Bullying victimization constitutes a serious public health concern in school students around the world. According to Olweus, bullying victimization is defined as the experience of intentional, repetitive, and aggressive actions by peers towards an individual who has difficulty defending himself or herself [1
]. A previous study found adolescents to be the most common targets of bullying behaviors [2
]. It was reported that approximately 32% of students across 38 countries had experienced being bullied [3
]. In China, the issue also seems to be prevalent, as approximately half of students have reported lifetime bullying victimization [4
Several types of bullying behaviors have been identified, such as physical, verbal, relational, and cyber behaviors. Traditional types of bullying include physical (hitting, kicking, pushing, etc.), verbal (name-calling, taunting, etc.), and relational (spreading rumors, excluding, etc.) bullying [6
]. Physical and verbal aggression are direct forms of bullying that involve face-to-face confrontation, whereas relational aggression is regarded as indirect bullying [7
]. Cyber bullying refers to an emerging form of bullying through the application of computers, cell phones, and other electronic devices [8
The literature has suggested that gender differences exist in the occurrence of different types of bullying victimization. Olweus reported that, in Norway, boys were more likely to be victims of physical and verbal bullying, whereas girls were more vulnerable to relational bullying [9
]. In a survey of Finnish children, boys reported more physical victimization and girls reported more relational victimization, but there was no significant difference in verbal victimization [10
]. The same results have been found in British studies [11
The impact of bullying victimization on children and adolescents has been linked to a wide range of negative mental health outcomes. Studies have found that victims of bullying are particularly vulnerable to serious mental health problems such as anxiety and depression [13
]. More seriously, bullying victimization is a significant risk factor for non-suicidal self-injury (NSSI), suicidal ideation and behaviors [15
]. Physiological responses to stress, cognitive distortion, and oversensitivity to social cues may be potential mechanisms for the associations between bullying victimization and mental health problems [18
]. As mental disorders in children and adolescents have been increasing over recent decades [19
], bullying prevention programs may be important in promoting mental health of victims of bullying [20
Considering that some individuals may be the victims of multiple types of bullying while others may be the victims of only one type, some recent studies have explored the application of latent class analysis (LCA) to better understand the co-occurrence of different types of bullying victimization [21
]. Similar to cluster analysis, LCA is a person-centered analysis approach that aims to classify samples into several mutually exclusive classes depending on their observed responses to multiple manifest indicators [23
]. In an LCA model, a latent categorical variable is created to determine individuals’ class membership, and their associations with predictor variables or outcome variables can be tested. This approach has been validated in numerous studies of health-related behaviors [24
]. LCA has been conducted in previous studies to examine patterns of bullying victimization in US adolescents [21
]. However, patterns of bullying victimization among Chinese youth and the associations of bullying victimization with mental health problems remain unclear. Therefore, we conducted this large-scale survey and employed LCA to examine patterns of involvement in different types of bullying victimization among Chinese adolescents and to evaluate the associations between bullying victimization and an array of mental health outcomes including anxiety symptoms, depressive symptoms, NSSI, suicidal ideation, and suicide attempts.
In the present study, we employed an LCA approach to explore patterns of bullying victimization in a large sample of Chinese adolescents. Notably, we found that distinct patterns of victimization existed between boys and girls, with different numbers and characteristics of latent classes by gender in the LCA models. Moreover, class differences were also found in terms of students’ anxiety symptoms, depressive symptoms, NSSI, suicidal ideation and suicide attempts. This result demonstrated the association of bullying victimization with mental health problems.
As expected, most of the students (84.2% of boys and 93.7% of girls) had minimal probabilities for each victimization item and were classified into the low victimization class. The proportion of this reference class was higher than that reported in American studies [21
]; in other words, Chinese students may generally have a relatively low probability of being bullied. One possible explanation is that the cultural characteristics and social environment of China have long been influenced by Confucian culture. The values of benevolence, righteousness, and propriety advocated by Confucian culture require people to love, help and respect each other, which establishes basic moral principles for the society and plays a role in restraining the bullying behaviors of Chinese adolescents. [37
]. On the other hand, the proportion of boys in the low victimization class was lower than that of girls, which is a similar finding to previous studies that bullying victimization was more common among boys [38
For the remaining students, characteristics of the item probabilities in the latent classes differed between boys and girls. There was a small number of students (0.56% of boys and 0.72% of girls) who were classified into the high victimization class, wherein boys had a high probability greater than 0.70 for each item. For girls in this class, the item probabilities were found to be heterogenous; they had high probabilities greater than 0.70 for items 1, 2, and 4, representing verbal and relational victimization and had a moderate probability approaching 0.50 for item 7, representing cyber victimization, whereas the probabilities for the other items were relatively low, ranging from 0.20 to 0.30. The moderate victimization class accounted for 2.8% of boys and 5.6% of girls and showed a similar contour of the probability plot to that of the high victimization class of the same gender. There was an extra group among the boys that we identified as the verbal victimization class. It comprised 12.4% of boys, all of whom reported “being called mean names, made fun of, or teased in a hurtful way” but had minimal probabilities for the other items.
Based on the classification and item probabilities, patterns of bullying victimization among students can be summarized. For those at moderate to high risk of bullying victimization, boys were vulnerable to all types of bullying behaviors, while girls were more likely to be victimized by verbal and relational bullying than physical bullying. Moreover, verbal victimization was likely to occur alone in a certain group of boys, but this phenomenon was not observed in girls. These findings are not completely consistent with those of previous studies. For example, research among adolescents in ten Europe countries reported that boys were more likely to be physically and verbally bullied, whereas girls were more prone to be victims of relational bullying [41
]. Similar results were found in American and Dutch adolescents [42
]. A study from South Korea noted that boys were more likely to experience physical bullying, while girls were more likely to experience relational, verbal and cyber bullying [44
]. The differences in conclusions may arise from the different sample sources and statistical analysis methods. Similar to our study, Wang et al. used an LCA approach to examine the co-occurrence of victimization subtypes among US adolescents [21
]. Specifically, they classified samples into three latent classes labeled “all-types victims”, “verbal/relational victims”, and “non-victims”. However, their study showed that the overall pattern of victimization was the same across genders, which was inconsistent with our finding. Another study [45
] of peer victimization conducted an LCA and identified a four-class pattern for middle school students and a three-class pattern for high school students. With regard to gender differences, boys were more likely to be in the “verbal and physical victimization” class, while girls were more likely to be in the “verbal and relational victimization” class. According to the LCA findings in Nylund’s study [22
], victims were likely to experience multiple types of bullying rather than a single type. Nevertheless, some boys in our study were only verbal victims without experiencing any other type of victimization. In sum, discrepancies in these LCA results to some extent reveal the diversity and complexity of bullying victimization patterns among adolescents.
Mean levels of anxiety symptoms and depressive symptoms and probabilities of NSSI, suicidal ideation, and suicide attempts were compared across the latent classes. We found that, in general, students with higher involvement in bullying victimization had poorer mental health. This finding is consistent with previous studies demonstrating that bullying victimization is positively associated with mental health problems such as anxiety, depression, NSSI, suicidal ideation, and suicide attempts [3
]. Importantly, boys in the verbal victimization class reported more severe problems than those in the low victimization class, signifying that even a single type of victimization could adversely affect victims’ mental health. In addition to the negative consequences included in our study, bullying victimization may also lead to psychosomatic complaints, difficulty sleeping, low self-esteem, poor academic achievement, and poor psychosocial adjustment [47
]. Therefore, zero-tolerance policies should be adopted for the prevention of and intervention in peer bullying.
To our knowledge, this is the first study to use LCA to examine patterns of bullying victimization in a large sample of Chinese adolescents and to find distinct patterns between genders. However, several limitations should be noted. First, our sample was limited to Guangdong Province in China. Thus, the findings may not be generalized to the whole country. Second, due to the cross-sectional nature of the study, it is difficult to draw causal inferences based on the associations between class membership and mental health outcomes. Third, as the data were collected based on students’ self-reports, we could not completely avoid recall bias and reporting bias. Finally, covariates that may predict individuals’ bullying victimization class memberships were not included in the LCA models.