 % held a master’s degree or higher. Parents indicated a modal family income of $40,001–$65,000 per year. In this sample, 88 % of fathers and 85.3 % of mothers were Caucasian, 1.3 % of fathers and mothers were Asian American or Pacific Islander, 1.3 % of fathers and mothers were Hispanic, 5.4 % of fathers and 8 % of mothers were biracial, and 3 % of parents did not report ethnicity.

Measures

History of Destructive Interparental Conflict (Moderator Variable)

Parents’ history of conflict during the past year was assessed using the Conflicts and Problem-Solving Scale (CPS; Kerig 1996). Parents rated how often in the past year they had major and minor disagreements on a scale from 1 = once a year or less to 6 = just about every day. Owing to the greater severity of major disagreements, response items for the major disagreements question were weighted by 2 and then responses across both items were summed together, forming a Conflict Frequency/Severity subscale. Parents additionally reported on the frequency with which they engaged in specific conflict behaviors in the past year on a scale from 0 = never to 3 = often. These conflict strategy questions formed eight subscales: Avoidance, Give in, Verbal Aggression, Physical Aggression, Seek Help, Child Involvement, Resolution, and Cooperation. The CPS Child Involvement Subscale reflects parents’ disagreements about their children, the extent to which their children are exposed to interparental conflict, and how much parents themselves pull their children into their own disagreements and is therefore appropriate for parents of children of all ages. For the purpose of data reduction, composite variables were created for mothers and fathers using standardized subscale scores. Negative subscales were averaged (Avoidance, Give in, Verbal Aggression, Physical Aggression, Seek Help, and Child Involvement). Cronbach’s αs were .84 for fathers and .83 for mothers. Mothers’ and fathers’ composite scores were significantly correlated, r = .89, p < .001, and therefore were averaged together to create a single destructive conflict history composite for parents. The mean composite score was −.05 (SD = .66).

Parental Depression Symptoms (Moderator Variable)

Depression symptoms were assessed using the Center for Epidemiological Studies-Depression Scale (CES-D; Radloff 1977). The CES-D is a 20-item scale designed to measure depressive symptomatology in the general population. The CES-D has well-established psychometric properties, including high internal consistency, test–retest reliability, and convergent validity with clinical and self-report measures of depression (Radloff 1977). Cronbach’s αs were .85 for fathers and .86 for mothers. The mean score was 9.17 (SD = 7.23) for mothers and 8.15 (SD = 7.07) for fathers. Scores of 16 or above are considered reflective of potentially serious depression (Ensel 1982). Using this cut-off, 11 % of fathers and 18 % of mothers had scores above this clinical cutoff.

Procedures

Parents consenting to participate were mailed packets containing consent forms and questionnaires to be completed at home prior to the laboratory visit. Upon arrival at the laboratory, parents engaged in two interactions: an interaction with their infant absent (dyadic context) and an interaction with their infant present (triadic context). The order of interactions was randomly counter-balanced across families whenever feasible. In some cases, parents and/or their infants were not comfortable being separated at the start of the laboratory visit and in those cases, in order to accommodate the families, we allowed them to engage in the triadic conflict first.

Observation of Marital Interactions in the Laboratory

For both interactions, parents separately indicated three topics that were most typically problematic for their relationship and then chose a topic that they were both comfortable discussing. Parents were asked to choose a different topic for their second interaction than what they discussed in their first interaction. Parents were instructed to attempt to reach a resolution to their problem and to share their feelings and perspectives on the issues. Because previous research has demonstrated that children are particularly sensitive to discussions about themselves or other child-related issues (Grych and Fincham 1990), parents were asked to not discuss these during their interaction when their child was present. Parents were instructed to interact with and tend to their baby as needed as they normally would if they were at home discussing the issue. In all cases, a standard set of developmentally appropriate toys was available for infants and spread out on the floor between the parents’ chairs. Families were left alone during their interaction, which was videotaped for later coding. After 7.5 min, the research assistant checked with the parents to see if they were finished with their discussion. Parents requesting additional time were given an additional 2.5 min. Seven parents in the dyadic interaction requested additional time, whereas twelve parents in the triadic interaction did so, which was not a significant difference, t = −1.21, p > .05. The average length of time was 8.01 and 8.37 min in the dyadic and triadic interactions respectively, which was not a significant difference, t = −1.86, p > .05. Following procedures developed by the Cummings’ lab, immediately following each of the interactions, parents independently completed ratings of how strongly they felt each of the following emotions during their interactions: loving feelings, happiness anger, worry, scaredness, sadness, helplessness, and hopelessness. Emotions were scored on a scale from 1 to 9, with 1 = absence of the emotion, 5 = mid-range level of feeling, and 9 = most intense feeling.

Coding Observations of the Marital Interactions (Dependent Variables)

An adapted version of the marital daily records (MDR) protocol was used to code observations of marital interactions (Cummings et al. 2002). The MDR has good convergent validity with widely used self-report measures of interparental conflict and marital relations (author cite). Conflict behaviors were defined as follows: (a) conflict, the level of tension, hostility, dissension, antagonism, or negative affect; (b) defensiveness, trying to avoid blame or responsibility; (c) contempt, lack of respect, insult, mockery, sarcasm, or derision of partner; (d) withdrawal, an avoidance of the interaction or of the problem discussion in some way; (e) demand, hounding or nagging partner; (f) communication skills, level of appropriate and positive expressive skills; (g) support-validation, appropriate and positive listening and speaking skills that convey supportiveness or understanding; (h) problem solving, the ability to constructively define a problem and work toward a mutually satisfactory solution; and (i) humor, trying to make a joke or finding something funny about the situation. For each of the behaviors degree of behavior intensity was coded on a scale from 1 to 9, with 1 = absence of the expression, 5 = mid-range level, and 9 = most intense expressions. The degree of emotional intensity of each of four emotions (positivity, anger, sadness, and anxiousness) and the overall degree (1–9) of conflict resolution were additionally coded for each partner on the same 1–9 scale. The primary adaptation to the coding system included coding behaviors on a 1–9 scale based on the Couples’ Interaction Global Coding System, rather than the original 0–2 scale on the MDR (Julien et al. 1987), allowing us to capture more variability in the behaviors.

Each discussion was coded once by one of five undergraduate research assistants. To minimize potential coding bias or carry-over effects, coders coded only one type of interaction (triadic or dyadic) for each family. The coders received extensive training by the principal investigator. A subset of 25 interactions was used to assess the coders’ agreement with the principal investigator’s codes using intraclass correlation coefficient (ICC) (3, 1). ICCs for conflict emotions, behaviors, and resolution ranged from .70 to 1.0 (with the exception of female demand which was .59), with a mean alpha of .81.

For the purposes of analyses, parents’ conflict tactics, emotions, and resolution were sorted into three categories according to conceptual criteria based on the research literature on marital conflict and child adjustment: constructive, destructive, and depressive conflict behavior. Constructive patterns of conflict included problem solving, communication skills, support-validation, humor, positivity, resolution, and self-reported loving feelings and happiness. Destructive patterns of conflict included conflict, contempt, defensiveness, demand, and observed and self-reported anger. Finally, depressive conflict patterns included withdrawal, observed sadness and anxiousness, and self-reported feelings of worry, scaredness, sadness, helplessness, and hopelessness. Confirmatory factor analyses confirmed the existence of these three factors, with behaviors loading adequately on their respective factors (details of the factor analysis may be obtained from the first author). Each of the conflict composites had high internal consistency: alphas for mothers and fathers were .89 and .84 for destructive conflict, .73 and .64 for depressive conflict, and .85 and .85 for constructive conflict, respectively. These conflict dimensions derived from observed conflict have good convergent and predictive validity with widely used self-report measures of marital conflict, marital relations, and depressive symptoms (Du Rocher Schudlich et al. 2004). Means and standard deviations for each parents’ conflict expressions within the dyadic and triadic interactions are presented in Table 1.

Table 1 Parents’ interparental conflict expressions in dyadic and triadic interactions
Full size table

Results

Preliminary Analyses

Given that families had differing lengths of time for their discussions, we correlated discussion length with the conflict composites and found no significant associations. Because a small subset of families who were assigned to the dyadic interaction requested to complete the triadic interaction first instead (21 %), we explored whether interaction order was significantly associated with interparental conflict. Independent samples t-tests revealed there were no significant effects of interaction order. Given the lack of significant findings for these variables and in order to preserve power, we did not control for any of them in the rest of the analyses. Finally, we also considered infant age as a potential covariate. Correlational analyses indicated no significant associations between discussion length, infant age, and interparental conflict. However, given that previous findings have documented significant associations between infant age and parental depression and conflict (e.g. Belsky and Rovine 1990; Crockenberg and Leerkes 2003; Houts et al. 2008), we included it as a covariate in primary analyses.

Data Analysis Plan

To account for the dependency of mothers’ and fathers’ data, dyadic multilevel modeling (MLM) was used (SPSS v 20) to estimate: (a) the associations between interparental conflict and interaction context; (b) whether interparental conflict history moderated associations between current interparental conflict and interaction context; and (c) whether parental depressive symptoms moderated associations between current interparental conflict and interaction context. We used a two-level model, with the dyad as the unit of analysis. Separate models were estimated for each of the conflict composite outcomes. We used the method of restricted maximum likelihood estimation and the Satterwhite approximation to determine degrees of freedom. All predictors entered into the model were grand mean centered. We first compared unconstrained models, where separate estimates were made for fathers and mothers in one model, with constrained models where the estimates for fathers and mothers for the intercept as well as for the predictors were set to be equal. This approach compares a full model allowing for parent gender differences with a simpler nested model constraining parent gender effects to be equal across equations. The deviances of the two models can be compared, yielding a χ2 statistic with degrees of freedom equal to the difference in parameters. A non-significant χ2 is consistent with the null hypothesis that there are no significant parent gender differences in the conflict predictors. A significant χ2 was found for each test, thus all final models freely estimated the parent gender effects, using correlation compound symmetry for the covariance structure. For each of the 6 types of parental conflict expression outcomes, we examined a single model to assess interaction context associations, while simultaneously controlling for the following covariates: infant age, parental destructive conflict history and depression symptoms, as well as moderation effects for interaction context by conflict history and parental depression symptoms. All significant interactions were probed using standard pick-a-point techniques, which have been validated in multilevel models (Preacher et al. 2006).

Associations Between Interaction Context and Observed Interparental Conflict

Multilevel modeling (MLM) analyses revealed significant associations between the interaction context and depressive conflict behaviors for both mothers and fathers (see Table 2). Controlling for infant age, interparental conflict history and parental depression symptoms, and interaction terms, mothers and fathers both demonstrated greater depressive conflict in the dyadic compared to the triadic interaction. Mothers additionally demonstrated greater constructive conflict in the dyadic compared to the triadic interaction. No significant differences were found in mothers’ and fathers’ destructive conflict across interaction context.

Table 2 MLMS of the associations between parents’ conflict history, depression symptoms, and mothers’ and fathers’ interparental conflict expressions
Full size table

Moderating Role of Interparental Destructive Conflict History

Significant main effects were found for interparental conflict history predicting current interparental conflict behaviors (see Table 2). Interparental conflict history predicted increased paternal and maternal destructive conflict, as well as increased paternal depressive conflict. Additionally, interparental conflict history predicted decreased paternal and maternal constructive conflict. Significant main effects were qualified by significant interaction effects found between interparental conflict history and interaction context for paternal destructive anger. The significant interactions were probed by testing the simple slopes defining the relationship between current conflict context at levels of interparental conflict history 1 SD below and 1 SD above the mean. Decreased paternal destructive conflict in the triadic versus dyadic interaction was significant only for parents reporting a low conflict history, B = 4.15, SE = 1.35, t = 3.07, p < .01 (see Fig. 1). Fathers with a high conflict history did not demonstrate observable differences in their destructive conflict across the dyadic and triadic contexts.

Fig. 1
figure 1

Interparental conflict history moderates associations between paternal destructive conflict and conflict interaction context

Full size image

Moderating Role of Parental Depression Symptoms

Significant main effects were found for parental depression symptoms predicting current interparental conflict behaviors (see Table 2). Paternal depressive symptoms predicted greater paternal and maternal destructive conflict, greater paternal depressive conflict, as well as decreased paternal constructive conflict. No significant main effects were found for maternal depression symptoms. Significant main effects were qualified by significant interaction effects found between maternal depression symptoms and interaction context in predicting paternal depressive conflict. A significant interaction was also found between paternal depression symptoms and interaction context in predicting maternal depressive conflict. The significant interactions were probed by testing the simple slopes defining the relationship between current conflict context at levels of interparental conflict history 1 SD below and 1 SD above the mean. Lower paternal depressive conflict in the triadic compared to the dyadic interaction was significant only for families with mothers reporting low depression symptoms, B = 2.75, SE = .87, t = 3.16, p < .01 (see Fig. 2). For families with mothers reporting high depression symptoms, fathers did not demonstrate observable differences in their depressive conflict across the dyadic and triadic contexts. Similarly, probing interaction effects for paternal depression by interaction context, lower maternal depressive conflict in the triadic compared to the dyadic interaction was significant only for families with fathers reporting low depression symptoms, B = 3.27, SE = 1.15, t = 2.28, p < .01 (see Fig. 3). For families with fathers reporting high depression symptoms, mothers did not demonstrate observable differences in their depressive conflict across the dyadic and triadic contexts.

Fig. 2
figure 2

Maternal depression symptoms moderate associations between paternal depressive conflict and conflict interaction context

Full size image
Fig. 3
figure 3

Paternal depression symptoms moderate associations between maternal depressive conflict and conflict interaction context

Full size image

Discussion

This study tested the hypotheses that couples’ conflict behavior would be dampened in their infant’s presence and that differences in conflict behavior across dyadic and triadic contexts would be moderated by parents’ conflict history and level of depression symptoms. Understanding the specific ways in which parents manage their conflict when their children are present versus absent is essential owing to the differential impact of specific conflict expressions on infants. Overall, there were no significant differences in terms of destructive conflict behavior in dyadic versus triadic contexts. Differences did emerge, however in terms of maternal constructive conflict, which was higher in the dyadic than triadic interaction. Furthermore, maternal and paternal depressive conflict were lower in the triadic than dyadic interaction. Overall it appears that parents did not seem to protect their children by decreasing destructive conflict behaviors, as family systems theory might suggest. Nor did they generally use conflict tactics that were much more negative in their children’s presence as Papp et al. (2002) found. Instead, observed conflict was fairly similar across dyadic and triadic contexts, with mothers only decreasing their constructiveness and both parents minimizing their depressive behaviors in children’s presence. These findings thus partially support our own hypotheses, and are mostly consistent with findings by Frosch et al. (1998).

Consistency with work by Frosch et al. may be due in part to methodological and child age similarities. Comparable to Frosch et al. (1998), our study examined laboratory observations of interparental conflict, whereas Papp et al. (2002), who found heightened conflict in triadic interactions, utilized parental self-reports. Furthermore, whereas Papp et al.’s (2002) parents were free to disagree about whatever subjects naturally occurred, our study employed restraints on topics; parents were required to choose topics that they were both comfortable discussing and were prohibited from discussing child-related conflicts, owing to ethical issues. By choosing topics they were more comfortable with and by avoiding highly charged topics such as child-rearing, which may be most likely to arise in the child’s presence, parents may have been able to avoid engaging in more negative behaviors.

Additionally, age differences in the children studied may have contributed to the difference in findings. Previous work (Deal et al. 1999; Papp et al. 2002) has focused primarily on conflict in the presence of school aged children; thus, our findings, which are more consistent with Frosch et al. (1998), may reflect parents’ underlying attitudes and beliefs about the extent to which infants are impacted by witnessing conflict. It is possible that, given the young age and limited language abilities of their children, parents may believe that their infants are not likely to understand the content of the conflict and therefore infer that their infants are not likely to be impacted by their conflict. It may not be until children are older and can better express themselves that parents begin to feel a greater need to minimize their conflict in front of their children as Deal et al. found. Although infants may not fully understand the content of interparental exchanges, previous research suggests that they are sensitive and react negatively to the emotional tone (Du Rocher Schudlich et al. 2011) of interparental conflict, thus making the ways in which parents handle their conflict, even at the infancy stage, highly salient to children’s development.

Although parents did not appear to shield their children from more overt destructive conflict, they did appear to shield them from depressive conflict. Both mothers and fathers displayed less depressive conflict in the triadic than in the dyadic context, which is consistent with Frosch et al. (1998) findings on decreased negativity. Depressive conflict included more tender and vulnerable emotional expressions (e.g. sadness, anxiety, hopelessness, helplessness, etc.). Anger and frustration are common experiences in family life (Cummings and Davies 2010), but parents’ expressions of more tender, vulnerable emotions may be less often expressed or witnessed, except in cases of parent emotional dysregulation (e.g. Cumberland et al. 2003). In other words, parents may view these more tender emotions and behaviors associated with depressive conflict as representing a vulnerability, which they may be less comfortable expressing in front of their children. This is supported by evidence that mothers and fathers tend to use more angry and less sad or fearful conflict expressions in children’s presence at home, with these more vulnerable expressions predicting children’s emotional insecurity (Cummings et al. 2002). These less common expressions may also be easier to contain for parents than more conflictual behaviors, such as hostility or demand, which can take on a life of their own once parents become embroiled in and sensitive to conflict over time (Cummings and Davies 2010). Furthermore, as noted by Frosch et al. (1998), parents may have greater difficult remaining as highly engaged, intimate, and resolution-focused during discussion when their infants are present. That is, the infant’s presence may serve to distract parents from conflict discussion, or may even be used as a means to avoid a high level of engagement in conflict with their partner, especially as the interaction becomes more distressing. High levels of engagement, support, and conversational depth are typically prerequisites to sharing more tender or vulnerable emotions (Johnson and Denton 2002). Previous research has documented the particularly detrimental effects of depressive conflict on children (Du Rocher Schudlich and Cummings 2003; Huffman and Cummings 2002) and infants (Du Rocher Schudlich et al. 2011), making this an important element of conflict to understand in terms of how parents manage it in front of their children.

Mothers, but not fathers, also displayed lower constructive conflict in triadic than dyadic interactions. This partially supports our hypotheses and is consistent with Deal et al.’s finding that parents displayed both less negative and less positive conflict behavior in children’s presence. Consistent with family systems theory, this also indicates an overall lower level of interparental engagement during triadic interactions. Again, the infant’s presence may have prevented the high level of interparental engagement necessary for constructive conflict behaviors, such as support and validation of the other’s expressions and attempts at problem solving, to occur. Thus, although mothers were able to shield their infants from depressive conflict expressions, they also made less positive attempts at resolving conflict in their infant’s presence. This may hold implications for infant’s development, as previous research suggests that witnessing positively resolved interparental conflicts is beneficial to children’s social and emotional development (Cummings and Wilson 1999; McCoy et al. 2009).

Hypotheses regarding the moderating effects of parents’ conflict history were supported. Consistent with previous research (Driver et al., 2012), parents with higher destructive conflict histories displayed more destructive and less constructive behavior in both dyadic and triadic context than did families with lower conflict histories. Conflict history also intensified fathers’ specific conflict behaviors. Decreased paternal destructive conflict in the triadic versus dyadic interaction was significant only for parents reporting a low conflict history. Fathers with a high conflict history did not demonstrate observable differences in their destructive conflict across the dyadic and triadic contexts. Thus, fathers with a heightened destructive conflict history were less able to shield their children from destructive conflict, perhaps due to an erosion of subsystem boundaries. As family systems theory holds, boundaries between marital and parent–child subsystems are weakened for couples experiencing high levels of destructive conflict; frustration from previous negative conflicts creates an atmosphere of negativity and anger that carries over into triadic interactions.

Findings regarding parental depression symptoms also were partially consistent with hypotheses and family systems theory. Partner moderation effects of depression symptoms by interaction context on parents’ conflict were found for both mothers and fathers. Lower paternal depressive conflict in the triadic compared to the dyadic interaction was significant only for families with mothers reporting low depression symptoms. Similarly, lower maternal depressive conflict in the triadic compared to the dyadic interaction was significant only for families with fathers reporting low depression symptoms Although high parental depression symptoms were not associated with greater partner depressive conflict in triadic contexts, having lower parental depression symptoms was associated with greater shielding of children from partner depressive conflict. In general, parents seemed to recognize the need to withhold depressive conflict expressions from their children; however, this was more difficult when partners were experiencing depression symptoms. Interestingly, maternal depression symptoms were associated with fathers having greater difficultly shielding their own conflict, and vice versa. This highlights spillover from interpersonal dysphoria into the marital relationship per interactionist theories of depression (Joiner et al. 1999). Parents from families with high partner depression symptoms did not demonstrate changes in their depressive conflict across the dyadic and triadic contexts. This is also consistent with family systems theory that depression may erode marital subsystem boundaries, making it difficult to withhold difficult and negative marital discussions from children.

Limitations and Future Directions

Although findings from this study suggest that interparental conflict behaviors may occur in response to the context of interaction or as a function of conflict history or depression symptoms, the correlational nature of this data prohibits determination of causality in these relations. Further research is needed to disentangle the precise nature and interactions of various family subsystems. Given the differential findings across methods, ideally, future research would utilize multiple methods, including behavioral observations as well as home self-report to ensure limited bias from any one individual methodological approach.

Future research should also consider other factors that may erode or support subsystem boundaries or potentially interact with the family factors assessed in the present study. For example, parent–child attachment, interadult attachment, and poor quality of parent–child relationships may all potentially be associated with interparental conflict management in children’s presence. Interparental discussions in this study were limited to topics both partners felt comfortable discussing and, during the triadic interaction, could not include child-related topics. This may limit the generalizability of these findings to the full range of naturally-occurring couple disagreements. It should also be noted that, because parents were instructed to avoid child-related topics only in the triadic interaction, our ability to compare across dyadic and triadic interactions may have been constrained. An additional limitation was the low reliability for the observed conflict variable demand in females (ICC = .59). This particular variable may not be as strong when analyzed individually, However, the destructive conflict composite that it was a part of demonstrated adequate reliability (α = .89), thus tempering the concern. Another important limitation of the current study is the limited diversity in families who participated. Families were primarily Caucasian, middle-class, and represented a non-clinical community sample. Different relations might be found for clinical samples seeking individual, couple, or family therapy. Thus, caution should be used when generalizing these findings.

Implications

These findings have several implications for marital and family therapists. In this study, parents were not observed to make many efforts to alter destructive conflict behaviors in order to shield their infants. This may be due to a belief that infants are not affected negatively by destructive forms of interparental conflict; however, research has shown that infants are in fact quite sensitive to destructive interparental conflict (Rocher Schudlich et al. 2011). Accordingly, parents and clinicians must be made aware that the specific behaviors employed during conflict have meaningful consequences for children, even during infancy. Although it is inevitable that children will be exposed to some level of interparental conflict, greater training and education could empower parents to make small adjustments in their conflict strategies in their children’s presence that would have meaningful effects on their children’s long-term emotional development.

Furthermore, the emotional security theory (Davies and Cummings 1994), along with recent research (McCoy et al. 2009), highlights the negative effects of destructive conflict on children as well as the positive effects of conflict that is constructively resolved. However, our findings suggest that conflict resolution is likely lessened in infants’ presence for families reporting high depression symptoms and conflict history, both of which predict impaired interparental conflict expressions overall, as well as increased paternal destructive conflict in triadic interactions. The net result is that infants in these families are not only exposed to greater stress associated with parental depression symptoms and interparental hostility, but are also directly exposed to more negative interparental conflicts, thus placing them at greatest risk for emotional security difficulties. Accordingly, early intervention efforts may be especially needed for families experiencing these issues.