Ethnic Disparities and the Psychological Trauma of Maltreated Children: Evidence from Three Multi-ethnic Counties in China

This study aims to discuss the effect of ethnicity on child maltreatment trauma in China and to contribute to international knowledge on the quality of life of children. The data come from a survey of 1763 rural children (Mage = 12.34, 50.0% boys) in three multi-ethnic counties in western China that was conducted from November 2019 to January 2020 with the modified versions of the “Juvenile Victimization Questionnaire” Scale. There are three important findings of this study. First, child maltreatment is significantly related to child depression in China. The depression score and rate of severe depression symptoms (SDS) are 2.09 times and 3.82 times higher, respectively, for maltreated children than children without maltreatment. Second, the effects of maltreatment on child depression differ significantly among the ethnic groups. The negative effect of maltreatment is most influential among the Han population and least influential among the Zhuang population. Third, ethnic disparities are also found in the effects of the influencing factors on child depression. The effect of intergenerational relationships on child depression is significant only in the Han and Tibetan populations, while the negative effect of peer support is found only among Han, Tibetan, and Miao children. Based on the confirmation of ethnic disparities in trauma due to maltreatment, this study suggests that it is necessary to establish a high-quality psychological intervention system in China’s multi-ethnic counties.


Child Maltreatment Trauma and Its Influencing Factors
CM is a worldwide risk factor for psychological health, especially depression. Evidence from the United States shows that the likelihood of depression is 37.1% higher for neglected children than for children who are not neglected (Bennett et al., 2010). In Canada, women who were abused in childhood were four times more likely to suffer from depression than women who were not abused (Ouellet-Morin et al., 2015). Evidence from Australia also shows that 15.7% of depression in men and 22.8% of depression in women are related to childhood maltreatment (Moore et al., 2015). Research has also found that the effect of CM on psychological health varies among different subtypes. Although most studies have found that emotional abuse and neglect are the most influential categories (Crow et al., 2014;Vugt et al., 2013), other studies still suggest that different subtypes have equivalent psychiatric and behavioral effects, which range from anxiety and depression to rule-breaking and aggression (Vachon et al., 2015).
Children's psychological trauma is not only worsened by the frequency of maltreatment but also significantly mitigated by some protective factors. Among these protective factors, family relationships, including spousal and intergenerational relationships, are recognized as the most important predictors (Leung & Fung, 2021). Weak family cohesion and IPV between parents were found to be predictive of greater depression and PTSD symptoms and low life satisfaction in children (Fung, 2021;Kaur & Kearney, 2013;Yucel & Yuan, 2016). Parent-child relationships, especially mother-child relationships, have also been shown to be highly correlated with maltreatment trauma. Children with depression generally receive less care and have an unsecure attachment with maternal caregivers (Lowell et al., 2014). Even for children who have been severely abused by their parents, good parent-child relationships can still help them eliminate their depressive mood, and they will even consider beating as "an act of love" (Katz et al., 2020). However, there are also studies that challenge the effect of family relationships. For example, based on a survey in Pennsylvania, Wang and Kenny (2013) found that the family relationship was insignificant in moderating the relationship between parents' harsh verbal discipline and adolescents' depressive symptoms.
Social support is another important protective factor. Researchers have proved that the promotion of individual and environmental (e.g., family, school, community) resources and the enhancement of positive interaction between individuals and the environment effectively protect children from depression and other negative psychological symptoms (Qi et al., 2020;Zhou et al., 2020aZhou et al., , 2020bZhou et al., , 2020c. The ecological perspective likewise suggests that social support factors are crucial in promoting adolescent positive development, enhancing child resilience and competencies, and increasing opportunities for child growth (Shek et al., 2019). Specifically, family support and peer support are generally considered to be the main protective variables, as they provide securement and attachment for children Li et al., 2019;Papafratzeskakou et al., 2011;van Harmelen et al., 2016). Evidence from Louisiana even suggests that the support from the parent abuser is necessary to repair child trauma, especially when the abuser is the mother (Fleckman et al., 2019). The influence of school/teacher support on maltreatment trauma is also inconclusive. Some studies have indicated that adolescents with school intervention are four times less likely to display internalizing problems compared to their counterparts (Gabalda et al., 2010), while other studies have argued that the influence of school/teacher support is limited due to a lack of intervention skills (Alisic, 2012). Furthermore, with the development of professional community-based intervention techniques, communities are playing an increasingly role in reducing children's mental disorders in developed countries (Swenson et al., 2010).
In addition to these two protective factors, maltreatment trauma is also related to some demographic factors, such as a child's gender and age. Differences in maltreatment trauma have been found between girls and boys (Evans et al., 2013), and maltreatment trauma seems to decrease along with a child's maturity (Dunn et al., 2017). Other studies also suggest that children's maltreatment trauma is highly correlated with their parents' mental health and socioeconomic status. Children whose parents have psychological problems or alcohol dependence are more likely to suffer from maltreatment trauma (Ingersoll & Hambrick, 2011;Jose & Cherayi, 2020). Similarly, children who were raised in a family with limited education experiences and inadequate household income are found to be more likely to have mental disorders (Chigogora et al., 2019;Zhou et al., 2020aZhou et al., , 2020bZhou et al., , 2020c.

Ethnic Disparities in Maltreatment Trauma
Currently, it is well documented in academia that maltreatment trauma is influenced by ethnicity. Many studies have shown that some ethnic groups have more severe trauma symptoms than other ethnicities in the United State. For example, with similar traumatic experiences, African Americans and Latinos often report lower depression compared to Whites (Roxburgh & MacArthur, 2014), and European Americans report greater anxious arousal symptoms compared to Latinos (Newcomb et al., 2009). Such ethnic disparities in maltreatment trauma have also been found in other parts of the world. The suicidal ideation of Asians with a history of childhood maltreatment is higher than the suicidal ideation of Whites, Hispanics and Blacks (Beristianos et al., 2016), and Chinese Han girls have more serious mental problems than their ethnic minority peers with a similar severity of abuse (Wan et al., 2019). However, some studies have argued that there is no substantial ethnic difference in psychological trauma after maltreatment (Pérez Benítez et al., 2010). Other studies have offered a more neutral statement and suggest that the impact of ethnicity on the consequences of maltreatment involves tendency rather than severity. Neglected White children showed extensive mental health disorders, Black children showed more anxiety and dysthymia, and Hispanic children showed an increased risk of alcohol problems (Wildom et al., 2012).
There are mainly two theoretical approaches to explain the ethnic disparity in CM trauma. The first approach involves the prevalent attitudes toward discipline/physical punishment among different ethnicities. Religion and cultural beliefs shape peoples' childrearing norms and their definition of CM (Ben-Arieh & Haj-Yahia, 2006;Coulton, et al., 2018). For example, it is suggested that the Spanish show a strong tolerance of CM due to the influence of patriarchy (Drake & Jonson-Reid, 2011). It is also explained that some physical punishment behaviors can be tolerated and not be regarded as abuse as they are frequently used as traditional ways to discipline children in Asian families (Kesner et al., 2016;Zhai & Gao, 2009). The internalized rationalization of violent behaviors reduces the possibility of maltreatment trauma (Duru et al., 2019), as the child is likely to attribute the abuse to his or her own mistakes rather than to error on the part of the abuser (Plus et al., 2019).
The other approach explains the ethnic disparity of CM trauma as the result of unequal child service resources among different ethnicities. Existing studies indicate that the benign interaction between the individual and the environment is important to restrain CM trauma (Qi et al., 2020), and the unavailability of sufficient child service resources will hinder the development of child resilience (Zhu & Shek, 2020). Evidence from the United States shows that compared to Latinos, Asians, and Anglos, it is more difficult for African-American children to access child service resources in terms of cases open to service, out-of-home placement, and family reunification (Lu et al., 2004). It is also reported that in Israel, it is more difficult for Arab children than for Jewish children to receive continuous child welfare services (Ben-Arieh & Haj-Yahia, 2006). Psychological trauma caused by maltreatment can be suppressed by the improvement of social communication skills. However, children raised in an ethnic community with insufficient professional child services and resources are likely to adopt internalized emotion-oriented coping or experiential avoidance strategies when encountering adversity (Flett et al., 2012;Shenk et al., 2014).
Despite its rapid development, China is still deeply influenced by Confucianism and patriarchal culture, under which the child is the personal property of the parent rather than an independent individual, and child obedience to parents is regarded as important to the maintenance of family order (Peng et al., 2015). Chinese people generally regard physical punishment as a necessary parenting technique (Zhao et al., 2017). A multi-country comparison found that Chinese people have the highest threshold for labeling behaviors as maltreatment (Mesman et al., 2020). Such high tolerance for childhood violence may reduce child post-traumatic depression, as children might rationalize such violent behaviors. In addition, the current child welfare system in China is still underdeveloped, and professional child and family services are very limited (Wan et al., 2020). This might make it difficult to conduct interventions for at-risk children, increasing the possibility of post-traumatic symptoms.
There are very few studies on Chinese ethnic disparities in maltreatment trauma, but evidence has suggested that different ethnicities in China may respond to stressful events or adversities differently. For example, Tibetan survivors were reported to have less PTSD symptoms than Han survivors after the 2008 Wenchuan Earthquake because of the support and comfort gained from Tibetan Buddhism rituals Yuan et al., 2009). Religion influences its followers in how they perceive stressful events and how they address problems, which could further lead to varied outcomes of adversity (Pargament et al., 1998). Moreover, most ethnic minorities in China live in economically underdeveloped areas where the high quality of social services are insufficient and inaccessible (Wan et al., 2019). This makes it is difficult for the ethnicities located in less developed regions to receive professional counseling services when they encounter adversities.

Hypotheses
Based on the evidence from the literature, we proposed the following three research hypotheses.
H 1 : Child maltreatment poses a general threat with respect to child depression in multi-ethnic areas. A child with maltreatment experiences is more likely to suffer from depression.
H 2 : Ethnic disparities exist in the psychological trauma caused by child maltreatment in multi-ethnic areas. There are differences in the strength of the association between child maltreatment and depression in different ethnic groups.
H 3 : The protective effect of family relationship and social support factors on child maltreatment trauma differs among ethnicities.

Procedure
The data of this study come from an empirical survey conducted in three multi-ethnic counties in western China from November 2019 to January 2020. All three of these counties are on the list of the 52 poorest counties in China. County R is a Miao-populated county in which the Miao accounts for 43.5% of the total population. Other ethnic minorities, including the Dong, Yao and Zhuang, together account for 32.4% of the population. County T is a Tibetan community in northwestern China. The proportion of Tibetans in the territory is approximately 38.4%, and other ethnic minorities account for another 8.8% of the population. County H is mainly constituted by the Hui minority group. The Hui and Dongxiang, along with other ethnic minorities, account for 57.0% of the total population in this area.
The survey in this study adopted a multistep cluster sampling strategy. First, in each county, one economically developed town, one medium-economically developed town, and one less-developed town were randomly selected. In each town, one elementary school and one junior high school were randomly selected. Considering children's ability to answer questions, we selected for investigation only students from the fifth to the ninth grade. In each grade, we randomly selected one class of students to participate in the survey. A total of 1885 questionnaires were sent out, and 1809 questionnaires were returned. Among them, 1763 questionnaires were valid, which was a valid response rate of 93.5%. The ethnic minority rates of the respondents in R, T, and H counties were 81.0%, 57.0%, and 66.1%, respectively, which were close to the overall ethnic ratios in each county. The ages of the children in our sample ranged from 9 to 16 years, and the average age was 12.34 years.
Ethical standards were strictly reviewed by Nankai University before the investigation. First, we entered each school with a letter of introduction from the local education department. Moreover, we obtained written consent before each investigation with a child. Second, trained investigators presented the entire process of investigation, answered inquiries and maintained order on the spot without any intentionally misleading language for the questions. All the investigators were trained, and they carefully read the written investigation manuals before the survey. Third, considering the potential interference of the teachers, the teachers were asked to be absent from the investigation site. Fourth, the children were required to sit at a distance to avoid mutual interference among classmates. Finally, the time for the investigation was sufficient so that all students could successfully complete the questions.

Child Maltreatment
This study used a modified version of the "Juvenile Victimization Questionnaire" (JVQ) Scale to measure CM (Radford et al., 2011, pp. 125-130). Three subtypes of CM, including physical abuse, emotional abuse and neglect, were investigated in the survey. The sexual abuse category was excluded because children in East Asia usually conceal their experiences to avoid family stigma (Zhai & Gao, 2009). Physical abuse was measured by the following items: (1) physical punishment such as standing, kneeling and palming; (2) slapping the face or kicking; (3) beating or kicking important body parts such as the head and chest; (4) beating with a belt, stick or blunt object; (5) striking with hard objects; and (6) burning or scalding. Emotional abuse was measured by the following six items: (1) deliberate insulting; (2) verbally insulting the child's dignity; (3) refusing to speak to the child; (4) mocking or a lack of respect toward the child; (5) witnessing or hearing violence; and (6) being threatened with violence. Neglect was measured by (1) inadequate daily care, (2) a delay or absence of medical care, (3) ignoring the child's feelings or emotions, (4) the indulgence of bad behavior, (5) showing a reluctance/unwillingness to assist in the child's education, and (6) a lack of concern about the child's safety.
The items were measured by asking the question, "Have you suffered the following experiences in your family in the past year?". The response options ranged from 0 to 4 (0 = never, 1 = occasionally, 2 = sometimes, 3 = often and 4 = always) on a Likert scale. The Cronbach's alpha coefficients of physical abuse, emotional abuse, and neglect were 0.74, 0.83, and 0.76, respectively. The KMO coefficients for these three subtypes of maltreatment were 0.80, 0.87, and 0.83, respectively. Bartlett's significance was 0.00, which showed good reliability. The scores for all three maltreatment subtypes ranged from 0 to 4. When the score was higher, the children were maltreated more frequently.

Depression
This study mainly focuses on child depression because it is most frequently discussed in academia. The DSM-5 Depression Scale (11-17 years old) was used in our survey to measure the children's depression (American Psychiatric Association, 2013). The DSM-5 scale has been used worldwide. Related studies have applied and validated this scale in the Chinese context (Zhang et al., 2018). For all 9 items that measure depression (see Table 2), the respondents were asked to reply to the statement, "Did you experience the following symptoms in the past two weeks?". There were four possible categories of responses to this question (0 = none, 1 = several days, 2 = more than once a week, 3 = almost every day). The Cronbach's alpha and KMO coefficients of the 9 items were 0.84 and 0.91, respectively, and Bartlett's significance was 0.00. The depression score ranged from 0 to 27; when the score was higher, the child's depression was more severe. According to McLeod et al. (2016), the measurement of severe depression symptoms (SDS) includes the following two dimensions: (1) subthreshold depression symptoms (in which the child reports having at least one of the core symptoms of major depression but has not been diagnosed with a major depressive episode), and (2) major depressive disorder (in which the child meets the diagnostic criteria for a major depressive episode).

Influencing Factors
The influencing factors consisted of CM as the risk factor and the parent-child relationship and social support as the protective factors. Specifically, the parent-child relationship was measured through the three variables of the "IPV of parents" (0 = neither parent suffers, 1 = one parent suffers, 2 = both parents suffer IPV), "parent-child trust" (0 = neither parent is trusted, 1 = one parent is trusted, 2 = both parents are trusted), and "parent-child interaction" (interaction that includes playing with, talking to, and educating children; 0 = neither parent actively interacts with the child, 1 = one parent actively interacts with the child, 2 = both parents actively interact with the child). The social support factors included family support (support from parents and relatives), peer support (support from classmates and friends), community support (support from neighbors and community social workers), and school/ teacher support (support from teachers and school counselors). These support factors were measured by asking whether the respondents had received psychological counselling services from the above people (0 = no, 1 = yes).
As suggested by the literature, six other factors were also included as control variables in this study. The child's gender (0 = male, 1 = female) and age (continuous variable) were first included as these two factors are often reported to be predictive of CM trauma. The socioeconomic status of the family might be also influential. Considering that children might be unable to accurately report the household income of their family, we instead used the father/mother's education experiences (0 = no schooling, 1 = primary school, 2 = junior high school,3 = high school, 4 = undergraduate, 5 = graduate student) and father/mother's occupation (0 = jobless, 1 = non-public servant, 2 = public servant) to examine the possible influence of parents' socioeconomic status. Non-public servant mainly includes farmers and migrant workers, while public servant mainly includes civil servants, teachers, doctors and village officials. Parents' mental health condition, namely, whether they have mood disorders or substance-abuse disorders, might also be influential (Huang et al., 2019). Therefore, we also included the factors of father/ mother's depressive mood (whether father/mother shows depressive symptoms, 0 = yes, 1 = no) and father/mother's alcohol dependence (whether father/mother drinks more than 3 times or was drunk once a week, 0 = yes, no = 1) as control variables in this study.

Analytical Method
This study mainly adopted four research methods. The descriptive statistics were mainly used to compare the data, especially to analyze the SDS rates among different ethnic groups with the Chi-square test. A mean analysis was used to examine the impact of maltreatment on the depression score (DS) with the F test. A factor analysis was mainly used for the measurement of CM and depression, which focused on the reliability of their sub-indicators. A linear regression analysis was used to measure the risk factors and protective factors for psychological trauma. Our main statistics were the B, standard error (S.E.) and p-value.

Sample Description
Seven ethnicities were included in our statistics. The data in Table 1 show that the Dongxiang (prevalence = 82.8%, Mean = 1.06) and Hui (prevalence = 76.9%, Mean = 1.01) populations had the highest maltreatment risk. Additionally, Tibetan (prevalence = 64.3%, Mean = 0.62) and Han (prevalence = 66.7%, Mean = 0.64) children had the lowest risk. For the family relationship, Zhuang family cohesion was limited; compared to the other ethnic groups, their IPV ratio (8.5%) was significantly higher, while their parent-child trust ratio (32.5%) and active interaction ratio (66.7%) were relatively lower. The parent-child relationship of Han and Tibetan families seems to be the best. Their parent-child trust ratio (Han: 37.1%; Tibetan: 38.4%) and parent-child interaction ratio (Han: 80.2%; Tibetan: 82.7%) were evidently higher than the other ethnic groups. Moreover, the Han, Tibetan and Hui populations showed a relatively higher rate of social support, and their average rates were 36.9%, 39.4% and 37.2%, respectively, compared to the average rates of the Miao (28.8%), Zhuang (28.8%) and Dongxiang (29.0%) populations. In terms of these control variables, the distributions of the gender and age factors in each ethnicity were similar. However, Miao and Zhuang children reported higher rates of their father's alcohol dependence. Moreover, the Hui and Dongxiang populations reported relatively higher rates of their parents' unemployment and mood disorders and lower levels of their parents' education. Table 2 shows the effect of maltreatment on child depression. The total DS of children with maltreatment experience was 2.09 times the total DS of children

Main independent variables
Child maltreatment ( without maltreatment experience (p = 0.00). The SDS rate of the children with maltreatment experiences also increased to 8.4% from 2.2% (p = 0.00). Among the three subtypes of maltreatment, the effect of neglect on child depression was the most significant. The DS of children who suffered from neglect was 1.93 times the DS of children who did not suffer from neglect (p = 0.00). The SDS rate of this group increased from 3.5% (without neglect experiences) to 10.4% (p = 0.00). The effect of emotional abuse was also significant. The DS of victims of emotional abuse increased by 69.3% (p = 0.00), and their SDS rate increased by 2.03 times (p = 0.00). The effect of physical abuse was comparatively less influential; the DS of children who suffered from physical abuse increased by 63.2% (p = 0.00), and their SDS rate increased by only 75.0% (p = 0.00). Therefore, H 1 cannot be rejected. Table 3 shows the results of the effect of maltreatment on child depression in each ethnic group. The data show that among the ethnic groups, Han children seemed to be most likely to suffer from maltreatment trauma. The DS of Han children with maltreatment experience increased by 1.70 times (p = 0.00), and their SDS rate increased by 6.55 times (p = 0.00) compared to the rates in children without maltreatment experience. The effect of maltreatment on ethnic minority children, including the Miao, Tibetan, Hui, and Dongxiang, was also influential. The DS of these ethnic minority children who suffered from maltreatment generally increased by 1.81 to 2.03 times (p ≤ 0.01) compared to the DS of their peers. Comparatively, the effect of maltreatment on children's depression was the lowest among Zhuang children. The DS of Zhuang children with maltreatment experience increased by only 23.6% (p = 0.32) and their SDS rate increased by only 1.73 times (p = 0.30) compared to the rates in Zhuang children without maltreatment experience. Ethnic disparities were also found in the effects of the three subtypes of maltreatment on child depression. The effects of the three subtypes of maltreatment were significant for both Han children's DS (p ≤ 0.01) and their SDS rate (p ≤ 0.05). Significant effects of the three subtypes of maltreatment on these two dependent variables were also found in Miao children (p ≤ 0.05). For Tibetan, Hui, and Dongxiang children, the effects of the three subtypes of maltreatment were significant on these children's DS (p ≤ 0.01) but not on their SDS rates (p > 0.05). For Zhuang children, the effect of neglect on DS and SDS was significant (p ≤ 0.05), but the effects of emotional abuse and physical abuse were not influential (p > 0.1). Thus, H 2 cannot be rejected. Table 4 shows the effects of the influencing factors on child depression. The model without the interaction effect confirms that child depression increased with the frequency of all subtypes of maltreatment (B = 2.45-3.26, p ≤ 0.01). Good intergenerational relationships had a protective effect on child depression, and increased parent-child trust (B = − 0.56 to − 0.49, p ≤ 0.01) and parent-child interaction (B = − 0.78 to − 0.55, p ≤ 0.01) could reduce children's depressive mood. However, the social support factors were not effective in reducing maltreatment trauma. Family support (p ≥ 0.13), community support (p ≥ 0.76), and school/teacher support (p ≥ 0.23) were all insignificant, and peer support even aggravated children's depressive disorders (B = 0.78-0.82, p ≤ 0.01). In terms of these control variables, girls (B = 0.59-0.78, p ≤ 0.01), late adolescents (B = 0.43-0.48, p ≤ 0.01) and children whose parents have mental disorders (father: B = − 0.72 to − 0.61, p ≤ 0.05; mother: B = − 0.88 to − 0.61, p ≤ 0.05) were more likely to experience depression.

Ethnic Disparities in the Influencing Factors
The results in Table 4 also show the moderation effect of ethnicity. Significant differences in the effect of physical abuse on child depression among the ethnicities were not found. However, the effect of emotional abuse on child depression among the Hui (B = − 2.39, p = 0.00) and Dongxiang (B = − 2.21, p = 0.00) were significantly less compared to the effect of emotional abuse on child depression among the Han. That is, Han children were more likely to suffer from depression than Hui and Dongxiang children after experiencing emotional abuse. Moreover, the effect of neglect on child depression among the Dongxiang (B = − 2.38, p = 0.00) was also significantly less than the effect of neglect on child depression among the Han.
The results in Table 5

Discussion
Based on an empirical survey of 1763 children in three ethnic counties in China, three important conclusions are found. First, child maltreatment poses a general threat to children's psychological health in multi-ethnic areas. Our data show that all three subtypes of maltreatment, including neglect, emotional abuse, and physical abuse, have negative effects on children's depression. The negative influence of maltreatment on children's depressive mood may be universal among different countries and ethnicities. Similar to the research of Vugt et al. (2013), this result also shows that the effects of neglect and emotional abuse on Chinese children's psychological  3.46 (6.65)** 4.6 (11.5) + trauma are more significant than the effects of physical abuse. This finding suggests that the healing of psychological trauma rather than physical injury is more important in maltreatment trauma therapy. This could be related to the high tolerance of physical discipline in Chinese culture, under which children might rationalize such violent behaviors (Plus et al., 2019). This, in turn, reduces the risk for child posttraumatic depression (Duru et al., 2019). On the other hand, the significant effects of neglect and emotional abuse on child depression could be related to the fact that today, many children in China have been separated from their parents as left-behind children. Parental absence could result in more limited parent-child trust and is not conducive to developing an intimate parent-child relationship, which could make children more vulnerable (Kaur & Kearney, 2013;Lowell et al., 2014). Second, ethnic disparities in maltreatment trauma are found in multi-ethnic regions. Among the different ethnic groups in our survey, the effect of maltreatment on child depression is most influential among Han people and least influential among Zhuang people, while other ethnic groups are in between these two groups. These ethnic disparities are also found in the effects of all three subtypes. Psychological trauma cannot be completely attributed to the severity of maltreatment itself (i.e., compared to other ethnic groups, the Hui, Zhuang, and Dongxiang, who have higher rates of maltreatment, do not show more psychological trauma). For different ethnic groups, perceptions or tolerance of child abuse (Drake & Jonson-Reid, 2011), the spiritual comfort derived from religion  and the availability of child service resources (Qi et al., 2020) could together lead to differences in CM trauma. This also shows that the internalization of psychological trauma may be a process of self-repression (van Harmelen et al., 2010), and children's emotion regulation ability after adversity plays a mediating role in maltreatment and its consequences (Crow et al., 2014).
Third, ethnic disparities are also found in the effects of the protective factors on maltreatment trauma in multi-ethnic regions. The effect of family relationships on maltreatment trauma is deeply influenced by family cohesion. A good family relationship can help children develop secure attachment, moral competence, and prosocial values (van Harmelen et al., 2016;Zhou et al., 2020aZhou et al., , 2020bZhou et al., , 2020c, which could all help them better adjust their negative feelings and emotions. Specifically, the Han and Tibetan populations, who have more harmonious parent-child relationships, have a reduced likelihood of maltreatment trauma. For the Hui, Zhuang and Dongxiang populations, their lower intergenerational cohesion makes the potentially protective influences ineffective. These findings revise Wang and Kenny's conclusion that the reason for the limited protective effect of family is not only the severity of maltreatment but also parent-child cohesion. The protective effect of social support as suggested in the literature has not been found in our study. In multi-ethnic areas of China, social support factors are not influential among the Hui, Zhuang, and Dongxiang populations, and peer support factors increase the possibility of CM trauma among the Han, Tibetan and Miao populations. The reason for this may be non-professional psychological intervention skills, as Alisic (2012) mentioned. Currently, because China still fails to provide adequate social services for children in multi-ethnic areas, the formal support (including personnel, funds, training courses and technology) for Chinese teachers and social workers is very limited (Wan et al., Table 4 Influencing factors for child depression in multi-ethnic regions  2020). Without proper training or education, the current counseling services provided by teachers and community workers are neither self-initiating nor professional in most multi-ethnic communities. Peer support may even enhance children's depression due to the effects of empathy when the left-behind rate is high (Leshem et al., 2016). It is worth noting that social demographic factors such as age, gender, and parental depression have significant effects on child post-traumatic depression. Our results support that compared to boys, girls have higher risks for CM (Evans et al., 2013). We also find that maltreatment trauma does not necessarily decrease as a child matures (Dunn et. al., 2017). The increasing academic pressure and burden experienced by Chinese upper-grade children could make them more vulnerable than younger children (Zhou et al., 2020a(Zhou et al., , 2020b(Zhou et al., , 2020c. Our results also support that child depression is highly related to parents' depressive mood (Ingersoll & Hambrick, 2011). These findings suggest that in practice, there is a need to pay more attention to those children at higher risk, including those who have parents with psychological disorders.

Policy Implications
This study has profound implications for child policy in China. Traditionally, CM has been regarded as a necessary way to discipline children and maintain family order (Peng et al., 2015). Therefore, CM trauma is easily overlooked and not taken seriously by many Chinese parents in multi-ethnic areas. Moreover, the current child welfare system in China is orphan-based, and over 200 million Chinese nonorphans are actually excluded (Wan et al., 2020). The current child policy in China has not been effective in intervening in the family parenting process and in regulating the abusive behaviors of parents. Since 2010, the Chinese government has begun to cooperate with UNICEF and has established 120 child welfare demonstration villages in five provinces; they aimed to provide universal welfare services such as temporary guardianship and psychological counseling for all children (UNICEF, 2015). However, by 2015, only 100 thousand children were covered by this program (UNICEF, 2015), and most Chinese children are unable to receive high-quality child welfare services (Jiang et al., 2019). Some unofficial projects, such as Hong Kong's "P.A.T.H.S. Project", have been experimented with in certain cities and have shown good intervention effects, but these still have not been widely implemented in mainland China (Zhu & Shek, 2020). Thus, professional family and child services are difficult to obtain, especially in underdeveloped multi-ethnic areas (Man et al., 2017). For the Han and Tibetan, the improvement of parent-child cohesion could reduce children's depression to a certain extent, while the abused children among the Hui and Dongxiang who are raised in high-risk families might only be able to rely on themselves to deal with adversity. It is urgent to provide universal child welfare services to children in multi-ethnic areas.

Contributions and Limitations
Our study mainly aims to reveal the potential Chinese ethnic disparities in CM trauma as few studies have explored the relationship between ethnicity and CM trauma in multi-ethnic regions. The results of this study not only indicate a correlation between ethnicity and maltreatment trauma in China but also respond to important issues in the international debate. Our results show that the current social support for abused children in the multi-ethnic regions of China is highly insufficient. It is urgent to reform the current residual orphan-based child welfare system and provide universal child welfare services to children in multi-ethnic areas.
Admittedly, there are some limitations of this study. First, considering the difficulties in obtaining reliable and valid data, we do not further our study to draw a link with ethnic culture and family values; therefore, we are unable to determine the extent to which the existing ethnic disparities in CM trauma are related to cultural factors. Second, we do not further the study to address the issue of whether there might be a correlation between who the maltreatment is from and where the support is from. Third, given the sensitivity of reporting sexual abuse experiences in East Asia, this study did not include sexual abuse in the survey. Fourth, due to the unavailability of official and public data, this study did not fully consider the influence of the potential differences of the general depression rate among ethnicities. In the future, studies should further explore the ethnic disparities in sexual abuse and their influence on child depression, and they should also examine the possible influence of regional factors such as different depression rates and regional social expenditures on child services.