“I'd rather not talk about it:” Emotion Parenting in Families of Children with an Anxiety Disorder

Abstract

This study examined emotion socialization practices in families of children with an anxiety disorder (AD; n = 28) and families of children who had no diagnoses (ND; n = 28), and considered gender differences. Youth (aged 8-13) and both parents discussed times when the child felt anxious, angry, and happy, for five minutes each. Fathers of AD children engaged in less explanatory discussion of emotion overall and exhibited less positive affect and more negative affect when interacting with sons than did fathers of ND children. Similar patterns emerged for mothers, although specific results varied by emotion type and child gender. Children with an AD demonstrated less positive affect overall and engaged in fewer problem-solving emotion regulation strategies when discussing anxious and angry situations than children in the ND group. In both AD and ND groups, fathers appeared to have greater involvement in emotion-related discussions with sons versus daughters. The results highlight parents' contributions to the emotional development of their children, the ways in which socialization may go awry in families of AD children, and the implications for children's emotional functioning.

The development of emotional competence--the ability to recognize, communicate, and regulate emotional experiences-- is important to children's socioemotional adjustment (Cicchetti, Ackerman, & Izard, 1995; Saarni, Mumme, & Campos, 1998). Emotional competence helps with personal well-being, relations with others, and one's self-presentation to the social world (Thompson & Calkins, 1996). Children with a poor understanding of or difficulty regulating emotions may be at increased risk for internalizing and externalizing symptomatology (Cicchetti et al., 1995; Cole, Michel, & O'Donnell-Teti, 1994; Zeman, Shipman, & Suveg, 2002).

The Role of the Family in the Development of Emotional Competence

Factors that may contribute to the development of emotional competence include child-specific (e.g., temperament) and family factors (Bradley, 2000; Sanson, Hemphill, & Smart, 2004). Eisenberg, Spinrad, and Cumberland (1998) suggest that parents influence children's emotional development through discussion of emotion, reactions to children's emotions, and family expressiveness. Directly, emotion-related discussions may explicitly teach the child ways of understanding and managing emotional experiences. Indirectly, emotion discussions help children learn to use emotion language and encourage emotion understanding. To the extent that these discussions are adaptive, they will contribute to emotion skill development. Parental reactions to children's emotions may influence their reliance on specific emotion management strategies. For example, rigid parental reactions may encourage a child's overreliance on inhibition as a form of emotion management. Finally, the frequency, intensity, and duration of positive and negative emotional expressiveness within the family may impact a child's formation of schemas about emotionality (Dunsmore & Halberstadt, 1997). In expressive homes, children may internalize that emotions are acceptable to show, whereas in inhibitive homes, children may learn that one should carefully monitor one's emotional expressions.

Empirical work supports the theoretical tenets: the explanatory discussion of emotion (Denham, Zoller, & Couchoud, 1994; Dunn, Brown, & Beardsall, 1991), parental responsiveness to children's emotions (Denham, Mitchell-Copeland, Strandberg, Auerbach, & Blair, 1997), and family expressiveness of emotion (Denham & Kochanoff, 2002; Denham et al.,1994; Dunn, et al., 1991; Garner, 1995) predict the understanding and regulation of emotion in normative youth. For example, with respect to family expressiveness of emotion, Denham and colleagues (1997) found that children with more affectively positive parents were more affectively balanced themselves, and that children with more affectively negative parents were rated as less socially competent by teachers. In addition, Denham and colleagues (1997) found that parents who reported the maintenance of positive affect during challenging circumstances were more likely to have children who were more adept at understanding emotions. Thus, it appears that parental modeling of increased positive affect and decreased negative affect may lead to enhanced child emotional understanding and social competence.

Gender Differences in Emotion Socialization

Most research on parental socialization of emotion has focused on mothers. Findings indicated that mothers (a) spoke more about emotions generally, but less about negative emotions in particular, with their daughters versus their sons (Dunn et al., 1987, Fivush, 1989) and (b) tended to discuss the emotion itself with their daughters but the causes and consequences of emotions with their sons (Fivush, 1989). Other research found that compared to boys, girls anticipated a more positive response from their mothers when expressing sadness (Fuchs & Thelen, 1988; Zeman & Garber, 1996). The few studies of fathers suggest that their role differs from that of mothers. Through direct observation of mother-child and father-child discussions of happy, angry, sad, and scared situations fathers spoke less overall, discussed the causes of emotions to a lesser extent, and used significantly fewer emotion words compared to mothers (Fivush, Brotman, Buckner, & Goodman, 2000). Similarly, when responding to vignettes with their child in mind, in comparison to mothers, fathers of 6- to 11-yr-olds endorsed minimizing the expression of sadness more and encouraging the expression of sadness less (Cassano & Zeman, 2007). Other research found that fathers (not mothers) showed greater attention to preschool girls' submissive expressions compared to boys (Chaplin et al., 2005). Some research has found that both mothers and fathers may be more accepting of anger expression in boys (versus girls) and fear in girls (versus boys) (Birnbaum & Croll, 1984). This literature suggests that studies of the socialization of emotion should include mothers and fathers (Cassano, Adrian, Veits, & Zeman, 2006; Bogels & Phares, 2008; Phares, Fields, Kamboukos, & Lopez, 2005).

Though the reviewed literature suggests that parental emotion socialization practices are unidirectional it is important to acknowledge that they are not. That is, emotional socialization practices are likely influenced by child-specific variables such as temperament, as well as patterns of interaction occurring within the larger family system. Indeed, an individual's behavior is influenced by transactional patterns within the family system (Hughes & Gullone, 2008; Minuchin, 1985). For example, a temperamentally difficult child who expresses intense negative affect may actually elicit negative emotional expression from a parent, which in turn may serve to reinforce the child's negative emotionality. Similarly, positive and pleasant interactions among the marital dyad likely contribute to similar communications with children.

Emotion Socialization and Anxiety Disorders

Research examining emotion-related functioning in youth with ADs is just beginning. AD youth are a particularly useful group in which to study family processes that contribute to emotional development, given the familial influences on the disorder (see Eley, 2001 and McLeod, Wood, & Weisz, 2007 for reviews). Notably, fathers have been included even less in research with disordered than with normal samples (Bogels & Phares, 2008; Cassano et al., 2006).

With respect to emotion socialization specifically, Suveg and colleagues coded family discussions in which AD and ND youth and their mothers talked about times when the child felt worried, angry, and sad (Suveg, Zeman, Flannery-Schroeder, & Cassano, 2005). Findings indicated that AD children spoke a greater percentage of time than their mothers, whereas the opposite was true for the ND dyads. Mothers of AD children used significantly fewer positive emotion-related words and demonstrated significantly more discouragement of emotion discussion than did mothers of ND children, and AD children and their mothers reported significantly less expressiveness of emotion in their families compared to ND children and their mothers. Given these findings, it is not surprising that AD youth exhibit deficits in emotion understanding, regulation, and emotional self-efficacy (Southam-Gerow & Kendall, 2000; Suveg & Zeman, 2004).

This study examined emotion socialization practices in families of children with an anxiety disorder and families of children who had no diagnoses, and considered gender differences. The study aims to contribute to the field of family psychology by including behavioral observations and thus helping to identify both verbal and nonverbal means of emotion interaction between youth and his/her mothers and fathers. The research has potential to contribute further to the field of family psychology by including both an AD and ND group, thus enabling the identification of the ways in which socialization processes may go awry and contribute to or maintain youth anxiety. It was hypothesized that, relative to parents of ND children, parents (both mothers and fathers) of AD youth would: (a) use a fewer number of words overall, (b) discourage the discussion of emotions by their children, (c) engage in less explanatory discussion of emotions and (d) exhibit more negative affect and less positive affect. Further, it was hypothesized that, relative to ND children, children with an AD would: (a) engage in less explanatory discussion of emotion, (b) exhibit less positive affect across all emotions, and (c) endorse a fewer number of problem-solving and a greater number of maladaptive emotion regulation decisions in the anxious and angry scenarios. With respect to child gender, it was hypothesized that, when speaking with their daughters versus sons, fathers would: (a) talk less overall (b) engage in less explanatory discussion of emotion and (c) exhibit less discouragement when discussing an anxious situation but more discouragement when discussing an angry situation, whereas mothers would (a) also engage in less explanatory discussion of emotion and also (b) discourage anger discussion less. Lastly, it was expected that (a) no significant group or gender differences would be found in the happy scenario.

Method

Participants

Participants were 28 AD youth (13 males, 15 females) and their mothers and fathers, and 28 ND youth (13 males, 15 females) and their mothers and fathers. AD and ND children were matched for age and gender. We included children who were in this age range because by age eight basic emotion competence skills (e.g., decoding, understanding, vocabulary; Saarni, 1999) should be developed. The majority of families in both groups was Caucasian and from middle- to upper-middle socioeconomic status (see Table 1). The 28 AD youth were diagnosed with generalized anxiety disorder (GAD; n = 16), separation anxiety disorder (SAD; n = 7), and social phobia (SP; n = 11). Some children met criteria for more than one principal anxiety disorder. Two children had comorbid mood disorders and 12 had comorbid externalizing disorders. None of the youth in the ND group met criteria for any psychological disorder. Chi-square analyses revealed non-significant group differences in demographic variables.

Table 1. Demographic Information.

	Anxiety Disorder	Non-Clinical
Age (Mean ± SD)	10.1 ± 1.45	10.04 ± 1.49
Gender
Male	13	15
Female	13	15
Race/Ethnicity
Caucasian	26	21
African-American	1	1
Biracial	1	3
Annual Family Income
<$39,999	1	3
$40,000-59,999	2	4
$60,000-80,000	9	11
>$80,000	14	9
Missing Information	2	1
Family Structure
Living with biological parents	27	28
Living with adoptive parents	1	0

Measures

The Anxiety Disorder Interview Schedule for Children and Parents (ADIS-IV-C/P)

The ADIS-IV-C/P (Silverman & Albano, 1996) has child and parent interviews. The emphasis is on the diagnosis of anxiety disorders, but the presence/severity of other disorders is assessed. Impairment ratings are given separately by the child and parents, and each are considered in deriving composite diagnoses. Psychometric properties have been established (Silverman, Saavedra, & Pina, 2001; Wood et al., 2002). Before administering the interviews independently, diagnosticians were required to reach a reliability of .80 (Cohen's kappa). A check of randomly selected interviews during the study indicated that raters maintained their initial reliability.

Emotion Discussion task

Children and parents participated in a 15-minute emotion task in which they discussed (5 minutes each) a time when the child felt anxious, angry, and happy. This study used observation (not self-report) given research suggesting that child self-report of emotion may not fully capture their experience (Casey, 1993; Hubbard et al., 2004). In addition, the present study examined emotion socialization during discussions about happiness to evaluate emotion socialization in a positive and negative context.

Families recalled a time when the child felt very anxious, angry, and happy, and discussed the details (what happened, what each member did). Following several training sessions (practice coding, discussions regarding the coding), the videotaped discussions were coded by an advanced doctoral student in clinical psychology and the 1^st author, both of whom were blind to diagnostic status. Variables were coded on a 5-point Likert scale (except the total number of words and the last variable that was coded categorically): (a) total number of emotion words, (b) parental discouragement of emotion discussion (when the parent changed the topic abruptly or at a clearly inappropriate juncture, ignored or belittled the youth who was discussing an emotionally-evocative situation), (c) explanatory discussion of emotion (any discussion relevant to the causes, consequences of emotion for each member of the triad), (d) affective tone of the discussion (overall negative and positive tone of the participant's participation in the discussion, which was rated by considering the participant's tone of voice, facial expression, and body language), and (e) child emotion regulation decisions. Child emotion regulation strategies were coded as either (a) problem-solving where the child attempted cognitive or behavioral strategies to constructively manage his or her feelings (e.g., “When I was worried about not making the soccer team, I just tried as hard as I could and thought that if I didn't make it this year, I could try again next year.”), (b) maladaptive where the child attempted to avoid or distance him- or herself from the situation, endorsed revengeful or other behaviorally or relationally aggressive strategies, or other strategies that were clearly maladaptive (e.g., “When I was worried because my friend could not go to camp with me and I didn't know anyone else who was going, I didn't go either.”, “When I was worried about going to camp because I didn't know any of the other kids there, I stayed home and told myself what a baby I was.”), or unscorable when the child provided a response that could not be categorized into either problem-solving or maladaptive. These variables reflect important components of emotion socialization and functioning. All emotion discussion variables were reliably coded (For approximately 1/3 of the tapes ICCs ranged from .66 for Mother's Positive Affect to .99 for Total Words for all participants) except for child's negative affect (ICC = .41). Due to the unreliability of the code, this variable was not analyzed.

Procedure

AD families were referred to a child anxiety clinic for participation in a treatment study by multiple sources (e.g., schools, pediatricians, media ads). ND families were community volunteers recruited for normative comparison through media advertisements. During the initial evaluation, parents and the child signed consent/assent forms and participated in interviews conducted separately with the parents and the child. Diagnosticians were blind to family status (treatment seeking or community volunteer). Children and parents then engaged in the emotion discussion tasks. Prior to each discussion, family members identified a recent situation that evoked strong emotion (e.g., happiness, anxiety, anger) in the child. After all family members agreed on a topic, they were instructed to discuss this situation in depth. Discussions were videotaped and limited to five minutes per emotion (order of emotions counter-balanced). Following the evaluation, AD families received treatment and ND families were paid $150 for their time.

Of 80 community volunteers who presented consecutively, only those 28 youth who were free of DSM-IV diagnoses (as determined by the ADIS-IV-C/P) and whose mothers and fathers were present during the emotion discussion tasks were included in the ND sample. Twenty-eight youth who met criteria for a principal diagnosis of GAD, SAD, or SP, whose mothers and fathers were present during the emotion discussion tasks, and who matched the ND youth on age and gender were then identified and included in the AD sample.

Results

Analyses examined the total number of words spoken (parents), parental discouragement of emotion discussion, explanatory discussion of emotion (parents and child), and the affective tone of the discussion (positive and negative; parents and child). To investigate Parent data, 3 (Emotion; Happy, Anxious, Angry) × 2 (Group; AD, ND) × 2 (Child Sex) Repeated Measures Analyses of Variance (RM-ANOVAs) were conducted separately by sex of the parent. To examine child data, 3 (Emotion) × 2 (Group) × 2 (Sex) RM-ANOVAs were conducted. Chi-square analyses were conducted separately for each emotion scenario to examine group differences in emotion regulation decisions for the situations (child data only). A power analysis using G Power (Buchner, Erdfelder, & Faul, 1997; http://www.psycho.uni-duesseldorf.de/aap/projects/gpower/) was conducted using an effect size in the moderate range, which yielded an acceptable level of power (.79). Measures of effect size (i.e., r) were obtained for all analyses where appropriate and interpreted to Cohen's (1998) criteria: .10 (small effect), .20 (medium effect), and .30 (large effect).

Mother Analyses

Number of words

A significant Emotion × Group × Sex interaction emerged for the number of words spoken, F(2, 48) = 5.12, p < .01; follow-up ANOVAs revealed that mothers in the ND group used a significantly greater number of words in the angry scenario when talking with their sons than did mothers in the AD group; F(1, 23) = 5.02, p < .05, r = .43 (see Table 2).

Table 2. Means and Standard Deviations for Emotion Variables by Group and Child Sex for Mothers and Fathers.

Mothers
Emotion	Anxious		Angry		Happy
	Anxious M (SD)	Non-Disordered M (SD)	Anxious M (SD)	Non-Disordered M (SD)	Anxious M (SD)	Non-Disordered M (SD)
Total Number of Words
Girls	371.87 (158.80)	293.79 (151.00)	359.13 (155.32)	278.86 (146.36)	300.87 (116.43)	293.00 (149.72)
Boys	277.62 (126.29)	329.00 (139.48)	273.77 (115.45)	435.91 (250.69)	263.08 (84.34)	293.76 (175.17)
Discouragement of Emotion Discussion
Girls	1.26 (0.70)	1.21 (0.58)	1.80 (9.41)	1.35 (0.63)	1.26 (0.46)	1.14 (0.36)
Boys	1.46 (0.66)	1.09(0.30)	1.46 (0.52)	1.09 (0.30)	1.15 (0.38)	1.00 (0.00)
Explanatory Discussion of Emotion
Girls	2.86 (1.06)	2.42 (0.85)	3.06 (1.33)	2.64 (0.84)	2.33 (0.72)	2.57 (1.01)
Boys	2.23 (0.83)	3.27 (0.90)	2.46 (1.12)	3.00 (1.09)	2.61 (1.26)	2.45 (1.29)
Positive Affective Tone
Girls	3.40 (0.63)	3.42 (0.76)	3.00 (0.85)	3.21 (0.70)	3.53 (0.74)	3.92 (0.62)
Boys	2.84 (0.69)	3.81 (0.61)	3.00 (0.58)	3.72 (0.47)	3.23 (0.60)	4.09 (0.83)
Negative Affective Tone
Girls	1.60 (0.63)	1.57 (0.76)	2.20 (0.86)	1.71 (0.73)	1.40 (0.51)	1.28 (0.47)
Boys	2.38 (0.96)	1.27 (0.47)	2.15 (0.38)	1.27 (0.47)	1.69 (0.63)	1.18 (0.40)
Fathers
Total Number of Words Spoken
Girls	187.26 (94.59)	265.71 (175.29)	198.20 (149.60)	229.14 (129.69)	169.80 (91.27)	255.21 (151.80)
Boys	283.46 (183.08)	289.00 (122.38)	317.38 (169.60)	268.27 (142.33)	303.69 (181.84)	323.72 (162.74)
Discouragement of Emotion Discussion
Girls	1.20 (0.41)	1.50 (0.76)	1.53 (0.64)	1.21 (0.43)	1.20 (0.41)	1.14 (0.36)
Boys	1.61 (0.77)	1.09 (0.30)	2.00 (0.58)	1.18 (0.40)	1.15 (0.38)	1.00 (0.00)
Explanatory Discussion of Emotion
Girls	2.00 (0.65)	2.42 (0.94)	2.00 (1.06)	2.50 (1.09)	1.93 (0.96)	2.21 (0.97)
Boys	2.30 (0.75)	3.00 (1.00)	2.30 (0.75)	2.72 (1.10)	2.07 (0.49)	2.81 (1.16)
Positive Affective Tone
Girls	3.20 (0.77)	3.07 (0.83)	3.00 (0.93)	3.14 (0.77)	3.60 (0.83)	3.50 (0.94)
Boys	3.00 (0.82)	3.81 (0.75)	3.00 (0.58)	3.63 (0.50)	3.30 (0.63)	4.00 (0.77)
Negative Affective Tone
Girls	1.80 (0.77)	1.78 (0.89)	2.13 (0.74)	1.64 (0.74)	1.60 (0.63)	1.57 (0.76)
Boys	2.07 (0.86)	1.09 (0.30)	2.07 (0.76)	1.27 (0.47)	1.84 (0.55)	1.18 (0.40)

Parental discouragement of emotion discussion

On the Discouragement scale, a significant Emotion × Sex interaction emerged, F(2, 48) = 3.18, p < .05; mothers discouraged emotion discussion more with daughters in the angry vs the happy [F(1, 28) = 9.11, p < .01, r = .58] and anxious [F(1, 28) = 11.29, p < .01, r = .47] scenarios. When speaking with their sons, mothers discouraged emotion discussion more in the angry vs the happy scenario; F(1, 23) = 6.05, p < .05, r = .23.

Explanatory discussion of emotion

On the Explanatory Discussion scale, a significant Emotion × Group × Sex interaction emerged, F(2, 48) = 4.64, p < .01; with sons, mothers in the ND group engaged in more explanatory discussion of emotion when discussing an anxious situation than did mothers in the AD group, F(1, 24) = 7.39, p < .01, r = .47.

Tone of discussion

A significant Emotion × Sex interaction emerged on Positive Tone, F(2, 48) = 3.41, p < .04. Follow-up univariate ANOVAs indicated that with sons, mothers exhibited more positive affect in the happy scenario than in the anxious [F(1, 22) = 10.77, p < .000, r = .20] or angry [F(1, 22) = 5.23, p < .05, r = .19] scenarios. With daughters, mothers also showed more positive affect in the happy versus anxious [F(1, 27) = 5.93, p < .05, r = .22] and angry [F(1, 27) = 16.37, p < .00, r = .39] scenarios but also more positive affect in the anxious than in the angry scenario, F(1, 27) = 12.39, p < .00, r = .21. A significant Group × Sex interaction also emerged, F(1, 49) = 3.95, p < .05; follow-up univariate ANOVAs indicated that with their sons, ND mothers exhibited more positive affect than did AD mothers, F(1, 22) = 13.97, p < .001, r = .62. On the Negative Tone scale, a significant Emotion × Sex interaction emerged, F(2, 48) = 4.00, p < .05; follow-up analyses indicated that with daughters, mothers exhibited more negative affect in the angry scenario than in the happy [F(1, 27) = 18.64, p < .000, r = .42] or anxious scenarios [F(1, 27) = 11.58, p < .000, r = .24], whereas with sons, mothers exhibited greater negative affect in the anxious versus the happy scenario [F(1, 22) = 10.47, p < .004, r = .24] and in the angry versus happy scenario [F(1, 22) = 6.52, p < .05, r = .22]. A Group × Sex interaction also emerged, F(1, 49) = 4.84, p < .05; with sons, AD mothers exhibited more negative affect than did ND mothers across all emotions, F(1, 22) = 17.67, p < .000, r = .67.¹

Father Analyses

Number of words

A RM-MANOVA revealed a significant main effect for Sex; fathers spoke more with their sons than their daughters, F(1, 49) = 4.59, p < .05, r = .30.

Parental discouragement of emotion discussion

For Discouragement, a significant Emotion × Group × Sex interaction emerged, F(2, 48) = 3.37, p < .05; follow-up ANOVAs indicated that when speaking with their sons, fathers in the AD group discouraged discussion in the angry scenario more than fathers in the ND group; F(1, 23) = 7.36, p < .01, r = .48.

Explanatory discussion of emotion

For Explanatory Discussion, a significant main effect for Group emerged, F(1, 49) = 6.33, p < .05, r = .34; fathers in the ND group engaged in more explanatory discussion of emotion across all scenarios than fathers in the AD group.

Tone of discussion

A significant main effect for Emotion emerged on the Positive Tone scale, F(2, 48) = 9.31, p < .000; subsequent within-subjects contrasts indicated that fathers exhibited more positive affect in the happy scenario compared to the anxious [F(1, 49) = 5.68, p < .000, r = .20] and angry [F(1, 49) = 17.60, p < .000, r = .26] scenarios. A significant Group × Sex interaction also emerged, F(1, 49) = 4.19, p < .05; follow-up univariate ANOVAs indicated that with sons, fathers in the ND group exhibited more positive affect than did fathers in the AD group, F(1, 22) = 10.16, p < .000, r = .56. On the Negative Tone scale, a significant main effect for Emotion emerged, F(2, 48) = 3.10, p < .05; subsequent within-subjects contrasts indicated that fathers exhibited more negative affect in the angry versus the happy scenario; F(1, 49) = 6.28, p < .05, r = .17. A Group × Sex interaction also emerged, F(1, 49) = 4.09, p < .05; follow-up univariate ANOVAs indicated that with sons, fathers in the AD group exhibited more negative affect than fathers in the ND group; F(1, 22) = 16.21, p < .001, r = .65.

Child Analyses

Explanatory discussion of emotion

No significant main effects or interactions emerged.

Tone of discussion

Significant main effects for Emotion, F(2, 48) = 22.92, p < .00, and Group, F(1, 49) = 14.66, p < .00, emerged for Positive Tone. Children exhibited more positive affect in the happy scenario compared to the anxious [F(1, 49) = 37.69, p < .000, r = .35] and angry [F(1, 49) = 38.10, p < .000, r = .37] scenarios. Across emotions, children in the ND group exhibited more positive affect than children in the AD group, F(1, 49) = 14.67, p < .000, r = .48.

Emotion regulation decisions

Chi-square analyses indicated a significant difference for the anxious, χ² (1, N = 49) = 6.44, p < .05, and angry scenarios, χ² (1, N = 45) = 8.70, p < .01. Based on the odds ratio, AD children were 5.25 and 15.23 times more likely to indicate a maladaptive versus adaptive response during the anxious and angry scenarios, than ND youth.

Discussion

Theory suggests that children's emotional development is influenced by patterns of interaction with family members, particularly parents. Previous supportive research involves ND youth and their mothers (e.g., Denham et al., 1994; Garner, 1995). This study provides support for the role of parents in children's emotional development and extends the findings by (a) documenting how emotion socialization practices vary in families of anxious youth and (b) detailing a specific role for fathers. Overall, parents of AD youth differed from parents of ND youth in some aspects of emotion socialization, with the results varying by emotion type and child gender. In both AD and ND groups, fathers appeared to have greater involvement in emotion-related discussions with their sons versus daughters.

Emotion Socialization in Families of Anxious Youth

The hypothesis that parents in the ND group would speak more than parents in the AD group was minimally supported: ND mothers used a greater number of words than AD mothers when discussing an angry situation with their sons, but not for the other emotions. Perhaps mothers of an AD child are uncomfortable discussing a negative emotion. If true, then we would expect to see a difference during the anxious discussion, which was not the case. However, given that parents of AD youth were seeking treatment for their child's anxiety, they may have deemed it important to discuss the anxious situation, at least at some level.

The second and third hypotheses that parents of AD youth would engage in less explanatory discussion of emotion and discourage emotion discussion more than parents of ND children were partially supported. Fathers in the ND group engaged in more explanatory discussion (i.e., spent time discussing the causes and consequences of emotions) across all scenarios than did AD fathers. So too for ND mothers, but to a lesser extent, as they were more likely than AD mothers to engage in explanatory discussion of emotion with their sons during the anxious scenario, but not during the angry or happy scenarios. Regarding discouragement of emotion discussion, only one group finding emerged – fathers of AD youth were more likely to discourage emotion discussion in the angry scenario than fathers of ND youth. The findings that parents of AD youth were at times less likely to engage in explanatory discussion of emotion is important however given data linking parental discussion of emotion to children's emotion understanding and regulation skills (Denham et al., 1994; Dunn et al., 1991). Youth who are not given ample opportunities to process emotional situations via discussions with their parents are at a disadvantage given that this is one way that children acquire emotion-related skills.

Our hypothesis that parents of AD youth would exhibit more negative affect and less positive affect than ND parents received partial support. When interacting with sons, mothers and fathers in the AD group exhibited more negative affect and less positive affect than mothers and fathers in the ND group, respectively regardless of the emotion being discussed. The findings were not expected to be specific to boys. It could be that parents of AD youth are overly concerned with socializing gender stereotypic emotional functioning (e.g., girls more emotionally expressive than males) and thus by exhibiting more negative and less positive affect discouraged boys' emotional expressions. Regardless, the findings are important, given the role of parental modeling of positive affect in appropriate emotion socialization. Parental modeling of increased positive affect and decreased negative affect, particularly during stressful situations, may lead to enhanced emotional understanding and social competence in their children (Denham et al., 1997). This uniformity in emotional reactions across emotion discussions is unlikely to help children differentiate amongst emotions (Denham et al., 1994) or learn when it may be appropriate to exhibit one emotion versus another. Some research suggests that boys are less adept at using display rules (e.g., hiding negative emotion when it seems socially appropriate) than girls (Davis, 1995; Saarni, 1984); perhaps AD boys would show a remarkable deficit in this area – future research should explore this question.

Together, the findings that parents of AD youth engaged at times in less explanatory discussion of emotion and more discouragement of emotion discussion and exhibited decreased parental positive affect are meaningful. Eisenberg et al. (1998) argued that children who are raised by parents who do not discuss negative emotions openly may lack information about emotion regulation and may deduce that certain emotions should not be expressed. Also, if parents of AD children are focusing on anxious emotions without necessarily elaborating on or providing additional information about the causes or consequences of these emotions, it is unlikely that such youth will learn emotion regulation methods. These children may rely on less adaptive strategies to negotiate anxiety-provoking situations, which may maintain their anxiety.

We expected parents of AD youth to engage in maladaptive emotion parenting, and we anticipated that AD youth would exhibit deficits in some domains of emotion functioning. Although we anticipated that AD youth would engage in less explanatory discussion of emotion than ND youth, this was not found. Yet, the finding is consistent with previous research (Suveg et al., 2005). It remains unclear why, given less adaptive socialization exposure, AD and ND children would evidence similar levels of explanatory discussion. Perhaps emotion difficulties become evident when a need arises for children to translate their knowledge of emotion experience into adaptive emotion regulation strategies. The current and previous findings support this notion. AD children engaged in less problem-solving and more maladaptive emotion regulation strategies when discussing anxious and angry situations than ND children. Recall that the situations that were discussed were real, and children were asked to discuss what they did during each emotion-provoking situation. It appears that emotion skill deficits are more evident in the actual enactment of emotion regulation strategies than when discussing emotions.

AD youth were expected to display less positive affect across all discussions than ND youth, and the findings were supportive. Perhaps AD youth experience less positive affect in general, as well as in emotionally-laden situations. However, research suggests that low positive affect is more related to depression than anxiety (e.g., Joiner, Catanzaro, & Laurent, 1996; Suveg, Hoffman, and Zeman, 2008), although these emotional disorders show high comorbidity. Last, given that parents of AD youth also exhibited less positive affect than parents of ND youth, it is possible that youth were modeling their parents' affect. Research is needed to determine whether the decreased positive affect of AD youth is present in all situations or whether it is specific only to emotion-laden situations.

Overall, results suggest that parents of AD youth engage in some maladaptive emotion parenting and that AD youth engage in less problem-solving and more maladaptive emotion regulation strategies as compared to ND youth when experiencing negative emotions. Although not evaluated in this study (larger samples of youth would be needed), it is possible that child emotion dysregulation mediates the relationship between maladaptive emotion parenting and child anxiety. Delineation of the mechanisms by which maladaptive parenting relates to poor outcomes in youth is a worthy area of study that has potential to further contribute to the field of family psychology.

The Role of Parent and Child Gender in Emotion Socialization

The present findings highlight a role for fathers and child gender in emotion socialization. Consistent with the literature and our hypotheses, parents engaged in covert and overt methods of gender-based emotion socialization. As predicted, fathers spoke more across all scenarios with their sons versus their daughters. However, neither fathers nor mothers engaged in less explanatory discussion when in conversations with daughters versus their sons. It appears that while fathers are talking less overall with their daughters, specific discussion regarding the causes and consequences of emotion is similar for both boys and girls. With respect to mothers, the finding of equal explanatory discussion of emotion is inconsistent with previous research (Fivush, 1989). The inconsistent findings may be due to age differences in the samples; where gender differences in maternal explanatory discussion of emotion are most salient with younger groups (as was utilized in the Fivush study).

Previous research using adult retrospective reports found that fathers were more likely to punish the expression fear by sons, as compared to daughters (Garside & Klimes-Dougan, 2002). We anticipated that fathers would discourage discussion during the anxiety scenario more with boys than girls, but this was not found. It may be that families seen in an anxiety clinic perceive pressure to discuss this particular emotion.

Mothers exhibited more negative affect in the angry scenario with girls, whereas with boys, mothers exhibited more negative affect in the anxious scenario. These findings are consistent with previous research that found socialization pressures for particular emotions to vary by gender (e.g., Chaplin et al., 1995). The display of negative affect by mothers may be a covert form of socialization -- sending messages about their acceptance of the emotion discussion or the child's behavior via facial expression, tone of voice, and body language. Mothers may inadvertently engage in gender-linked emotion socialization by subtly attending to or ignoring a child's expression of an emotion.

Parents engaged in overt forms of gender-linked emotion socialization. We anticipated mothers and fathers to engage in more discouragement in the angry scenario with girls than boys and this was partially supported. Though there were no differences for fathers, mothers generally discouraged discussion in the angry scenario for both boys and girls, and this was particularly the case with girls. For AD and ND samples, it appears that there is strong pressure for the socialization of anger. Normatively, it might be appropriate to teach children to control or otherwise manage their anger expressions – be it via parental affective tone or active discouragement. As was found here, the pressures to socialize anger expression might be particularly strong with girls, given that anger is an emotion more stereotypically expressed by boys. Indeed, girls who express anger more freely than is considered typical may falsely be identified as having an externalizing type of problem.

However, there are less adaptive implications of discouraging girls' expressions or discussions of anger. Girls may be at a disadvantage: if anger is suppressed there may be physiological arousal (Gross & Levenson, 1997) and nonadaptive expressions of emotion. It is possible that girls express their anger in ways that are less obvious to others, in order to be consistent with gender stereotyped expectations. Such pent-up anger in girls may contribute to relational aggression (Crick & Grotpeter, 1995). The role of anger socialization in girls and its potential contribution to the development of relational aggression warrants further investigation.

This study included a discussion of a happy scenario to examine whether similar emotion socialization mechanisms operate within the context of positive emotions. As hypothesized, there was not salient gender or other socialization pressures/differences when discussing a happy scenario. Rather, parents (both of AD and ND youth) let children discuss the situation without much involvement. The interaction was fairly pleasant and participants were likely to feel comfortable and at ease.

The results contribute to the field of family psychology in a number of ways. First, the results help to articulate a specific role for fathers in youth emotional development. Second, the findings highlight some of the ways in which parents engage in emotion socialization and how such practices vary depending on the gender of the parent and the child. Finally, the results ascertain how emotion socialization processes go awry in families of children with an AD potentially contribute to or maintain the youth's anxiety. The results also have clinical implications. Anxiety treatments are beginning to address the emotion-related deficits in AD youth. Emotional competence is important, so it is reasonable that an intervention that maintains the demonstrated efficacies of CBT but enhances the content to address emotion-related deficits may improve outcomes (Southam-Gerow & Kendall, 2002; Suveg, Kendall, Comer, & Robin, 2006; Suveg, Southam-Gerow, Goodman, & Kendall, 2007). Some child anxiety treatments include parents (e.g., Bögels & Siqueland, 2006; Kendall, Hudson, Gosch, Flannery-Schroeder, & Suveg, 2008), although how parents are involved has varied (see Barmish & Kendall, 2005). These results offer ideas for parental roles in treatment.

Potential limitations merit consideration. The present study was a snapshot of parent-child interactions, which may compromise generalizability (e.g., to peer interactions). The verbal and nonverbal responses of mother, father, and child were analyzed separately and the effects that one family member's responses may have had on another family member's responses but these were not examined. Theory and research suggest that individual behavior is influenced by transactional family patterns (Bell & Chapman, 1986; Hughes & Gullone, 2008; Minuchin, 1985), and this is an important topic for future work. It may be that the discouragement of emotion discussion by one parent impacts the number of words spoken by another parent, or that the child's tone affects the parents' tone. Siblings, who may play an active role in emotion-related discussions or whose presence may impact family interactions, were not included. Process studies utilizing sequential analyses are needed to examine the impact of one family member's responses on another family member's responses. Coding schemes that might have better captured family processes could have been created for this study. For example, a scheme that examines parental and child emotional responsiveness to one another might have more fully captured emotion socialization processes. Last, the sample was homogenous, and it is not known whether the results here would generalize to other populations.