Understanding Parent-Child Relationship as A Developmental Process: Fluctuations Across Days and Changes Over Years

Abstract

Toward advancing the understanding of relations among family relationships when children transition into adolescence, this study investigated whether parent–child relationship (PCR) quality assessed at the daily level changed developmentally and/or fluctuated due to daily experiences. Specifically, this study examined (1) whether parents’ daily perceptions of marital relationship (MR) quality was associated with their own and/or their partners’ PCR on the same day and the following day, and (2) whether relations among these daily influences changed over a two-year period as children developed. Participants recruited included 237 two-parent families with preadolescent or adolescent children (52% girls). Both fathers and mothers completed daily diaries about MR and PCR for 15 consecutive days each year for three consecutive years. Results indicated that daily PCR did not change developmentally but was subject to day-to-day variations based on parents’ daily MR: parents’ distressed MR was related to their own lower emotional quality in PCR on the same day (supporting the spillover hypothesis), but higher emotional quality in PCR on the next day (supporting the compensatory hypothesis). Compensatory association was also found between father-reported average MR and mother-reported daily PCR. Furthermore, the same-day spillover and cross-day compensatory effects tended to decrease developmentally, as children transitioned into adolescence. The findings illustrated the interdependent, changing, and dynamic patterns of family relationships and underscored the importance of differentiating the father-child and mother-child relationship.

Are adolescents really the trouble makers who constantly bring stress and strains to their relationship with their parents? Some empirical studies have shown a decreasing trend of closeness and increasing trajectory of negativity in parent–child relationship (PCR) during adolescence (Laursen & Collins, 2009; Marceau, Ram, & Susman, 2015), whereas other scientific inquiries indicate that adolescents continue to feel close and show respect to their parents (Steinberg, 2001). The mixed findings regarding PCR-change across adolescence years warrant more research attention, given the critical role of a close, warm, and supportive PCR in shaping adolescents’ healthy development (e.g., Eisenberg, Morris, McDaniel, & Spinrad, 2009; McElhaney, Allen, Stephenson, & Hare, 2009). Little of the existing research, however, has been based on a dynamic, micro-level, or process-oriented perspective on PCR, resulting in multiple limitations in the articulation of developmental processes related to PCR. First, intra-familial change, or within-family, long-term changes in PCR has been mostly assessed yearly with assessments completed in a laboratory setting rather than in the home. Second, intra-familial variability, or within-family, short-term state fluctuations of PCR has rarely been captured and examined in past research. Lastly, even less research has situated the short-term variability in the context of long-term changes to achieve a more comprehensive understanding of PCR. Therefore, the present study aimed to investigate how the daily fluctuations in PCR, indicated by the effect of same-day and previous-day marital relationship (MR), relate to systematic changes over time during children’s transition into adolescence.

Over-year Changes and Daily Fluctuations of PCR

Empirical examinations of PCR changes at a macro level (i.e., over years) indicate that families with children at different ages exhibit different levels of PCR quality (see Collins & Laursen, 2004 for a review). From middle childhood to early adolescence, a change trajectory may occur in PCR: some findings suggest that closeness and support in the parent–child subsystem slightly decline, while negativity increases (Marceau et al., 2015; Seiffge-Krenke, Overbeek, & Vermulst, 2010). Such trajectory of PCR may be a function of changes associated with children’s developmental stages. Compared with younger children, older children and adolescents have a larger peer group, spend more time with peers, and receive increased support from their peers (Markiewicz, Lawford, Doyle, & Haggart, 2006), thus are less likely to turn to parents for advice and support. Besides, while younger children tend to see their parents as omniscient, authoritative figures, older children and adolescents start to de-idealize their parents (Steinberg & Silverberg, 1986). As a result, adolescents are more likely to challenge parents’ orders and opinions, creating more conflicts in the parent–offspring dyads. Furthermore, parents tend to hold expectations and assign responsibilities to their physically mature, yet emotionally or cognitively incompetent, adolescents (Ge & Natsuaki, 2009; Mendle, Harden, Brooks-Gunn, & Graber, 2010). The mismatch between parental expectations and adolescent competencies may lead to negative interactions in parent–offspring dyads. Thus, it is possible that changes of PCR over time are merely a result of child age and its associated developmental characteristics.

Alternatively, PCR may be more likely to fluctuate based on parents’ daily experiences, rather than to change based on children’s age. In other words, daily family stressors, such as distressed marital relationship (MR), may take a toll on parents’ emotional resources to maintain a good relationship with the child on the same day or even the following day. The daily MR quality, indicated by levels of marriage satisfaction, frequency of disagreement, and engagement in constructive conflicts, may be a particularly salient predictor of the fluctuations in PCR (Cummings & Davies, 2002; Erel & Burman, 1995; Nelson, Boyer, Villarreal, & Smith, 2017; Sears, Repetti, Reynolds, Robles, & Krull, 2016; Sherrill, Lochman, DeCoster, & Stromeyer, 2017). On days when individuals experience more strain and less support in marital relations, they report more negative affect or depressive symptoms (Delongis, Capreol, Holtzman, Brien, & Campbell, 2004; Smith, Breiding, & Papp, 2012). Such emotional negativity may exert tensions on parent–child interactions, which may elevate parent–child conflict and lead over time to children’s emotional problems (Chung, Flook, & Fuligni, 2009). Thus, daily PCR may vary as much, or even more, based on daily family experiences than simply on children’s age.

Two competing hypotheses have been proposed to describe the interrelatedness between MR and PCR: the spillover and compensatory hypotheses. The spillover hypothesis suggests a positive MR–PCR association such that parents’ experience of negativity in their interactions with the spouse may be transferred, that is, spilled over to their interactions with the child, resulting in tensions in PCR (Almeida, Wethington, & Chandler, 1999; Gerard, Krishnakumar, & Buehler, 2006) and disruptions in parenting (Kaczynski, Lindahl, Malik, & Laurenceau, 2006; Krishnakumar & Buehler, 2000). On the contrary, the compensatory hypothesis indicates a negative association between marital distress and parenting difficulties (Belsky, Youngblade, Rovine, & Volling, 1991; Brody, Pellegrini, & Sigel, 1986; Engfer, 1988). That is, parents may be less harsh and more involved in parenting to compensate for marital distress and conflict (e.g., Brody et al., 1986). Overall, more empirical findings, especially those examining short-term effects of MR on same-day PCR, have supported the spillover than the compensatory hypothesis (e.g., Almeida et al., 1999; Nelson et al., 2017; Sears et al., 2016; Sherrill et al., 2017). A few studies providing evidence for the compensatory hypothesis (e.g., Kouros, Papp, Goeke-Morey, & Cummings, 2014; Nelson, O’Brien, Blankson, Calkins, & Keane, 2009; Ponnet et al., 2013), however, suggest heterogeneity of the MR–PCR association and deserve a closer examination.

In fact, both age-related changes and daily-experience fluctuations may play an indispensable role in affecting the trajectory of PCR. It may be time to examine the two factors together and view PCR as a developmental process. Specifically, within-family variability of daily PCR, indicated by daily variations of MR, may exhibit between-family heterogeneity on a daily level and eventuate in over-year changes in PCR. This view of examining PCR on different time scales (i.e., daily and yearly) as well as different analytical levels (e.g., within- vs between-family) closely echoes the goal of developmental research (Nesselroade, 1991; Ram & Gerstorf, 2009). Hence, the present study aimed to explore how daily fluctuations in MR–PCR link 1) exhibit heterogeneity at the daily level and 2) relate to systematic changes over time. Next, we elaborate on each of the two aims.

Heterogeneity in the MR–PCR Link

Mother–child and father–child relationship.

Whether the MR–PCR within a family shows a spillover or compensatory association may depend on which relationship dyad is examined. Existing studies have found that father–child relationships are more vulnerable to poor marital functioning than mother–child relationships (e.g., Cummings, Merrilees, & George, 2010; Kaczynski et al., 2006; Krishnakumar & Buehler, 2000). Stroud and colleagues (2011) observed parent–child dyadic interactions in laboratory tasks and found that marital functioning was only related to fathers’, but not mothers’, responsiveness to children’s signals for attention. Nelson and colleagues (2009) found that fathers tended to experience more spillover from marital dissatisfaction than mothers. Furthermore, longitudinal studies have corroborated cross-sectional results, in that pathways between interparental conflict and parenting difficulties were significant only for fathers but not mothers (Davies, Sturge-Apple, Woitach, & Cummings, 2009).

The stronger link between marital and father–child subsystem, compared with the mother–child subsystem, may be explained by the differences in how fathers and mothers view their role as a spouse and as a parent. As the emotional gatekeeper and conflict conciliator of the family, mothers tend to regard a poor mother–child relationship as a personal failure (De Luccie, 1995; Pleck, 1983). As a result, they may devote more effort to blocking the spillover of marital distress to their relationship with the child. Fathers, on the other hand, tend to have more difficulties separating their role as a spouse from that as a parent, thus they are more likely to provide positive fathering under supportive MR and negative fathering under distressed MR. It is also important to note that findings supporting the compensatory relationship in the MR-PCR link only exist in maternal, rather than paternal, link (Belsky et al., 1991; Brody et al., 1986; Engfer, 1988). Collectively, empirical evidence indicates that MR quality may be differently related to mother–child and father–child relationships.

Within-person and cross-person association.

Studies of the association between marital and parent–child subsystem have been mostly focused the within-person transmission of affect or behaviors (or actor effects), leaving the cross-person influence among family members (or partner effects) unexamined. Living under the same roof, fathers and mothers together experience marital up-and-downs, and at the same time share joys and troubles of raising a child. These shared experiences inevitably result in parents’ reciprocal influences on one another (Cox & Paley, 1997). As such, the study of both within-person and cross-person effects broadens the scope of the investigation of family influence processes, to include additional influences that may be important but are neglected in many studies of family dynamics. For example, positive marital quality reported by the mother influences the father’s responsiveness to the child, above and beyond father’s self-report MR quality (Ponnet et al., 2013); for fathers who use more destructive marital behaviors, their partners (i.e., mothers) are more likely to provide unsupportive parenting one year later (Gao, Du, Davies, & Cummings, 2018). Furthermore, parents may not necessarily acquiesce in their partner’s feelings of MR quality, it is therefore plausible that parent–child relationship may be differently affected by an individual’s own and their partner’s perceptions of MR. Toward the goal of achieving a better characterization of the daily family dynamics, the present study examined how the influence of MR on PCR is manifested in the within-person as well as the cross-person association. While within-person effects represent how one’s perception of emotional quality in the marriage influences his or her relationship with the child, cross-person effects indicate the influence on one’s spouse’s relationship with the child.

Same-day and next-day effect.

The association between MR and PCR may also change depending on the time they are assessed. With a daily diary design, Kouros, Papp, Goeke-Morey, and Cummings (2014) examined how parents’ emotional quality in the marriage was related to their emotional quality with the child both on the same day and the next day. Results showed that although mothers with lower levels of MR reported poorer relationship with the child on the same day (supporting the spillover hypothesis), they tended to report higher levels of relationship quality with the child the following day (supporting the compensatory hypothesis). However, spillover hypothesis was supported by both the same-day and next-day model in another daily diary study examining the transmission of tension from marital to parent–child dyads (Almeida et al., 1999). Complicating the picture even further, Sherrill et al (2017) found that the occurrence of interparental conflict predicted the occurrence of parent–child conflict 1 time period later and 1 full day later, but not 2 time periods later (less than 1 full day). These findings suggested that daily PCR might be affected differently by parents’ MR on the same day or previous day. By assessing both the same-day and next-day effects of MR on PCR, the present study aimed to clarify these mixed findings.

Changes of the MR–PCR Link Over Time

A complete understanding of the daily fluctuations in MR-PCR link should be placed within the perspective of long-term changes, in addition to the consideration of the daily-level couple processes (e.g., paternal vs. maternal MR-PCR link). Middle childhood to adolescence is a period of transitions and strains in the family system due to changes in developmental tasks. The fluctuation and variability happening at this stage may result in adaptation and reorganization of different family subsystems (Cox & Paley, 2003), which may in turn contribute to the development of psychopathologies (Davies & Cicchetti, 2004). Despite its significance for family functioning and children’s developmental outcomes, how the MR–PCR link changes over time, especially when children transition into adolescence, has received limited empirical attention (Carlson, Pilkauskas, Mclanahan, & Brooks-Gunn, 2011; Grych, 2002). One aim of the present study, therefore, was to examine how the daily MR–PCR link changed over two years of assessment. We hypothesized a stronger association between MR and PCR as children leave childhood and enter adolescence. Adolescents are more likely to challenge parents’ authority, disobey parents’ orders, and engage in risk-seeking activities as their way of striving for autonomy (Collins & Laursen, 2004). More consistent discipline, in return, is required from their parents, which may be a major source of parental stress. When parents are under stress, positive MR is particularly crucial in buffering the negativity whereas negative MR is more evident in exacerbating the stress (Belsky, 1984). Thus, we expected the association between MR and PCR to be stronger as children develop in this age period.

In order to capture the daily fluctuations in family relationships and explicate the associated between-person differences (i.e., maternal and paternal relations) as well as over-year changes, a method that allows examination of family relationships at both the micro- and macro-level is needed. In the present study, a measurement-burst design (Gerstorf, Hoppmann, & Ram, 2014) was adopted in which we asked participants to complete a daily diary for 15 days and repeated this assessment burst yearly for 3 years. Such a design allowed us to take advantage of both the high ecological validity of the daily diary approach (Bolger, Davis, & Rafaeli, 2003; Laurenceau & Bolger, 2005) and the developmental implications of the longitudinal designs. With more closely spaced and repeated assessments, the daily diary method permits investigations of day-to-day dynamics, that cannot be assessed with traditional designs (Bolger et al., 2003; Laurenceau & Bolger, 2005; Repetti, Reynolds, & Sears, 2015). Meanwhile, it is equally informative to examine how interrelatedness among different sets of family relationships change systematically over a long period of time. Specifically regarding the study of the MR–PCR link, little compelling evidence has adequately addressed how the daily effect of MR on PCR may change over time, especially from middle childhood to adolescence.

Present Study

Happy and healthy PCR provides an advantageous foundation for children’s emotional and behavioral development, thus an understanding of fluctuations and changes in PCR has the potential to inform intervention and prevention programs that target on improving children’s adjustment. There is also a need for study designs and statistical techniques that consider daily fluctuations of family relationships in the context of over-year changes and paternal-vs-maternal differences. By adopting a measurement-burst design, the present study first aimed to examine the extent to which PCR assessed at the daily level changed over years and varied based on daily experiences (Aim 1). We hypothesized that variability in daily PCR was more attributable to MR fluctuations everyday than age changes over years. Then, we aimed to take a closer examination of the daily MR-PCR link within a family (Aim 2). Two specific research questions were examined to explore Aim 2: 1) how was MR related to same-day, as well as next-day, PCR for fathers and mothers? Based on previous research, we expected to observe spillover association for the same-day effect and compensatory association for the next-day effect. We also expected to see differences between maternal and paternal MR–PCR link. 2) How did the daily MR–PCR link change over years from middle-childhood to adolescence? And were such over-year changes different for fathers and mothers? We hypothesized stronger MR–PCR associations as children developed. No hypothesis was formed for father–mother differences in these changes, given the mixed findings of past research and the relative lack of studies using daily diary data.

Method

Participants

Couples and their child were drawn from a longitudinal study in a small city in the Midwest. Participants were recruited from the community through flyers, newspaper/TV/radio advertisements, community events, and letters distributed to local schools and neighborhood residents. To be eligible to participate, the couples had to be living together for at least 2 years with an 8- to 16-year-old child who lived with them for majority of the time. 299 couples participated at Wave 1 (W1), among which 250 remained at Wave 2 (W2) and 248 were retained at Wave 3 (W3). Families lost to attrition at either W2 or W3 did not differ from the retained families on any of the study variables (i.e., marital and parent–child relationship) at W1 or any demographic variables measured, including child gender, marital status, family income, or mother education. Fathers from families that dropped out at both W2 and W3 had lower education than those who stayed in the study, t = - 2.26, p < .05.

237 families (out of the 299) at W1 completed the father-reported and mother-reported daily diaries, in addition to the laboratory portion of the study. Therefore, the final analyses of W1 diary data were conducted on these 237 families. Most couples were married (97.5%) for an average of 13 years (SD = 6.35). Parents’ mean age was 40.15 years (SD = 6.52, range = 25 to 70) and 37.82 years (SD = 6.03, range = 24 to 70) for fathers and mothers, respectively. Children were 7 to 17 years old (M = 10.68 years, SD = 2.40; 51.5% girls), including 160 in middle childhood (age 7 to 11), 55 in early adolescence (age 12 to 14), and 22 in late adolescence (age 15 to 17). The distribution of children’s age was not skewed (skewness = .646) despite of the imbalance across age groups.

Participants were representative of the community from which they were drawn (Papp, Kouros, & Cummings, 2009). Based on mothers’ reports, the median family income was in the US$40,001–$65,000 range (n =166); 3 families reported a combined family income less than US$10,000, 14 reported a family income between US$10,001 and US$25,000, 52 reported a family income between US$25,001 and US$40,000, 32 reported a family income between US$65,001 and US$80,000, 28 reported a family income above US$80,000, and two mothers did not report this information. 90.7% of the mothers were European American, 4.6% were African American, 3.4% were Hispanic, .4% was Native American, and two mothers did not report their ethnicity or race. With regard to fathers, 90.7% were European American, 5.5% were African American, and 3.8% were Hispanic. The analytic sample (i.e., families where both the mother and the father completed diaries) was 201 families for W2 and 190 families for W3.

Procedure

Couples and their child visited the laboratory at each of the three waves spaced one year apart. During their visits, mothers and fathers completed questionnaires, including items about marital satisfaction, marriage length, and child gender. In addition, couples were instructed to complete a daily paper diary entry everyday consecutively for 15 days, beginning the next day following the laboratory visits. Occasional exceptions were allowed in which parents might miss a day or two when they were out of town. In such situations, parents were instructed to simply extend the period of diary days so that in the end, they filled out a total of 15 diary entries. All families completed the diary at the end of each day and parents were instructed not to discuss their answers with each other. Fathers and mothers mailed back their dairies separately when 15 entries were completed. Each family received $140 for their participation. The study procedure was approved by the University of Notre Dame’s review board (approval number is not available); the name for the larger project was Couples and Kids Project.

Measures

Relationship quality.

MR and PCR were assessed via daily diaries. Concerning the measurement of MR, on a scale ranging from 0 (negative) to 10 (positive), both parents responded each day to the question “what is the emotional quality of your relationship with your spouse that day?” With regard to PCR, on the same scale, they also rated the overall emotional quality of their relationship with their child that day. Consisted with a functionalist perspective on emotions (Davies & Cummings, 1994), we assumed that participants would best reflect and report their own perceptions for the emotional quality with their spouse and the child, and that this task did not require, and would not necessarily benefit from, special training. Therefore, we did not provide training for completing the diaries about emotional qualities of daily interactions.

The mean scores for mothers’ MR and PCR were approximately 6.85 and 7.25; the mean scores for fathers were approximately 6.70 and 6.90 (see Table 1 for specific descriptive statistics for MR and PCR at each wave). At the daily level, correlations between mothers’ and fathers’ daily reports of MR (r = .30) and PCR (r = .18) were small to moderate, indicating that fathers and mothers were having somewhat distinct perceptions of their relationship quality with the spouse and with the child. Significant within-person variability existed in the MR (from 54.9% to 69.8%) and PCR ratings (from 49.9% to 60.5%) across the 15 assessment days, supporting the validity of daily diary method to capture daily fluctuations in family relationship quality, as shown in Table 1. Because data from both members of the couple are required for testing cross-person effects, only diaries from days when the other partner also completed a diary entry were kept for analyses. As a result, 2.4%, 1.8%, and 1.4% diaries were excluded from final analyses, resulting in 6936, 5924, and 5622 diary entries in total for three waves, respectively.

Table 1.

Descriptive Statistics of Mother and Father Person-Averaged MR and PCR Quality at Each Wave

		Mother							Father
		MR1	MR2	MR3	PCR1	PCR2	PCR3		MR1	MR2	MR3	PCR1	PCR2	PCR3
	M (SD)	6.88 (2.02)	6.80 (2.10)	6.85 (2.01)	7.21 (1.82)	7.19 (1.84)	7.31 (1.72)		6.72 (1.98)	6.71 (2.04)	6.69 (2.03)	6.90 (1.85)	6.89 (1.83)	6.93 (1.78)
	Validity1	69.8%	54.9%	63.6%	60.5%	52.4%	60.0%		59.6%	55.7%	58.5%	56.5%	49.9%	51.4%
		Correlations
		Mother
Mother	MR1	1.00
	MR2	0.62	1.00
	MR3	0.47	0.52	1.00
	PCR1	0.70	0.39	0.29	1.00
	PCR2	0.41	0.56	0.35	0.53	1.00
	PCR3	0.33	0.31	0.58	0.46	0.59	1.00
		Mother & Father							Father
Father	MR1	0.55	0.40	0.37	0.30	0.24	0.23		1.00
	MR2	0.44	0.61	0.39	0.26	0.32	0.13		0.56	1.00
	MR3	0.44	0.36	0.54	0.18	0.23	0.23		0.63	0.61	1.00
	PCR1	0.41	0.26	0.35	0.43	0.34	0.34		0.71	0.41	0.53	1.00
	PCR2	0.42	0.46	0.32	0.30	0.47	0.28		0.53	0.77	0.54	0.56	1.00
	PCR3	0.37	0.32	0.36	0.26	0.38	0.41		0.57	0.49	0.77	0.67	0.60	1.00

Note. The correlations were all significant at the .001 significance level. The correlations are from three blocks. The Mother block and the Father block contain the correlations for mothers and fathers, respectively over time. The Mother & Father block contains the correlations between mothers and fathers at each time point and over time. MR1: marital relationship averaged across 15 daily diary reports for each parent at Wave 1 (similarly, MR2 represents averaged scores at Wave 2, and MR3 for Wave 3); PCR1: parent–child relationship averaged across 15 daily diary reports at Wave 1 (similarly, PCR2 represents averaged scores at Wave 2, and PCR3 for Wave 3).

¹Validity refers to the amount of variability in parents’ ratings of MR and PCR accounted by within-person variation.

Global marital satisfaction.

Mothers and fathers reported on the Marital Adjustment Test (MAT; Locke & Wallace, 1959) that assessed marital disagreements, relationship cohesion and communication, and overall satisfaction with the relationship. Possible scores of the MAT ranged from 2 to 158, with higher scores reflecting greater marital satisfaction. Although mother-reported marital satisfaction (M = 114.90, SD = 23.45) did not differ from that of father-report (M = 115.27, SD = 21.41) in W1, there were significant differences between mothers’ (W2: M = 109.92, SD = 25.51; W3: M = 108.06, SD = 25.20) and fathers’ MAT (W2: M = 115.44, SD = 23.06; W3: M =113.69, SD = 24.78) in W2 (t = −2.98, p < .01) and W3 (t = −3.59, p < .001). Alphas for fathers’ and mothers’ MAT scores across the three waves ranged from .70 to .80.

Data Analysis Plan

Dyadic multilevel models (Bolger & Laurenceau, 2013) were used to examine the two main research aims using the PROC MIXED procedure in SAS version 9.1 (Littell, Milliken, Stroup, & Wolfinger, 1996). Each model contained three levels: the daily level (Level 1, within-family), the wave level (Level 2, within-family and between-wave), and the family level (Level 3, between-family). Level 1 included separate indicator variables for mothers and fathers in order to simultaneously create intercept and slope parameters for these paired individuals. Time was included (such that 0 represented the first daily diary) as a covariate at Level 1 in order to control for systematic changes in daily PCR over the period when participants completed 15 daily diaries (Wang & Maxwell, 2015).

To test Aim 1 regarding the daily variability and yearly changes in PCR, we fit an intercept-only model (Model 1, see below for Level 1 equation) with variable Wave (recoded so that 0, 1, 2 represented W1, W2, and W3) added in Level 2 to predict the level of daily PCR (i.e., β_0ji and β_5ji). Because the effects of time (β_4ji and β_9ji for mothers and fathers) were controlled in the model as suggested by Wang and Maxwell (2015), the intercept estimates for mothers’ and fathers’ PCR at Level 1 then represented their parent–child relationship quality on the first day of the study at W1 (i.e., time = 0, wave = 0). Parameter estimates for Wave (added in Level 2) quantified the intrafamilial changes of mothers’ and fathers’ first-day PCR over three years, thus capturing the yearly changes in daily PCR. Random effects of the intercepts for both parents were estimated, representing the extent to which daily mother-child and father-child relationship changed over years; we also examined the Level-1 residual variance for mothers and fathers, representing the extent to which daily PCR varied based on daily experience. Equations for Levels 2 and 3 were presented in full in the Supplemental Material, given space limits of the journal.

$$PC{R}_{tji}=mothe{r}_i\left[{\beta }_{0ji}+{\beta }_{4ji}tim{e}_{tji}\right]+fathe{r}_i\left[{\beta }_{5ji}+{\beta }_{9ji}tim{e}_{tji}\right]+e_{tji}$$ Model 1, Level 1

Two conditional models were used to examine Aim 2, concerned with understanding the daily associations between MR and PCR. One model was used for the Same-day effect of MR on PCR (Model 2-SameDay) and another one for the Next-day effect of MR on PCR (Model 2-NextDay). In both models, Level-1 included within-family daily diary ratings that were calculated by subtracting individual’s wave-average score from their daily scores. By disaggregating these within-family, within-wave effects from the within-family, between-wave (Level 2) and between-family effects (Level 3), these models allowed us to test the change in PCR associated with a daily (either same-day or previous-day) one-unit increase in MR (within-family, within-wave effect) that would otherwise be blended with the between-wave or between-family effects. Level-1 for the lagged-day model (Model 2–NextDay) is as follows:

$$\begin{gathered} PC{R}_{(t+1)ji}=mothe{r}_i[{\beta }_{0ji}+{\beta }_{1ji}\left(AM{R}_{tji}-AM{R}_{\cdot ji}\right)+{\beta }_{2ji}\left(PM{R}_{tji}-PM{R}_{\cdot ji}\right)+{\beta }_{3ji}PC{R}_{tji} \\ +{\beta }_{4ji}tim{e}_{tji}]+fathe{r}_i[{\beta }_{5ji}+{\beta }_{6ji}\left(AM{R}_{tji}-AM{R}_{\cdot ji}\right)+{\beta }_{7ji}\left(PM{R}_{tji}-PM{R}_{\cdot ji}\right)+{\beta }_{8ji}PC{R}_{tj} \\ +{\beta }_{9ji}tim{e}_{tji}]+e_{tji} \end{gathered}$$ Model 2–NextDay, Level 1

where AMR_tji and PMR_tji were individuals’ own (Actor) and spouse’s (Partner) report of MR for individual i on diary day t in wave j, respectively. AMR_·ji and PMR_·ji were the averaged MR scores across the 15-day period for each individual i and his or her spouse in wave j, which were included for person-mean centering. Two dummy codes, mother (1: mother; 0: father) and father (1: father; 0: mother), were simultaneously included. PCR quality (PCR_(t+1)ji) was predicted by the previous day’s rating of MR, controlling for the autoregressive effect of PCR (PCR_tji). The fixed effects of β_1ji and β_2ji represented mothers’ actor and partner effects of MR on next-day PCR, while β_6ji and β_7ji quantified fathers’ daily actor and partner next-day effects.

At Level-2 of Model 2-NextDay, we examined Wave effect, or the effect of intraindividual changes in daily MR-PCR link. MR at Level 2 was centered by subtracting each person’s total average across all waves of data (AMR_··i and PMR_··i) from each person’s wave average (AMR_·ji and PMR_·ji); thus, terms involving Level-2 MR variables quantified mothers’, as well as fathers’, wave-level actor and partner effects of MR on next-day PCR. Special attention should be paid to the interpretation of coefficients involving Level-2 MR variables, due to the detrending statistical procedure to control for the effect of time. These within-person, between-wave effects represent the relations between levels of MR at a specific time point (i.e., on the first day, when time = 0) and average levels of PCR pooling over waves.

At Level 3, the grand-mean centered averaged MR scores (averaged between two dyad members across all diary entries from three waves) were included, such that Level 3 captured pure between-family effect of MR on daily PCR. Child age was added to Level-3 to account for the between-family differences in the MR-PCR link. An interaction term of Child age (Level 3) × Wave (Level 2) interaction was also added to explore the within-family actor effects of MR on daily PCR (i.e., β_1ji for mothers and β_6ji for fathers), providing information of whether over-year changes in within-family, lagged-day MR-PCR associations were different for families with children at different ages. Other covariates (i.e., marital satisfaction, relationship length, child gender, and child age) were included to Level-3 to control for between-family differences in the intercepts. Terms involving Level-3 MR variables quantified mothers’ family-level actor and partner effects of MR on next-day PCR, along with fathers’ family-level actor and partner lagged-day effects. Similarly with the Level-2 MR, the interpretation of the effects of Level-3 MR requires consideration of the detrending statistical procedure at Level-1. These effects capture the between-family, between-wave effects of the average level of MR at a specific time point (i.e., first day of daily diaries at Wave 1, time =0, wave =0) on the average levels of PCR pooling over families.

The specification for Model 2-SameDay was the same as Model 2-NextDay, except that the autoregressive effect of PCR was not included:

$$\begin{gathered} PC{R}_{tji}=mothe{r}_i\left[{\beta }_{0ji}+{\beta }_{1ji}\left(AM{R}_{tji}-AM{R}_{\cdot ji}\right)+{\beta }_{2ji}\left(PM{R}_{tji}-PM{R}_{\cdot ji}\right)+{\beta }_{4ji}tim{e}_{tji}\right] \\ +fathe{r}_i\left[{\beta }_{5ji}+{\beta }_{6ji}\left(AM{R}_{tji}-AM{R}_{\cdot ji}\right)+{\beta }_{7ji}\left(PM{R}_{tji}-PM{R}_{\cdot ji}\right)+{\beta }_{9ji}tim{e}_{tji}\right]+e_{tji} \end{gathered}$$ Model 2–SameDay, Level 1

Random effects of the intercepts and slopes of the daily actor MR-PCR link for both parents were estimated in Model 2-SameDay and Model 2-NextDay, representing the between-family variability around the average or fixed effects. All models were estimated using the Maximum Likelihood estimation to handle missing data, assuming the missingness was at random or completely at random (Fitzmaurice, Laird, & Ware, 2004).

Results

Preliminary Analyses

Means, standard deviations, and correlations among the wave-level study variables are presented in Table 1. On average, both fathers and mothers reported moderately high levels of relationship quality with the spouse and the child at each wave of assessment. Families in the analytic sample where both parents completed at least one daily diary (n = 237, 201, 190, respectively for W1, W2, and W3) did not significantly differ from the families that did not complete or only one parent completed daily diaries (n = 62, 49, and 58, respectively) in any of the three waves on the following variables: parents’ age, race, education, marital status, or family income. Marriage length and marital satisfaction, however, were found to be different. Specifically, couples in our analytic sample (M = 12.85, SD = 6.56) were married longer than the families that had no report or only reports from one parent (M = 10.48, SD = 6.49) at W1, t (294) = 2.50, p < .05. In addition, mothers in the analytic sample at W1 reported higher marital satisfaction (M = 114.54, SD = 23.84), t (297) = 3.23, p < .001. Therefore, marriage length and marital satisfaction, after grand-mean centering, were included as covariates in all models tested.

Aim 1: Daily Variability vs. Over-Year Changes in PCR

The averaged mother–child relationship quality (M = 7.21, 7.19, and 7.31, respectively at each wave) and father–child relationship quality (M = 6.90, 6.89, and 6.93, respectively at each wave) were at similar level over years (see Table 1). Results of Model 1 showed that neither mothers’ or fathers’ PCR at the first day of daily diaries changed over two years of assessment (ps > .1). Additionally, we examined the random effects of both parents’ intercepts of PCR in Model 1. Results showed that there were no Level 2 (Wave level) variance in mother-child or father-child relationship quality (ps > .1), indicating that the wave-average PCR did not vary across three waves. However, Level-1 residual variance of PCR was significant (mother–child relationship, σ² = 2.40, p < .001; father–child relationship, σ² = 2.26, p < .001), indicating that PCR varied significantly from day to day. Specifically, approximately 72% of the variance in mother–child relationship and 67% of the variance in father–child relationship were found at Level 1. In summary, PCR varied significantly across days but not across years.

Aim 2: MR and Same Day PCR

The existence of within-person, across-day variability in PCR enabled us to examine our second aim: whether father’s and mother’s MR accounted for variability in their own, as well as their spouse’s, relationship quality with the child on the same day. Controlling for marital satisfaction, marriage length, child gender, and day of assessment, both mothers’ (b = .353, p < .001) and fathers’ (b = .428, p < .001) within-family MR were positively associated with their own PCR on the same day (see also Table 2). That is, parents, either fathers or mothers, reported better relationship with their child on days when they reported higher emotional quality with their spouse than their personal average level.

Table 2.

Parameter Estimates of Models Examining the Influence of Daily MR, Wave-level MR, and Family-level MR on Daily PCR: Same-day and Next-day Models

			Same-Day Model		Next-Day Model
			b (SE)	b (SE)	b (SE)	b (SE)
			Mother rating of PCR at time t	Father rating of PCR at time t	Mother rating of PCR at time t+1	Father rating of PCR at time t+1
			Fixed effects
		(within-family) Previous-day PCR	–	–	0.188 (0.012)***	0.150 (0.013)***
		Daily MRt	0.353 (0.021)***	0.428 (0.023)***	−0.050 (0.017)**	−0.040 (0.021)+
		Daily MR_ partnert	0.001 (0.015)	−0.019 (0.015)	0.012 (0.017)	−−0.007 (0.018)
Intercept ×
		Wave-level MR	0.652 (0.090)***	0.635 (0.077)***	0.558 (0.080)***	0.546 (0.069)***
		Wave-level MR_partner	−0.111 (0.083)	−0.009 (0.084)	−0.082 (0.071)	0.014 (0.076)
		Family level MR	0.753 (0.071)***	0.817 (0.062)***	0.663 (0.062)***	0.713 (0.057)***
		Family level MR_partner	−0.124 (0.061)*	0.041 (0.067)	−0.115 (0.052)*	0.011 (0.060)
Wave ×
		Daily MR t	−0.001 (0.012)	−0.039 (0.013)**	−0.005 (0.014)	0.041 (0.015)**
		Daily MR_ partner t	−0.0003 (0.012)	0.018 (0.012)	−0.017 (0.014)	−0.009 (0.014)
		Wave-level MR	−0.094 (0.072)	−0.015 (0.067)	−0.105 (0.063)+	−0.012 (0.060)
		Wave-level MR_partner	−0.023 (0.073)	−0.097 (0.066)	−0.006 (0.064)	−0.102 (0.059)+
		Family level MR	−0.015 (0.044)	−0.011 (0.039)	−0.027 (0.038)	−0.023 (0.034)
		Family level MR_partner	0.002 (0.039)	−0.020 (0.043)	0.009 (0.034)	−0.006 (0.038)
Child age ×
		Daily MR t	−0.022 (0.009)*	0.002 (0.010)	0.005 (0.007)	−0.003 (0.008)
		Daily MR_ partner t	0.001 (0.004)	−0.0004 (0.004)	−0.005 (0.005)	−0.001 (0.005)
		Wave-level MR	−0.019 (0.022)	0.017 (0.020)	−0.020 (0.019)	0.017 (0.018)
		Wave-level MR_partner	0.029 (0.022)	−0.054 (0.020)**	0.031 (0.019)	−0.043 (0.018)*
		Family level MR	−0.005 (0.027)	−0.062 (0.021)**	0.012 (0.023)	−0.078 (0.019)***
		Family level MR_partner	0.009 (0.023)	0.018 (0.025)	−0.016 (0.019)	0.052 (0.023)*
Wave × Child age ×
	Daily MR t		0.015 (0.005)**	−0.0003 (0.005)	0.003 (0.006)	0.003 (0.006)
Other effects and controls
	Intercept		2.700 (0.423)***	0.924 (0.407)*	5.705 (0.113)***	5.650 (0.113)***
	Day		0.012 (0.003)***	0.015 (0.003)***	0.012 (0.004)**	0.023 (0.004)***
	Wave		0.133(0.251)	0.258 (0.246)	0.030 (0.030)	0.027 (0.030)
	Child age		−0.070 (0.141)	0.254 (0.135)+	−0.022 (0.022)	−0.032 (0.022)
	Wave * child age		−0.003(0.015)	−0.004(0.014)	−0.002 (0.013)	−0.006 (0.013)
	Marital satisfaction		−0.001 (0.002)	−0.002 (0.002)	−0.004 (0.002)+	−0.0002 (0.002)
	Child gender		0.087 (0.099)	0.026 (0.088)	0.074 (0.084)	−0.014 (0.078)
	Relationship length		−0.003 (0.008)	0.013 (0.008)	−0.005 (0.007)	0.013 (0.007)
			Random effects (variance estimates)
Level 3 Intercept variance			0.418 (0.059)***	0.252 (0.042)***	0.277 (0.042)***	0.196 (0.034)***
Level 3 Daily MR variance			0.046 (0.006)***	0.048 (0.007)***	0.007 (0.002)***	0.009 (0.003)**
Level-1 Residual variance			1.789 (0.036)***	1.550 (0.032)***	1.989 (0.037)***	1.755 (0.034)***
Level-1 Daily correlation			0.183 (0.011)***		0.084 (0.009)***

⁺p < .10

^*p < .05

^**p < .01

^***p < .001

Note. Daily MR is the within-wave, person-centered rating of marital relationship, calculated by subtracting individual’s wave-average score from their daily scores (AMR_tji − AMR._ji). Effects involving Daily MR indicate within-family effects of daily MR on PCR. Wave-level MR is individuals’ wave-centered score of marital relationship, calculated by subtracting each person’s total average across all waves of data from each person’s wave average (AMR._ji − AMR.._i). Effects involving wave-level MR indicate within-family, between-wave effects of MR on PCR. Family-level MR is the grand-mean centered marital relationship (averaged across all families). Effects involving family-level MR indicate between-family effects of MR on PCR. Daily MR_partner, wave-level MR_partner, and family-level MR_partner are spouse’s ratings of marital relationship, which correspond to (PMR_tji − PMR._ji), (PMR._ji − PMR.._i), and (PMR.._i − PMR..) in Model 2-SameDay and Model 2-NextDay. Effects involving Daily MR_partner indicate within-family partner effects of daily MR-PCR link, while effects involving wave-level MR_partner indicate between-wave partner effects and effects involving family-level MR_partner indicate between-family partner effects.

Wave-level (Level 2) and family-level (Level 3) MR for both parents were also positively associated with daily PCR, consistent with findings in the within-family daily associations (Level 1). That is, at the first day of diaries, parents tended to report better relationship quality with their child in years when they reported higher relationship quality with the spouse than their wave-average level (b = .652, p < .001, and b = .635, p < .001, for mothers and fathers respectively); moreover, parents with higher MR in general are expected to have higher PCR at the first day of Wave 1(b = .753, p < .001, and b = .817, p < .001, for mothers and fathers respectively).

The specification of our model also allowed us to test the partner (or cross-person) effects of MR on daily PCR: whether one parent self-reported MR (Level 1, 2, or 3) was related to the other parent’s (i.e., partner’s) relationship quality with the child. A significant partner effect was found at Level 3: father-report average MR was negatively associated with mother-report PCR (b = - .124, p < .05), supporting the compensatory hypothesis in this partner association. That is, mothers whose spouse reported lower average MR quality than other couples tended to report higher PCR with the child on the first day of Wave 1. No significant partner effects were found at Level 1 or Level 2.

Changes of daily MR-PCR link over years.

We examined whether the same-day association between MR and PCR changed over years. Results showed that the within-family effect of fathers’ MR on their own PCR on the same day decreased over years (b = − .039, p < .01). Thus, although significant associations were found between fathers’ daily MR on their PCR of the same day, the strength of the associations decreased over years.

A significant Child age (Level 3) × Wave (Level 2) × MR (Level 1) interaction was significant for mothers (b = .015, p < .01). We probed this three-way interaction by examining the Wave (Level 2) × MR (Level 1) two-way interaction at different child age. As shown in Figure 1, for families who entered the study with an 8-year-old (1 SD below the mean age) child, effect of mothers’ daily MR on their relationship quality with the child of the same day decreased over time. On the contrary, for families who entered the study with a 13-year-old (1 SD above the mean) teenager, the association between mothers’ daily MR on same-day PCR became stronger over time.

Figure 1.

The Effect of Mothers’ Daily MR on Their Own Same-day PCR (i.e., the Spillover Effect) Was Moderated by Wave and Child Age: The Wave (Level 2) × Child Age (Level 3) × Daily MR (Level 1) Interaction for Mothers

Aim 2: MR and Next Day PCR

Next, we tested the time-lagged model. MR predicted next-day PCR, controlling for the autoregressive effect of the previous day’s PCR quality (see Table 2 for results). Contrasting with same-day effects, mothers’ within-family MR negatively predicted their own PCR on the next day (b = − .050, p < .01), controlling for marital satisfaction, marriage length, child gender, and day of assessment. For fathers, the same negative relation was found at a trend approaching significance, b = − .040, p = .054. In other words, when parents, either fathers or mothers, reported lower relationship quality with their spouse than their personal average level, better relationship quality with their child was reported on the following day. These relations were in support of the compensatory hypothesis.

Changes of daily MR-PCR link over years.

We examined whether time-lagged association between MR and PCR changed over years. The strength of the association between fathers’ MR and their own PCR on the next day decreased over years (b = .041, p < .01). Thus, although on average, lower MR was related to higher father–child relationship quality on the subsequent day, this negative association became less strong over time.

Discussion

Adopting a measurement-burst design, the present study evaluated the extent to which daily assessed parent–child relationship (PCR) changed developmentally over two years and/or varied across a 15-day period within families as a function of family context, that is, daily marital relationship (MR) quality. Specifically, we examined whether daily variations in PCR quality were related to father-reported and mother-reported daily MR quality. Findings suggested that daily PCR did not change over the years of children transitioning into adolescence. Moreover, across all years PCR varied significantly from day to day based on parents’ daily experiences in marital interactions. Closer examination of the daily relations between marital and parent–child subsystems revealed that the impact of MR quality on PCR quality (a) varied as a function of same-day and/or next-day effects, (b) differed for fathers and mothers, and (c) changed over time across the two-year period of the present study. Overall, by situating the micro-level, day-to-day variability of PCR in the context of both micro- and macro-level daily and developmental changes, our study adds novel contribution to the understanding of PCR as a contextual and developmental process.

Variability and Changes in PCR

PCR assessed at daily level did not show significant changes developmentally over the two-year period studied (i.e., the assessment of macro-level change), yet exhibited significant within-person variations from day to day as a function of contextual family influences (i.e., the assessment of micro-level fluctuation). Our results suggested that, at least from parent’s perspective, the emotional quality of their relationship with the child did not change as children transitioned into adolescence. Although diminished closeness in parent–adolescent dyads have been reported in past studies (Laursen & Collins, 2009; Marceau et al., 2015), our result did not indicate loss of affection within the parent–offspring dyad within this age period, contrary to depictions of adolescence as a period of “storm and stress”. Future studies of the notion that children become more difficult as they enter adolescence are needed at the daily level, given that such belief held by parents has been found to predict teenagers’ behavior problems and worse parent–adolescent relationships (Jacobs, Chhin, & Shaver, 2005).

PCR varied within families as a function of contextual factors (i.e., MR quality) across the 15 days of daily diary assessments. That is, rather than changing developmentally in characteristic patterns over the two-year period, PCR was more likely to vary everyday based on daily contextual factors. The daily fluctuations of PCR associated with family influences have been demonstrated by several previous daily diary studies. For example, mothers were found to report more oppositional interactions with their child on days characterized by higher levels of stress (e.g., work stress, home stress, and relationship stress with the parent; Nelson et al., 2017). Interparental conflict were found to predict subsequent parent–child conflict (Kouros et al., 2014; Sherrill et al., 2017) and relate to more frictional, irritable parental behaviors (Sears et al., 2016). These studies examining daily-level influencing factors of PCR suggest that MR quality is an important contextual predictor of PCR quality.

MR and PCR Assessed at Daily Level: Spillover or Compensatory?

Providing further evidence for the influence of MR on PCR at the daily level (see also, Kouros et al., 2014; Sherrill et al., 2017), we found significant same-day and next-day MR–PCR associations. Interestingly, evidence supporting both the spillover (i.e., positive MR–PCR link) and compensatory (i.e., negative MR–PCR link) hypotheses emerged, suggesting that the influence of MR on PCR may depend on multiple factors, such as the actor (within-person or cross-person), the dyad (father-child or mother-child), and the time scale (same-day or next-day).

Significant spillover associations were found when examining the within-person, same-day effect: both parents’ daily and average MR were positively related to their own PCR of the same day. This is consistent with findings of both cross-sectional (Nelson et al., 2009; Ponnet et al., 2013; Stroud et al., 2011) and longitudinal studies (Davies et al., 2009; Gao et al., 2018). Demonstrating family relations at a daily level in our study supplements findings based on longer time frames, lending further support for the spillover effect of ones’ perceived relationship experience with their spouse on that with their child. Poor marital relationships may take a toll on one’s relationship quality with the child on the same day, and reiteration of such spillover effect may eventually contribute to long-term effects.

On the other hand, evidence for the compensatory effect was found in both cross-person (between father average MR and mother PCR) and next-day (mother MR–PCR) associations. Specifically, fathers’ low average MR quality was related to mothers’ better daily PCR quality, and lower levels of mothers’ MR were related to their own higher levels of PCR the next day. In other words, mothers tended to report better daily relationship quality with their child when their spouses reported poor averaged MR quality, or when they themselves reported poor MR on previous day. Although some research has supported the compensatory processes in family relationships with observational or survey methods (e.g., Brody et al., 1986; Nelson et al., 2009), our study contributes to the limited evidence for this process using a daily diary design (see also, Kouros, Papp, Goeke-Morey, & Cummings, 2014).

Interestingly, the compensatory effect was more evident in the cross-person (i.e., partner effects) than within-person associations (i.e., actor effects). For example, in the same-day effect model, mothers’ relationship with the child was found to be affected differently by their own daily experience in the marriage (i.e., spillover) than by their spouse’s average experience in the marriage (i.e., compensatory). Whereas mothers might have a hard time preventing the negativities they experienced in their relationship with the spouse from affecting their relationship with the child (i.e., the significant spillover effect), they were more capable of stopping, or even compensating for, the transmission of their spouse’s feelings in the marriage to their own PCR. To better understand the manifestation of compensatory effects in the cross-person association, it is important to discuss the conceptual underpinning of the cross-person association in the present study: why would one’s partner’s perception of marital quality affect one’s own relationship with the child? Consistent with Larson and Almeida’s (1999) emotional transmission hypothesis, we contend that one’s emotional quality in marital relationship would guide their behaviors in family relationships: parents who report lower emotional quality in the marriage may exhibit more withdrawing or aggressive behaviors; or parents may be likely to show affection and support to their spouse on days when they do not feel as close. As a result, the spouse may feel distressed and emotionally drained, which may further affect his or her relationship quality with the child. In addition to the account of emotional transmission hypothesis, we propose that the partner effects may also function as a result of the disruption in the coparenting system. Particularly, the “parent vs. spouse” role differentiation may be more evident in the cross-person MR-PCR link than in the within-person link for mothers. Therefore, it may be easier for mothers to separate their role as a mother and as a wife when they encounter negativities from their husbands than from themselves. Spouse’s report of poor marital quality may be perceived particularly by mothers as a potential threat to the coparenting subsystem, so that they are more likely to be aware of their parental role and build better relationship with the child (Le, McDaniel, Leavitt, & Feinberg, 2016). Such an account may also help explain the emergence of compensatory effects for mothers in the next-day model: activation of the coparenting subsystem takes time, thus it is not until the next day that the mothers realize the threat from their coparening partner and begin makin g efforts to compensate for the potential negativity in marital interactions.

These compensatory processes need to be interpreted in conjunction with gender differences: poor marital quality predicted improvements in mother–child, but not father–child, relationship quality on the following day. This finding lends support for the fathering vulnerability hypothesis (Cummings et al., 2010) that posits a greater impact of marital distress on fathers’ parenting than mothers’ parenting. Fathers’ greater susceptibility to poor marital functioning at the daily level found in our study contributes to the literature exhibiting different patterns of family relationship processes for fathers and mothers. One potential explanation is that fathers tend to take fewer caregiving responsibilities, view parenthood as less a fundamental role, and are more likely to blend the boundary between marital and parent–child subsystems (Belsky, Gilstrap, & Rovine, 1984). The less clearly defined boundary may lead to greater transmission of poor relationship quality from fathers’ MR to PCR quality. Mothers, on the other hand, may take a bigger role in organizing caregiving opportunities or practices (De Luccie, 1995), thus they are more likely to make intentional effort to counteract the unfolding cascade of negativity into child-rearing contexts (Belsky et al., 1991; Denham, Bassett, & Wyatt, 2010) by engaging in more positive parenting.

Collectively, the findings underscored the importance of differentiating family relationship processes for fathers and for mothers. These gender differences, however, have not been consistently reported. For example, in a landmark meta-analysis, Erel and Burman’s (1995) did not find parent gender differences in the MR–PCR link. Similarly, Ponnet and his colleagues (2013) reported that the strength of the association between family subsystems was similar for both fathers and mothers, including analyses of both within-person and cross-person links. These inconsistencies in the literature may possibly reflect differences in the time scales on which relationships were measured (i.e., yearly vs. daily) and specific constructs that were of the focal interest (e.g., marital conflict vs. marital relationship quality, specific parenting behaviors vs. parent–child relationship quality).

Influence of MR on PCR: Developmental Processes

We hypothesized that the daily MR–PCR link would become stronger as children entered or went through adolescence (Almeida et al., 1999; Seiffge-Krenke et al., 2010; Waite & Lillard, 1991). Contrary to our hypothesis, opposite results emerged from our study. Specifically, poor relationship quality of fathers was less likely to spillover to fathers’ own relationship with the child on the same day as children got older; fathers were also less likely to compensate for their poor marital relationship quality from the previous day by reporting better relationship with the child over a three-year period. Taken together, these results suggest that fathers’ relationship with the child may be less, either positively or negatively, affected by their relationship with the spouse when children enter or go through adolescence. Fathers may have developed better abilities at compartmentalizing their role as a husband and as a father over the period assessed, such that a negative relationship with the spouse may not exert as much negative impact on the same-day father–adolescent relationship, or result in improved father–adolescent interactions the next day.

Although the present study with three yearly assessments only allowed an investigation of a linear developmental trend of the MR–PCR link, our results indicate that the spillover effect between marital and parent–child subsystem may follow a quadratic change pattern for the maternal same-day MR–PCR association: the spillover effect decreased over years of middle-to-late childhood but increased again over years of adolescence. Probing the Child age × Wave × MR three-way interaction showed that the developmental trend of daily maternal MR-PCR association was different for families that entered the study with children at different ages. While mothers’ spillover effect of MR on their same-day PCR decreased for families with a child transitioned from childhood to adolescence (i.e, from 8 to 10 years old), this effect increased for families with a teenager over years (i.e., from 13 to 15 years old). Future longitudinal daily diary studies that follow families for an extended period of time will help facilitate our understanding of the developmental trend of family processes.

Limitations and Future Directions

The limitations of our study merit consideration. First, our sample was predominantly European American intact families, which may limit the generalizability of our findings to other samples and other populations, such as divorced families and populations with diverse racial compositions. Second, we used a one-item measure to assess parents’ relationship quality with the spouse and with the child, which may pose risks to construct validity. However, the one-item measure is a commonly used method in daily diary research because it is less time consuming and ensures compliance (Bolger et al., 2003). Adding evidence to the construct validity of our study, significant within-person response variations were found over diary days, indicating successful assessment of the subtle fluctuations in relationship quality. Third, although participants were instructed to complete one diary every day before bed, and record the date when diaries were completed each day, we can’t rule out the possibility that in some instances more than one diary was completed on the last day of the burst. Future studies are encouraged to use time stamps on paper-pencil diaries to track date/time or to employ electronic diaries to automatically capture time of completion. Lastly, in addition to the influence of family-level variables (e.g., relationship length, marital satisfaction, and child gender included in our models), family relationship quality may be susceptible to the influence of some daily-level characteristics, such as the number of interactions/conflict one had with the spouse and with the child and stressful life events happened on that day. A more refined measure that includes information on multiple aspects of daily life would contribute to better understanding of the family dynamics involving the transmission of relationship quality.

The novel findings of the present study suggest multiple directions for future research. Explanations of the spillover and compensatory effect should be considered. Existing literature has proposed several mechanisms by which spillover occurs (such as scapegoating, detouring, or drained emotional resources; see Erel & Burman, 1995). Nevertheless, little empirical effort has been devoted to testing any of the proposed accounts (of one notable exception, see Davies et al. 2009) or examining the mediating pathway at a daily level. Even less is known about the mechanism of the compensatory effect. We suspect that the occurrence of the compensatory effect may involve the activation of the coparenting subsystem. Marital negativities reported by the spouse may pose threat to the solidarity of the parenting-team, triggering disturbance in the coparenting system. In turn, the activation of a parent’s coparenting system may make the parental role more salient to him/her and subsequently provoke more positive parenting behaviors and better PCR. Given the role of coparenting as the executive system in a family (Minuchin, 1974), it remains for future research to test coparenting as the mechanism underlying the transmission of relationship quality between marital and parent–child subsystems. Finally, a more complete understanding of these processes in the future would benefit from including children in the assessments. For example, past empirical findings have suggested that children affect marital functioning (Emery & Kitzmann, 1995; Schermerhorn, Chow, & Cummings, 2010) and the quality of PCR. Children’s acting-out behaviors in response to interparental tensions could increase emotional stress in parents’ relationship with both the spouse and the child. Children’s role and perspective in the transmission of relationship quality in the family remain to be further examined and clarified by future studies.