Considering the increasing researches of ethnic socialization, and the close communication among different ethnic minority groups, our current study priority was to clarify how ethnic socialization affect ethnic minority adolescents’ identity and their self-esteem. Researches on ethnic-racial socialization nearly emerged over 47 years ago and predominantly focused on American ethnic minorities (Wilkinson, 1974; Bowman and Howard, 1985; Peters, 1985; Rodriguez et al., 2009). More recently, this field has expanded to other minority groups, such as American adopted children (Mohanty and Newhill, 2011; Mohanty, 2013; Rosnati and Ferrari, 2014; Hu et al., 2017), as well as European Roma ethnic minority youth (Dimitrova et al., 2018). According to Hughes and Johnson (2001), ethnic socialization is considered as how parents shape children’s learning about their own race and their relationships with other ethnic groups. Parents are the crucial agent of ethnic socialization, besides peers (Rivas-Drake et al., 2009), neighborhood and community (Smith et al., 2003) are also play important role. Communication with children about their ethnicity is the main and salient part of ethnic family education. Conceptually, ethnic socialization messages have often been categorized into three categories: cultural socialization, preparation for bias, and promotion of distrust (Hughes et al., 2006; Juang et al., 2017). The latter two categories are aimed at raising awareness and preparation for negative things that minority ethnic adolescents may encounter in the future (Hughes and Chen, 1997), while the first category message is to promote positive messages concerning the ethnic group’s culture (Boykin and Toms, 1985; Hughes et al., 2006; Atkin and Yoo, 2019; Ulerio and Mena, 2020).

The past studies mainly conducted among the adolescents who aged from 12 to 18 years old (Hughes and Johnson, 2001; Hughes, 2003; McHale et al., 2006). According to Erikson (1968), ego identity is the most important topic during this period, adolescents continually explore their social role and seek the meaning of themselves in this stage of the life cycle. Ego identity is established after exploring multiple options and making a commitment to the choice made (Marcia, 1966, 1967).

Recent studies of ethnic socialization have mainly focused on cultural socialization messages, and have revealed that such messages are positively correlated with minority adolescents’s performance (Bynum et al., 2007; Seol et al., 2016), such as self-esteem (Lee et al., 2018) and ethnic identity (Juang and Syed, 2010). According to Gartner et al. (2014), cultural socialization messages are correlated with higher ethnic identity and subsequently promote self-esteem among Asian-American adolescents. Regarding the association between preparation for bias messages and ethnic identity, there is a lack of consensus in the empirical results. According to Rivas-Drake (2011), Latin college students who perceived preparation for bias messages showed negative public regard, which was the main dimension of ethnic identity. Gartner et al. (2014) found that the association between preparation for bias and ethnic identity was not significant among Asian-American adolescents. These correlations have not been sufficiently verified in existing studies. Perceived promotion of distrust messages in the process of ethnic socialization is aimed at raising children’s vigilance when they get along with people from other ethnic backgrounds. The promotion of distrust is negatively correlated with ethnic identity (Gartner et al., 2014). In addition, Yin et al. (2016) investigated 10,390 Chinese minority students and found that their ethnic socialization messages showed unique cultural features that could be mainly divided into four categories: cultural socialization, promotion of distrust, preparation for bias, and promotion of harmony. Notably, the promotion of harmony dimension was aimed at promoting a positive relationship with other groups, and it was positively correlated with ethnic identity (Yin et al., 2016).

Ethnic identity is defined as self-identification as a group member, a sense of belonging, and an attitude toward the group (Phinney, 1992, 2003). In addition, self-identity is an important component of self-concept, especially among adolescents (Erikson, 1968). Furthermore, ethnic identity is one aspect of social identity for minority adolescents (Tajfel, 1981) and also a common phenomenon across groups (Phinney, 1992). Over the past three decades, ethnic identity has been the most commonly researched outcome of parents’ ethnic socialization (Brittian et al., 2013). Based on the Theory of identity development (Phinney and Ong, 2007), ethnic identity is developed from the process of commitment and exploration. Commitment mainly refers to the strong attachment and personal investment toward a group. Exploration means the search for ethnic information and activities, such as reading and talking about things related to the group, understanding ethnic culture, participating in cultural activities. These process correspond with the information of ethnic socialization, especially for the culture socialization. Hence, it is reasonable to speculate that ethnic socialization may have a effect on ethnic identity. Additionally, previous research also widely provide empirical support for the correlations between ethnic socialization and identity (Brega and Coleman, 1999; Umaña-Taylor and Fine, 2004; Brittian et al., 2013; Gartner et al., 2014; Reynolds and Gonzales-Backen, 2017; Dimitrova et al., 2018). Besides, according to Rodriguez et al.’s (2009) model, ethnic identity is also a predictor of minority adolescents’ psychological and educational outcomes.

Self-esteem, a one-dimensional construct, is defined as a general sense of one’s worth (Rosenberg, 1965). It refers to how people value themselves, especially their feelings about their own social status, and how much coordination exists between their ideal and real selves (O’Donnell et al., 2008; Cieślak and Golusiñski, 2018). Ethnic identity, which is a sense of belonging and identification to a group. Social identity theory (Tajfel and Turner, 1986) pointed out that individuals always have an potential desire for social identity which is developed from the processes of comparison, categorization, and identification. Ethnic minorities will compare the inner-group with the other groups and then conclude the difference and similarities. Categorization will make them feel that they are similar to others. Hereby, a sense of identity will be generated, which results in the preference of the inner group. Self-esteem is affected to a certain extent by individual group identity or social identity which means belonging and pride. Existing studies also noted that ethnic identity could be a predictor of self-esteem (Smith et al., 1999; Umaña-Taylor and Fine, 2004; Umaña-Taylor and Updegraff, 2007; Mohanty, 2013; Gonzales-Backen et al., 2015).

Overall, Previous studies have reported that age, gender, and parents’ education could be predictors of ethnic socialization (Bowman and Howard, 1985; González et al., 2006; Hughes et al., 2006; Priest et al., 2014). Parents may tend to engage in some certain type of messages according to child’s age (Hughes and Chen, 1997), more variation in age is necessary to observe an effect (Brown et al., 2007). Meanwhile, where findings of ethnic socialization varied, gender was a considerable factor (Bowman and Howard, 1985; Caughy et al., 2002; Brown et al., 2007; Chavous et al., 2008; Cunningham et al., 2012). Additionally, parents is the primary ethnic socialization agent, whose education level affected their parental style (White-Johnson et al., 2010). These variables may moderate the relationships between ethnic socialization, ethnic identity, and self-esteem. Further, Brown et al. (2009) examined the role of gender and provided supporting evidence for the moderating effect on the association between ethnic socialization and academic grades. Similarly, a significant gender difference in path from cultural socialization to identity was showed in the research from Gartner et al. (2014).

Previous studies mainly adopted the whole perspective when considering the influence of ethnic socialization and viewed it as one latent variable (Supple et al., 2006; Lee et al., 2018; McDermott et al., 2018). However, considering multidimensional ethnic socialization, different ethnic messages may have different correlations with ethnic identity. Studies have proposed the three categories of messages should be considered as three observed variables rather than one whole latent variable as “ethnic socialization” (Liu and Lau, 2013; Gartner et al., 2014). Gartner et al. (2014) proposed a method that considers the influence of ethnic socialization deeply and separately. Most studies on ethnic socialization have predominantly been conducted in the United States (94%) (Umaña-Taylor and Updegraff, 2007; Priest et al., 2014; Ulerio and Mena, 2020). Therefore, it is necessary to examine ethnic socialization theory in the context of Eastern countries (Atkin and Yoo, 2019). Additional studies on ethnic socialization from diverse groups would contribute to the understanding of the process (Hughes et al., 2006). The Mulao minority group is a traditional ethnic group in China that retains their unique culture and customs (Wu, 2016). Focusing on this group’s ethnic socialization process could provide more empirical evidence to the associations between ethnic socialization and adolescents’ mental performance from different cultural backgrounds.

The first aim of this study was to examine the relationships between ethnic socialization, ethnic identity, and self-esteem among Chinese Mulao minority adolescents, which could reveal the influence of ethnic socialization.

Second, we examined the links between ethnic socialization and self-esteem via ethnic identity to ascertain its stability in terms of different genders, grades, and parents’ education. Based on the theoretical and empirical evidence, we hypothesized the following: (1) Positive ethnic socialization messages (cultural socialization, promotion of harmony) would be positively correlated with ethnic identity. (2) Negative ethnic socialization messages (preparation for bias and promotion of distrust) would be negatively associated with ethnic identity. (3) Ethnic identity would be positively correlated with self-esteem and mediate the relationships between ethnic socialization and self-esteem. (4) Gender, grade, and parental education would moderate the relationship between perceived ethnic socialization, ethnic identity, and self-esteem.

The participants were Mulao ethnic minority adolescents residing in the Luocheng Mulao Autonomous County. The original sample consisted of 530 students. A total of 61 students were eliminated due to more than 50% of total items not being answered, or more than 90% of answers were the same, or answered regularly. Finally, 469 valid samples remained, with a response rate of 88%.

The mean age of the sample students was 14.43 years (range = 11–17 years; SD = 1.07), consisting of 220 (44.4%) boys and 249 (55.6%) girls. In addition, 131 (28.0%) students were freshman students, 132 (28.1%) were junior students, and 206 (43.9%) were senior students. The description of sample characteristics is presented in Table 1.

Participants were recruited from three junior high schools in different townships of Luocheng Mulao Autonomous County, which is their major residential area. The school authorities were contacted to obtain necessary consent for this study and permission was also secured from our affiliated university’s institutional review board. Participants were given 20 minutes to complete the survey during class time under the guidance of research team members, and questionnaire was collected on the spot. Only private information about age, sex, grade, and parents’ education was marked, and students were free to answer or withhold consent.

The Statistical Package for Social Science (SPSS) version 22.0 was used to compute descriptive statistics, correlations, and analysis of variance (ANOVA). Structural equation modeling (SEM) was used to evaluate the hypotheses, using maximum likelihood with Analysis of Moments Structures (Amos 22.0). Model evaluation based on criteria such as CMIN/DF (χ² divided by degree of freedom), where smaller values are preferable, Goodness-of-Fit Index (GFI) ≧0.9, which shows the relative amount of variance and covariance, comparative fit index (CFI) ≥ 0.9, and root mean square error of approximation (RMSEA) ≤ 0.8 has been recognized as one of the most informative criteria in covariance structure modeling.

The bivariate correlations, means, and standard deviations of the primary study variables are presented in Table 2. The results indicated that cultural socialization messages were positively correlated with other variables (r = between 0.20 and 0.48, p < 0.005), and the promotion of harmony messages were positively correlated with other variables (r = between 0.24 and 0.49, p < 0.005), except for the promotion of distrust. Notably, preparation for bias was positively correlated with ethnic identity (r = 0.21, p < 0.05) and negatively correlated with self-esteem (r = −0.074, p < 0.005). Messages entailing distrust were only significantly correlated with lower self-esteem (r = −0.15, p < 0.005).

One-way ANOVA revealed that Mulao students’ perceived promotion of distrust information varied significantly by gender; boys reported more distrust information than girls (t (467) = 3.72, p < 0.0001). One-way ANOVA results indicated significant effects of grades on the promotion of harmony information and ethnic identity. Senior students reported more harmony information than freshman and junior students [F(2, 466) = 6.27, p < 0.0001]. Senior students’ ethnic identity was higher than that of freshman students [F(2, 466) = 3.48, p < 0.05]. Meanwhile, one-way ANOVA results revealed a significant effect of parental education on ethnic identity; that was, students whose parents were educated to senior high school or above reported stronger ethnic identity than those whose parents were educated to primary school or junior high school level [F(2, 466) = 4.29, p < 0.05].

To estimate the relationships between ethnic socialization, ethnic identity, and self-esteem, confirmatory factor analysis (CFA) was conducted using AMOS 22.0, and full information maximum likelihood methods (FIML) were utilized. Because the correlation between the promotion of distrust and ethnic identity was not significant, the pathway in the model was not established. The pathway from preparation for bias to ethnic identity was deleted because the standard regression weights were not significant (r = 0.01). Standard solutions for SEM are presented in Figure 1. Indices of GFI were CMIN = 17.59, DF = 6, CMIN/DF = 2.93, GFI = 0.99, CFI = 0.98, and RMSEA = 0.06.

To examine the mediation effect, we used a bias-corrected bootstrapping confidence interval (95% CI) to identify significant indirect pathways. (1) The indirect effect of promotion of harmony on self-esteem through ethnic identity was significant (CI = 0.45 to 0.93), suggesting that promotion of harmony messages is associated with higher ethnic identity and subsequently related with higher self-esteem among Mulao minority students. (2) The indirect effect of cultural socialization on self-esteem through ethnic identity was significant (CI = 0.08 to 0.54), indicating that ethnic identity fully mediates the pathway from cultural socialization to self-esteem.

Multigroup analysis based on AMOS 22.0 was used to evaluate whether the associations between variables were stable across gender, grades, and parental education. Comparing the Chi-square of the constrained model with structural residuals constrained to equality across groups to a baseline model with the same coefficients freely across groups, significant variance of Chi-square (Δχ²) indicates variation across groups (Byrne, 2009; Peng and Lu, 2020).

Results indicated that the variance of Chi-square (Δχ²) between the constrained and baseline models across gender was 25.43 (p = 0.09 > 0.05), and 48.61 (p = 0.07 > 0.05) across parental education, suggesting that the associations between variables were stable across boys and girls, and students with different parental education. However, the variance of Chi-square (Δχ²) between the constrained and baseline models across grades was 66.31 (p = 0.00 < 0.0001), indicating a significant variance in the associations across different grades. Figure 2 shows the model for three grades separately.

Further analysis used critical ratios for difference between parameters; an absolute value larger than 1.96 emphasized a significant difference between the pathway coefficients (Byrne, 2009; Peng and Lu, 2020). The results revealed that the pathway from cultural socialization to ethnic identity was significant only for senior students (B = 0.32, P < 0.0005). The pathway from promotion of distrust to self-esteem was significant for freshman students (B = -0.27, P < 0.0005).