Household Chaos, Parental Responses to Emotion, and Child Emotion Regulation in Middle Childhood

Abstract

Parents’ responses to children’s negative emotional states play a key role in the socialization of emotion regulation skills in childhood. Much of the prior research on child ER has focused on early development using cross-sectional designs. The current study addresses these gaps by using a longitudinal design to examine individual differences of ER at two times points in middle childhood. We examined the development of children’s ER by testing hypotheses about the interplay of parent response to emotions and household chaos in the prediction of individual differences in children’s ER. Participants were the mothers of children at 6 and 9 years of age among 224 families in a socioeconomically diverse sample that was part of an ongoing longitudinal study. Mothers completed questionnaires regarding themselves, their children, and their home environment. Mothers’ reports of better child ER at both time points were positively associated with mothers’ more supportive responses and negatively associated with mothers’ less non-supportive responses, as well as lower household chaos. Chaos statistically moderated the link between non-supportive parental responses to emotion and child ER, but only at 6 years of age. The strength of the link between child ER and non-supportive parental responses to emotions was strong only at lower levels of household chaos. At the beginning of middle childhood, family processes linking parent responses to child emotions and children’s developing ER may not function at higher levels of household chaos.

One essential component of successful development is learning to regulate emotional responses in socially appropriate and adaptive ways (Denham, Bassett, Zinsser, & Wyatt, 2014; Denham et al., 2003; Eisenberg, Spinrad, & Morris, 2002; Halberstadt, Denham, & Dunsmore, 2001; Morris, Silk, Steinberg, Myers, & Robinson, 2007; Morris, Criss, Silk, & Houltberg, 2017). Emotion regulation (ER) is defined as children’s management of their emotional experiences and expressions that allow then to function appropriately with an optimal level of engagement (Dunsmore, Booker, & Ollendick, 2013; Kim-Spoon, Cicchetti, & Rogosch, 2013; Rogers et al., 2016). These critical ER skills are particularly emphasized in developmental psychopathology research, which often finds that difficulty in regulating negative emotions such as anger and sadness is linked to emotional and behavioral problems (Cicchetti, Ackerman, & Izard, 1995; Eisenberg et al., 2001; Frick & Morris, 2004; Morris et al., 2007; Steinberg, & Morris, 2003). ER development is supported and enhanced through learning that occurs in socialization interactions and relationships with parents (Morris et al., 2007; Morris et al., 2017; Rutherford, Wallace, Laurent, & Mayes, 2015)—yet, these processes may be distinct depending on the broader family context (e.g., resources, stressors). The first goal of the current study was to examine individual differences in parent responses to emotion and children’s ER at two different time-points in middle childhood (from 6 to 9 years of age). Specifically, we tested whether better children’s ER skills would be positively associated with mothers’ supportive responses to emotion, and would be negatively associated with mothers’ nonsupportive responses to emotion. The second goal was to test whether the links between parent responses to emotion and children’s ER varied as a function of the specific contextual factor of household chaos—an understudied, potentially modifiable, environmental factor.

Parent Responses to Emotion and Children’s Emotion Regulation Development

Parents’ coping with and responses to their own child’s negative emotions have been identified as a key feature of the family context, because it consistently relates to children’s developing ER skills and deficits (Fabes, Poulin, Eisenberg, & Madden-Derdich, 2002; Morgan, Izard, & King, 2010, Pintar Breen, Tamis-LeMonda, & Kahana-Kalman, 2018; Song & Trommsdorff, 2016). Parents play an important role in emotion communication and socialization of ER skills in childhood, from the time children first begin to express their needs and desires (Eisenberg, Cumberland, & Spinrad, 1998; Eisenberg, Fabes, & Murphy, 1996; Fabes et al., 2002; Pintar Breen et al., 2018). Children model the emotions and ER processes learned both directly and indirectly from parents (Eisenberg et al., 1998; Eisenberg et al., 1996; Morris et al., 2007; Morris et al., 2017).

More specifically, parents can influence how children respond to and cope with emotionally evocative situations by reacting to children’s emotions, discussing emotions, and expressing their own emotions (Eisenberg, Cumberland, & Spinrad, 1998; Hurrell, Hudson, & Schniering, 2015; Perry et al., 2020). Their reactions to their child’s displays of negative emotions provide rich opportunities for emotion socialization and can directly influence children’s developing emotion management skills. Parent-child discussions of emotions assist children’s development of emotion-related knowledge, language and skills (Eisenberg et al., 1998; Hurrell et al., 2015). Parents help their children to verbally label their own emotions, empathize with and validate their children’s emotions, and help children to problem-solve by providing specific suggestions for coping with negative emotions (Eisenberg, 1998; Gottman, Katz, & Hooven, 1997; Morris et al., 2007; Perry et al., 2020). Through these various socialization experiences with parents, children hone their emotional expression and regulation skills and develop internal representations of emotions that define and support their regulation across a wide range of situations (Morris et al., 2007; Morris et al., 2017; Saarni et al., 1998; Saarni, Campos, Camras & Witherington, 2007). A number of studies have shown that individual differences in parents’ responses to children’s emotions can affect children’s regulation of their own emotional states. For example, in a small longitudinal study of 5- to 8-year-olds, parents’ emotion-coaching behaviors were associated with children’s ER (Gottman et al., 1996). In a subsequent study of 9- to 15-year-olds, youth who had more effective self-regulation of emotions and behavior reported higher levels of parental emotional support (Morris & Age, 2009). These two studies exemplify the wide range of developmental periods and constructs that have been examined in the parenting and child ER development literature.

Several consistent findings have emerged in the literature regarding parental supportive and non-supportive responses to child emotions—the focus of the current study. Parents who respond to their child’s emotions in supportive ways contribute positively to the development of children’s social and emotional competencies, including ER (Eisenberg, Fabes, Schaller, Carlo, & Paul, 1991; McQuade & Breaux, 2017; Morris et al., 2017). Supportive parental responses focus on children’s emotions and encourage appropriate expression of emotions and problem-solving strategies (Fabes et al., 2002). These kinds of supportive parenting behaviors to emotion are more common among parents who have greater awareness of children’s emotions and who exhibit more effective emotion socialization practices—and their children exhibit better social-emotional adjustment and regulation skills. These findings have been observed across early development - in toddlers, preschoolers, and school-age children (Acar et al., 2019; Morris et al., 2007).

One interpretation is that parents who are warm and responsive are more aware of their children’s emotions and therefore are more effective in helping the child learn to regulate and appropriately express their emotions (Houltberg, Morris, Cui, Henry, & Criss, 2016; Morris et al., 2007; Morris et al., 2017). Further, if parents are supportive when children express negative emotions, children may be less likely to become distressed and better able to manage their emotional reactions in the future. They learn appropriate strategies for handling their emotionally driven behaviors by processing parents’ messages and other relevant information (Acar et al., 2019; Fabes, Leonard, Kupanoff, & Martin, 2001). ER strategies—such as reframing a frustrating situation, redirecting attention, validating a child’s emotions, and helping the child to problem solve to regulate an emotion—have been linked to children’s coping and ER across early and middle childhood (Eisenberg et al., 1996; Gottman & Declaire, 1997; Morelen, Shaffer, & Suveg, 2016; Morris et al., 2007; Morris et al., 2011).

In contrast, parents’ non-supportive reactions, such as ignoring the emotion or using punitive control strategies, may impede the development of social-emotional competencies and ER in children (Fabes et al., 2001; McElwain, Halberstadt, & Volling, 2007; McQuade & Breaux, 2017; Morris et al., 2017). Non-supportive parental responses to emotion include devaluing the child’s problem, using punishment, and showing distress when the child expresses negative emotions (Fabes et al., 2002). In response to minimizing and punishing parenting behaviors, children may learn to suppress negative emotions rather than appropriately express them, and otherwise struggle with internal regulation of emotion experiences that can lead to an increase in negative emotionality and arousal (Gross & Levenson, 1993; Morris et al., 2007; Perry et al., 2020). Also, the parents who feel distressed when faced with their children’s negative emotional expressions are unlikely to support their children and more likely to control their children’s negative emotional displays by punishing or minimizing them, which in turn may lead children to suppress their emotions until they lose control. Thus, children may display decreased social and emotional competence (Fabes et al., 2002).

Prior evidence suggests that parental minimization of their child’s emotions is associated with children’s avoidance and angry outbursts (Eisenberg, Fabes, & Murphy, 1996; Snyder, Stoolmiller, & Wilson, 2003; Morris et al., 2007). Punitive parental responses to emotion have been associated with children’s poorer social-emotional competence and ER, with greater reliance on avoidance or revenge-seeking behaviors (Eisenberg & Fabes, 1994; Jones, Eisenberg, Fabes, & MacKinnon, 2002; Morris et al., 2007). Parents’ distressed responses to children’s negative emotions also have been linked with children’s poorer emotional and behavioral adjustment and functioning (Cassidy, Parke, Bukowsky, & Braungart, 1992). One important caveat is that most of the literature on supportive and non-supportive emotion socialization practices has relied on studies of white families. Emotion socialization behaviors, and their associations with variation in child ER, may vary depending on cultural context.

Household Context: Order or Chaos

Parental emotion socialization behaviors operate within a broader household context that also impacts children’s development. In the current study, we focused on an important and potentially modifiable, yet understudied, aspect of this context—household chaos. Prior evidence suggests that children’s ER development is optimized in home environments that are typically calm and predictable. In such contexts, children may be better situated to experience and learn from emotion socialization and other parent-child relationship processes that support ER development (Coldwell, Pike, & Dunn, 2006; Bowes, Maughan, Caspi, Moffitt, & Arsenault, 2010). In contrast, homes that are crowded, noisy, and lacking in routines (i.e., are “chaotic”) are more often emotionally and physiologically distressing, and there may be fewer opportunities, or more opportunities but less capacity, for socializing adaptive ER strategies and skills (Berry et al., 2016; Coldwell et al., 2006; Dumas, et al., 2005; Garrett-Peters et al., 2016; Garret-Peters, Marsh, Dobson, & Maddison, 2020; Mokrova, & Vernon-Feagans, 2019; Vernon-Feagans, Willoughby, Garrett-Peters, & The Family Life Project Key Investigators, 2016; Wachs, 2005).

Higher levels of chaos consistently are associated with harsher parenting, more parenting stress, more reliance on ineffective and reactive discipline, and less parental sensitivity and responsiveness to children’s needs (Dumas et al., 2005; Garrett-Peters et al., 2019; Smith, Prinz, Dumas,& Laughlin, 2001; Tucker, Sharp, Van Gundy, & Rebellon, 2018; Vernon-Feagans et al., 2016; Wachs, 2005; Wachs & Corapci, 2003; Zvara et al., 2014). The literature also strongly implicates household chaos as a correlate and statistical predictor of children’s and adolescents’ social-emotional and cognitive development outcomes (e.g., behavioral and emotional problems, poorer social cue detection, lower IQ, poorer physical health; Dumas et al., 2005; Hughes & Ensor, 2009; Petrill, Pike, Price, & Plomin, 2004; Pike et al., 2006; Tucker et al., 2018; Vernon-Feagans et al., 2016; Wachs & Corapci, 2003).

The interpretation of these links between household chaos, parenting, and children’s functioning is that if they are chronic, higher levels of distractions and unpredictability in chaotic households may alter (i.e., modulate or statistically moderate) the links between parenting behavior and children’s functioning. These links reflect the socialization and self-regulation processes that are fundamental to children’s healthy development (Blair et al., 2007; Deater-Deckard, Wang, Chen, & Bell, 2012; Erickson, Drevets, & Schulkin, 2003; Evans et al., 2005; Vernon-Feagans et al., 2016). With regard to the current study’s emphasis on child ER, there is reason to believe that the statistical association between parenting behaviors and children’s self-regulation may operate differently at higher versus lower levels of household chaos (Hughes & Ensor, 2011; Hughes, Roman, & Ensor, 2014; Wachs & Evans, 2010, Vernon-Feagans et al., 2016).

Two possibilities for the direction of moderation have been positied (Coldwell et al., 2006). First, the link between negative parenting behavior and children’s functioning may be strengthened in chaotic environments. This is because chaotic environments increase arousal and reactive behavior. Second, the link between positive parenting behaviors and children’s functioning may be weakened in chaotic environments. This is because uncertainty, noise, and distractions may interfere with effective behavioral and emotion socialization processes.

There also are two alternative predictions not compatible with findings reported by Coldwell and colleagues (2006). Third, negative parenting links may be weakened in chaotic environments, because those processes are overridden by the effects of a chaotic home environment. Fourth, positive parenting links may be strengthened in chaotic environments, because it is in such contexts that positive parenting becomes even more important. Few studies have tested for these interaction effects, but the evidence so far supports the first prediction: that higher levels of chaos can amplify the potentially deleterious effects of non-supportive and harsh parenting on child and adolescent adjustment and functioning (Chen, Deater-Deckard, & Bell, 2014; Coldwell et al., 2006; Tucker et al., 2018).

The Current Study: Middle Childhood

Our study focuses on identifying the potential interactive effects of household chaos and parental responses to children’s emotions, with a focus on middle childhood. Parental strategies to teach ER differ with the age of the child (Eisenberg et al., 1999). In early childhood, parents typically begin to teach regulation strategies, and children’s self-regulation skills gradually improve across early and middle childhood (Eisenberg & Morris, 2002). In early childhood, much of the children’s ER happens with direct support from parents’ interventions and actions. As children develop more sophisticated cognitive and emotional skills later in childhood, they gradually become better at regulating their emotions independently (Lissa et al., 2019; Morris et al., 2007). This developmental progression and the emotion socialization history within the family may be particularly important for supporting the child’s successful transition to all-day schooling, a context that requires autonomous self-regulation of emotions without direct support from parents (Nelson et al., 2013; Sroufe, 1996).

Most of the work cited above has focused on early childhood and the transition to middle childhood, but there has been less attention paid to the associations between parent responses and children’s ER across middle childhood, with no prior studies examining this in the context of household chaos. Furthermore, there is very little prior research examining longitudinal data. In light of theory and the literature reviewed above (most notably, the predictions offered in Coldwell et al., 2006), in the current study we tested three hypotheses, each involving statistical prediction of parent-reported ER skills at 6 years of age and again at 9 years of age (with and without age 6-year predictors included):

H1: Better child ER skills will be positively associated with supportive responses to emotion, and will be negatively associated with non-supportive responses to emotion.

H2: Non-supportive parent responses to emotion will be negatively associated with children’s ER in more chaotic households.

H3: Supportive parent responses to emotion will be positively associated with children’s ER in less chaotic households.

Although we did not anticipate finding evidence of other interaction patterns (e.g., negative parenting links being weaker at higher levels of chaos; positive parenting links being stronger at higher levels of chaos), our analysis approach permitted detection of such effects if they were present. In addition, the emphasis in the current two-wave study (6- and 9-years) was on parent-reported chaos and parental responses to emotions as statistical predictors of variance in parent-reported child ER—in line with some of the socialization theories and empirical studies of ER development cited above (e.g., Hurrell et al., 2015; McQuade & Breaux, 2017; Perry et al., 2020; Rogers et al., 2016). However, bidirectional “child effects” on parenting also are likely to be operating (Eisenberg, 1998; Morris et al., 2007). Poorly regulated child emotions and behaviors (e.g., emotional reactivity; attention deficits; hyper-impulsivity) are well-established as statistical predictors over time of harsher and less supportive parenting behavior (Te Brinke, Deković, Stoltz, & Cillessen, 2017; Zvara, Sheppard, & Cox, 2018). Therefore, we also conducted supplemental analyses, testing for potential additive and interactive effects of child ER and household chaos, in the prediction of parental responses to emotions.

Method

Participants

Participants in the current analyses were the mothers of 6- and 9-year old children with data collected in an ongoing longitudinal study on ER and cognition-affect development (Perry, Dollar, Calkins, & Bell, 2018; child physiology and cognitive skills were assessed in the larger study, but those data were not used in the current analyses). The current sample of maternal participants is comprised of two of three cohorts (75% of the entire study; one cohort lacked data collection at age 6 years). The cohorts were recruited during infancy equally using mailing lists, media advertisements, flyers, and word of mouth from two research locations in the United States (a rural town in south central Appalachia and a mid-sized city in the mid-Atlantic region). Each research location recruited half of the participants in the longitudinal study. The recruitment criteria for infants included full gestation, typical birth weight, and having no prenatal or birth complications. Prior to the 6-year assessment, these two cohorts also had lab visits at 5, 10, 24, 36, and 48 months.

At age 6, the sample included 277 mother-child pairs, but there were 53 who were missing some or all data at age 9. We compared this sub-group who had missing data at 9-years (n = 53) to the sub-group that had complete data at both waves (n = 224). There were no statistically significant mean differences for child ER, parent responses to emotion, home environment, parental education, and parental employment status (p-values for independent samples t-tests ranged from .224 to .915).

We used listwise deletion of missing data across both waves, to select a sample for analyses that included 224 mothers. Listwise deletion is the simplest method for handling missing data, but it can yield biased estimates compared to more complex multiple imputation methods. Yet listwise deletion biases estimates only when data are not missing completely at random (MCAR). We used Little’s algorithm for testing MCAR (Rhemtulla & Little, 2012), and the data met the assumption of MCAR. Therefore, we retained the simple listwise deletion method.

Participants were recruited from a rural town in Virginia and a mid-size city in North Carolina in the United States. For the current study, 50.4% of children were male, and the remaining 49.6% were female. Children were 6.63 years old on average at the 6-year time point (SD = 0.42) and 9.30 years old on average at the 9-year time point (SD = 0.40). Mothers reported on the race of themselves and their children; these were reported to be the same within each family. The participants were predominantly White (78%), with 14% identifying as African American, 4% as multi-racial, and 1% as Asian.

Parental education varied widely. Of the mothers, 2.2% had not completed high school, 6.7% had a high school diploma, 24.7% had a two-year college degree or had completed some college, 35.9% had completed a four-year degree, and 30.5% completed an advanced graduate or professional degree. Among the fathers, 1.8% completed only elementary school, 7.6% had not completed high school, 14.8% had a high school diploma, 20.6% completed a two-year college degree or some college, 21.6% had a four-year degree, and 33.6% completed an advanced degree.

Procedure

At the beginning of each lab visit, signed consent was obtained from the mothers, and verbal assent was obtained from the children. Prior to the lab visits, mothers completed a set of questionnaires regarding themselves, their child, and their home environment, and demographics. For each assessment, mothers received an honorarium, and children were given an honorarium and a small gift. The lab visit was 2.5 hours. For 238 mothers at age 6 and 200 mothers at age 9, questionnaires were collected in the context of the research lab visits. For 39 mothers at age 6 and 49 mothers at age 9, questionnaires were mailed to the families because they had moved away from the local area or were unavailable for a research lab visit. Those mothers completed the questionnaires at home and mailed them back with a signed consent form. For mailed questionnaires, mothers received a small honorarium.

Data analyses were conducted using SPSS v.23 (IBM Corporation). Out of our final sample (n = 224), we compared the sub-group of mothers who completed the questionnaire during a lab visit (n = 195 at age 6, n = 190 at age 9) to the sub-group that completed them by mail (n = 29 at age 6, n = 34 at age 9) on the main study variables. There were some statistically significant mean differences. When compared to the “lab questionnaire” mothers, the “mailed questionnaire” mothers reported less non-supportive responses at 6-years (t(222) = 2.19, p = .03, Cohen’s d = 0.45); at 9-years, they reported more supportive (t(222) = 2.78, p = .006, d = 0.60) and less non-supportive responses(t(222)= −2.01, p = .046, d = 0.39). We ran analyses with questionnaire completion method (e.g., completed in lab versus home) included as a binary covariate and results did not change; results are reported without this covariate.

Measures

Parent Responses to Emotion.

Mothers completed the Coping with Child Negative Emotions Scale (CCNES; Fabes, Eisenberg, & Bernzweig, 1990; Fabes et al., 2002). Prior research on the CCNES measure reports strong internal consistency and construct validity, as well as high test-retest reliability (Eisenberg et al., 1999; Fabes et al., 2002). This questionnaire assesses how parents respond to their children’s emotional distress. It includes 12 scenarios in which their children experience negative emotions (e.g., angry or sad) in public and private situations, and mothers were asked to indicate how they would respond to each scenario. Each scenario has six responses, and mothers rated each response on 7-point Likert-type scale regarding their likelihood of reacting that way (i.e.,1 = Very Unlikely, 4 = Medium, 7 = Very Likely).

The measure includes six subscales: emotion-focused (i.e., strategies that focus on the emotion causing child distress and help child identify source of stressors), problem-focused (i.e., behaviors that help the child solve the problem that caused the distress), expressive encouragement (i.e., responses that reinforce expression of emotions to validate feelings), minimization (i.e., reduces the perceived seriousness and importance of the situation), punitive (i.e., verbal or physical punishment to stop the child’s negative emotion expression), and distress (i.e., becoming emotionally upset oneself in response to child negative emotions). Following previous research, supportive and non-supportive composite scales were calculated by combining three subscales (Denham & Kochanoff, 2002; Fabes et al., 2002; Shadur & Hussong, 2019); supportive responses (36 items, Cronbach’s α in the current sample = .91 for both 6-years and 9-years) included emotion-focused, problem-focused, and expressive encouragement, and non-supportive responses (36 items, Cronbach’s α = .82 for 6-years and α = .86 for 9-years) included minimization, punitive, and distress responses.

Child Emotion Regulation (ER).

Mothers completed the Emotion Regulation Checklist (ERC; Shields & Cicchetti, 1997) which assesses their children’s ER and emotionality according to the parents’ perceptions. ERC is an age-appropriate measure for six to twelve years old (Shields & Cicchetti, 1995; Shields & Cicchetti, 1997). We used the emotion regulation subscale that includes eight items regarding expression of emotions, empathy, emotional self-awareness, and the tendency to modulate emotional arousal (e.g., “Is a cheerful child”, “Can say when he/she is angry”, “Can wait for something when asked to do”). The items were rated on a 4-point Likert-type scale (1 = Never, 2 = Sometimes, 3 = Often, 4 = Almost Always); these were summed to create an ER scale score (8 items, Cronbach’s α = .63 for the children at both 6-years and 9-years).

Household Chaos.

Mothers completed the short version of the Chaos, Hubbub and Order Scale (CHAOS; Matheny, Wachs, Ludwig, & Phillips, 1995) which is designed to assess the level of confusion and disorganization in the child’s home environment. The scale has been validated against observer ratings of environmental conditions (Deater-Deckard et al., 2009; Matheny et al., 1995). The short version consists of six items rated on a 5-point Likert-type scale (rather than the original instrument’s binary (yes/no) scale from the Matheny et al. questionnaire). The short version was used in two studies in the UK: the Twin Early Development Study (Pike et al., 2006) and the Sisters and Brothers Study (Coldwell et al., 2006). Mothers were asked to indicate how much each statement describes their home environment. The questionnaire includes six items (e.g., “You can’t hear yourself think in our home”, “It’s a real zoo in our home”, and “The atmosphere in our house is calm” [reverse scored]). They were measured on a 5-point scale (1 = definitely untrue, 5 = definitely true) and were averaged to create an overall score (6 items, Cronbach’s α = .65 for 6-years and α = .66 for 9-years).

Covariates.

We controlled for SES by using individual indicators of mother’s and father’s years of education (1 = Grade school, 2 = some high school, 3 = high school graduate, 4 = some college or 2-year college, 5 = college graduate of 4 years of college, 6 = Master’s degree, 7 = Ph.D. or M.D. or other doctoral degree ), and employment status (0 = Not employed, 1 = Employed). We also controlled for the sex of the child (1 = Male, 2 = Female).

Results

Descriptive statistics and bivariate correlations were computed. These are reported in Table 1. All variables showed some skewness (i.e., skewness statistic from ± 0.33 to 0.86) but approximated a normal distribution spanning the entire range of each scale. Of particular interest was the mother-reported household chaos score, given its key role as a tested moderator in the current study. It was slightly skewed toward the “lower chaos” end of the scale, with the M and SDs at both time points being below the scale midpoint of 3.00. Nevertheless, the distribution was wide, spanning nearly the entire range of possible scores (one standard deviation covered two-thirds of a point on the 5-point scale). The distribution was very similar to the distribution reported in several other studies in the U.S. and the U.K. that have used the same version of the instrument (Coldwell et al., 2006; Deater-Deckard et al., 2009; Deater-Deckard, Chen, Wang, & Bell, 2012). Household chaos, along with the two mother-reported response to emotions scales and the mother-reported child ER scale, showed moderate longitudinal stability from 6- to 9-years (i.e., rs from .55 to .67, p < .01).

Table 1.

Correlations and Descriptive Statistics

	Age 6				Age 9
Variable	1	2	3	4	5	6	7	8
1. Chaos Age 6	1.00
2. Supportive responses Age 6	−.03	1.00
3. Non-supportive responses Age 6	.14*	−.34**	1.00
4. Emotion regulation Age 6	−.21**	.21**	−.18**	1.00
5. Chaos Age 9	.67**	−.07	.18**	−.19**	1.00
6. Supportive responses Age 9	.02	.55**	−.30**	.24**	−.11	1.00
7. Non-supportive responses Age 9	.21**	−.23**	.64**	−.20**	.24**	−.26**	1.00
8. Emotion regulation Age 9	−.14*	.20**	−.18**	.56**	−.23**	.31**	−.27**	1.00
M	2.23	5.51	2.60	27.66	2.23	5.38	2.67	27.41
SD	0.68	0.67	0.58	2.66	0.65	0.65	.60	2.78

^*p< .05

^**p< .01 (all two-tailed tests)

All data were based on maternal reports of their own reactions to child negative emotions, their child’s ER, and chaos. Bivariate correlations between chaos, mothers’responses to emotion, and child ER showed that in general there was covariation (within and across the two waves) between higher chaos, lower supportive maternal responses, higher non-supportive maternal responses, and lower child ER (Table 1). The exception was a nonsignificant correlation between chaos and mothers’ supportive responses. Bivariate correlations with study covariates (e.g., maternal and paternal education; employment status; child sex) showed that higher maternal and paternal education were correlated with lower chaos at both 6- and 9-years. Higher maternal education and being employed were associated with better child ER just at 6-years.

To test the study hypotheses, we centered all predictors by standardizing, then used multiple standard (simultaneous) and hierarchical regression analyses to estimate the additive and interactive statistical effects (i.e., maternal supportive or non-supportive responses[analyzed together], household chaos) in the prediction of variance in child ER. Three sets of equations were estimated: one standard regression equation for variables at 6-years, a second standard regression equation for variables at 9-years without 6-year variables included as predictors, and a third hierarchical regression equation for variables at 9-years with 6-year variables included as predictors. We first ran each equation with all covariates included (i.e., maternal and paternal education and employment status; child sex), then dropped nonsignificant covariates, re-estimated the equation, and interpreted the equation with only significant covariates included.

Cross-sectional Results: 6-years

Results for 6-years are shown in Table 2. Regarding the anticipated direct associations between child ER, mothers’ responses and household chaos (H1), mothers’ reports of better ER were predicted by mothers’ less non-supportive responses to emotion, more supportive responses to emotion and less household chaos (as well as maternal employment as a covariate). Regarding the interaction effects of chaos and supportive responses (H2), the expected two-way interaction between non-supportive responses to emotion and chaos was significant, though the interaction term itself accounted for only a small portion of the variance in ER. Regarding the interaction effects of chaos and non-supportive responses (H3), the expected two-way interaction between supportive responses to emotion and chaos was not significant.

Table 2.

Standard Regression, Parent Responses Predicting Emotion Regulation at 6- and 9-years

6 years old F(6, 222) = 6.31, p = .000, R2 = .15	B	S.E	β	t	p
Maternal employmenta	.31	.14	.15	2.30	.023
Chaos (z)	−.16	.06	−.17	−2.61	.010
Non-supportive responses (z)	−.11	.07	−.11	−1.62	.107
Supportive responses (z)	.16	.07	.16	2.38	.018
Chaos X Non-supportive responses(z)	.14	.07	.14	2.00	.046
Chaos X Supportive responses (z)	−.08	.07	−.07	−1.09	.278
9 years old F(5,223) = 8.46, p = .000. R2 = .16	B	S.E	β	t	p
Chaos (z)	−.16	.06	−.16	−2.55	.011
Non-supportive responses(z)	−.18	.07	−.18	−2.66	.008
Supportive responses (z)	.25	.07	.25	3.82	.000
Chaos X Non-supportive responses (z)	.05	.06	.05	.83	.409
Chaos X Supportive responses (z)	−.02	.07	−.02	−.32	.750

Notes.

^a0 = unemployed; 1 = employed. z = variable standardized. The interaction term itself accounted for only a small portion of the variance in ER.

We conducted simple slopes analyses to interpret the significant two-way interaction effect. Results are shown in Figure 1, presented as standardized slopes (i.e., parent responses to emotions predicting ER) at different levels of the moderator (i.e., chaos). The association between higher non-supportive maternal responses to emotion and lower child ER was most substantial at lower levels of chaos: at 1 SD above the mean for chaos, β = .04, p = .692, and at 1 SD below the mean, β = −.35, p <.001.

Figure 1.

Non-supportive parent responses (z-score) predicting child emotion regulation at different levels of household chaos at 6-years. Solid lines reached statistical significance (p <.05).

Cross-sectional Results: 9-years

Results for 9-years are presented in Table 2. Regarding the hypothesized direct associations (H1), mothers’ reports of better children’s ER were statistically predicted by mothers’ less non-supportive responses, more supportive responses, and less household chaos. Regarding the hypothesized interaction effects (H2 and H3), neither of the anticipated two-way interactions between chaos and maternal responses to emotion was significant.

Longitudinal Results: 9-years, Controlling for 6-year Variables

Next, we examined whether the results for 9-years (reported above) would change when controlling for 6-year variables. We used hierarchical multiple regression to control for the predictors at 6-years prior to entering 9-year predictors. Results are reported in Table 3, and the findings for age 9-years reported above did not change. With regard to hypothesized direct associations (H1), mothers’ reports of better children’s ER at 9 years of age (while controlling for variables at 6-years) were statistically predicted by better children’s ER at 6-years, lower chaos at 9-years, mothers’ lower non-supportive responses at 9-years, and higher supportive responses at 9-years. Regarding the hypothesized interaction effects (H2 and H3), the anticipated two-way interactions between chaos and maternal responses to emotion were not significant.

Table 3.

Hierarchical Regression, 6-Year and 9-Year Predictors of Emotion Regulation at 9-Years

	Step 1					Step 2
Step 1: 6-year predictors F(6, 223) = 17.93, p = .000, R2 = .33	B	S.E	β	t	p	B	S.E	β	t	p
Emotion Regulation at 6-years (z)	.56	.06	.54	9.14	.000	.53	.06	.51	8.74	.000
Chaos at 6-years (z)	−.02	.06	−.02	−.30	.768	.10	.07	.10	1.33	.184
Non-supportive responses at 6-years (z)	−.07	.06	−.06	−1.06	.290	.07	.08	.07	.95	.341
Chaos X Non-supportive responses at 6-years (z)	.00	.06	.00	.03	.966	−.01	.07	−.01	−.21	.834
Supportive responses at 6-years (z)	.07	.06	.07	1.11	.267	−.02	.07	−.02	−.24	.812
Chaos X Supportive responses at 6-years (z)	.07	.07	.06	1.01	.312	.06	.07	.06	.91	.366
Step 2: Adding 9-year predictors F(11, 223) = 12.25, p = .000, R2 = .36						B	S.E	β	t	p
Chaos at 9-years(z)						−.16	.07	−.16	−2.20	.029
Non-supportive responses9-years (z)						−.17	.08	−.16	−2.18	.031
Chaos X Non-supportive responses at 9-years						.05	.06	.05	.79	.430
Supportive responses at 9-years (z)						.17	.07	.17	2.44	.015
Chaos X Supportive responses at 9-years (z)						−.08	.06	−.07	−1.19	.236

Notes.

^a0 = unemployed; 1 = employed. z = variable standardized

Prediction of Parenting Responses from Child ER

As a supplemental set of analyses, we examined whether there was any evidence of potential statistical bidirectional “child effects” of child ER on parenting responses to emotions. We estimated the three equations described above, with a key modification: Child ER was examined as a statistical predictor of maternal non-supportive and supportive responses to emotions as dependent variables. We used multivariate multiple regression to estimate the additive and interactive effects of statistical predictors (i.e., child ER, household chaos) of maternal supportive and non-supportive responses in the same equation. Three equations were estimated—at 6-years, at 9-years, and at 9-years with 6-year variables also included as predictors. The results are presented in Supplemental Results. Overall, the pattern of results was very similar to those reported above for the statistical prediction of child ER—clear evidence of direct associations between child ER and parent responses, and little evidence for a consistent interaction effect between child ER and chaos at 6-years and 9-years.

Discussion

The current study adds to extant findings and theory regarding the development of ER in childhood by testing hypotheses about the interplay of parental perceptions of their own responses to emotions and household chaos, in the prediction of individual differences in children’s ER. The current study addresses two major gaps in the literature. First, much of the relevant prior research on child ER has focused on early development, with comparatively little research examining individual differences across middle childhood, especially while also examining the parenting and broader household context. Second, nearly all the related prior studies have used cross-sectional designs, which do not permit estimation of temporal patterns in the covariation between constructs across multiple time points. The current study addresses both gaps by using a longitudinal design to examine individual differences at two times points in middle childhood (6 and 9 years of age). Study hypotheses investigating variation in children’s ER were tested at each time point separately and longitudinally. To our knowledge, no prior studies have examined associations between parent responses and child ER across middle childhood in the context of a wide range of levels of household chaos.

Consistent with the first hypothesis regarding direct associations between parent responses, chaos, and child ER (H1), we found that mothers’ reports of children’s better ER were statistically predicted at 6-years, and also at 9-years (regardless of whether the 6-year variables were included in the equation), by mothers’ more supportive responses, less non-supportive responses, and less household chaos. However, all significant effects were only observed within each time point; there were no significant longitudinal associations between parenting responses (at age 6) and ER (at age 9). It also is noteworthy that the overall pattern of results was consistent across the individual subscales within the non-supportive and supportive parent response composites, with regard to statistical prediction of child ER (see Supplementary results).

Overall, the findings for H1 with regard to “main effects” align with previous studies, suggesting that chaos and non-supportive parental responses to emotions are associated with poorer ER, but parents’ supportive responses serve a beneficial role in ER development, at two points in middle childhood (Dumas et al., 2005; Evan et al., 2005; Vernon-Feagans, 2016). Parent-child interactions regarding children’s negative emotions are an important aspect of emotion socialization, and they provide a key context for children to acquire emotion knowledge and regulation skills (Eisenberg et al., 1996; McElwain et al., 2007; Pintar Breen et al., 2018). Parents indirectly and directly teach their children about the meaning of emotions and how to respond to their own and other’s emotions, and these are critical experiences for healthy emotion socialization in families (Eisenberg et al., 1996; Pintar Breen et al., 2018).

Supportive responses include striving to understand the child’s negative emotions, helping the child solve the problem, and validating the child’s feelings. Our interpretation of the current study’s findings is that supportive responses can promote children’s developing ER by helping children feel emotionally safe, better enabling the regulation of emotions, and ensuring socially acceptable expression of emotions (Houltberg et al., 2016; Morris et al., 2017). In contrast, when parents are harsh and controlling, children have more difficulty regulating their emotions because parental responses to children’s emotions can increase distress (Morris et al., 2017). Non-supportive emotion parenting behaviors can create an environment where emotions are not discussed, are regarded as insignificant, are discouraged, or are overtly judged in a negative way (Denham, Zoller, & Couchoud, 1994; Morelen et al., 2016). However, the current study cannot address one major gap in the literature: whether these interpretations of emotion socialization processes generalize broadly across cultural contexts. There may be cultural differences in how these processes operate—for example, the ways in which parents and children interact, how children express their emotions, and how parents and children alike interpret and react to each other’s emotions (Emde, Biringen, Clyman, & Oppenheim, 1991; Morris et a.., 2007; Saarni, 1990). In particular, studies that focus on African-American (Nelson et al., 2013; Smith & Walden, 2001), Latinx (Pintar Breen, Tamis-LeMonda, & Kahana-Kalman, 2018), and Korean (Song & Trommsdorff, 2016) families have observed different distinct patterns of supportive and non-supportive emotion socialization behaviors and associations with children’s ER development.

Turning to the proposed interaction effects involving household chaos and non-supportive parent responses (H2), results showed that mothers’ reports of household chaos moderated the link between non-supportive responses and children’s ER (H2), but only at 6 years of age and the effect size was small. Post-hoc probing showed that the pattern was the opposite of what we had expected for H2. Rather than the link between non-supportive responses and poorer child ER being stronger at higher levels of chaos, this link was stronger at lower levels of chaos. This pattern was unexpected. We return later to this finding to offer an initial speculative interpretation. Regarding the expected interaction between chaos and supportive parental responses to emotions (H3), the anticipated interaction effect was not significant.

Thus, the current study’s findings for non-supportive (but not supportive) parent responses suggest that as children enter the middle childhood years at 6 years of age—but not toward the end of middle childhood at 9 years of age—chaos may interfere with the child’s ability to learn appropriate ER skills in parenting environments that are low in non-supportive behaviors. In the supplemental analyses in which we tested for potential child effects of ER on parental responses to emotions, the interaction effects between supportive responses and chaos as well as non-supportive responses and chaos were observed again only for chaos at 6-years. Thus, whether testing for statistical “parent” or “child” effects in the association between parental responses to emotions and children’s ER, chaos may disrupt or overwhelm ER socialization and self-regulation processes (in children and parents alike), weakening the statistical effects at higher levels of chaos and strengthening their effects at lower levels of chaos (Deater-Deckard, 2014; Evan et al., 2005; Vernon-Feagans et al., 2016). That is, at higher levels of chaos, the potential effects of parent reactions to their child’s emotions seem to be overwhelmed, while the link between parent response and child ER appears to be operating at lower levels of chaos.

Bringing together the statistical tests we conducted, the results, and the prior empirical evidence and theories, it is probably most useful to think about the family processes we are studying as bidirectional, with child ER and parental responses to child emotion influencing each other over time and across various situations (Morris et al., 2007). Though the emotion socialization processes highlighted above are thought to reflect parenting influences on children’s development, children are not passive recipients of those experiences. Rather, they also affect their environments, and their behaviors evoke differentiated responses from their family members. Children’s characteristics, including how often they express emotions and how well they regulate those emotions, can affect the ways in which parents respond to their children’s emotions. If the child is not well-regulated and has frequent emotional outbursts when experiencing negative situations, the parent may have more difficulty responding supportively to their child’s negative emotions. Furthermore, these bidirectional processes of child and parent self-regulation may be most evident when household chaos levels are lower but become overwhelmed and no longer functional at higher levels of chaos (Deater-Deckard, 2014).

Age Differences in Effects

There was no evidence to suggest household chaos moderated the links between parent responses to emotions and child ER at 9 years of age, regardless of whether the 6-year variables were included in the equation. Why might chaos operate as a moderator of ER socialization processes at 6 years of age, but not 9 years of age? There are several potential interpretations. First, this may reflect a real developmental shift in the potential modulating role of the broader household environment. As children move through middle childhood (e.g., from 6 to 9 years of age, as in our study), they gradually increase the size of their peer networks—social relationships that contribute to ER development beyond the effects of the home environment (e.g., Contreras et al., 2000; Penela et al., 2015; Trentacosta & Shaw, 2009). In addition, as children move across middle childhood, they continue to become more familiar with and integrated into a school context that provides novel influences on ER skill development (Bardack & Widen, 2019; Kwon, Hanrahan, & Kupzyk, 2017). Thus, when comparing parenting processes and links with ER development at 6-years compared to 9-years, the home environment may be a more substantial source of socialization.

A second potential explanation is that children become more resilient to, and less affected by, a chaotic household environment as they develop across middle childhood. For example, children growing up in chaotic environments may find it harder to learn developmentally and situationally appropriate ways to express and regulate their emotions when they are younger. As individual differences in children’s parenting environments and their own ER skills become more stable across childhood, they may also become less easily affected by the potential disruptive effects of an unpredictable home environment. Thus, as they get older, the direct association between child ER and parent responses to negative emotions may be a consistent, unmoderated process that is not affected by broader household context factors such as the level of chaos. Finally, the lack of significant interaction effects at age 9-years may simply reflect that the moderating effect of household chaos is a weak effect that does not replicate when tested repeatedly.

Caveats and Conclusions

There are some limitations that should be considered. First, maternal reports were the sole source of information. Therefore, the findings and effect sizes are likely affected to some degree by method and informant bias. This is consequential, because although the current results may be informative with regard to maternal perceptions (of the child, of their own responses to the child’s emotions, and of the levels of household chaos), they may not generalize to other informants’ views, or to broader and more “objective” measurement of family processes and contexts over time. Also, there may be systematically varying (but unmeasured) factors such as depression and anxiety, that alter maternal perceptions of their responses and their child’s emotion regulation skills. We would encourage readers to interpret the current results in that light—that our findings pertain to how it is that mothers perceive and understand these family processes. Future research should utilize multiple methods and informants (e.g., tasks; observers; other caregivers – e.g., fathers and grandparents) to minimize the impact of a mono-method bias on the results, and to permit formal testing of the degree to which subjective perceptions of these processes matter. Recent research suggests that fathers’ emotion socialization skills benefitted children’s behavior (Havighurst, Wilson, Harley, & Kehoe, 2019) and that mothers’ and fathers’ emotion socialization strategies might differentially impact behavior across siblings (Yaremych & Volling, 2018).

Second, the internal consistency coefficients for the ER and household chaos scales were in the .6 range, which are below the standard threshold for what is considered a reliable measure (Nunnally & Bernstein, 1994). This means that our ability to detect significant effects was compromised and should be taken into consideration when interpreting our results.

Third, even though we utilized a two-wave longitudinal design, the data were correlational, so inferences of causality were not possible. Future research should consider including three or more waves of data (to more thoroughly address potential temporal precedence of effects in longitudinal data), and to better analyze the potential bidirectional processes within mother and child dyads over time (e.g., multivariate latent growth or latent difference score modeling).

Finally, although there was some diversity in the study sample, it was quite homogeneous and not representative with respect to race and ethnicity of families in the United States. The processes we examined operate in families and for individuals who are embedded in cultures. Emotion socialization messages may be delivered in different ways and with somewhat different meanings, in distinct cultural and ethnic groups (Halberstadt & Lozada, 2011; Morris et al., 2017). Culture affects the ways in which parents and children interact and express emotions, and how adults and children alike interpret and react to others’ emotions (Emde, Biringen, Clyman, & Oppenheim, 1991; Morris et a., 2007; Saarni, 1990). It will be essential in future research to include adequate representation of ethnic diversity in the samples, to address this shortcoming in the current study and literature.

These limitations aside, the current study adds to the growing literature on parenting and family processes and children’s ER development in middle childhood. Although the moderating effect of household chaos was not present at both time points in the current study, the effect was significant at age 6-years (even when controlling for family SES indicators), providing preliminary evidence to fuel future research on chaos and child ER development. Our results suggest that higher levels of chaos may overwhelm or subsume the proximal parent-child relationship processes of ER socialization. The current study adds additional impetus for considering household chaos as a critical feature of family resources and functioning, and as a potential target for prevention and intervention efforts given that many aspects of household routines and the physical environment can be modified with adequate supports in place (Cripe & Venn, 1997; Wachs & Evans, 2010; Weitzman et al., 2013; Whitesell, Teti, Crosby, & Kim, 2015).

Data availability statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.