Adolescents’ Emotion Regulation Strategies,Self-Concept,and Internalizing Problems

Abstract

This study examined the relationships among adolescents’ emotion regulation strategies (suppression and cognitive reappraisal), self-concept, and internalizing problems using structural equation modeling. The sample consisted of 438 early adolescents (13 to 15 years old) in Taiwan, including 215 boys and 223 girls. For both boys and girls, suppression was negatively associated, and cognitive reappraisal was positively associated with self-concept. Self-concept negatively predicted adolescents’ internalizing problems. The findings support the hypothesis that self-concept mediates the relationship between emotion regulation and internalizing problems. These findings support generalization of one part of Cicchetti’s and Toth’s model of the development of internalizing problems to an Asian culture.

Keywords

emotion regulation emotional suppression cognitive reappraisal self-concept internalizing problems depression

Internalizing problems are behaviors that are focused inward, such as withdrawal from relationships, anxiety, depression, or somatic problems (Achenbach & Edelbrock, 1978). High levels of internalizing problems may lead to the development of depression, which is the most common psychiatric disorder in the United States (Kessler, Berglund, Demler, Jin, & Walters, 2005). Rates of depression increase beginning in adolescence, and depression begun in adolescence may continue across the lifespan (Hankin et al., 1998). The present study in Taiwan focuses on early adolescence, when internalizing problems increase but before major depression typically develops (Hankin et al., 1998).

Cicchetti and Toth’s Model of the Development of Depression

One of the most influential theoretical models of the development of depression was proposed by Cicchetti and Toth (1998). This complex model is based on attachment theory (Bowlby, 1982) and ecological theories of development (Bronfenbrenner, 1979). Cicchetti and Toth propose that across infancy and childhood, biological, socioemotional, cognitive, and representational systems interact. By adolescence, some children have developed coherent organization across these developmental domains, whereas some children have developed what Cicchetti and Toth call “depressotypic organization,” which is incoherent organization of these important areas of functioning. An important factor in the development of depressotypic organization in this model is the interaction between parent-child relationships and infant and child biological systems. Cicchetti and Toth emphasize that as the biological systems of regulation develop, parent-child relationships affect and are affected by the development of emotion regulation. Emotion regulation is a set of complex skills which are necessary for effective adaptation and successful social interactions in everyday life (Macklem, 2008). There is considerable discussion and disagreement about how to define, model, and assess emotion regulation (see the 2004 March/April volume of Child Development for a series of articles focusing on these issues). In this first exploratory study in Asia, we tested a model focusing on one part of the emotion regulation system. We focused on emotion regulation strategies, and we used one of the dominant models of emotion regulation strategies by Gross (2002). Gross, in his process model of emotion regulation, describes two common emotion regulation strategies used by adolescents and adults: cognitive reappraisal and suppression. Cognitive reappraisal occurs early in the process of emotion regulation. After we attend to a situation that may arouse an emotion, we process the situation cognitively and attach meaning to aspects of the situation. According to Gross, adolescents and adults are able to cognitively reappraise or change how they think about a situation to regulate the emotional impact. After personal meanings are attached to the situation, then physiological, experiential, and behavioral responses begin. We can use suppression or inhibit the expression of outward signs of these responses. Sometimes suppression is an appropriate strategy; for example, when the expression of anger is socially not acceptable. However, overuse of this strategy is related to poorer adjustment (Bonanno, Papa, Lalande, Westphal, & Coifman, 2004).

In Cicchetti and Toth’s model, the child’s attachment relationships to important adults, such as the mother and father, affect the development of emotion regulation. Later in development, they propose that the attachment relationship and the emotion regulation system affect the development of cognitive representational models of self. If the parent is not sensitive and responsive to the child and the child forms an insecure attachment to the parent, then the child’s systems of emotion regulation become dysregulated and a negative representational model of self develops.The present study focused on just one important part of the representational model, self-concept. Although there are multiple aspects of self-concept (Marsh, 1994), the present study focused on the two aspects that may be most relevant in Taiwan: general self-concept and school self-concept, the latter because education is highly valued in Chinese culture. Early adolescence is an important time to explore the role of self-concept because self-concept changes as adolescents become increasingly self-conscious and aware of the way they are perceived by others (Vartanian, 2000). According to Cicchetti and Toth, emotion dysregulation and negative representational models of self affect the development of internalizing problems and increase the risk of the development of depression.

Developmental Contexts That May Affect the Model

Cicchetti and other developmental theorists emphasize that developmental processes are affected by contexts outside the child’s immediate family (Cicchetti & Toth, 1998; Cummings, Davies, & Campbell, 2000). Two important contexts that may affect the relationships between emotion regulation, self-concept, and internalizing problems are gender and culture. Gender and culture may affect how adolescents regulate their emotions and cultures do not always share the same beliefs about emotion expression (Matsumoto, 1990). For example, young men in Taiwan believe it is more important to suppress emotions when they are upset than European-American men (Chia, Moore, Lam, Chuang, & Cheng, 1994). Gender roles for expressing emotions also can differ across cultures. Brody (1997) found that although American young adult males express less positive and negative emotions than American young adult females, there is no difference in the expression of emotion for Asian American or Asian male and female young adults, except that Asian men express more shame and less annoyance than Asian women. American men use suppression more than women (Flynn, Hollenstein, & Mackey, 2010; Gross & John, 2003), whereas there are no differences for cognitive reappraisal (Gross & John, 2003). There also are gender differences in self-concept in the United States. General self-concept is consistently higher for American adolescent boys than girls (Kling, Hyde, Showers, & Buswell, 1999; Marsh, Byrne, & Shavelson, 1988; Marsh, Ellis, Parada, Richards, & Heubeck, 2005; Marsh, Parada, & Ayotte, 2004). The results for school self-concept are mixed. Studies have found that American girls have higher school self-concept than boys (Marsh, 1989; Marsh et al., 1988), lower school self-concept (Marsh et al., 2004), or found no difference (Skaalvik & Rankin, 1990). There are also gender differences for internalizing problems. Adolescent girls report more depression than boys in developed countries in both the West and the East (Allgood-Merten, Lewinsohn, & Hops, 1990; Cicchetti & Toth, 1998; Culbertson, 1997; Leadbeater, Kuperminc, Blatt, & Hertzog, 1999; Lewinsohn, Gotlib, & Seeley, 1997).

The present study explored whether the relationships among emotion regulation strategies, self-concept, and internalizing problems are the same for early adolescent boys and girls in Taiwan. It is important to explore whether models developed in Western cultures generalize to other cultures (Schneider, 1998). Taiwan has a culture that is very different from the United States, where most of the research on emotion regulation, self-concept, and internalizing problems has taken place. The majority of early adolescents in Taiwan are of Chinese heritage. Traditional Chinese culture emphasizes distinct gender roles (Chia, Allred, & Jerzak, 1997). However, Western values have affected Chinese culture so that today, gender roles for male and female adolescents are less distinct than for the older generation (Chia et al., 1997; Shek, 2006).

Evidence for the Model in Western and Eastern Samples

The theoretical model described by Cicchetti and Toth is a mediated model so that emotion regulation affects the development of self-concept which then affects the development of internalizing problems. There is good empirical evidence for a relationship between emotion regulation and internalizing problems in children. Children who have dysregulated emotion are more likely to experience internalizing problems (see Zeman, Cassano, Perry-Parrish, & Stegall, 2006 for a review). Children with internalizing problems do not express anger when appropriate, inappropriately express anger or sadness (Zeman et al., 2006), or generate less effective emotion regulation strategies (Garber, Baafladt, & Zeman, 1991).

There is also evidence in Western samples that the two emotion regulation strategies included in the present study’s model, cognitive reappraisal and emotion suppression, predict internalizing problems. Experiments in the United States with adult samples have found that cognitive reappraisal reduces negative physiological, experiential, and behavioral responses whereas suppression reduces behavioral responses such as emotion expression but increases negative physiological arousal and emotions (Campbell-Sills, Barlow, Brown, & Hofmann, 2006; Gross, 1998). Possibly, because of these different effects of the two emotion regulation strategies, cognitive reappraisal is associated with fewer internalizing problems whereas suppression is associated with more internalizing problems in Dutch and American adolescents and adults (Garnefski & Kraaij, 2006; Gross & John, 2003; Moore, Zoellner, & Mollenholt, 2008; Troy, Wilhelm, Shallcross, & Mauss, 2010). Only one study has found different associations among cognitive reappraisal, suppression, and internalizing problems for men and women, where suppression predicted internalizing in American men but not in women (Flynn et al., 2010). There have been no studies, that we are aware of, that have focused on the relationships between cognitive reappraisal, suppression, and internalizing problems in Eastern cultures.

In this Asian sample, we predict that cognitive reappraisal will be negatively associated with internalizing problems, as has been found in Western samples. However, we predict, in contrast to the results found for Western samples, that suppression will not be positively associated with internalizing problems. We propose that suppression may not represent emotion dysregulation in Taiwan because suppression may have different meanings and effects in this culture. For example, Kramer, Kwong, Lee, and Chung (2002) reported that children and adolescents in Asian families are usually taught to be polite, quiet, shy, humble, and deferential. Emotional outbursts are discouraged and failure to meet these expectations may bring shame and loss of face to the children (Kramer et al., 2002). In particular, expression of anger is not valued in Asian cultures and is less likely to be expressed in Asian cultures than in Western cultures (Cole, Bruschi, & Tamang, 2002). Therefore, cultural expectations may encourage Asian adolescents to suppress their negative emotions in social situations. If suppression of emotions is encouraged and valued in Taiwan, then suppression may not represent emotion dysregulation.

Self-concept as a mediator

However, relatively few studies have examined processes, such as self-concept, that may mediate between emotion regulation and internalizing. Based on Cicchetti and Toth’s model, we hypothesize that the relationship between emotion regulation and internalizing problems is mediated by self-concept. There is only one study, that we are aware of, focusing on the relationships among cognitive reappraisal, suppression, and self-concept. Cognitive reappraisal was positively associated, and suppression was negatively associated with general self-concept in American adults (Gross & John, 2003). In the present study, cognitive reappraisal is expected to be positively associated with both general and school self-concept. However, because suppression is valued in Taiwan, suppression may be positively associated with self-concept. Because previous research found no differences in the expression of emotion for Asian men and women (Brody, 1997), we expected the relationships between emotion regulation and self-concept to be the same for boys and girls in our sample.

There is more evidence for the other link in the mediated model, the association between self-concept and internalizing problems. Adolescents’ general and school self-concepts are associated negatively to internalizing problems such as depression in American and Swedish samples (Allgood-Merten et al., 1990; Egan & Perry, 1998; Lewinsohn et al., 1997; Marsh et al., 2004; Ybrandt, 2008). However, there may be different relationships between self-concept and internalizing problems for boys and girls. Studies of adolescent boys and girls in different cultures have found mixed results. General self-concept was associated with American boys’ reports of internalizing problems in 6th grade but not in 7th grade whereas general self-concept was not associated with girls’ internalizing problems in either grade (Leadbeater et al., 1999). In contrast, Ybrandt (2008), found a stronger negative relation between positive self-concept and internalizing problems for Swedish girls than boys. For both male and female Japanese adolescents, both positive general and school self-concept predicted fewer internalizing problems (Nishikawa, Sundbom, & Hagglof, 2010). In the present study, we expect the same result as the Japanese sample, that is, both aspects of self-concept will be negatively associated with internalizing problems for both boys and girls.

We are not aware of any studies that have tested our entire hypothesized mediated model using structural equation modeling in either Western or Eastern samples. Based on Cicchetti and Toth’s theoretical model and previous research, we hypothesized that the association between emotion regulation and internalizing problems will disappear when self-concept is included as a mediator in the model and that the model will be the same for boys and girls.

Method

Participants and Procedures

The study was approved by the human subjects review board of the authors’ university in the United States. In June and September of 2009, junior high school teachers from Hualien and Taitung counties in Taiwan were invited to participate in the study. After teachers gave permission for data collection in their classrooms, students received an introductory letter and consent forms for the study. Students whose parents gave consent completed a 15-minute questionnaire on their emotion regulation strategies, self-concept, and internalizing problems during a free-study period. Teachers completed a questionnaire of their perceptions of the student’s internalizing problems if they had known the student for at least a year. Of the 819 students who were asked to be in the study, 487 participated. Of the 487, 8 students were dropped because no teacher questionnaire was completed, and 4 students were dropped because their questionnaires were not finished.

The final sample of the study consisted of 438 adolescents, including 215 boys and 223 girls. There was no missing data for these adolescents. The sample consisted of 19 seventh graders, 233 eighth graders, and 186 ninth graders, ranging in age from 13 to 15 years. Most of the adolescents were of Han Chinese ethnic origin, with 39.5% identifying as Taiwanese aborigines. Chinese was the first language of 83.4% of the students and the second language for 16.6 %. Most of the adolescents (48%) lived with both parents; 16.2% lived with their parents and their grandparents, 12.6% lived with only one parent, less than 1% split time between parents, 14.8% lived with other relatives, and 8.2% lived with other people. Most fathers and mothers had a high school diploma (44.2% and 47%) or did not finish high school (36.6% and 36.6%). 11.8% of fathers and 8.9% of mothers had a bachelor’s degree, 1.2% and 1.9% had a master’s, less than 1% and 1.9% had a doctoral degree, and 5.7% and 5.6% did not answer.

Measures

Because two of the questionnaires used in the study, the Emotion Regulation Questionnaire (Gross & John, 2003) and the Social Skills Improvement System (Gresham & Elliott, 2008), have no Chinese versions, the original English measures were translated into Chinese by a bilingual Chinese-English speaker who lives in Taiwan. A Chinese scholar in the English department of a university in Taiwan and an American who has worked as a professional translator more than 10 years in Taiwan assessed the Chinese translation. The questionnaires were then back translated into English by a bilingual Chinese-English speaker who lives in the United States. The original English and the back-translated English versions were then compared. The meanings of the items in the two versions were very similar, providing evidence for the quality of the translation.

Adolescents’ emotion regulation strategies

Adolescents’ use of emotion regulation strategies was assessed using the Emotion Regulation Questionnaire (Gross & John, 2003). The questionnaire measures two emotion regulation strategies, cognitive reappraisal and suppression. The 6-item cognitive reappraisal subscale measures the extent to which an individual modifies his or her emotional response by changing the way he or she views a particular situation, for example, ‘‘When I’m faced with a stressful situation, I make myself think about it in a way that helps me stay calm.” The 4-item suppression subscale assesses the extent to which an individual inhibits his or her external expression of internal emotional feelings, for example, “When I am feeling negative emotions, I make sure not to express them.” Each item is rated using “1” (strongly disagree) to “7” (strongly agree). In a series of five studies (Gross & John, 2003), the two factor structure of the questionnaire was supported, the measure had acceptable internal consistency, (Cronbach’s α = .73 and .79 for suppression and cognitive reappraisal, respectively), adequate test-retest reliability across a 3-month period (.69 for both subscales), and acceptable convergent and discriminant validity. In another sample, Magar, Phillips, and Hosie (2008) reported that α = .72 for suppression and .81 for cognitive reappraisal.

Adolescents’ self-concept

Two subscales from the Chinese version of the Self-Description Questionnaire II (Marsh, Kong, & Hau, 2001), were used to assess adolescents’ self-concept. The Chinese version of the questionnaire measures 12 dimensions of self-concept for adolescents aged 12 to 18 years. Yeung and Lee (1999) used 4 of the 12 subscales (verbal, math, academic, and general self-concept) in a sample of 487 early adolescents (Grades 7, 8, and 9) in Southern China. Confirmatory factor analysis supported the four-factor structure in the Chinese sample. Given time limitations, only the general and school self-concept subscales were used in the present study. Items were rated from 1 (false—not like me) to 6 (true—like me). The general self-concept subscale has 10 items measuring adolescents’ perceptions of themselves as effective and capable individuals, for example, “Overall, most things I do turn out well.” The school self-concept subscale has 11 items measuring adolescents’ perceptions of themselves in academic work, for example, “I am good at most school subjects.” Alphas for the general and school self-concept subscales using samples from the United States were .84 and .89 (Marsh et al., 2004) and .86 and .87 for a Chinese sample (Yeung & Lee, 1999). The 4-week test-retest reliability was .70 and .79 for general and school self-concepts, respectively (Gilman, Laughlin, & Huebner, 1999). In the present Taiwanese sample, α = .78 and .85 for general and school self-concept. Overall scores for the two subscales were calculated by averaging the items.

Student and teacher report of internalizing problems

Adolescents’ internalizing problems were assessed using the internalizing subscale of the Social Skills Improvement System (Gresham & Elliott, 2008). Students answered 10 items using ratings from 1 (not true) to 4 (very true), for example, “I feel sad.” Factor analyses by Gresham and Elliot (2008) provided evidence for the construct validity of the internalizing factor. The subscale had acceptable internal consistency, α = .88, and test-retest reliability, r = .65 (Gresham & Elliott, 2008). Alpha was .80 in the present sample. Teachers rated 7 items on the Teacher version of the questionnaire using ratings ranging from 1 (never) to 4 (almost always). The teacher subscale has good reliability; the developers of the measure report that α = .90, test-retest reliability = .81, and the inter-rater reliability coefficient = .50 (Gresham & Elliott, 2008). In the present sample, α = .88. Internalizing scores were calculated by averaging the ratings of the items.

Results

Descriptive statistics for each of the variables are reported for boys and girls in Table 1. There were no mean differences between males and females for cognitive reappraisal, suppression, general and school self-concept, and internalizing problems (teacher and adolescent report), MANOVA, F(6,431) = .62, p = .72. The focus of the study was to explore the relationships among the two emotion regulation strategies, self-concept, and internalizing problems. Structural equation modeling (AMOS, Version 7 for Windows) was used to test the models, using maximum likelihood estimation. A two-step approach was used (Anderson & Gerbing, 1988; Blunch, 2008). First, the measurement model was tested and then the structural model (including both the measurement and path models) was tested.

Table 1.

Means, Standard Deviations, and Bivariate Correlations for Girls and Boys

																	Boys
Girls\Boys	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	M	SD	Minimum	Maximum
1. Suppression	—	.16*	.02	−.01	.03	−.06	.13	.11	.08	.13	.08	.21*	.83*	.72*	.74*	.38*	3.93	1.39	1.00	7.00
2. Reappraisal	.07	—	.31*	.27*	−.12	−.12	.72*	.79*	.76*	.73*	.72*	.72*	.24*	.19*	−.05	.13	4.90	1.29	1.00	7.00
3. General self-concept	−.18*	.15*	—	.72*	−.18*	−.20*	.13	.24*	.16*	.22*	.22*	.42*	−.02	.16*	−.09	.05	3.93	0.86	1.00	6.00
4. School self-concept	−.12	.13	.72*	—	−.22*	−.23*	.13	.21*	.18*	.18*	.22*	.31*	.00	.08	−.10	.03	3.70	0.94	1.00	6.00
5. Internal (student)	.16*	−.04	−.43*	−.32*	—	.08	−.12	−.09	.00	−.06	−.18*	−.07	−.07	.07	.06	.04	1.82	0.53	1.00	3.80
6. Internal (teacher)	.11	−.13*	−.16*	−.20*	.17*	—	−.10	−.15*	.01	−.07	−.03	−.22*	−.08	.02	−.07	−.08	1.78	0.58	1.00	3.57
7. ERQ Q1	.15*	.64*	.07	.02	.05	−.09	—	.53*	.45*	.39*	.38*	.43*	.19*	.12	−.01	.07	4.92	1.78	1.00	7.00
8. ERQ Q2	.07	.71*	.12	.14*	.03	−.07	.32*	—	.63*	.40*	.50*	.50*	.19*	.07	.00	.07	4.85	1.73	1.00	7.00
9. ERQ Q3	.05	.74*	.16*	.15*	−.02	−.07	.45*	.60*	—	.48*	.40*	.43*	.12	.18*	−.10	.10	4.97	1.72	1.00	7.00
10. ERQ Q4	.05	.73*	.10	.08	.00	−.13	.45*	.30*	.43*	—	.51*	.43*	.19*	.21*	−.10	.07	5.08	1.78	1.00	7.00
11. ERQ Q5	−.01	.69*	−.03	.00	−.02	−.01	.27*	.45*	.36*	.47*	—	.41*	.16*	.03	.00	.00	4.62	1.71	1.00	7.00
12. ERQ Q6	.03	.59*	.21*	.15*	−.17*	−.20*	.17*	.33*	.29*	.35*	.30*	—	.25*	.23*	.01	.25*	5.03	1.68	1.00	7.00
13. ERQ Q7	.86*	.07	−.13*	−.06	.13	.13	.12	.04	.03	.06	.01	−.01	—	.47*	.44*	.25*	3.92	1.98	1.00	7.00
14. ERQ Q8	.82*	.15*	−.06	−.03	.09	−.01	.18*	.13*	.08	.09	.01	.19*	.57*	—	.23*	.38*	4.40	1.84	1.00	7.00
15. ERQ Q9	.82*	−.03	−.25*	−.20*	.19*	.15*	.06	−.01	.02	−.03	−.05*	−.09	.57*	.48*	—	.25*	3.47	2.04	1.00	7.00
16. ERQ Q10	.06	.02	−.04	−.06	.09	.05	.06	.00	.04	.05	−.06	.00	.05	.07	.03	—	3.95	2.00	1.00	7.00
Girls M	3.75	4.93	4.00	3.82	1.81	1.79	4.81	4.86	4.94	5.26	4.73	5.04	3.98	4.26	3.20	3.55
SD	1.31	1.14	0.80	0.88	0.56	0.50	1.69	1.58	1.60	1.55	1.67	1.63	1.91	1.88	1.88	2.01
Minimum	1.00	1.33	1.00	1.82	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Maximum	7.00	7.00	6.00	5.91	3.60	3.57	7.00	7.00	7.00	7.00	7.00	7.00	7.00	7.00	7.00	7.00

Note: The correlations for the girls (n = 223) are below the diagonal and the boys (n = 215) are above the diagonal. ERQ = Emotion Regulation Questionnaire.

p < .05.

The Measurement Model for Emotion Regulation

Because the Emotion Regulation Questionnaire had not been used before in an Asian sample, the first step of testing the measurement model was to test the measurement model for emotion regulation. Multigroup confirmatory factor analysis was used to test measurement invariance of the Emotion Regulation Questionnaire for two groups, boys and girls (see Table 2). First, the 10-item model was run with all parameters allowed to vary across gender. However, item 10, “When I am feeling positive emotions, I am careful not to express them.” did not load well on the hypothesized suppression factor for girls (see Table 3). The factor loading of .07 was well below the recommended level of .40 (Di Iorio, 2005). As the culture of Taiwan emphasizes establishing harmonious relationships with others, which is especially true for girls, and consequently expressing positive emotions, we hypothesized that the item did not fit the values of the culture. Consequently, we dropped the item and tested Model 2, a 9-item model with all parameters freed to vary across gender. The second model confirmed the two factor structure and had acceptable fit (see Table 2). Third, the model was tested with all factor loadings constrained to be equal across gender. The fit of Models 2 and 3 then was compared using chi-square difference tests (see Table 2), and no difference was found, Δχ²(7) = 10.99, p = .14, and the Comparative Fit Index (CFI) value did not decrease by more than .01 (Cheung & Rensvold, 2002). Because there was not a significant difference between the models, Model 3 was accepted given that the constrained model was the more parsimonious model. Fourth, the model was tested with all factor loadings and intercepts constrained to be equal across gender. The fit of Models 3 and 4 was compared, and no difference was found, Δχ²(9) = 4.35, p = .89. Moreover, the CFI value increased from .924 to .931, and the Root Mean Square Error of Approximation (RMSEA) value decreased from .06 to .05. Because Model 4 was the more parsimonious model than Model 3, Model 4 was accepted. Fifth, the model was tested with all factor loadings, intercepts, factor covariances, and factor variances constrained to be equal across gender. There was no difference between Models 4 and 5, Δχ²(3) = 4.90, p = .18, and the CFI value did not decrease by more than .01. Thus, following Cheung and Rensvold’s (2002) guidelines, the results of the model comparisons suggested that Model 5 was the preferable model. This final model fit the data well, χ²(71) = 120.60, p < .05, CFI = .93, RMSEA = .05; n = 438, leading to the conclusion that the model does apply across gender and does display measurement invariance. All of the factor loadings for the final model were above .50 (see Table 3). The latent correlation between the two subscales was low, .15 for both boys and girls. All item-to-total correlations were above .35, and internal consistency was adequate, .81 for cognitive reappraisal and .72 for suppression.

Table 2.

The Measurement Model for the Emotion Regulation Questionnaire

								Difference tests compared with the previous model
Tested models	χ ²	df	p	χ²/df	CFI	RMSEA	AIC	Δχ²	df	p
Model 1 (10 items)	189.04	69	.00	2.74	.893	.06	271.04
Model 2 (unconstrained, 9 items)	100.37	52	.00	1.93	.930	.06	212.37
Model 3	111.36	59	.00	1.89	.924	.06	209.36	10.99	7	.14
Model 4	115.70	68	.00	1.70	.931	.05	195.70	4.35	9	.89
Model 5 (the chosen final model)	120.60	71	.00	1.70	.928	.05	194.60	4.90	3	.18

Table 3.

Factor Loadings for Boys and Girls for the Emotion Regulation Questionnaire

	Factor loadings when Q10 was included		Factor loadings after Q10 was dropped
Item	Boys	Girls	Boys	Girls
Cognitive reappraisal
I. I control my emotions by changing the way I think about the situation I’m in.	.56	.56	.65	.70
2. When I want to feel less negative emotion, I change the way I’m thinking about the situation.	.75	.75	.73	.72
3. When I want to feel more positive emotion, I change the way I’m thinking about the situation.	.74	.74	.73	.73
4. When I want to feel more positive emotion (such as joy or amusement), I change what I’m thinking about.	.62	.62	.61	.71
5. When I want to feel less negative emotion (such as sadness or anger), I change what I’m thinking about.	.61	.61	.65	.59
6. When I’m faced with a stressful situation, I make myself think about it in a way that helps me stay calm.	.55	.55	.58	.55
Expressive suppression
7. I control my emotions by not expressing them.	.77	.77	.85	.95
8. When I feel negative emotions, I make sure not to express them.	.68	.63	.60	.63
9. I keep my emotions to myself.	.58	.67	.57	.63
10. When I feel positive emotions, I am careful not to express them.	.57	.07

Testing the Structural Model

Because measurement invariance was obtained, factor loading invariance for emotion regulation was assumed when testing the rest of the model. Multiple-group analysis for two groups, boys and girls, was used to test the hypothesized structural model and examine gender differences. The structural model included both direct and indirect paths, with direct paths from cognitive reappraisal and suppression to self-concept and to internalizing problems, and from self-concept to internalizing problems (see Figure 1). First, the hypothesized model was run with all parameters allowed to vary by gender, except the factor loadings for the Emotion Regulation Questionnaire. Second, the model was rerun with all factor loadings set to be equal across gender, and the fit of the two models was then compared using chi-square difference tests. Because the fit of Models 1 and 2 was significantly different (see Table 4), Δχ²(2) = 6.43, p < .05, the hypotheses of equivalent factor loadings was rejected. The source of variance was identified by testing a series of nested models which allowed one parameter to vary across gender at a time and then comparing the fit of these models with the fit of the first model. Hence, the third model tested constrained the factor loadings for general self-concept and emotion regulation to be equal but freed other parameters to vary across gender. The fit of Models 1 and 3 then was compared (see Table 4), and no significant difference was found, Δχ²(1) = 2.11, p = .15, suggesting that Model 3 fits the data as well as Model 1 and that the factor loadings for self-concept were invariant across gender. The same result was found when the factor loadings for school self-concept and emotion regulation were constrained to be equal and the other parameters were freed to vary across gender. The fourth model constrained the factor loadings for student-reported internalizing problems and emotion regulation to be equal but freed other parameters to vary across gender. The fit of Models 1 and 4 was then compared, and the models differed significantly, Δχ²(1) = 4.88, p < .05, so the null hypothesis of no gender difference in the factor loadings for internalizing problems was rejected.

Figure 1.

The complete structural equation model for suppression, cognitive reappraisal, self-concept, and internalizing problems, χ²(147) = 308.36, p < .05, χ²/df = 2.10, CFI = .90, RMSEA = .05, AIC = 430.36

Table 4.

Tested Models

								Difference tests compared with Model 1
Models	χ ²	df	p	χ²/df	CFI	RMSEA	AIC	Δχ²	df	p
Model 1	283.54	125	.00	2.27	.90	.05	449.54
Model 2	289.97	127	.00	2.28	.90	.05	451.97	6.43	2	.04
Model 3	285.65	126	.00	2.27	.90	.05	449.65	2.11	1	.15
Model 4	288.42	126	.00	2.29	.90	.05	452.42	4.88	1	.03
								Difference test between Models 3 and 5
Model 3	285.65	126	.00	2.27	.90	.05	449.65
Model 5 (the chosen final model)	308.36	147	.00	2.10	.90	.05	430.36	22.71	21	.36

As Model 3 fits the data as well as Model 1 and Model 3 was the more parsimonious model, Model 3 was chosen as the preferable model. In Model 3, all parameters were freed to vary by gender, except the factor loadings for self-concept and emotion regulation. The fifth model constrained the factor loadings for self-concept and emotion regulation, the intercepts, structural path coefficients and the variance of the factors, but freed the covariances and variances of the errors and the factor loadings for internalizing problems to vary across gender. The fit of Models 3 and 5 was compared (see Table 4), and no significant difference was found, Δχ²(21) = 22.71, p = .36. Because Model 5 fits the data as well as Model 3 and because Model 5 was the most parsimonious model, model 5 was chosen as the most preferable model (see Table 5 and Figure 1). In this final model, pairwise parameter comparisons indicated that girls’ factor loadings for student-reported internalizing problems were significantly larger than boys’ (z = 2.39, p < .05) whereas there was no difference between the loadings for boys and girls for teacher-reported internalizing problems (z = -.76, p > .05).

Table 5.

Unstandardized Parameter Estimates and Bootstrap-Based Confidence Intervals for the Structural Equation Model

		Standardized parameters		Confidence intervals
Paths	Unstandardized parameters	Girls	Boys	Lower	Upper
Path estimates
Cognitive reappraisal to self-concept	0.22*	.31*	.31*	0.12	0.36
Suppression to self-concept	−0.09*	−.15*	−.15*	−0.17	0.00
Self-concept to internalizing problems	−0.16*	−.76*	−.76*	−0.23	−0.06
Cognitive reappraisal to internalizing problems	0.00	−.02	−.02	−0.05	0.03
Suppression to internalizing problems	0.01	.10	.10	−0.01	0.05
Factor loadings
Teacher report to internalizing problems	1.00	.31*	.27*	1.00	1.00
Student report to internalizing problems	Girls = 2.21, Boys = 0.97,^a	.60*	.28*	Girls = 1.25, Boys = 0.37	Girls = 6.82, Boys = 2.55
School self-concept to self-concept	1.00	.79*	.80*	1.00	1.00
General self-concept to self-concept	1.03*	.92*	.89*	0.85	1.34
Q1 to Cognitive reappraisal	1.00	.59*	.60*	1.00	1.00
Q2 to Cognitive reappraisal	1.20*	.74*	.75*	1.00	1.49
Q3 to Cognitive reappraisal	1.18*	.75*	.72*	1.01	1.43
Q4 to Cognitive reappraisal	1.00*	.65*	.60*	0.80	1.27
Q5 to Cognitive reappraisal	0.99*	.60*	.61*	0.77	1.30
Q6 to Cognitive reappraisal	0.92*	.55*	.59*	0.73	1.18
Q7 to Suppression	1.38*	.85*	.82*	1.09	1.82
Q8 to Suppression	0.99*	.64*	.61*	0.76	1.25
Q9 to Suppression	1.00	.64*	.56*	1.00	1.00

Indicates a significant (p < .05) gender difference in the estimate for that parameter.

p < .05.

Model 5 fit the data well, Δχ²(147) = 308.36, p < .05, χ²/df = 2.10, CFI = .90, RMSEA = .05, Akaike Information Criterion (AIC) = 430.36. The final model accounted for 61% of the variance in internalizing problems for boys and girls. For both boys and girls, neither cognitive reappraisal (B = –.004, SE = .02, β = –.02, p = .81) nor expressive suppression (B = .013, SE = .01, β = .10, p = .31) was related to internalizing problems. Cognitive reappraisal was positively related to self-concept, B = .22, SE = .05, β = .31, p < .01, and expressive suppression was negatively related to self-concept, B = –.09, SE = .04, β = –.15, p < .05. Self-concept was associated negatively with internalizing problems, B = –.16,SE = .04, β = –.76, p < .01.

Evaluating mediation effects

In the preliminary correlation analysis, suppression was positively associated with student report of internalizing problems, r = .11, p = .03, but not with teacher report, r = .01, p > .05. Cognitive reappraisal was negatively associated with teacher report of internalizing problems, r = –.13, p = .01, but not with student report, r = –.04, p > .05. However, in the structural equation model which included the possible mediator and both direct and indirect paths from emotion regulation to internalizing problems (Figure 1), there were no longer associations between cognitive reappraisal, suppression, and internalizing problems. The results from these analyses support the hypothesis that self-concept mediates the relationships between cognitive reappraisal, suppression, and internalizing problems.

Bootstrapping methodology, a newly developed technique based on resampling methods, also was used to evaluate mediation (Shrout & Bolger, 2002; Fiske, Gilbert, & Lindzey, 2009). With 500 bootstrap samples, parameters of interest were estimated from each of the pseudo-samples, and bias-corrected bootstrap confidence intervals for the indirect effects were calculated. The final model was run with bootstrapping, at a level of significance of .05. For both girls and boys, the unstandardized indirect effect, direct effect, and total effect of cognitive reappraisal on internalizing problems were –.035, –.004 and –.039, respectively. The indirect effect (mediated effect) indicates that when cognitive reappraisal increases by 1, internalizing problems decrease by .035. The 95% bias-corrected bootstrap confidence interval for the indirect effect of cognitive reappraisal on internalizing problems ranged from –.065 to –.011. Because the confidence interval did not include zero and the bootstrap p value was .003 (MacKinnon, Lockwood, & Williams, 2004), the null hypothesis of no indirect effect of cognitive reappraisal on internalizing problems was rejected, supporting the hypothesis that self-concept mediates the relationship between cognitive reappraisal and internalizing problems. Likewise, the unstandardized indirect effect, direct effect, and total effect of suppression on internalizing problems were .015, .013, and .028 for both girls and boys. Due to the indirect effect, when suppression increases by 1, internalizing problems increase by .015. This increase is in addition to any direct (unmediated) effect that suppression may have on internalizing problems. Because the confidence intervals did not include zero and the bootstrap p value was .03, the null hypothesis of no mediated effects of suppression on internalizing problems was rejected, providing support for the hypothesis that self-concept mediates the relationship between suppression and internalizing problems.

Testing alternative models using only one indicator for internalizing problems and self-concept.

Because it was unclear in this Asian sample whether teachers or students were more accurate reporters of adolescents’ internalizing problems, the final model was compared with a model using only adolescents’ self-report and a model using only teacher report of internalizing problems. The final model including both informants fit the data as well as the model using adolescent-report, Δχ²(21) = 30.21, p > .05, and the model using teacher report, Δχ²(21) = 29.69, p > .05.

Likewise, to answer whether the exclusion of school or general self-concept would result in a significant improvement in the model fit, the final model including both aspects of self-concept was compared to a model excluding school self-concept and a model excluding general self-concept. The model including both aspects of self-concept fit the data as well as the model only including general self-concept, Δχ²(22) = 6.82, p > .05, but there was a significant difference between the model including both aspects of self-concept and the model including only school self-concept, Δχ²(22) = 45.78, p < .05. However, the model including only school self-concept did not fit the data well, χ²(125) = 262.59, p < .05, χ²/df = 2.10, CFI = .88, RMSEA = .05, AIC = 372.59, so this model was rejected.

Summary

In summary, the hypothesized model was supported. The measurement model for the emotion regulation items was adequate in this Asian sample for both boys and girls and the full structural model testing both measurement and path models had a satisfactory fit for both boys and girls. Self-concept mediated the relationships between cognitive reappraisal and suppression to internalizing problems.

Discussion

Cicchetti and Toth’s (1998) theoretical model proposed that emotion regulation indirectly affects internalizing problems through the development of cognitive representational models of self. In this study, we found support for our hypothesis that one aspect of cognitive representational models of self, self-concept, mediates the relationship between emotion regulation strategies and internalizing problems. Consistent with previous empirical research, in the present study, there were significant bivariate correlations between cognitive reappraisal, expressive suppression, and internalizing problems. The hypothesized model predicted that these associations would be reduced when self-concept was included in the model, which would provide evidence that self-concept was mediating the relations between emotion regulation and internalizing problems. In the final structural equation model, there were no significant direct paths from cognitive reappraisal and suppression to internalizing problems. Instead, cognitive reappraisal positively and expressive suppression negatively predicted self-concept which then negatively predicted internalizing problems. Possibly because cognitive reappraisal reduces negative physiological, experiential, and behavioral responses (Campbell-Sills et al., 2006; Gross, 1998), adolescents who frequently use this strategy may develop a more positive self-concept. In contrast, adolescents who rely on suppression may reduce behavioral responses but still be affected by physiological and experiential responses which may in turn affect self-concept.

The study was a first step in exploring whether one part of Cicchetti and Toth’s model developed using Western samples is supported in Asian cultures. Although previous research using Western samples, summarized in the introduction, found a positive relationship between suppression and internalizing problems and a negative relationship between suppression and self-concept, we had hypothesized that neither of these relations would be found in this sample in Taiwan. We hypothesized that suppression may not represent emotion dysregulation in this culture. Both of these hypotheses were not supported. Instead, we found a positive bivariate correlation between suppression and internalizing problems which was reduced and no longer significant in the structural equation model testing for both direct and indirect effects of expressive suppression. We also found in both the bivariate correlations and the structural equation model, a negative association between suppression and self-concept. If suppression predicts negative outcomes in both Western and Eastern samples, possibly the greater acceptance of suppression in Asian cultures reduces the strength of this effect. The present study did not have an American sample so cultural comparisons of the strength of these associations are not possible. Future studies using both Western and Eastern samples could test whether the strength of these associations is the same in Western and Eastern samples.

The tested model was the same for boys and girls in Taiwan except for the measurement model for internalizing problems. These results support previous research which found no differences in the expression of emotion in Asian men and women (Brody, 1997), and the same associations between self-concept and internalizing problems for Japanese male and female adolescents (Nishikawa et al., 2010). The gender difference in the measurement model for internalizing problems may be because the girls and their teachers agreed to some extent (r = .17, p < .05) and boys and teachers did not (r = .08, p > .05) on their reports of whether the student showed internalizing problems (see Table 1).

Although the study was not longitudinal, it had several strengths. First, the structural equation model made it possible to test the fit of the data to the theoretical model while considering both the structural and the measurement models. Second, because internalizing problems were based on both student and teacher report, the chances that the relationship between student report of self-concept and internalizing problems was the result of shared-method variance were reduced. Third, the study used measures that have previous research evidence of acceptable reliability and validity. Confirmatory factor analysis supported the two factor structure of the Chinese version of the emotion regulation questionnaire after dropping an item that did not fit Chinese cultural values. A weakness of the study was that answering questions about emotion regulation strategies retrospectively does not adequately assess the actual emotion regulation strategies used in the context of emotionally charged situations (Silk, Steinberg, & Morris, 2003). A more valid means of assessing emotion regulation strategies is to use experience sampling (subjects record information on emotion regulation every couple of hours or so when a device signals them; Silk et al., 2003). However, experience sampling is costly in larger samples needed to test models of mediation. An advantage of using a questionnaire to assess emotion regulation strategies in the present study was that a large sample of adolescents could be collected.

The Chinese translation of the internalizing questionnaire also had acceptable internal consistency for both teachers’ and early adolescents’ self-reports of student internalizing problems in Taiwan. Considering that internalizing problems are not easily observable, it is not surprising that the correlation between girls’ and teachers’ reports of the students’ internalizing problems was small and that there was no relation between the reports of boys and teachers. Teachers may not be able to accurately report their students’ internalizing problems, especially their male students who may hide problems more than female students. And early adolescents may not be accurate reporters of their own internalizing problems because they may still be developing cognitive skills for reflecting on their own thinking. However, even young children can be accurate reporters of emotion regulation and depression (Bowie, 2010; Eiser & Morse, 2001). Because of strengths and weakness of both teacher and student report, using both teacher and student report of internalizing problems provided a better measurement model than using just teacher or student report.

Although a mediated model was supported, the design of the study did not make it possible to determine whether adolescents’ use of cognitive appraisal or expressive suppression affected self-concept or vice versa or whether self-concept affected internalizing problems or vice versa. However, a 6-year longitudinal study of American adolescents using a related construct, self-worth, found that adolescents’ self-worth predicted changes in their depressive symptoms whereas depressive symptoms did not predict changes in self-worth (Garber & Cole, 2010). An important next step would be longitudinal studies to examine relationships between changes in cognitive reappraisal and expressive suppression and changes in self-concept across childhood and adolescence. Another valuable study would be an experiment randomly assigning early adolescents to an intervention to encourage the use of cognitive reappraisal versus a control group. Comparing self-concept before and after the intervention would provide evidence of whether cognitive reappraisal positively affects the development of self-concept.

Because a considerable proportion of the students did not participate, the self-selection of participants may have reduced the sample variance for internalizing problems and reduced the likelihood of finding relations between emotion regulation, self-concept, and internalizing problems. However, the sample had an excellent range of parent education indicating that, despite refusals, the sample still had a good cross-section of families in Taiwan.

The study did not provide a complete test of Cicchetti and Toth’s model of the development of depression. Our model did not include important parts of their model such as biological factors and parent-child relationships. We also focused on just one aspect of the emotion regulation system, using just one model of emotion regulation strategies developed by Gross (2002). We also included just one aspect of the representational model of self, self-concept. Testing a more complete model including more aspects of the emotion regulation system and a more complete model of the representational model of self would provide a better test of Cicchetti & Toth’s theoretical model.

However, the current study is the first to model direct and indirect pathways among emotion regulation strategies, self-concept, and internalizing problems in early adolescents in either Eastern or Western cultures. The findings of the study are helpful for guiding future prevention research using experimental methods to explore methods of preventing the development of internalizing problems and later depression. For example, Donegan and Rust (1998) found a positive effect of an emotional education program focusing on identifying emotions, changing ways of thinking about emotions, and discussing behavioral consequences on students’ self-concept. Future experimental research could test the effectiveness of interventions designed to support the development of cognitive reappraisal strategies and to reduce the use of suppression in early adolescents. The present study suggests that possible outcomes of interest could be adolescents’ self-concept and internalizing problems.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Bios

Manying Hsieh received her PhD from Indiana University, Bloomington. Her research interests include child and adolescent development, emotional development, and problem behaviors. She is an assistant professor of the Department of Child Care and Family Studies at Asia-Pacific Institute of Creative Technology, Taiwan.

Anne Dopkins Stright is an associate professor of human development at Indiana University, Bloomington. Her research interests focus on the effects of family rocesses on children’s social and academic development.

References

Achenbach

Edelbrock

(1978). The classification of child psychopathology: A review and analysis of empirical efforts. Psychological Bulletin, 85, 1275-1301.

Allgood-Merten

Lewinsohn

Hops

(1990). Sex differences and adolescent depression. Journal of Abnormal Psychology, 99, 55-63.

Anderson

Gerbing

(1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103, 411-423.

Blunch

(2008). Introduction to structural equation modeling: Using SPSS and AMOS. Los Angeles, CA: Sage.

Bonanno

Papa

Lalande

Westphal

Coifman

(2004). The importance of being flexible: The ability to both enhance and suppress emotional expression predicts long-term adjustment. Psychological Science, 15, 482-487.

Bowie

(2010). Emotion regulation related to children’s future externalizing and internalizing behaviors. Journal of Child and Adolescent Psychiatric Nursing, 23, 74-83.

Bowlby

(1982). Attachment and loss: Attachment (Vol. 1). New York, NY: Basic Books.

Brody

(1997). Gender and emotion: Beyond stereotypes. Journal of Social Issues, 53, 369-394.

Bronfenbrenner

(1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press.

10.

Campbell-Sills

Barlow

Brown

Hofmann

(2006). Acceptability and suppression of negative emotion in anxiety and mood disorders. Emotion, 6, 587-595.

11.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233-255.

12.

Chia

Allred

Jerzak

(1997). Attitudes toward women in Taiwan and China: Current status, problems, and suggestions for future research. Psychology of Women Quarterly, 21, 137-150.

13.

Chia

Moore

Lam

Chuang

Cheng

(1994). Cultural differences in gender role attitudes between Chinese and American students. Sex Roles, 31, 23-30.

14.

Cicchetti

Toth

(1998). The development of depression in children and adolescents. American Psychologist, 53, 221-241.

15.

Cole

Bruschi

Tamang

(2002). Cultural differences in children’s emotion reactions to difficult situations. Child Development, 73, 983-996.

16.

Culbertson

(1997). Depression and gender: An international review. American Psychologist, 52, 25-31.

17.

Cummings

Davies

Campbell

(2000). Developmental psychopathology and family processes: Theory, research, and clinical implications. New York, NY: Guilford.

18.

Di Iorio

(2005). Measurement in health behavior: Methods for research and evaluation. San Francisco, CA: Jossey-Bass.

19.

Donegan

Rust

(1988). Rational emotive education for improving self-concept in second-grade students. Journal of Humanistic Education and Development, 36, 248-256.

20.

Eiser

Morse

(2001). Can parents rate their child’s health-related quality of life? Results of a systematic review. Quality of Life Research, 10, 347-457.

21.

Egan

Perry

(1998). Does low self-regard invite victimization? Developmental Psychology, 34, 299-309.

22.

Fiske

Gilbert

Lindzey

(2009). The handbook of social psychology. New York, NY: Wiley.

23.

Flynn

Hollenstein

Mackey

(2010). The effect of suppressing and not accepting emotions on depressive symptoms: Is suppression different for men and women? Personality and Individual Differences, 49, 582-586.

24.

Garber

Braafladt

Zeman

(1991). The regulation of sad affect: An information-processing perspective. In Garber

Dodge

K. A.

(Eds.), The development of emotion regulation and dysregulation (pp. 208-240). New York, NY: Cambridge University Press.

25.

Garber

Cole

(2010). Intergenerational transmission of depression: A launch and grow model of change across adolescence. Development and Psychopathology, 22, 819-830.

26.

Garnefski

Kraaij

(2006). Relationships between cognitive emotion regulation strategies and depressive symptoms: A comparative study of five specific samples. Personality and Individual Differences, 40, 1659-1669.

27.

Gilman

Laughlin

Huebner

(1999). Validation of the Self-Description Questionnaire II with an American sample. School Psychology International, 20, 300-307.

28.

Gresham

Elliott

(2008). Social Skills Improvement System–Rating Scales. Bloomington, IN: Pearson Assessments.

29.

Gross

(1998). Antecedent- and response-focused emotion regulation: Divergent consequences for experience, expression, and physiology. Journal of Personality and Social Psychology, 74, 224-237.

30.

Gross

(2002). Emotion regulation: Affective, cognitive, and social consequences. Psychophysiology, 39, 281-291.

31.

Gross

John

(2003). Individual differences in two emotion regulation processes: Implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85, 348-362.

32.

Hankin

Abramson

Moffitt

Silva

McGee

Angell

(1998). Development of depression from preadolescence to young adulthood: Emerging gender differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107, 128-140.

33.

Kessler

Berglund

Demler

Jin

Walters

(2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey replication. Archives of General Psychiatry, 62, 593-602.

34.

Kling

Hyde

Showers

Buswell

(1999). Gender differences in self-esteem: A meta-analysis. Psychological Bulletin, 125, 470-500.

35.

Kramer

Kwong

Lee

Chung

(2002). Cultural factors influencing the mental health of Asian Americans. Western Journal of Medicine, 176, 227-232.

36.

Leadbeater

Kuperminc

Blatt

Hertzog

(1999). A multivariate model of gender differences in adolescents’ internalizing and externalizing problems. Developmental Psychology, 35, 1268-1282.

37.

Lewinsohn

Gotlib

Seeley

(1997). Depression-related psychosocial variables: Are they specific to depression in adolescents? Journal of Abnormal Psychology, 106, 365-375.

38.

MacKinnon

Lockwood

Williams

(2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39, 9-128.

39.

Macklem

(2008). The importance of emotional regulation in child and adolescent functioning and school success. In Macklem

(Ed.), Practitioner’s guide to emotion regulation in school-aged children (pp. 1-12). New York, NY: Springer.

40.

Magar

Phillips

Hosie

(2008). Self-regulation and risk-taking. Personality and Individual Differences, 45, 153-159.

41.

Marsh

(1989). Age and sex effects in multiple dimensions of self-concept: Preadolescence to early adulthood. Journal of Educational Psychology, 81, 417-430.

42.

Marsh

(1994). Using the National Longitudinal Study of 1988 to evaluate theoretical models of self-concept: The Self-Description Questionnaire. Journal of Educational Psychology, 86, 439-456.

43.

Marsh

Byrne

Shavelson

(1988). A multifaceted academic self-concept: Its hierarchical structure and its relation to academic achievement. Journal of Educational Psychology, 80, 366-380.

44.

Marsh

Ellis

Parada

Richards

Heubeck

(2005). A short version of the Self-Description Questionnaire II: Operationalizing criteria for the short-form evaluation with new applications of confirmatory factor analyses. Psychological Assessment, 17, 81-102.

45.

Marsh

Kong

C.-K.

Hau

K.-T.

(2001). Extension of the internal/external frame of reference model of self-concept formation: Importance of native and nonnative languages for Chinese students. Journal of Educational Psychology, 93, 543-553.

46.

Marsh

Parada

Ayotte

(2004). A multidimensional perspective of relations between self-concept (Self-Concept Questionnaire II) and adolescent mental health (youth self-report). Psychological Assessment, 16(1), 27-41.

47.

Matsumoto

(1990). Cultural similarities and differences in display rules. Motivation and Emotion, 14, 195-214.

48.

Moore

Zoellner

Mollenholt

(2008). Are expressive suppression and cognitive reappraisal associated with stress-related symptoms? Behaviour Research and Therapy, 46, 993-1000.

49.

Nishikawa

Sundbom

Hagglof

(2010). Influence of perceived parental rearing on adolescent self-concept and internalizing and externalizing problems in Japan. Journal of Child and Family Studies, 19, 57-66.

50.

Schneider

(1998). Cross-cultural comparison as doorkeeper in research on the social and emotional adjustment of children and adolescents. Developmental Psychology, 34, 793-797.

51.

Shek

(2006). Chinese family research: Puzzles, progress, paradigms, and policy implications. Journal of Family Issues, 27, 275-284.

52.

Shrout

Bolger

(2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422-445.

53.

Silk

Steinberg

Morris

(2003). Adolescents’ emotion regulation in daily life: Links to depressive symptoms and problem behavior. Child Development, 74, 1869-1880.

54.

Skaalvik

Rankin

(1990). Math, verbal, and general academic self-concept: The internal/external frame of reference model and gender differences in self-concept structure. Journal of Educational Psychology, 82, 546-554.

55.

Troy

Wilhelm

Shallcross

Mauss

(2010). Seeing the silver lining: Cognitive reappraisal ability moderates the relationship between stress and depressive symptoms. Emotion, 10, 783-795.

56.

Vartanian

L. R.

(2000). Revisiting the imaginary audience and personal fable constructs of adolescent egocentrism: A conceptual review. Adolescence, 35, 639-661.

57.

Ybrandt

(2008). The relation between self-concept and social functioning in adolescence. Journal of Adolescence, 31, 1-16.

58.

Yeung

Lee

(1999). Self-concept of high school students in China. Educational and Psychological Measurement, 59, 431-450.

59.

Zeman

Cassano

Perry-Parrish

Stegall

(2006). Emotion regulation in children and adolescents. Journal of Developmental and Behavioral Pediatrics, 27, 155-168.