African American and European American Mothers’ Beliefs about Negative Emotions and Emotion Socialization Practices

SYNOPSIS

Objective

Mothers’ beliefs about their children’s negative emotions and their emotion socialization practices were examined.

Design

Sixty-five African American and 137 European American mothers of 5-year-old children reported their beliefs and typical responses to children’s negative emotions, and mothers’ emotion teaching practices were observed.

Results

African American mothers reported that the display of negative emotions was less acceptable than European American mothers, and African American mothers of boys perceived the most negative social consequences for the display of negative emotions. African American mothers reported fewer supportive responses to children’s negative emotions than European Americans and more nonsupportive responses to children’s anger. African American mothers of boys also reported more nonsupportive responses to submissive negative emotions than African American mothers of girls. However, no differences were found by ethnicity or child gender in observed teaching about emotions. Group differences in mothers’ responses to negative emotions were explained, in part, by mothers’ beliefs about emotions.

Conclusions

Differences in beliefs and practices may reflect African American mothers’ efforts to protect their children from discrimination.

INTRODUCTION

Parental socialization of children’s emotions, specifically children’s understanding, expression, and regulation of emotion, is important for children’s development as evidenced by associations with children’s emotional and social competence (Eisenberg, Cumberland, & Spinrad, 1998; Gottman, Katz, & Hooven, 1997). Ethnicity and culture largely influence children’s development, particularly through the goals, beliefs, and practices parents use regarding children’s socialization (Bronfenbrenner, 1977; Cole & Tan, 2007; Super & Harkness, 1986). For example, cultural groups support children’s adaptation to the larger society by enacting shared socialization practices that maintain order, accomplish goals, and promote culturally-defined competencies (Cole & Tan, 2007; Ogbu, 1981). Despite this view, relatively few investigators have studied ethnic or cultural group differences in emotional socialization, and even fewer have investigated the reasons behind emotion socialization differences. The purpose of the current paper was to examine African American and European American mothers’ emotion beliefs and socialization practices and to test a mediation model where ethnic differences in emotion socialization practices were explained by maternal beliefs about the acceptability and negative social consequences of their children’s display of negative emotions. Ethnicity, Context, and Emotion

Ogbu’s (1981) cultural-ecological perspective provides a culturally specific model of childrearing with causal relations from the competencies that a population desires in youth to the socialization practices that caregivers use to ensure children acquire those competencies. Moreover, Ogbu posits that the same child outcome may be viewed as more or less competent in different groups depending on the tasks or demands faced by that group on an everyday basis. Existing emotion socialization literature generally suggests that parental encouragement of children’s experience and expression of emotion results in enhanced emotional competence (Eisenberg et al., 1998; Gottman et al., 1997). However, as Ogbu (1981) suggested, emotional competence must be considered with regard to the cultural context. Depending on a child’s age and goals, a behavior, such as expressing anger to assert dominance, may lead to a successful outcome in one cultural context but not in another (Halberstadt, Denham, & Dunsmore, 2001).

Although there is a scarcity of research on ethnic differences in emotion socialization, the roles and purposes of emotion seem to vary across groups. For example, there is evidence that overt emotional expression is linked to an orientation toward individuality and independence characteristic of European Americans in the United States, as opposed to a collectivist or familial orientation where emotions tend to be suppressed (Markus & Kitayama, 1991). Boykin’s (1986) Triple Quandary Theory suggests that African American families may have Afro-cultural experiences that emphasize emotional expression, oral modes of communication, and communication styles between parents and children that reflect a social-emotional orientation (Blake, 1993). However, African American families also have minority experiences where they must develop defense mechanisms in response to social stratification (Boykin, 1986). Despite their traditional cultural influences that value emotional expression, African American families have been influenced by a history of oppression that has led African Americans to value emotional self-control and limited self-disclosure (Consedine & Magai, 2002; Plasky & Lorion, 1984). Indeed, empirical research has demonstrated that European Americans express more negative emotions than African Americans, both in the report of day-to-day expression of negative affect and in the use of emotion to deal with conflict (Consedine & Magai, 2002). The explanation of these ethnic differences is consistent with García Coll et al.’s (1996) view that minority families develop an adaptive culture in response to historical and current demands.

The adaptive culture of African American families influences daily family process as well as children’s developmental competencies (García Coll et al., 1996). Parents’ beliefs and practices that discourage the expression of their children’s negative emotions in response to oppression can be thought of as firm emotional control strategies that parallel firm behavioral control. African American mothers report that behavioral control characterized by child obedience and compliance is an important skill for their children to have to interact successfully with the majority culture (Kelley, Power, & Wimbush, 1992). Some authors have suggested that this “surrendered” orientation is particularly endorsed among mothers of African American boys (Smith, 1982). In fact, African American boys are more likely to be punished for misbehavior at school than girls or boys of other ethnicities (Gregory, 1997), are viewed as more threatening (Stevenson, Herrero-Taylor, Cameron, & Davis, 2002), and report more discrimination (Fischer & Shaw, 1999). Parents’ behavioral control strategies may prepare children to handle discrimination by being cautious and defensive (Demo & Hughes, 1990) and prepare children for bias (Hughes et al., 2006) to succeed in a society where every behavior will be scrutinized (Dodge, McLoyd, & Lansford, 2005). Parents emotional control strategies may operate similarly. That is, African American mothers may worry that their children’s display of negative emotions will be misinterpreted or viewed as unacceptable by the majority culture and may respond to this concern by discouraging their children’s display of negative emotions, especially for boys. Evidence that faces of African American males are perceived as more angry and hostile than faces of European American males (Kang & Chasteen, 2009), particularly among individuals that are ethnically prejudiced (Hugenberg & Bodenhausen, 2003), suggests that concerns of this type are warranted. Next we review different types of negative emotions and the scant literature that has examined ethnic differences in emotion-related beliefs and behaviors.

Types of Negative Emotions

Specific negative emotions vary in many ways. Active expressions of anger have been discussed as dominant, whereas passive or more internalizing expressions of fear or sadness have been labeled submissive (Halberstadt & Eaton, 2003). It is possible that parents may have different beliefs and responses to different types of negative emotions. For example, mothers use more supportive responses to children’s sadness than anger, and they punish children for the expression of anger more than sadness, especially among boys (Klimes-Dougan et al., 2007). Thus, children’s submissive emotions may have a different meaning to parents than children’s dominant emotions, particularly in the contexts of ethnicity and gender.

Emotion Beliefs and Socialization

Emotion beliefs

The thoughts and feelings parents have about their own emotions and their children’s emotions may range from expressing openness and respect for the experience of negative emotions to expressing a belief that negative emotions are toxic and harmful (Gottman, Katz, & Hooven, 1996). Although ethnic comparisons of beliefs about emotions have been rare, African Americans report that the display of negative emotions, specifically disgust and sadness, is less acceptable than do European Americans (Matsumoto, 1993), and consistent with our conceptualization, it has been suggested that this difference may be a function of the lower social power afforded to minority groups (Matsumoto, 1989).

Emotion socialization

Emotion socialization processes generally include parents’ response to children’s emotions, their emotional expression, and the provision of opportunities for children to observe, experience, and discuss emotion (Cole & Tan, 2007; Eisenberg et al., 1998). Because the task of coping with negative affect is developmentally difficult for children (Ramsden & Hubbard, 2002), the ways parents respond when their children are sad, angry, or scared convey important information to the child about the display of negative emotion in general and about the appropriateness of negative emotion in particular situations. Additionally, parents’ abilities to directly teach their children about emotions and relate emotional experiences to the child’s own life have been linked to a greater understanding of emotions among children (Denham, Zoller, & Couchoud, 1994; Garner, Jones, Gaddy, & Rennie, 1997).

Although the importance of emotion socialization practices to children’s emotional development has also been demonstrated among African American families (Bocknek, Brophy-Herb, & Banerjee, 2009; Garner, 2006), studies of ethnic differences in emotion socialization practices are limited. Two retrospective studies of parents’ emotion socialization practices found that African Americans were more likely than European Americans to report that parents downplayed or punished negative feelings (Leerkes & Siepak, 2006; Montague, Magai, Consedine, & Gillespie, 2003). In another study, no ethnic differences were found for parents’ attempts to encourage or restrict children’s emotions (Lunkenheimer, Shields, & Cortina, 2007), although a small sample size and the inclusion of biracial children may have limited power.

Links between Emotion Beliefs and Socialization Practices

The importance of parents’ beliefs has been emphasized in the literature and a causal path has been proposed whereby emotion beliefs influence emotion socialization practices (Dunsmore & Halberstadt, 1997; Eisenberg et al., 1998). For example, parents who report that they accept and value their children’s emotions have been found to be more likely to encourage children to display negative emotions (Wong, Diener, & Isabella, 2008), to report using fewer hostile or dismissive responses to children’s negative emotions (Wong, McElwain, & Halberstadt, 2009), and to discuss a salient emotional event with children (Halberstadt, Thompson, Parker, & Dunsmore, 2008). Similarly, mothers who held positive beliefs about the usefulness of infant negative emotion were more sensitive toward their infants when they were distressed (Leerkes, 2010). Together, these findings suggest that if parental beliefs influence parenting practices, then some variations in practices may be explained by variations in beliefs. We explored the extent to which parenting behaviors can be explained by parenting beliefs by suggesting that the association between ethnicity and emotion socialization would be mediated by parents’ emotion beliefs. In other words, ethnic differences in socialization would be explained by differences in beliefs about the display and potential consequences of children’s negative emotions.

The Current Study

The few studies that have examined emotion socialization practices among African American mothers have concentrated on disadvantaged samples, such as low-income families (Bocknek et al., 2009; Garner et al., 1997) or families living in violent neighborhoods (Cunningham, Kliewer, & Garner, 2009). In the current study, we examined a normative sample of African American families from a range of economic circumstances to better understand the role of ethnicity in emotion beliefs and socialization practices, separate from income or education. We utilized the economic diversity in the sample to test whether income or education interacted with ethnicity to predict differences in beliefs and socialization practices between African American and European American mothers.

We aimed to address a gap in the emotion socialization literature by examining the relation between beliefs about negative emotions and emotion socialization practices in African American and European American mothers. We also examined the role of child gender in light of previous research on the particularly strict consequences for African American boys’ noncompliance (Gregory, 1997; Smith, 1982). Finally, we were informed by the cultural-ecological perspective to move beyond simply comparing ethnicities to understand similarities and differences. To do this, we tested a mediation model in which ethnic differences in mothers’ emotion socialization practices were expected to be explained by ethnic differences in mothers’ beliefs about the appropriateness and consequences of children’s display of negative emotions. We also explored the possibility that this mediation pathway differed for African American mothers of boys compared to other mothers considering the negative experiences and harsh consequences that African American boys disproportionately face for displaying negative emotions compared to other children (Fischer & Shaw, 1999; Gregory, 1997; Smith, 1982). Thus, we also tested a mediated moderation model where differences in mothers’ emotion socialization practices related to their child’s ethnicity and gender were explained by differences in mothers’ beliefs about children’s display of negative emotions.

We addressed the following questions: (1) Do mothers’ beliefs about negative emotions differ by ethnicity and child gender? We expected African American mothers, especially mothers of boys, to be less accepting of the display of negative emotions and to anticipate negative social consequences when children display negative emotions. (2) Do mothers’ emotion socialization practices differ by ethnicity and child gender? We examined the possibility that African American mothers would report using more emotion control strategies, such as less supportive and more nonsupportive responses when their children display negative emotions and would engage in less emotion teaching during mother-child interaction due to an emphasis on limited emotional expression, and that these differences would be amplified for African American mothers of boys. (3) If ethnic and gender differences in emotion socialization practices are apparent, do mothers’ beliefs about negative emotions explain these differences? We anticipated mothers’ beliefs about the appropriateness of negative emotion display and the social consequences of negative emotions would mediate the links between ethnicity and gender and mothers’ emotion control strategies. Finally, we explore differences in mothers’ beliefs and responses to submissive and dominant negative emotions and report variations separately.

METHOD

Participants

Children were recruited from child care centers and preschools. Families were enrolled in the study when children were 3 years old and participated further at 4 and 5 years of age. Of the 263 families that participated at age 3, 228 families had questionnaire data available at the 5-year visit. The measures of interest for the current study were available at the 5-year visit only. There were no significant differences by child gender or family income-to-needs ratio (total family income divided by the poverty threshold given family size) between families who continued and those who did not have data available at age 5; although families lost to attrition were more likely to be minority, χ²(1, N = 263) = 3.89, p < .05.

Given the goal of the current study, only families with a mother and target child who were both African American or both European American were included. Biracial families were excluded based on low numbers in the sample and the cultural variations that can exist within these families. The final sample included 200 families, one-third (n = 63) of whom were African American. Families were diverse in terms of income and education. Almost 32% of families had income-to-needs ratios less than 2.0 indicating low income. Median maternal education was a 4-year college degree, with 10% completing a high school degree or less and 22% having attended some college. Mothers were 35 years old on average; 52% of the children were female.

Procedure

The laboratory visit lasted approximately 2 hours. Mothers provided written consent and completed questionnaires during the session. Children were videotaped while engaging in multiple tasks, either with an experimenter or with their mothers. Families received $80 for the 5-year visit, and children selected a toy as thanks for their participation.

Measures

Demographics

Mothers completed a demographic questionnaire including child gender and ethnicity, maternal age, parents’ marital status, and family income. Ethnicity and child gender were recoded as 0 (European American; female) and 1 (African American; male).

Emotion beliefs: Display of negative emotions

The first emotion belief questionnaire, adapted from Matsumoto (1993), asked mothers to rate how acceptable they believed it was for their child to display anger, fear, sadness, and crying when he/she was alone, with family, with other children, in public, and with an authority figure on a 4-point scale (1 = not at all, 4 = very much). An exploratory factor analysis was conducted with a principal component analysis extraction and promax rotation. A one-factor solution was chosen based on the scree plot, eigenvalues (1 factor with a value over 7 times the next component), and variance explained (60%). The factor, consisting of 20 items, had an internal reliability of .96 with identical reliability values for African American and European American mothers. A composite was created, with higher scores indicating greater acceptance of the child’s display of negative emotions. For the exploratory analyses by type of emotion, 5 items described the display of a dominant emotion and 15 items described the display of submissive emotions. An item-level mean for dominant and submissive negative emotions was computed. Internal reliability was .87 for the dominant negative emotion subscale (African American α = .87, European American α = .84) and .97 for the submissive negative emotion subscale (African American α = .96, European American α = .96).

Emotion beliefs: The social consequences of negative emotions

The second questionnaire was developed for this study to assess the extent to which mothers believed there were negative social consequences associated with the display of negative emotions. Mothers rated 5 items on a 6-point Likert scale (1= strongly disagree, 6 = strongly agree). Examples of the items are “When my child shows anger, people may view my child as aggressive”, “If my child shows fear, people may think my child is a ‘scaredy-cat’”, and “If my child shows negative emotions openly, other people might give him/her fewer opportunities for success in life”. An exploratory factor analysis yielded a single factor with an eigenvalue > 1 that accounted for an adequate amount of variance (49%). All factor loadings were > .63, and internal reliability was acceptable (α = .73, African American α = .62, European American α = .77). These 5 items were averaged to yield a measure in which higher scores indicated a perception of more negative social consequences for the display of negative emotion. For the exploratory analyses by type of emotion, the single item about anger was maintained as the measure of beliefs about negative consequences in response to a dominant emotion, and the two items that referred to fear and sadness were averaged to yield a measure of beliefs about negative consequences in response to submissive emotions. For the latter, internal reliability was .72 (African American α = .65, European American α = .76). Two items that referred to negative emotions in general were excluded from the exploratory analyses.

Emotion socialization practices: Response to children’s negative emotions

Mothers reported on their responses to their children’s negative emotions using a revised version of the Coping with Children’s Negative Emotions Scale (CCNES; Fabes, Eisenberg, & Bernzweig, 1990). The CCNES includes 12 common situations in which the child is distressed (e.g., If my child loses some prized possession and reacts with tears, I would…). Mothers were asked to rate the likelihood that they would respond in each of 6 possible ways on a 7-point scale (1 = very unlikely, 7 = very likely). Revisions to the original CCNES measure include slight wording changes to the vignettes to add items that specifically address parent response to children’s anger. The measure yields a likelihood for 6 possible responses, or 6 subscales: problem-focused reactions (e.g., help my child think of places he/she hasn’t looked yet), emotion-focused reactions (e.g., distract my child by talking about happy things), expressive encouragement (e.g., tell him/her it’s OK to cry when you feel unhappy), distress reactions (e.g., get upset with him/her for being so careless and then crying about it), minimization reactions (e.g., tell my child that he/she is over-reacting), and punitive reactions (e.g., tell him/her that’s what happens when you’re not careful). Supportive (problem-focused, emotion-focused, expressive encouragement) and nonsupportive (distress, minimizing, punitive) items were averaged to create two composites. Internal reliabilities for the supportive and nonsupportive subscales in the current sample were .93 and .85, respectively (African American αs = .93, .87, European American αs = .92, .83). Three situations described the child’s expression of a dominant emotion (anger; items 1, 6, and 8 of the revised measure), and nine situations described the child’s expression of submissive emotions (sadness and fear; items 2, 3, 4, 5, 7, 9, 10, 11, and 12 of the revised measure). Thus, for the exploratory analyses by emotion type, an item-level mean was computed for dominant negative and submissive negative emotions. Internal reliabilities were .69 and .62, respectively, for the supportive and nonsupportive responses to dominant negative emotions (African American αs = .68, .57, European American αs = .67, .63) and .92 and .80, respectively, for the supportive and nonsupportive responses to submissive negative emotions (African American αs = .92, .85, European American αs = .91, .77).

Emotion socialization practices: Emotion teaching

Mothers and children participated in a joint storybook activity that lasted approximately 5 min. Mothers were asked to “read” an age-appropriate picture book, created for this study, depicting emotional situations with their children. On each page The Feeling Book contained a single emotion word (e.g., nervous; excited) with a corresponding picture illustrating the feeling. The interactions were videotaped, and two separate coding schemes were used, each coded independently by two trained coders.

The first coding scheme assessed direct emotion teaching during the interaction, and three core features of emotion socialization were rated: the emotional explanations provided by the mother, the direct provision of emotion information specific to the child’s experiences, and the extent to which the mother provided rich emotion cues by matching her voice, facial expressions, and gestures to the emotions in the book. The quantity and quality of mothers’ explanations and references to the child’s emotional experiences during the task were rated on a 5-point scale. A high score on these items involved frequent basic explanations or references (e.g., She’s nervous to be on stage; You felt happy yesterday!) in addition to occasional complex statements (e.g., Do you ever feel jealous when I hold baby Kayla? It’s okay to feel jealous sometimes but you should let Mommy know when you feel that way). Additionally, mothers’ ability to match their emotional expressions to the book emotions was rated on a 5-point scale. High scores included subtle tone variations between positive and negative emotions as well as occasional highly expressive verbal (e.g., Awww, he’s sad) or nonverbal matched cues (e.g., quivering to represent fear). Inter-observer agreement was calculated as the Intraclass Correlation between the two coders on 20% of the observations for emotion explanations, ICC = .66, references to the child’s emotions, ICC = .83, and emotion matching, ICC = .82.

The second coding scheme assessed the frequency with which mothers used positive and negative emotion words during the mother-child picture book interaction. Trained research assistants coded for the frequency of the mothers’ mental state language using procedures outlined by Jenkins, Turrell, Kogushi, Lollis, and Ross (2003). Separate positive (e.g., happy, love) and negative (e.g., sad, hurt, upset) scales were used. Approximately 25% of the videotapes (n=61) were coded independently by two coders. Inter-observer agreement was calculated as the Pearson correlations between the two coders, r = .93 positive emotion words, r = .91 negative emotion words. Mothers’ positive and negative emotion word totals were calculated by summing the count for each valence. The totals were prorated by total activity time to create a score that represented the frequency of emotion words per min.

Correlations between the five emotion teaching items ranged from .23 to .42, all ps < .01. An emotion teaching composite was created by standardizing and then summing the five scores. Data on dominant versus submissive emotions were not available for the emotion teaching composite because of the coding protocol used. Internal reliability for the composite was adequate (α = .72, African American α = .68, European American α = .73). Higher scores indicated more maternal emotion teaching during the parent-child interaction. Emotion teaching ratings were not made separately for different types of emotions; thus, this variable could not be divided into teaching about dominant and submissive emotions for the exploratory analyses.

RESULTS

Preliminary Analyses

Means and standard deviations for study variables can be seen in Table 1; correlations are shown in Table 2. Observational data on emotion teaching was excluded for six families where fathers engaged in the parent-child interaction task and for two families where a language other than English was spoken. The remainder of data from these families was used in analyses.

TABLE 1.

Descriptive Information of Study Variables as a Function of Ethnicity and Gender

			African American		European American		F
	Grand M (SD)	Range	Boys	Girls	Boys	Girls	Ethnicity	Gender	Ethnicity X Gender
Acceptable Display Beliefs	2.93 (.67)	1.05-4	2.66 (.80)a	2.59 (.63)a	3.02 (.58)b	3.14 (.60)b	18.34**	.25	.56
Dominant	2.43 (.67)	1-4	2.23 (.76)	2.07 (.67)	2.52 (.60)	2.60 (.61)
Submissive	3.12 (.72)	1-4	2.83 (.86)	2.79 (.71)	3.19 (.64)	3.32 (.66)
Negative Consequence Belief	2.56 (.94)	1-5	2.81 (1.00)a	2.33 (.78)b	2.55 (.94)ab	2.56 (.96)ab	.02	2.53	2.85†
Dominant	3.36 (1.52)	1-6	3.61 (1.75)	3.09 (1.61)	3.50 (1.34)	3.23 (1.55)
Submissive	2.19 (1.10)	1-5	2.53 (1.24)	1.95 (.96)	2.16 (1.15)	2.18 (1.03)
Supportive Responses	5.82 (.68)	3.15-7	5.60 (.79)	5.69 (.71)	5.97 (.61)	5.85 (.66)
Dominant	5.60 (.81)	2.50-7	5.20 (.91)a	5.34 (.77)a	5.82 (.70)b	5.66 (.81)b	14.59**	.01	1.49
Submissive	5.86 (.70)	3.25-7	5.70 (.82)a	5.74 (.75)a	5.99 (.62)b	5.87 (.67)b	5.93*	.43	.20
Nonsupportive Responses	2.22 (.52)	1.27-3.91	2.39 (.71)	2.11 (.45)	2.18 (.45)	2.22 (.51)
Dominant	2.20 (.81)	1-4.83	2.54 (.88)a	2.31 (.71)a	2.06 (.80)b	2.14 (.81)b	5.27*	.69	2.08
Submissive	2.27 (.53)	1.33-3.98	2.40 (.71)a	2.12 (.49)b	2.26 (.47)ab	2.29 (.52)ab	.02	2.04	3.40†
Emotion Teaching	.03 (3.42)	−8.36-8.22	.01 (3.61)	−1.11 (3.03)	.26 (3.66)	.36 (3.26)	1.16	.49	.76

Note. Means and standard deviations as a function of ethnicity and gender are adjusted for family income-to-needs ratio. Means in the same row that do not share subscripts differ significantly at p < .05. Bolded F values are those reported in the text.

^†p < .10.

^*p < .05.

^**p < .01.

TABLE 2.

Correlations Among Study Variables

	2	3	4	5	6	7	8	9	10	11	12	13
1. Acceptable Display Beliefs	.78**	.98**	−.15*	−.09	−.21**	.35**	.31**	.34**	−.13	−.19**	−.08	.21**
2. Dominant		.65**	−.07	−.08	−.07	.25**	.23**	.23**	−.07	−.15*	−.03	.18*
3. Submissive			−.17*	−.09	−.24**	.36**	.31**	.35**	−.13	−.18*	−.09	.20**
4. Negative Consequence Belief				.67**	.80**	−.25**	−.20**	−.25**	.33**	.11	.35**	−.00
5. Dominant					.35**	−.12	−.11	−.12	.18*	.10	.17*	−.07
6. Submissive						−.21**	−.13	−.22**	.28**	.10	.29**	−.01
7. Supportive Responses							.82**	.99**	−.30**	−.17*	−.30**	.05
8. Dominant								.73**	−.27**	−.21**	−.25**	.05
9. Submissive									−.29**	−.15*	−.29**	.04
10. Nonsupportive Responses										.66**	.97**	−.01
11. Dominant											.46**	−.09
12. Submissive												.01
13. Emotion Teaching

^*p < .05.

^**p < .01.

To assess whether ethnicity and income or ethnicity and education interacted to predict emotion beliefs or emotion socialization, two hierarchical regression analyses were conducted with the relevant main effects entered in the first block and the interaction term entered in the second block. The Ethnicity by Income and Ethnicity by Education interactions were nonsignificant. Thus, family income-to-needs ratio was simply included as a covariate in all analyses due to the moderate correlation between ethnicity and income, r = -.24, p < .01.

Ethnicity and education were not significantly correlated

Ethnic and Gender Differences in Emotion Beliefs

The first question addresses whether African American mothers, particularly of boys, held more restrictive beliefs about the display of negative emotions and perceived the display of negative emotions to carry more social consequences than European American mothers and African American mothers of girls. We used analyses of covariance (ANCOVA), controlling for income-to-needs, to test the hypotheses. Follow-up analyses investigated whether results were consistent across dominant and submissive emotions by testing the within-subject effect of an Emotion Type by Ethnicity interaction. When the interaction was nonsignificant, results were determined to be consistent across emotion type for all mothers and were reported with the total negative emotion scales. When the Emotion Type by Ethnicity interaction was significant, it was determined that the pattern of results were different across ethnicities for submissive and dominant negative emotions, and results are reported separately by emotion type.

A significant ethnicity main effect was found for beliefs about the display of negative emotions. African American mothers indicated it was less appropriate for their children to display negative emotions as compared to European American mothers (see Table 1). There was not a significant Ethnicity by Child gender interaction for display beliefs nor was there a significant within-subjects Emotion Type by Ethnicity interaction; thus, results were consistent across dominant and submissive emotions, and the total score was used in subsequent analyses.

As seen in Table 1, a marginally significant (p = .09) interaction between ethnicity and child gender controlling for income was found predicting negative social consequences beliefs. A nonsignificant within-subjects interaction between Emotion type and Ethnicity revealed that the pattern of ethnicity results was similar across submissive and dominant negative emotions. The trend-level Ethnicity by Child gender interaction predicting the negative consequences belief total score was probed given evidence that interaction effects are difficult to detect in nonexperimental research (McClelland & Judd, 1993). Consistent with our hypothesis, follow-up tests examining all possible comparisons indicated that African American mothers of boys perceived more negative social consequences for displaying negative emotions than African American mothers of girls, F(1, 62) = 4.65, p < .05.

Ethnicity and Gender Differences in Emotion Socialization Practices

Next, we examined whether African American mothers reported fewer supportive and more nonsupportive responses to their children’s negative emotions and did less emotion teaching than European American mothers, and whether these differences were more pronounced among African American mothers of boys. Results are presented separately for dominant and submissive negative emotions in cases where a within-subjects interaction between Emotion type and Ethnicity is significant.

Contrary to the emotion beliefs findings, the pattern of results for submissive and dominant negative emotions was found to differ among African American and European American mothers as indicated by a significant within-subjects interaction between Emotion type and Ethnicity for both supportive, F(1, 194) = 6.06, p < .05, and nonsupportive responses, F(1, 194) = 6.98, p < .01. Results of ANCOVAs showed a main effect for ethnicity on mothers’ reported supportive responses to both their children’s dominant and submissive negative emotions controlling for income. African American mothers reported fewer supportive responses to their children’s submissive negative emotions and reported many fewer supportive responses to their children’s dominant negative emotions than European American mothers (see Table 1). For nonsupportive responses, a main effect for ethnicity on mothers’ reported nonsupportive responses to children’s dominant negative emotions emerged controlling for income. African American mothers reported more nonsupportive responses to their children’s dominant negative emotions than European American mothers. A trend-level interaction between Ethnicity and Child gender was found for mothers’ reports of their nonsupportive responses to their children’s submissive negative emotions. After testing all possible comparisons, it was determined that African American mothers of boys reported marginally more nonsupportive responses to children’s submissive negative emotions than African American mothers of girls, F(1, 62) = 3.55, p = .07. There were no ethnicity or Ethnicity by Child gender differences in observed emotion teaching.

Beliefs as a Mediator of Ethnicity and Gender Differences in Emotion Socialization

We also examined whether ethnicity and ethnicity by child gender differences in emotion socialization practices were accounted for by maternal beliefs about emotions. Because ethnic differences were found for display beliefs, supportive responses to dominant and submissive negative emotions, and nonsupportive responses to dominant negative emotions, we used the SPSS multiple mediation macro provided by Preacher and Hayes (2004) testing display beliefs as the sole mediator in predicting supportive responses to both types of negative emotions and in predicting nonsupportive responses to dominant negative emotions. Because Ethnicity by Gender differences were found for social consequences beliefs and nonsupportive responses to submissive negative emotions, a mediated moderation model (see Morgan-Lopez & MacKinnon, 2006) was tested using the macro provided by Preacher, Rucker, and Hayes (2007). Social consequences beliefs were tested as the sole mediator of the Ethnicity by Gender interaction effect on nonsupportive responses to children’s submissive negative emotions.

Mediation analyses

The mediation macro addresses three important mediation criteria identified by Baron and Kenny (1986): a significant relation between the independent variable and the mediator, a significant relation between the independent variable and the dependent variable, and a nonsignificant relation between the independent variable and the dependent variable after including the mediator variable. Although these steps suggest some evidence of mediation, the strength of the effect should be tested. Thus, the mediation macro was used to test the significance of the indirect effect using bootstrapping procedures. Bootstrapping, which does not require distributional assumptions, is recommended in place of the Sobel test when sample sizes are moderate (Preacher & Hayes, 2004). The bootstrapping procedure repeatedly estimates the indirect effect by sampling with replacement from the dataset; in the current paper we utilized 1000 bootstrap resamples to construct a 95% bias-corrected confidence interval for the indirect effect (Preacher & Hayes, 2008). The multiple mediation macro was used because of the ability to include covariates. Confidence intervals that do not include zero suggest that the indirect effect is significant (Preacher & Hayes, 2008).

For mediation predicting supportive responses to dominant negative emotions, tests of the three mediation criteria showed that, controlling for income, ethnicity was significantly associated with display beliefs, β = -.28, b = -.40, t(197) = -4.10, p < .01, explaining a significant proportion of variance, R² = .11, F(1, 197) = 11.53, p < .01. Ethnicity was also significantly associated with supportive responses to dominant negative emotions, β = -.27, b = -.47, t(197) = - 3.83, p < .01, explaining a significant proportion of variance, R² = .07, F(1, 197) = 7.62, p < .01. After controlling for display beliefs, the relation between ethnicity and supportive responses to dominant negative emotions decreased but remained significant suggesting partial mediation, β = -.20, b = -.34, t(196) = -2.75, p < .01. Overall, 28% of the total effect (β = -.47) of ethnicity on supportive responses to dominant negative emotions was accounted for by the indirect effect (β = -.13) via display beliefs. Bootstrapping was conducted to formally test the significance of the partial mediation effect. The indirect effect of ethnicity on supportive responses to dominant negative emotions through display beliefs was significant; the bias corrected confidence interval did not contain zero, ranging from -.29 to -.02. Thus, the finding that African American mothers report less supportive responses to their children’s anger than European American mothers is partially accounted for by their belief that their children should not display negative emotions.

For mediation predicting supportive responses to submissive negative emotions, the three mediation criteria revealed that, controlling for income, ethnicity was significantly associated with display beliefs, β = -.28, b = -.40, t(197) = -4.10, p < .01, explaining a significant proportion of variance, R² = .11, F(1, 197) = 11.53, p < .01. Additionally, ethnicity was significantly associated with supportive responses to submissive negative emotions, β = -.17, b = -.26, t(197) = -2.42, p < .05, explaining a significant proportion of variance, R² = .04, F(1, 197) = 4.01, p < .05. After controlling for display beliefs, the relation between ethnicity and supportive responses to submissive negative emotions was nonsignificant, β = -.08, b = -.11, t(196) = -1.07, p = .29, suggesting that display beliefs mediated the link between ethnicity and supportive responses to submissive negative emotions. Overall, 58% of the total effect (β = -.26) of ethnicity on supportive responses to submissive negative emotions was accounted for by the indirect effect (β = -.15) via display beliefs. Bootstrapping was conducted to test the significance of the mediation effect. The indirect effect of ethnicity on supportive responses to submissive negative emotions through display beliefs was significant; the bias corrected confidence interval did not contain zero, ranging from -.30 to -.04. The finding that African American mothers report responding less supportively to their children’s fear and sadness than European American mothers is accounted for by their belief that their children should not display their negative emotions.

For mediation predicting nonsupportive responses to dominant negative emotions, mediation criteria revealed that, controlling for income, ethnicity was significantly associated with display beliefs, β = -.28, b = -.40, t(197) = -4.10, p < .01, explaining a significant proportion of variance, R² = .11, F(1, 197) = 11.53, p < .01. Ethnicity was also significantly associated with nonsupportive responses to children’s dominant negative emotions, β = .17, b = .29, t(197) = 2.33, p < .05, explaining a significant proportion of variance, R² = .04, F(1, 197) = 4.13, p < .05. After controlling for display beliefs, the relation between ethnicity and nonsupportive responses to dominant negative emotions was nonsignificant, β = .13, b = .22, t(196) = 1.72, p = .09, suggesting that display beliefs mediated the link between ethnicity and nonsupportive responses to dominant negative emotions. Overall, 24% of the total effect (β = .29) of ethnicity on nonsupportive responses was accounted for by the indirect effect (β = .07) via display beliefs. Bootstrapping was conducted to test the significance of the mediation effect. The bias corrected confidence interval contained zero, ranging from -.01 to .20. Thus, the indirect effect was not statistically significant. Although the most compelling evidence of mediation exists when the causal steps approach (Baron & Kenny, 1986) and the bootstrapping approach agree, some evidence of mediation exists but should be interpreted with caution due to weakness in the indirect effect. In other words, the finding that African American mothers report using more nonsupportive responses to their children’s dominant negative emotions than European American mothers is somewhat accounted for by African American mothers’ belief that their children should not display negative emotions.

Mediated moderation analyses

We used the mediated moderation macro to test if the path between the independent variable, ethnicity, and the mediator, mothers’ beliefs of social consequences, was moderated by child gender in predicting nonsupportive responses to children’s submissive negative emotions. For results to support mediated moderation, two significant effects must be present: the interaction effect must be a significant predictor of the mediator variable and the mediator must be a significant predictor of the dependent variable. The macro also tests the significance of the indirect effect at various values of the moderator using bootstrapping procedures with 1000 resamples. In the current study, the moderator, child gender, was dichotomous; therefore, a 95% bias-corrected confidence interval was provided for values of 0 (female) and 1 (male). A confidence interval that does not contain zero indicates a significant indirect effect (Preacher & Hayes, 2008). Family income, ethnicity, and child gender were controlled for in the following analyses.

Results showed that the interaction between ethnicity and child gender was a marginally significant predictor of social consequences beliefs, β = .39, b = .50, t(195) = 1.71, p = .09, although the proportion of variance explained was nonsignficant, R² = .02, F(1, 195) = 1.11, p = .35. The Ethnicity by Child gender interaction was a marginally significant predictor of nonsupportive responses to children’s submissive negative emotions, β = .42, b = .30, t(195) = 1.84, p = .07, although the proportion of variance explained was again nonsignificant, R² = .02, F(1, 195) = 1.10, p = .36. Finally, the relation between the Ethnicity by Child gender interaction and nonsupportive responses to submissive negative emotions controlling for social consequences beliefs (the proposed mediator) was nonsignificant, β = .29, b = .21, t(194) = 1.34, p = .18, providing some evidence of mediation. Forty-three percent of the total effect (β = .21) of ethnicity by child gender on nonsupportive responses was accounted for by the indirect effect via social consequences beliefs (β = .09).

Bootstrapping was conducted to test the significance of the indirect effect at specific values of the moderator. Among both girls and boys, the bias corrected confidence interval contained zero, ranging from -.13 to .02 for girls and -.02 to .16 for boys. Thus, the indirect effect was nonsignificant at both values of the moderator suggesting that the mediated moderation effect was not statistically significant. Some evidence of mediated moderation exists but should be interpreted with caution due to weakness in the indirect effect and the small amount of variance explained by the total effect. The finding that African American mothers report using more nonsupportive responses to their boy’s submissive negative emotions than African American mothers of girls is somewhat accounted for by their belief that their boys will encounter more negative social consequences if they display their negative emotions.

DISCUSSION

The current study examined African American and European American mothers’ beliefs about the display and consequences of negative emotions and their emotion socialization practices. A mediation model was proposed to explore one reason behind ethnic differences in emotion socialization practices, namely mothers’ beliefs about how appropriate it is for their preschool-aged children to express negative emotions.

African American mothers were less likely to believe that it was appropriate for children to display negative emotions in public or private settings than European American mothers. This finding is consistent with previous research that European Americans are more accepting of negative emotional displays than African Americans (Matsumoto, 1993). Minority mothers may worry that their children will be judged harshly by the majority ethnic group if they display negative emotions, especially in the context of oppression and discrimination (Consedine & Magai, 2002; Dodge et al., 2005). In contrast, ethnic group differences did not emerge for beliefs about the negative consequences of displaying negative emotions. Instead, a marginally significant interaction between ethnicity and child gender was found, with African American mothers of boys perceiving more negative social consequences for the display of negative emotions than African American mothers of girls. In an environment where African American males’ emotions are more likely than European American males’ to be scrutinized (Gregory, 1997; Hugenberg & Bodenhausen, 2003; Kang & Chasteen, 2009), mothers may perceive boys’ anger, sadness, and fear as signs of vulnerability that threaten their children’s success. While this interaction was hypothesized and is consistent with previous research, it is noteworthy that it only occurred with the belief that emphasizes penalties in the social context. It is possible that this orientation toward consequences from others in society for displaying distress emphasizes the need for African American boys to “fit in” and “be strong” within the majority culture, a task that does not seem to be as difficult or consequential for African American girls.

Next, we investigated the emotion socialization practices of African American and European American mothers. African American mothers reported fewer supportive responses to their children’s negative emotions, such as problem solving and encouraging the expression of emotion. This was the case for both types of negative emotions, but was particularly evident for dominant emotions. African American mothers were much less likely to support their children’s displays of anger than European American mothers, possibly due to the cultural emphasis on children’s obedience to parental authority in African American families. Furthermore, African American mothers reported more nonsupportive responses to their children’s anger and more nonsupportive responses to boys’ displays of sadness and fear than girls’. Nonsupportive responses refer to punitive, minimizing, and stressful reactions when children are distressed. African American mothers of boys may feel they need to take more drastic measures than mothers of girls to ensure that their children “get the message” when it comes to showing vulnerability through the display of sadness and fear. That African American mothers report using behaviors that emphasize the firm control of negativity is consistent with prior research on the importance of behavioral control in African American families and may occur for the same reason (Kelley et al., 1992). That is, African American mothers may socialize emotional control to keep their children, particularly boys, safe given discrimination from the majority ethnicity.

Despite ethnic differences in parents’ reports of their emotion socialization practices, no differences were found in mothers’ observed emotion teaching practices. These findings, coupled with prior evidence that African American and European American mothers do not differ in the extent to which they value emotions (Stelter & Halberstadt, in press) suggests that African American mothers may have a highly nuanced approach to emotion socialization. That is, they appear to value negative emotions and teach their children about recognizing such emotions and their causes and consequences in others, while simultaneously teaching their children that it may be better not to display their own negative emotions, particularly fear and sadness. This explanation parallels previous findings where African American mothers paid more attention to emotions during a mother-child storybook task than non-African Americans in what the authors describe as a display of “emotion vigilance” in a potentially hostile climate of prejudice (Garrett-Peters et al., 2008). In the context of ethnic discrimination, this pattern may be highly adaptive if it teaches young children to recognize when others are upset and to be cautious about the expression of their own upset. Further, the understanding of emotions has been proposed as an important precursor to the regulation of emotions (Kopp, 1989); thus, messages that provide children with emotional information and information about controlling negative displays may co-occur and complement one another.

Beyond testing ethnic differences, we examined a specific mediation pathway through which ethnic differences in emotion socialization practices were explained by differences in mothers’ beliefs about the display of negative emotions. The mediation pathway proved to be an effective way to understand some ethnic differences. Differences between African American and European American mothers’ supportive responses to their children’s submissive negative emotions were accounted for by their display beliefs, and differences in supportive responses to children’s dominant negative emotions were partially accounted for by mothers’ display beliefs. In other words, our results suggest that African American mothers provide fewer socialization practices that encourage children’s negative expression because they believe that the display of negative emotion is not acceptable for their children. We also found that ethnic differences in mothers’ nonsupportive responses to their children’s anger were somewhat accounted for by their display beliefs, although this indirect effect was not particularly strong. This finding, coupled with the partial mediation finding for supportive responses, suggests that there is more to understanding African American and European American mothers’ responses to the display of anger than we have considered in the current study. One possibility is that mothers’ responses to dominant negative emotions reflect a focus on obedience among African American families (Kelley et al., 1992). This prediction is supported by the fact that the three scenarios used to assess mothers’ responses to child anger seem to reflect challenges to parental authority. Perhaps a greater emphasis on parents’ beliefs about what their children’s anger expression reflects (e.g., goal blockage vs. disrespect or disobedience) or the measurement of parental response to additional types of anger scenarios may shed further light on the reason behind African American mothers’ control of anger.

To explain differences in nonsupportive responses to submissive negative emotions, we tested a mediated moderation pathway to account for the role of child gender. Specifically, we proposed that African American mothers of boys reported more nonsupportive responses to their children’s sadness and fear because they perceived that their children would experience more negative social consequences for distress displays. Traditional statistical approaches provided support for this pathway, but this was not shown to be a strong effect. Thus, conclusions regarding this finding must be made with caution. Beliefs about negative social consequences may provide a small indication of why African American mothers of boys provide more firm emotional control in response to their children possibly appearing “weak”, but other explanations may account for more variance. Vulnerability or weakness among African American males may have a broader symbolic meaning to parents, such as implications for the family as a whole, than what is captured in our three items describing negative consequences among peers.

Mothers’ beliefs about the appropriateness and consequences of displaying negative emotions were linked with mothers’ emotion socialization practices, both reported and observed, for the whole sample. Almost two-thirds of the possible correlations were significant, all in the expected direction, lending support to the view that parents’ beliefs about emotions guide emotion-relevant parenting behavior (Eisenberg et al., 1998). The continued examination of these beliefs, in addition to the extent to which parents value negative emotions, may be fruitful.

Despite the unique study aims and diverse sample, the current study has limitations. The first is that African American families were sampled from one geographic location in the Southeastern United States. It is possible that the historical context of ethnic relations in the South may create a stronger parental focus on children’s preparation for discrimination than in other parts of the United States. Second, our examination of submissive versus dominant negative emotions was limited by our measures and was therefore exploratory. Very few items were available that tapped parents’ beliefs and responses to children’s anger specifically. Future research should better anticipate possible differences between these two types of negative emotions and plan for such analyses by including a balanced number of items in their measures. Third, data on beliefs and socialization practices were only available from mothers. Fathers distinctively contribute to parenting and likely have a unique perspective on the role of emotion. Although some research has explored the joint influence of mothers’ and fathers’ emotion socialization efforts (e.g., McElwain, Halberstadt, & Volling, 2007), future research should also consider the effect one parent’s beliefs may have on the other’s beliefs and practices. Also, shared method variance may have contributed to significant findings between maternal report measures. Finally, future efforts to understand the role of emotion beliefs with qualitative methods could provide more information about additional emotion-related beliefs of importance.

IMPLICATIONS FOR PRACTICE, APPLICATION, AND POLICY

The current study looks at how ethnic differences in emotion socialization practices are explained by parents’ broader beliefs about children’s expression of negative emotions. These beliefs are likely adaptive in many ways for ethnic minority children, and it is advantageous for professionals working with families and children to understand their potential utility. The control of African American children’s anger and boys’ vulnerability may protect them from discrimination in a culture where the majority will likely be overly critical of their negative emotional displays. European American children do not face the same contextual constraints and are likely encouraged to express themselves through all types of negative emotions. A next logical step for this program of research is to consider the possibility that similar emotion socialization practices are interpreted differently and exert a different effect on social and emotional child outcomes for African American children than European American children because of ethnicity-specific beliefs about emotional expression and control as has been demonstrated for behavioral control (Bean, Barber, & Crane, 2006). Future research on this topic can help to illuminate the implications of variability in parents’ emotion socialization practices on children’s well-being taking into account the importance of the cultural context.