Abstract
This study examined the link between maternal emotion regulation (ER) and emotion parenting behaviors and child ER, particularly emphasizing the previously understudied potential associations between mothers’ ER and concurrent emotion parenting behaviors. Community-recruited participants included 64 mother–child (M = 9.5 years, 38 girls) dyads. Mothers completed measures on their own ER, their child’s ER, and their emotion parenting strategies. Children completed measures on their ER and mother–child dyads engaged in a conflict discussion task that was coded using an ER behavioral observation scale. Results indicated that observed maternal ER was negatively associated with unsupportive emotion parenting, whereas self-reported maternal emotion dysregulation was positively associated with unsupportive parenting and child emotion dysregulation and negatively associated with child adaptive ER. Partial support was provided for the mediating role of emotion parenting behaviors on the link between maternal and child ER. Implications regarding parenting programs and intervention efforts are discussed.
Ample literature has documented a variety of adaptive and maladaptive emotion parenting behaviors that promote or hinder children’s emotional development (e.g., Gottman, Katz, & Hooven, 1996; Morris, Silk, Steinberg, Myers, & Robinson, 2007). Though decades of research have examined ways that parents, and mothers in particular, shape children’s emotional and behavioral competencies through their parenting behaviors, predictors of these parenting behaviors remain understudied. In particular, empirical research thus far has not extensively addressed the ways in which mothers’ own emotional functioning relates to these processes and outcomes. Examination of the ways that mothers’ own emotion regulation (ER) skills relate to her parenting behavior can inform conceptual models of emotional development and also help identify targets for prevention/intervention programs. The primary goal of the present study is to examine the link between maternal ER and emotion parenting behaviors and child ER, particularly emphasizing the previously understudied potential associations between mothers’ ER and their concurrent emotion parenting behaviors.
The current study follows Thompson (1994) in defining ER as “the extrinsic and intrinsic processes responsible for monitoring, evaluating, and modifying emotional reactions, especially their intensive and temporal features, to accomplish one’s goals” (pp. 27-28). In other words, ER is conceptualized to encompass awareness of emotional experiences and ability to modulate emotional responses in accordance with situational demands or individual goals (Gratz & Roemer, 2004). ER is composed of both internal (e.g., physiological reactivity, cognitive efforts) and external (e.g., emotional expressivity, facial reactions, emotion-driven behavior) processes all working together to manage emotional intensity, duration, and displays. “Adaptive” ER implies that one has a sense of control over his or her emotions and that the emotional management efforts are adaptive/appropriate given his or her goals and the situational demands. Conversely, emotion dysregulation encompasses difficulties with ER that interfere with one’s ability to reach their goals and/or to meet situational demands (e.g., Cicchetti, Ganiban, & Barnett, 1991). Emotion dysregulation can entail maladaptive emotional duration, valence, and/or intensity that cause functional interference (e.g., Linehan, Bohus, & Lynch, 2007). Emotion dysregulation is not synonymous with completely unregulated emotion expression as ER efforts are likely still present even when emotions appear to be dysfunctional (Cole, Michel, & Teti, 1994). As such, ER and emotion dysregulation are considered to be separate, yet related, constructs. In the present study, self-reported emotion dysregulation captures when an individual perceives his or her emotions to be overwhelming, out of control, causing inappropriate behaviors, confusing, and/or interfering with goals (Gratz & Roemer, 2004).
From the emotion socialization literature, Gottman et al.’s (1996) notion of parental meta-emotion philosophy offers valuable theoretical explanations about how mothers’ own feelings may influence their own emotion parenting behaviors and their child’s ER. According to their model of parental meta-emotion philosophy, parents’ thoughts and feelings about emotions (including their ability to regulate their own emotions) influence parenting behaviors, and parenting behaviors influence child outcomes (e.g., child ER, child mental/physical health; Gottman et al., 1996). Furthermore, Eisenberg, Cumberland, and Spinrad (1998) developed a heuristic model of factors contributing to the socialization of emotions that included parental regulation as one of the parental factors contributing to emotion-related socialization behaviors. Building on Gottman et al.’s (1996) and Eisenberg et al.’s (1998) work, Morris et al. (2007) used an empirical literature review to develop the tripartite model of the impact of the family on children’s ER and adjustment. The tripartite model frames children’s ER development through three realms of familial influence: (a) observational learning, modeling, and social referencing; (b) parenting practices specifically related to emotion and emotion management; and (c) the emotional climate of the family via parenting style, the attachment relationship, family expressiveness, and the marital relationship. The tripartite model posits that the variables influencing children’s healthy emotional/psychological development are dynamic and interact with each other. The present study examines two of the three realms identified in this model: (a) Maternal ER is considered to be a part of observational learning, modeling, and social referencing as children observe and learn from their mothers’ ER efforts; and (b) Emotion parenting behaviors are the parenting practices specifically related to emotions.
Maternal Emotion Regulation and Child Emotion Regulation
Though the literature on child ER has grown rapidly over the past decades, scant research has empirically examined the link between parent and child ER. However, one study found that self-reported maternal emotion suppression was positively related to self-reported child (ages 9-19) emotion suppression (Bariola, Gullone, & Hughes, 2011) and that child age did not moderate the relation. A longitudinal study found that parental emotion dysregulation when the child was 9 to 13 years old predicted child emotion dysregulation when the child was 17 to 19 years, though this study included only boys (Kim, Pears, Capaldi, & Owen, 2009). Emotion dysregulation was measured through a composite score from parent, child, and observer report (Kim et al., 2009). Though not examining maternal ER specifically, a study examining maternal modeling of fear with their children (ages 9-12) found that the number of maternal-reported fears correlated with the number of child-reported fears, suggesting the influence of maternal modeling (Muris, Steerneman, Merckelbach, & Meesters, 1996). A study by Valiente et al. (2004) examined the relation between maternal (observed and reported) negative expressivity and child (M age = 7.67 years) emotional experience and expressivity (as measured via observed facial expression, heart rate, and self-report). Results revealed that maternal negative expressivity (observed and reported) was positively related to children’s expression of facial distress/sadness. Of note, there were no significant relations between subjective (self-report) and objective (observations, heart rate) assessment of emotional experiences. Beyond these few empirical studies, there clearly remains a gap in our understanding of the link between maternal and child ER. However, the studies reviewed below suggest that this link is not likely to be direct but possibly mediated through the relations between maternal ER and emotion-related parenting behaviors.
Maternal Emotion Regulation and Emotion Parenting
Observing maternal ER, including emotional expressivity, lability, and coping behaviors, is one way that children learn about ER (Eisenberg et al., 1998; Morris et al., 2007), but direct modeling is only one method of emotion socialization. Emotion parenting encompasses many behaviors such as discussing emotions and responding to children’s emotional displays, and empirical studies have investigated these emotion parenting behaviors and associated outcomes (e.g., Denham, Mitchell-Copeland, Strandberg, Auerbach, & Blair, 1997; Eisenberg et al., 1999). Though emotion parenting falls under the umbrella of general parenting behaviors, emotion parenting behaviors are ones specifically related to creating an emotional climate in which children observe, learn about, and experience emotions. From the emotion parenting literature, two broad categories have been identified—supportive and unsupportive. Supportive emotion parenting supports, validates, and helps understand a child’s emotional experience, whereas unsupportive emotion parenting criticizes, minimizes, or punishes a child’s emotional experience. Supportive emotion parenting behaviors provide a comfortable environment in which children can learn about emotions and are associated with adaptive child emotion outcomes (Gottman et al., 1996; McDowell, Kim, O’Neil, & Parke, 2002). Conversely, unsupportive emotion parenting behaviors create an uncomfortable emotion climate where emotions are not discussed, deemed as insignificant, discouraged, or expressed with high levels of negativity (Denham, Zoller, & Couchoud, 1994). Unsupportive emotion parenting behaviors have been linked with child emotion dysregulation (e.g., Eisenberg et al., 1999) in a bidirectional fashion (Morelen & Suveg, 2012).
Though scant, some literature supports a link between maternal ER and emotion parenting behaviors. For example, in a longitudinal study that followed parents and sons across 21 years, parental emotion dysregulation (as measured by a composite of parent, child, and observer report) predicted poor punishment strategies (as measured by a composite of parent and observer report; Kim et al., 2009). Though this study did not examine emotion parenting in particular, it does provide support for the notion that emotion dysregulation can result in parental use of more punitive strategies, which could reflect unsupportive emotion parenting if directed at a child’s emotion (e.g., “If you don’t stop crying, I’ll give you something to cry about”). Another study found that African American mothers who tended to respond to stressful situations with self-reported hopelessness (their measure of emotional coping) had a lower parent–child relationship quality and children with more ER difficulties (as assessed by mother and child report) than mothers who used more constructive coping strategies (Bynum & Brody, 2005). Gottman et al. (1996) used a semistructured meta-emotion interview with parents of children ages 4 to 5 years. Results demonstrated that parents with greater awareness of their own emotions engaged in more emotion parenting behaviors associated with adaptive child outcomes compared to parents who were less aware of their own emotions (Gottman et al., 1996). Given that adaptive ER involves monitoring and evaluating emotional experience (Thompson & Calkins, 1996), emotion awareness is likely an essential component of effective ER, providing support for the notion that parental ER does in fact relate to emotion parenting behaviors. Additionally, a study by Eisenberg et al. (2001) used structural equation modeling to look at the inner connectedness of observed parental (160 mothers, 9 fathers) warmth, emotional displays, and emotion parenting behaviors with their child’s (2nd-5th grade, M age = 9.4 years) ER and behavior problems (as measured by informant report). An observational task had children view slides of pleasant and unpleasant emotional pictures with an experimenter and then again with their parent present. Parents were instructed to discuss the slides with their child. Videotaped interactions were coded for child/parental facial expressivity in response to pleasant/unpleasant slides, parental warmth, and parental discussion of emotions. Results indicated that parents who displayed warmth in the observational task tended to display positive emotions and to engage in helpful emotion discussions that linked the displayed emotions to their child’s emotional experiences (Eisenberg et al., 2001).
Emotion Parenting and Child Emotion Regulation
Literature shows that supportive emotion parenting behaviors are more likely to correspond with adaptive child ER compared to unsupportive emotion parenting behaviors (e.g., Denham et al., 1994; Eisenberg et al., 1999). Recent research has further illuminated the interconnected nature of emotion parenting behaviors and child ER (e.g., Morelen & Suveg, 2012; Morris et al., 2011). For example, behavioral observations of parents and children (ages 7-12) in emotion discussions revealed that children were more likely to use adaptive ER if parents engaged in supportive emotion parenting behaviors compared to unsupportive emotion parenting behaviors (Morelen & Suveg, 2012). Another observational study that examined the interconnected nature of emotion parenting behaviors and child ER examined how maternal emotion parenting behaviors influence child ER and emotional displays in laboratory tasks designed to elicit child (ages 4-9) anger and sadness (Morris et al., 2011). Results demonstrated that when mothers engaged children in observed adaptive ER strategies (i.e., attention refocusing, cognitive reframing), children displayed low-intensity anger and sadness. Mothers who are readily able to regulate their own emotional experiences are likely better equipped to assist their children during emotionally challenging situations. Furthermore, the study by Eisenberg et al. (2001) discussed in the previous section also revealed that parental linking of emotions (i.e., discussing how their child’s emotions related to the displayed emotional slides) was negatively related to children’s dysregulated emotional expression. Their comprehensive model illustrates the interconnected nature of parental factors (e.g., emotional expressivity), emotion parenting behaviors (e.g., linking emotions), and children’s ER abilities.
Present Study
Overall, the literature review reveals that parents are important socialization agents for children’s emotional development, and children’s ability to regulate their emotions relates to emotional, social, and psychological outcomes. Classic models of children’s emotional development highlight the essential role that parents play in helping children build adaptive ER abilities (Eisenberg et al., 1998; Gottman et al., 1996), and more recently, the tripartite model (Morris et al., 2007) illustrated three core mechanisms through which families influence children’s ER development. All of these models document the important role of parental ER in the model, yet no known studies have specifically examined the implications of maternal ER on emotion parenting behaviors. Furthermore, only a handful of studies have examined the relation between maternal and child ER strategies (e.g., Bariola et al., 2011; Kim et al., 2009). As such, more research is needed to understand the relation between parent and child ER and the role that emotion parenting behaviors play in that relation.
The primary goal of the present study is to build on this research base and examine the relations between maternal ER, emotion parenting behavior, and child ER using multiple methods of assessment. Specifically, we examine emotion parenting behavior as one mediating mechanism through which maternal ER relates to child ER. Lastly, this study examines ER during middle childhood, an important developmental period given that ER strategies become more pronounced during this age (Cole & Kaslow, 1988). Middle childhood is a time when youth become more aware of social cues that guide decisions on how to manage their emotions (Zeman & Garber, 1996), yet past studies have focused predominately on emotion socialization in infancy and the preschool years (see Zeman, Cassano, Perry-Parrish, & Stegall, 2006, for a review).
The following hypotheses are put forth:
Unsupportive emotion parenting will mediate the link between (high) maternal caregiver emotion dysregulation and (high) child emotion dysregulation.
Unsupportive emotion parenting will mediate the link between (high) maternal caregiver emotion dysregulation and (low) child ER.
Supportive emotion parenting will mediate the link between (high) maternal caregiver ER and (low) child emotion dysregulation.
Supportive emotion parenting will mediate the link between (high) maternal caregiver ER and (high) child ER.
Method
Participants
Participants included 64 mother–child dyads. Children included 26 boys and 38 girls ranging in age from 8 to 11 years (M = 9.5, SD = 1.04). Caregivers included a primary, maternal figure (e.g., biological mother, adoptive mother, grandmother) with whom the child was residing. Of the 64 mothers, 59 (92%) self-identified as the biological mother, and for brevity’s sake, “mother” will refer to all maternal caregivers. The majority of mothers (95%) graduated from high school or completed equivalent requirements, and 33% completed college. The sample was ethnically diverse with 52% of the families identifying as African American, 41% Caucasian, 3% Latino, 2% Asian, and 2% “Other.” Additionally, 47% of families reported an income less than $20,000 per year, and 23% reported income greater than $60,000 per year. Regarding the mothers’ relationship status, 44% were married or cohabiting, 20% were divorced or separated, and 36% were never married.
Procedure
Newspaper advertisements and community flyers were used to recruit dyads. Inclusion criteria included maternal and child ability to read and write in English and that the caregiver lived with the child for at least the past 2 consecutive years. The study took place in a psychology research laboratory designed for assessments with behavioral observations. Following consent procedures, dyads engaged in a warm-up task, then four sequential interaction tasks, all of which were video recorded for later coding. Following the interaction tasks, participants completed questionnaires and then were paid $40 for participation. All procedures were approved by an institutional review board.
Coding and Reliability
The observational codes from a structured Conflict Discussion task (e.g., Rueter & Conger, 1995) were used in the present study. For this task, mothers and children independently rank ordered a list of potential conflict areas (e.g., chores, homework, time spent with friends), and the topic rated highest by both members of the dyad was the one chosen for discussion. If mothers and children rated topics differently, then the topic with the greatest sum (of mother and child rating) was chosen. The dyad was given 5 minutes to talk about the topic and asked to come up with a solution, if possible, to the problem. This task in particular was chosen for the study because it is widely used in the developmental and clinical literature to elicit stressed interactions or disagreements between parents and children (e.g., Caron, Weiss, Harris, & Catron, 2006; Donenberg & Weisz, 1997), thus allowing more opportunities to observe dysregulated emotion.
Coding team members independently rated a family, then the entire coding team (composed of seven doctoral students and a PhD-level professor) met to discuss their codes and any discrepancy between codes. Once consistency across coding styles was achieved (i.e., scores ±1 point of each other on a 7-point Likert-type scale), coders were paired into dyads and scored randomly selected videos to independently code and later meet to conference scores. Interrater reliability was computed on 30 randomly selected families with all coding dyads included in the computation of interrater reliability. See the appendix for the codes for all behavioral observation included in the present study.
Measures
Maternal Emotion Regulation
Mothers reported on their own emotion dysregulation via the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004). The DERS is a 36-item self-report measure of current, clinically relevant difficulties with ER. Mothers were asked to rate items on a 5-point Likert-type scale (1 = almost never, 4 = almost always) according to the frequency with which the statement applied to them. The measure yields an overall score (i.e., DERS Total) and six subscales: Nonacceptance of Emotional Responses (e.g., “When I’m upset, I feel ashamed with myself for feeling that way”), Difficulties Engaging in Goal-Directed Behavior (e.g., “When I’m upset, I have difficulty getting work done”), Impulse Control Difficulties (e.g., “When I’m upset, I lose control over my behaviors”), Lack of Emotional Awareness (e.g., “When I’m upset, I take time to figure out what I’m really feeling,” reverse scored), Limited Access to Emotion Regulation Strategies (e.g., “When I’m upset, I believe that there is nothing I can do to make myself feel better”), and Lack of Emotional Clarity (e.g., “I am confused about how I feel”). Higher scores are indicative of more ER difficulties. High reliability and acceptable construct and predictive validity (e.g., internal consistencies from .80 to .89) have been established for the overall scale (Gratz & Roemer, 2004). The overall score (DERS Total) was used as a measure of maternal emotion dysregulation for the present study (α = .94).
Maternal ER was also assessed through behavioral observations in the Conflict Discussion task. Mothers’ ER was rated using a global ER code that ranged from 1 (extremely dysregulated) to 7 (extremely regulated). Descriptions were provided for each value between 1 and 7 to exemplify each code point and facilitate reliable coding (see the appendix). For example, a score of 1 was described as “Parent’s emotions are predominantly dysregulated. There are almost no (or not any) instances of regulated emotion.” Intraclass correlations were used to evaluate the reliability of the behavioral observations codes. The intraclass correlation for the maternal ER code was .79.
Emotion Parenting Behavior
Self-reports of emotion parenting behavior were collected via the Coping With Children’s Negative Emotions Scale (CCNES; Fabes, Eisenberg, & Bernzweig, 1990). The measure assesses parental reactions to children’s negative emotions in the context of 16 hypothetically distressing situations (e.g., “If my child falls off his/her bike and breaks it, and then gets upset and cries, I would . . .”). Each individual item contains six theoretically distinct ways of responding to the distressing scenario, and parents rate how likely they are to respond in each of the emotion parenting ways using a 7-point Likert-type scale. Several investigators have used this measure and reported adequate internal reliability (Eisenberg et al., 1999; Eisenberg & Fabes, 1994; Fabes, Poulin, Eisenberg, & Madden-Derdich, 2002). Unsupportive emotion parenting subscales include punitive reactions (e.g., “tell the child to stop crying or he/she will not be able to ride the bike”), minimization reactions (e.g., “tell the child he/she is overreacting”), and distress reactions (e.g., “get angry with the child”). The internal consistencies for the unsupportive scales were .89, .87, and .61, respectively, and they were significantly correlated with each other (r between .53 and .86). Supportive emotion parenting subscales include emotion-focused reactions (e.g., “comfort your child and try to get him/her to forget about the accident”), expressive encouragement (e.g., “tell your child it is OK to cry”), and problem-focused reactions (e.g., “help your child figure out how to get the bike fixed”). The internal consistencies for the supportive scales were .81, .90, and .79, respectively, and they were all significantly correlated with each other (r = .55-.88).
Given the significant relations were found among the unsupportive emotion parenting subscales from the CCNES (Punitive Reactions, Minimizing Reactions, Distress Reactions) as well as among the supportive emotion parenting subscales from the CCNES (Emotion Focused Reactions, Expressive Encouragement, Problem Focused Reactions), the items from each subscale were combined into an unsupportive emotion parenting composite (α = .93) and supportive emotion parenting composite (α = .93), respectively. Of note, the internal consistency was conducted on the items contributing to each composite (i.e., 36 items for unsupportive and 36 items for supportive). Principal components analysis also supported the creation of an unsupportive emotion parenting composite (eigenvalue = 2.45) and supportive emotion parenting composite (eigenvalue = 2.33).
Child Emotion Regulation
Emotion Regulation Checklist (ERC; Shields & Cicchetti, 1997). Mothers completed the 24-item ERC (4-point Likert-type scale) based on their child’s typical methods of managing emotional experiences. The checklist has two subscales: (a) ER (i.e., assesses appropriate emotional expression, empathy, emotional self-awareness) and (b) Negativity/Lability (i.e., assesses inflexibility, lability, dysregulated negative affect). The ERC has acceptable reliability and validity (α = .83 for ER, α = .96 for Negativity/Lability; Shields & Cicchetti, 1997). Internal reliability was lower than expected for the present study for the ER subscale (α = .62) and acceptable for the Negativity/Lability subscale (α = .78). Two items were removed from the ER scale—Item 18 reverse scored (displays flat affect/emotion) and Item 23 (displays appropriate negative emotion, e.g., anger fear, frustration, distress, in response to hostile, aggressive, or intrusive acts by peers)—because they were found to lower reliability (α = .52 before items were reduced). Given that the majority (64%) of mothers’ highest degree was a high school diploma, it may have been that the language (e.g., “flat affect”) and wording on these items were misunderstood, contributing to the lower internal consistency for this scale.
The Children’s Emotion Management Scales for Sadness (CSMS), for Anger (CAMS; Zeman, Shipman, & Penza-Clyve, 2001), and for Worry (CWMS; Zeman, Cassano, Suveg, & Shipman, 2010). Children completed the CEMS for sadness (CSMS), anger (CAMS), and worry (CWMS) as a self-report measure of ER. The 12-item CSMS, 11-item CAMS, and 13-item CWMS use a 3-point Likert-type scale (1 = hardly ever, 2 = sometimes, 3 = often) to assess specific ER behaviors. For the present study, the Dysregulation subscale was used (e.g., “I do things like cry and carry on when I’m worried,” “I say mean things to others when I’m mad”). Preliminary studies indicate adequate internal consistency (i.e., alphas range from .62 to .77) and good test–retest reliability (Zeman et al., 2001). A composite Dysregulation score (α = .60) was formed for the present study by averaging the items from the Sadness, Anger, and Worry Dysregulation scales. Of note, the Dysregulation scale has few items (i.e., 3 items per emotion type), which may have contributed to the low reliability for this scale (Kline, 2000). The nature of the items themselves may also contribute to low reliability. For example, children prone to more internalizing problems may report that they “cry and carry on when sad” and/or “have difficulty controlling worried feelings” but that they do not “slam doors when mad.”
Child ER was also assessed through behavioral observations in the Conflict Discussion task and was rated using a global ER code that ranged from 1 (extremely dysregulated) to 7 (extremely regulated). Descriptions were provided for each value between 1 and 7 to facilitate reliable coding. For example, “Child shows one or two instances of dysregulated emotion but predominately remains in control of their emotions. Child’s behavior indicates that they are able to regulate their emotions most of the time,” is the description for a score of 5 (somewhat well regulated). The intraclass correlation for this code was .87.
Results
Preliminary Analyses
See Table 1 for means, standard deviations, and ranges for all measures. Pearson bivariate correlations were used to examine whether maternal income related to the measures. Income was significantly correlated with observed maternal ER (r = .40, p = .001), maternal self-reported emotion dysregulation (DERS Total; r = −.29, p = .02), maternal report of child emotion dysregulation (ERC Negativity/Lability; r = −.31, p = .01), and child report of emotion dysregulation (CEMS Dysregulation; r = −.46, p < .001). Regarding ethnicity, a one-way analysis of variance revealed that Caucasian mothers (M = 6.1) had significantly higher scores on observed ER than African American mothers, M = 5.2, F(3, 53) = 4.15, p = .01. Given the small number of participants identifying as Asian (n = 1), Hispanic (n =2), or “Other” (n =2), two groups (families of color n = 38, Caucasian families n = 38) were formed to have a dichotomous control variable for further analyses. T tests were used to examine whether there were child sex differences on the measures. There was a significant sex difference on maternal report of child ER (ERC ER), t(62) = −2.27, p = .03, such that daughters were rated as having higher ER (M = 23.6) compared to sons (M = 21.96). As such, ethnicity, income, and child sex were controlled in further analyses as relevant based on these preliminary analyses.
Means, Standard Deviations, and Ranges of Paper-and-Pencil Measures.
Note. DERS = Difficulties in Emotion Regulation Scale; CCNES = Coping With Children’s Negative Emotions Scale; ERC = Emotion Regulation Checklist; ER = Emotion Regulation; CEMS = Children’s Emotion Management Scale.
Pearson bivariate correlations were used to examine the strength of relation between paper-and-pencil measures and corresponding behavioral observations. Given that there were three main constructs of interest (maternal caregiver ER, emotion parenting behavior, child ER), a Bonferonni’s correction was applied reducing alpha to .02 (.05/3). The correlation between maternal self-report of dysregulation (DERS Total) and observed maternal ER in the conflict discussion was nonsignificant (r = −.22, p = .09), so the two measures were left separate. Regarding child ER, there was a marginal negative relation between observed child ER and maternal caregiver report of child emotion dysregulation (ERC Negativity/Lability, r = −.28, p = .03). Partial correlations (see Table 2) were then used to examine whether these relations differed when controlling for the relevant demographic variables mentioned above. The pattern of results stayed the same though the effect sizes got smaller (DERS Total and observed maternal ER r = −.17, p = .19; observed child ER and ERC Negativity/Lability, r = −.26, p = .07). Given the lack of a clear pattern of relations between self- and parent report measures and observed child ER, they were kept separate for the primary analyses.
Partial Correlations Among Study Variables.
Note. DERS = Difficulties in Emotion Regulation; ER = Emotion Regulation; CCNES = Coping with Children’s Negative Emotions Scale; ERC = Emotion Regulation Checklist; CEMS = Children’s Emotion Management Scales. Partial correlations controlled for ethnicity, income, and child sex.
p ≤ .07. *p ≤ .03. **p ≤ .005.
Primary Analyses
Bivariate correlations were used to test Hypothesis 1 (maternal caregiver emotion dysregulation would be negatively related to supportive emotion parenting behaviors and child adaptive ER and positively related to unsupportive emotion parenting behaviors and child emotion dysregulation) and Hypothesis 2 (maternal caregiver ER would be positively related to supportive emotion parenting behaviors and child-adaptive ER and negatively related to unsupportive emotion parenting behaviors and child emotion dysregulation). Hypotheses 1 and 2 were partially supported. Specifically, maternal self-report of dysregulation (DERS Total) was positively related to maternal report of unsupportive emotion parenting behaviors (CCNES Unsupportive Composite, r = .36, p = .007) and maternal report of child emotion dysregulation (ERC Negativity/Lability, r = .30, p = .02). Maternal self-report of dysregulation (DERS Total) was also negatively related to maternal report of child ER (ERC ER, r = −.36, p = .004). Observed maternal caregiver (adaptive) ER was negatively related to maternal report of unsupportive emotion parenting behaviors (CCNES Unsupportive Composite, r = −.30, p = .02) and marginally negatively related to child self-reported emotion dysregulation (CEMS Dysregulation, r = −.25, p = .050). Maternal caregiver ER, both self-reported and observed, was not related to supportive emotion parenting behaviors.
Partial correlations (see Table 2) were then used to examine the hypotheses while controlling for relevant demographic variables. Whereas the effect sizes got smaller, the pattern of results stayed the same except for the following changes: (a) The relation between maternal self-report of dysregulation (DERS Total) and maternal report of child emotion dysregulation (ERC Negativity/Lability) went from significant to marginal (r = .28, p = .05); (b) the relation between maternal self-report of dysregulation (DERS Total) and maternal report of child ER (ERC ER) went from significant to marginal (ERC ER, r = −.27, p = .05); and (c) the relation between observed maternal ER and child self-reported emotion dysregulation (CEMS Dysregulation) went from marginal to nonsignificant (r = −.16, p =. 24). Overall, there was partial support for Hypotheses 1and 2.
Regarding Hypothesis 3 (emotion parenting behaviors will mediate the link between maternal caregiver and child ER), two mediation models were tested based on the significant relations that mapped onto the predictions. Bootstrapping analyses were conducted using the SPSS Process Macro (Hayes, 2013) to test whether emotion parenting behavior mediated the link between maternal ER and child ER. Five thousand bootstrap resamples were used to generate 95% confidence intervals (CIs) that estimated the size and significance of the indirect effect. This approach generally follows the classic Baron and Kenny (1986) approach to testing mediation (i.e., testing whether the independent variable influences the mediator variable, which in turn influences the dependent variable); however, bootstrapping methods are considered advantageous for testing mediation in samples in which assumptions of normality may be violated or when power is low, such as with smaller sample sizes (Preacher & Hayes, 2004; see Hayes, 2009, for an explanation of statistical advances since Baron & Kenny, 1986). Demographic variables were entered as covariates in the mediation models when relevant, as described above. Only two mediation models were tested based on the pattern of correlations.
The first model examined the indirect effect of maternal-reported emotion dysregulation (DERS Total) on maternal report of child ER (ERC ER) through unsupportive emotion parenting (CCNES Unsupportive Composite). Unsupportive emotion parenting did not mediate the link between maternal emotion dysregulation and child ER (point estimation = − .005, 95% bias-corrected and accelerated [BCa] bootstrap CI [−.02, .003]). The second analysis examined the indirect effect of maternal-reported dysregulation (DERS Total) on maternal report of child dysregulation (ERC Negativity/Lability) through unsupportive emotion parenting (CCNES unsupportive emotion parenting composite). Maternal unsupportive emotion parenting mediated the link between maternal emotion dysregulation and child emotion dysregulation (point estimation = .031, 95% BCa CI [.005, .083]).
Discussion
An increasing number of researchers and clinicians have urged for research and treatment addressing the integral role that ER plays in children’s healthy psychosocial development (e.g., Cicchetti, Ackerman, & Izard, 1995; Cole, Dennis, Smith-Simon, & Cohen 2009; Suveg, Hoffman, Zeman, & Thomassin, 2009). Additionally, there is a general consensus that certain emotion parenting behaviors contribute to or impede children’s emotional development (e.g., Gottman et al., 1996; Ramsden & Hubbard, 2002). Despite such acknowledgements in the field of emotion research, few studies have specifically examined the relation between maternal ER and child ER. Furthermore, no known studies have examined how maternal ER relates to specific emotion parenting behaviors, despite a theoretical rationale for this association as presented in Morris et al.’s (2007) tripartite model. The present study filled a gap in the literature by examining maternal ER as one variable that may promote or hinder children’s emotional development. Additionally, this study examined relations between maternal ER, supportive and unsupportive emotion parenting behavior, and child ER using multiple methods of assessment. Results based on maternal report indicated that maternal ER was negatively associated with unsupportive emotion parenting, whereas maternal emotion dysregulation was positively associated with unsupportive parenting and child emotion dysregulation and negatively associated with child adaptive ER. Partial support was provided for the mediating role of emotion parenting behaviors on the link between maternal and child ER.
Regarding the link between maternal ER and emotion parenting behaviors, we posited that maternal ER would be positively related to supportive emotion parenting and negatively related to unsupportive emotion parenting, and conversely, maternal emotion dysregulation would be positively related to unsupportive emotion parenting and negatively related to supportive emotion parenting. There was partial support for these hypotheses. Mothers who experience emotions as overwhelming, intense, and/or uncontrollable likely have limited coping resources to use during stressful situations (Gratz & Roemer, 2004), which in turn, may make it difficult to support their child’s own emotional experiences in a constructive way. The results from this study indicate that mothers with self-reported ER difficulties are more critical about children’s emotions than mothers who do not have ER difficulties. This may be because mothers with ER difficulties view emotions as disruptive and unpredictable and therefore try to minimize their children’s experience of negative emotions in an attempt to prevent emotion dysregulation from escalating. For example, a mother with difficulty regulating anger/frustration may be more prone to engage in unsupportive emotion parenting behaviors (e.g., punishing a child for showing emotions) than a mother who is able to cope with her anger without directing it at her child. Commensurate with this explanation, research has documented a positive relation between ER difficulties, avoidance of emotional situations, and use of suppression as a response to emotional experiences (Gratz & Roemer, 2004). Conversely, mothers with observed adaptive ER abilities (i.e., ability to regulate emotions in conflict situations) likely have more emotional resources that enable them to remain calm in response to their child’s negative emotional displays (Gottman et al., 1996).
Interestingly, mothers’ emotional functioning (as measured by observed ER and self-reported emotion dysregulation) related only to unsupportive emotion parenting but not to supportive emotion parenting, though results for supportive emotion parenting were in the hypothesized direction. The results suggest that emotionally regulated mothers are not necessarily prone to engage in more supportive emotion parenting techniques (compared to mother low in adaptive ER) but rather are less likely to engage in minimizing, punitive, and critical emotion parenting strategies. Perhaps maternal ER capabilities are necessary but not sufficient for supportive emotion parenting behaviors. The correlation in our study between unsupportive and supportive emotion parenting (r = .02) highlights the fact that these constructs are not opposite ends of a single dimension but rather two separate constructs. Fewer unsupportive behaviors does not necessarily imply more supportive—it could just reflect an absence of emotion parenting behaviors. Furthermore, an increase in parenting programs teaching emotion parenting strategies supports the notion that supportive emotion parenting behaviors are not necessarily automatic behaviors for parents and that the skills underlying supportive emotion parenting can be taught (see Baker, Brassard, Schneiderman, Donnelly, & Bahl, 2011, for a review). Last, factors not examined in the present study (e.g., maternal emotional awareness, family emotional climate) might help explain the maternal and environmental factors that are associated with supportive emotion parenting behaviors. For example, maternal warmth is one variable that Eisenberg et al. (2001) found to be related to supportive emotion parenting behavior. Future research would benefit from further investigating what parental characteristics are associated with the tendency to engage in supportive emotion parenting behaviors.
There was also partial support for the hypothesis that examined links between maternal ER and child ER. Specifically, based on maternal report, maternal emotion dysregulation was negatively related to child ER and positively related to child emotion dysregulation. There are several explanations for the findings. One possible explanation for this link is that mothers model ER (or dysregulation) for their children (Morris et al., 2007). From this perspective, children whose mothers are emotionally dysregulated have more opportunities to observe maladaptive methods of ER in comparison to children whose mothers are emotionally well regulated. This interpretation is consistent with past research that found similar patterns of emotion management between mothers and children (Bariola et al., 2011; Muris et al., 1996). Another possible explanation is that mothers and children engage in evocative interactions where one member of the dyad becomes dysregulated, which results in the other becoming dysregulated and in emotional escalation for them both. Relatedly, the factors that may predispose a mother to adaptive/maladaptive ER (e.g., the personality trait of neuroticism, high negative affectivity) have heritable components (Emde et al., 1992; Rhee et al., 2012). Furthermore, children’s temperaments influence the reactions they evoke from their environment (e.g., Caspi, 1998). Put together, both passive and active gene by environment (G × E) effects (Scarr & McCartney, 1983) likely influence the link between maternal and child ER. Future research could examine possible G × E interactions.
Given that few of the potential direct relations between maternal and child ER were significant, this relation is likely indirect and influenced by contextual factors. The present study examined emotion parenting behaviors as one mechanism through which maternal ER relates to child ER. Partial support was found for this hypothesis. Specifically, based on maternal report, unsupportive emotion parenting behavior mediated the link between maternal and child emotion dysregulation. High maternal emotion dysregulation was positively related to unsupportive emotion parenting behaviors. Unsupportive emotion parenting behaviors, in turn, were positively related to child emotion dysregulation. Thus it seems that emotion parenting behaviors are one mechanism through which maternal ER difficulties relate to child ER difficulties. As previously discussed, a mother’s own ER abilities (or difficulties) influence the likelihood that she will engage in unsupportive/supportive emotion parenting behaviors. Subsequently, when children’s emotional experiences are punished, minimized, or criticized, children are either left with limited adaptive coping skills or learn maladaptive ways of expressing their emotions (e.g., minimizing it themselves, yelling, crying; Eisenberg & Fabes, 1994). Maladaptive coping strategies likely result in increased levels of negative affect and unpleasant cycles of interaction between mothers and children (Morelen & Suveg, 2012).
Summary, Limitations, and Future Directions
The findings contribute to the emotion socialization literature, as no previous studies have specifically examined the link between maternal ER and emotion parenting behavior. The closest study found was Gottman et al.’s (1996) seminal work on parental meta emotion that established a link between parents’ awareness of their own emotions and adaptive emotion parenting behaviors. The present study adds to prior knowledge to previous models by establishing that maternal ER is related to emotion parenting behavior; specifically, maternal ER is negatively related to unsupportive emotion parenting behavior. Consistent with the tripartite model framework (Morris et al., 2007), the findings illuminate that maternal ER does in fact relate to unsupportive emotion parenting behaviors, which in turn relate to child ER. The findings provide preliminary empirical support for the need to address maternal ER in interventions aimed to teach emotion parenting skills. Beyond that, the findings suggest that adaptive maternal ER is necessary, but not sufficient, for adaptive emotion parenting. Specifically, supportive emotion parenting behaviors did not relate to maternal ER or emotion dysregulation. Thus, although adaptive ER abilities decrease the likelihood of engaging in unsupportive emotion parenting behaviors, they do not necessarily increase the likelihood of engaging in supportive emotion parenting behaviors. Perhaps strategies for supporting children’s emotions (e.g., encouraging coping through problem solving, encouraging adaptive forms of emotional expression) are difficult for some parents. Therefore, parenting programs may benefit from the inclusion of emotion parenting skills and strategies for supporting children’s emotional experiences. The suggestions regarding the potential benefits for intervention programs are speculative; whether interventions would benefit from inclusion of parental ER skills is an empirical question that should be answered by future research.
The present study contributes to the literature by demonstrating links between maternal ER, emotion parenting, and child ER. The findings are strengthened by the diverse sample, the use of multiple methods of assessment, and the use of multiple informants. Despite these contributions, limitations are noted. First, the sample size was relatively small and included only mothers, which therefore limited the nature and scope of the analyses and required us to control for certain variables (e.g., income) rather than further investigate potential differences. Future research should recruit fathers and employ larger sample sizes to allow for investigation of whether/how the relations between maternal ER, emotion parenting, and child ER vary by parent/child sex, ethnicity, and other contextual variables that may influence the nature of these relations (e.g., socioeconomic status). Inclusion of fathers is important given research suggesting that behavior and outcomes associated with paternal socialization may be qualitatively different than those for maternal socialization (e.g., Fivush, Brotman, Buckner, & Goodman, 2000). Additionally, the present study only controlled for ethnicity (given that there was a significant mean difference between African American and Caucasian mothers on observed ER) whereas future research should consider how these relations function similarly/differently in diverse groups. For example, a growing body of literature suggests that “adaptive” emotion parenting techniques vary across ethnic groups (e.g., Morelen, Jacob, Suveg, Jones, & Thomassin, 2013; Nelson, Leerkes, O’Brien, Calkins, & Marcovitch, 2012). Second, the mediation model was comprised of same-method (maternal report) variables that may have biased the results. Despite this limitation, the multimethod nature of this study is considered a methodological strength given emerging research that highlights how different methods of assessment may reflect different aspects of ER (e.g., Cole et al., 2004). Our results were consistent with past research that found either nonexistent or weak relations between self-reported emotional experiences and observed emotional displays (Valiente et al., 2004). Furthermore, recent research on parent–child discrepancies in reported ER suggests that some discrepancies are expected and provide important information about how a child’s ER is perceived compared to how a child experiences the internal processes of ER (Hourigan, Goodman, & Southam-Gerow, 2011). As the science behind multimethod assessment continues to grow, it will be necessary to investigate how each method (e.g., informant based, physiological, observational) contributes unique information about ER (Suveg & Zeman, 2011). Furthermore, as our ability to find relations between the different methods of assessing ER in the current study was likely influenced by sample size, future multimethod studies will benefit from larger sample sizes. Third, the cross-sectional nature of the study and the dyadic nature of the observational codes make it impossible to make conclusions regarding the temporal sequence of the variables and the bidirectional nature of parent–child interactions. Future research should employ a longitudinal design and micro coding to capture better the order in which maternal ER displays, maternal emotion parenting behaviors, and child ER displays occur. The present study assumed the mediational order (maternal ER influencing maternal emotion parenting, which then influences child ER) based on a review of the literature but acknowledges the empirical necessity of longitudinal designs. Lastly, the present study used composite scores to examine emotion parenting behaviors, which limited our investigation of specific subtypes of emotion parenting behavior. Future research would benefit from a deeper understanding of whether specific emotion parenting behaviors are more/less helpful or detrimental than others.
Footnotes
Appendix
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.