Joint Developmental Trajectories of Bullying and Victimization from Childhood to Adolescence: A Parallel-Process Latent Class Growth Analysis

Abstract

Previous studies have shown that bullying and victimization can be experienced simultaneously by an individual and can change over time. Understanding the joint longitudinal development of the two is of great significance. We conducted a 4-year longitudinal study to examine the joint developmental trajectories of bullying and victimization, gender and grade differences in trajectory group membership, and changes in specific forms of bullying and victimization (verbal, relational, and physical bullying /victimization) in each trajectory group. A total of 775 children from China participated in our study. The average age of participants at the first wave was 10.90 years (SD = 1.12), and boys accounted for 69.5% of the sample. Based on mean scores, four distinct joint developmental trajectories of bullying and victimization were found: the involvement group (both bullying and victimization increased from low to high over time, accounting for 7.6% of the total), the desisted group (both bullying and victimization decreased from high to low over time, 6.1%), the victimization group (victimization remained at a high level, whereas bullying remained at a low level for 3 years, 13.2%), and the noninvolved group (bullying and victimization remained at a stable low level, 73.1%). Boys were more likely than girls to belong to the involvement group, desisted group, and victimization group, whereas girls were more likely than boys to belong to the noninvolved group. There was no significant grade difference in the trajectory group. All forms of bullying/victimization were consistent with the overall trend and showed similar levels. These results have important implications for the prevention of and interventions for school bullying.

Keywords

bullying victimization joint developmental trajectories children adolescents

School bullying is a kind of aggressive behavior in which some students often intentionally harm other students who are unable to defend themselves (Zych et al., 2020). It has become a serious public health problem worldwide (Kim et al., 2009). A recent meta-analysis that included 80 studies from different cultural backgrounds found that the prevalence of bullying and victimization was 35% and 36%, respectively (Modecki et al., 2014). Children who bully others often advocate violence, have little empathy (Isernhagen & Harris, 2004), and use physical, verbal, and/or relational aggression to harm other students. The victims are students who are often targets of bullying (Olweus, 1996). In most cases, they are from backgrounds of lower socioeconomic status, have lower self-esteem, and consider themselves unattractive (Isernhagen & Harris, 2004). In addition, aggressor-victims engage in bullying others while also being bullied.

Children involved in school bullying, whether as perpetrators or victims, often show long-term maladjustment, including school, health, behavior, and emotional problems (Hysing et al., 2019; Jetelina et al., 2019; Juvonen & Graham, 2014). Considering the seriousness and certainty of these consequences, various intervention programs have been developed to prevent school bullying (Randa & Hayes, 2018). However, most existing interventions have been less effective (Juvonen & Graham, 2014). The reasons may be varied and complex. As some researchers have noted, one possible reason is that general prevention programs neglect the heterogeneous nature of victims/perpetrators (Sumter et al., 2012). According to relevant theoretical perspectives, from childhood to adolescence, the development of peer bullying and victimization is characterized by heterogeneity, such as early and late onset, and social interactional continuity patterns (Ettekal & Ladd, 2017). Some researchers believe that the weak effects of present interventions indicate the need for more research to examine the dynamic process of bullying and victimization (Guerra et al., 2011). Understanding longitudinal changes in bullying and victimization can help clinicians and researchers develop effective prevention and intervention programs (Hill et al., 2017; Zych et al., 2020) and determine the optimal time to implement interventions (Jetelina et al., 2019). In summary, researchers should conduct more basic research that not only examines the dynamic changes in bullying and victimization but also considers the heterogeneity of such changes. Thus, a group-based modeling approach that considers longitudinal changes, such as latent class growth analysis (LCGA), is a suitable method to capture the moving picture of bullying and victimization.

At present, some longitudinal studies have used this method to explore the heterogeneous trajectories of bullying and victimization. Some studies explored trajectories of bullying perpetration (e.g., Cho & Lee, 2018; Espelage et al., 2018; Pabian & Vandebosch, 2016; Pepler et al., 2008; Reijntjes et al., 2013). In general, these studies identified three to five classes, such as a noninvolvement group, stable high group, and increasing and decreasing groups. Some others explored the trajectories of victimization (e.g., Biggs et al., 2010; Brendgen et al., 2016; Geoffroy et al., 2018; Goldbaum et al., 2003; Sumter et al., 2012), and the results also appeared in three to five classes. Almost all studies identified relatively stable high, moderate, and low groups, while some studies found increasing and decreasing groups (Biggs et al., 2010; Goldbaum et al., 2003).

These studies provided evidence of the heterogeneous development of both bullying and victimization. However, these studies were conducted with different samples. Few studies have examined both bullying and victimization trajectories in the same sample. In fact, this situation constitutes another challenge for prevention programming (Guerra et al., 2011). If we are concerned with only one of the two, such as victimization, and only intervene in that area, the child is likely to fall into victimization again over time because of bullying others. Hence, it is critical to examine the common correlates of bullying and victimization, which could provide a basis for universal intervention (Guerra et al., 2011).

Only a few studies have explored the trajectories of both bullying and victimization in the same sample (Barker et al., 2008; Cho & Lee, 2019; Haltigan & Vaillancourt, 2014). Using a second-order growth mixture model, Cho and Lee (2019) found three distinct subgroups among 2,844 Korean adolescents: a bully-victims to low bully-victims transition group (9.9%), a moderate bully-victims to victims transition group (6.8%), and a limited involvement/stable group (83.3%). Barker et al. (2008) conducted research in a sample of British adolescents from early to mid-adolescence. They examined the trajectories of bullying and victimization separately and found two bullying trajectories (low-decreasing, high-increasing) and three victimization trajectories (low, high-decreasing, high-increasing). Therefore, six distinct subgroups existed (2 × 3): the low-decreasing bullying and low victimization group (75%), the low-decreasing bullying and high-decreasing victimization group (7%), the low-decreasing bullying and high-increasing victimization group (3%), the high-increasing bullying and low victimization group (11%), the high-increasing bullying and high-decreasing victimization group (3%), and the high-increasing bullying and high-increasing victimization group (2%). Using the same method, Haltigan and Vaillancourt (2014) found two bullying trajectories (low-stable and moderate-increasing) and two victimization trajectories (low-declining and moderate-declining) among 695 Canadian children. Thus, four different subgroups were identified: low-stable bullying and low-declining victimization (named low/limited involvement, 73%), low-declining victimization and moderate-increasing bullying (bullies, 11%), moderate-declining victimization and low-stable bullying (initial/declining victims, 11%), and moderate-declining victimization and moderate-increasing bullying (victim-to-bully group, 6%).

Two of the three studies examined the developmental trajectories of bullying and victimization in two separate LCGAs and then combined them (Barker et al., 2008; Haltigan & Vaillancourt, 2014). Although informative, the application of this method makes it difficult to find a balance between the choice of the number of trajectory classes and the meaningful subgroups that exist in reality. For example, if three classes were found for both bullying and victimization in two separate LCGAs, nine subgroups will exist. However, some subgroups identified by this method account for only a small proportion of the total population (e.g., the proportions of three classes were 2% or 3% in Barker et al., 2008). These classes may exist only in this study because of the characteristics of the sample itself and cannot be generalized (the smallest class accounts for at least 5% of the total sample generally; Nylund, Asparouhov, et al., 2007). If generalizability is taken into account when choosing classes and fewer classes are deliberately chosen, the principle of class selection will be violated. In contrast, conducting parallel-process LCGAs is a more promising approach to find a balance between the choice of the number of trajectory classes and the meaningful subgroups that exist in reality; this method involves examining the joint developmental trajectories of bullying and victimization in the same LCGA. Parallel-process LCGAs, which are especially useful when it is necessary to test the simultaneous development process, have been used in previous research (Betts et al., 2014; Cruz et al., 2017). Thus, we use parallel-process LCGAs to test the number and shape of the joint trajectories of bullying and victimization. Considering that there may be gender differences (Geoffroy et al., 2018; Kim et al., 2009; Pepler et al., 2008) and grade differences in the involvement in bullying/victimization, we test whether trajectory membership differs by gender and grade.

Generally, bullying and victimization take three different forms, namely, physical, verbal, and relational bullying/victimization. Understanding different forms of bullying is crucial for detecting bullying and victimization because some forms of bullying (e.g., physical) are easily recognized and detected, whereas other forms (e.g., relational) are difficult to detect and may require special attention (Zych et al., 2020). Regarding whether forms of bullying change over time, there are two different perspectives.

According to the perspective of heterotypic continuity, aggression is a kind of age-dependent behavior. That is, the forms of aggression change with age (Björkqvist, 1994; Björkqvist et al., 1992; Underwood et al., 2009). More specifically, the forms of bullying that children use change from direct (e.g., physical bullying) to indirect (e.g., relational bullying) with age (Barker et al., 2008). There are two possible reasons for this change. First, with improvements in verbal ability and emotional intelligence, children acquire the ability to use more nonphysical means to harm their peers (Ettekal & Ladd, 2017). Second, as children grow older, physical bullying becomes less normative and less acceptable by society (Archer & Coyne, 2005) and is always accompanied by punishments from adults (Ettekal & Ladd, 2017). Empirical research has provided evidence that the form of bullying changes from physical to more indirect forms (e.g., Vaillancourt et al., 2007). In contrast to this perspective, some researchers believe that various skills and abilities make all forms of bullying available among older children and adolescents (Card et al., 2008). Thus, children use all forms of bullying freely and equally. A meta-analysis of 148 studies, mainly ranging from middle childhood to adolescence, found no significant age difference in the usage of direct and indirect bullying and noted that different forms of bullying showed a high correlation (Card et al., 2008). That is, the forms of bullying do not change with age.

One of the reasons for the mixed findings of previous studies is that heterogeneity has not been fully considered. That is, for some children, but not others, bullying forms change over time. Our study examines the changes in bullying/victimization forms in each subgroup identified by LCGAs, which is crucial to gain a clear picture of whether different bullying forms change over time.

Finally, bullying is a culture-dependent behavior. In other words, cultural factors must be considered when investigating children’s bullying and victimization. China is dominated by the collectivist values of Confucianism and its related philosophical systems. Its core values include strong avoidance of conflicts and maintaining social harmony (Chen et al., 2019). In this perspective, we assume that the proportion of children who are not involved in bullying and victimization is higher than that of their Western counterparts. In addition, in such a cultural system, the current views have not reached a consensus on which form of bullying is the most commonly used by Chinese children. In collectivist cultures, people are more likely to value interdependence in relationships and maintain harmonious relationships with others (Kawabata et al., 2012). Therefore, the level of relational bullying is lower than that of overt bullying. A cross-cultural study involving primary and secondary students in the United Kingdom and China shows that the relational bullying of Chinese students is lower than that of British students, but there is no difference in physical bullying (Ji et al., 2004). In another empirical study, authors found that among Chinese primary and secondary students, the level of verbal bullying was the highest, followed by the level of physical bullying, whereas the level of indirect bullying (relational bullying) was the lowest (Zhang et al., 2001). However, on the contrary, such a culture also lays the foundation for relational bullying (Kawabata et al., 2012). One study argued that indirect aggression is far less confrontational than direct overt aggression (Forbes et al., 2011). Also, relational aggression is more difficult for teachers and parents to detect; thus, children engaged in relational aggression could avoid punishment to some degree (Chen et al., 2019). Therefore, among Chinese children, the level of relational aggression is higher than that of direct overt aggression. A cross-cultural study across nine countries supported this view (Lansford et al., 2012). They found that Chinese children are more likely to engage in relational aggression than physical aggression. Overall, however, the physical and relational forms of aggression among Chinese children are lower than the grand mean in nine countries. Finally, one study indicated that the influence of culture on indirect aggression is the same as that on direct aggression (Forbes et al., 2009). That is, the level of indirect aggression is similar to that of direct aggression. Empirical research has shown no significant preference for relational aggression among Chinese college students (Forbes et al., 2009). The reasons for the mixed results are difficult to disentangle; thus, more research is needed.

We focus on the transition period from childhood to adolescence, that is, the period from primary school to secondary school, for two reasons. First, this period is the most serious and unstable period of bullying (Schäfer et al., 2005). Second, peer bullying during this period has been found to be particularly harmful. As this period is an important period for the development of peer relations, children often seek companionship and spend more time with peers, and peer status becomes more important than ever (Salmivalli, 2018). Therefore, both theoretical and intervention research is more beneficial for this period than for other periods.

The first aim of this study was to examine the number and shape of the joint trajectories of bullying and victimization as well as gender and grade differences in each class. The second aim was to examine whether forms of bullying and victimization change over time in each subgroup.

Method

Participants and Procedures

The data for this study come from an ongoing longitudinal study designed to provide a picture of the developmental status of children and adolescents in rural areas of China. A representative rural school located in Anhui Province was selected. This 9-year compulsory school includes primary and junior high schools. In the first survey (December of 2015), participants were in Grades 4, 5, and 6. Annual measurements were taken from 2015 to 2018 (December of each year). By December 2018, participants were in Grades 7, 8, and 9. We included participants who participated in four measurements, resulting in a sample of 775 children. There were 197, 261, and 317 children in Grades 7, 8, and 9, respectively. The average age of the participants at the first measurement was 10.90 (SD = 1.12) years. Boys accounted for 69.5% of the sample.

As with any longitudinal studies, sample attrition is unavoidable. The sample sizes for the four measurements were 1,587, 1,328, 1,041, and 775, respectively. The main reason for attrition in our study was the transition from primary school to secondary school. Although the participating school in the study is a 9-year school, students can choose to stay in the same school to attend junior high school or enter other junior high schools after graduation from primary school. Of course, other elementary school students will enter this junior high school, which is common in China. As the transition from Grade 6 to Grade 7 is a transition period from elementary school to junior high school, a large number of participants who went to other secondary schools were lost to follow-up every year. Attrition analyses were conducted to determine whether there was bias in this loss. The results revealed that compared with those children who were lost to follow-up, our study participants were more likely to be boys (odds ratio [OR] = 1.55, 95% confidence interval [CI] = [1.22, 1.97], p < .001). In terms of grade and bullying and victimization scores, there were no differences between participants and those lost to follow-up. Missing data were determined to be missing at random (MAR; Enders, 2010).

All research procedures were approved by the Research Ethics Review Board of the authors’ institution. Self-reporting was used for each data collection. Trained research assistants conducted all assessments during regular school hours using a standardized procedure. Before each survey, written informed assent was obtained from children and school administrators, and oral informed consent was obtained from parents. Participants were informed that their participation was completely voluntary and confidential.

Measures

Bullying perpetration

To assess bullying, we used the Chinese version of the Bully/Victim Questionnaire (Olweus, 1996; Zhang et al., 1999), which consists of six items. Each form of bullying is measured by two items. Sample items for physical, verbal, and relational bullying were (a) I hit, kicked, pushed, shoved or threatened some of my classmates; (b) I teased, called names, or made fun of some classmates; and (c) I spread rumors about some of my classmates and tried to make others dislike him or her (them), respectively. Children were asked how often, since the beginning of this semester, they bullied other students at school in any of the mentioned ways. Items were rated on a 5-point scale (1 = not at all, 2 = once or twice, 3 = 2 or 3 times a month, 4 = about once a week, 5 = several times a week). Responses across the six items were averaged, with higher scores representing higher levels of bullying perpetration. This measure showed reasonable internal consistency at each time point (Cronbach’s α: α_T1 = .66, α_T2 = .71, α_T3 = .85, α_T4 = .89).

Victimization

The Chinese version of the Bully/Victim Questionnaire was also used to assess victimization (Olweus, 1996; Zhang et al., 1999). Similarly, two items were used to assess each form of victimization. Sample items for physical, verbal, and relational victimization were (a) some of my classmates hit, kicked, pushed, shoved or threatened me; (b) some of my classmates teased, called names, or made fun of me; and (c) some of my classmates spread rumors about me and tried to make others dislike me, respectively. Children were asked how often, since the beginning of this semester, they were bullied by other students at school in any of the mentioned ways. Items were rated on a 5-point scale (1 = not at all, 2 = once or twice, 3 = 2 or 3 times a month, 4 = about once a week, 5 = several times a week). Responses across the six items were averaged, with higher scores representing higher levels of victimization. This measure showed good internal consistency at each time point (Cronbach’s α: α_T1 = .75, α_T2 = .78, α_T3 = .82, α_T4 = .86).

Plan of Analyses

First, a parallel-process LGCA (latent growth curve analysis) was used to determine the form of the average growth curve (i.e., linear curve or quadratic curve) and confirm that the growth factors (i.e., intercept and slope) had significant variances. Significant variances in the growth factors suggested that our sample had distinct subgroups in terms of bullying and victimization.

Second, we used a parallel-process LCGA to capture the joint developmental trajectories of bullying and victimization. This method extended the typical univariate LCGA to a parallel-process approach to account for multiple growth trajectories simultaneously (Cruz et al., 2017). The parallel-process LCGA is especially useful when it is necessary to examine the joint development process of two closely related constructs. This approach has been successfully used in previous studies (Betts et al., 2014; Cruz et al., 2017). Five models (two to six classes) were estimated using Mplus 7.4 (Muthén & Muthén, 1998–2015). To determine the optimal number of trajectory classes, multiple fit indices were used. First, the Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample-size-adjusted BIC (a-BIC) were taken into account. Models with lower AIC, BIC, and a-BIC values were considered better solutions. In addition, the Lo–Mendell–Rubin likelihood ratio test (LMR-LRT) compares the model with k trajectory classes with a model with k – 1 trajectory classes. A significant p value on the LMR-LRT indicates that a model with k trajectory classes has better fit than a model with k – 1 trajectory classes. Moreover, the entropy value was used to evaluate classification accuracy, which ranged from 0 to 1.0, with higher values indicating that individuals were more precisely classified. To increase the generalizability of our results, we also considered whether each class size consisted of at least 5% of the sample (Nylund, Asparouhov, et al., 2007). In addition to those fit indices, it is also important to consider the substantive interpretability of the trajectory classes (Nylund, Asparouhov, et al., 2007). In this step, we used robust maximum likelihood estimates (MLR) to handle missing data and estimate the model. Then, chi-square tests in SPSS 24.0 (IBM, Armonk, NY, USA) were used to investigate gender and grade differences in the membership in different trajectory groups. In this step, we used multiple imputation to handle missing data.

For the second aim of the present study, descriptive statistics (means) in SPSS were used to describe how forms of bullying/victimization change over time in different subgroups. Multiple imputation was also used to handle missing data in this step.

Results

The General Trends of Bullying and Victimization

The parallel-process latent growth curve analysis revealed that compared with the linear growth model, the quadratic model has a better model fit. Thus, both bullying and victimization followed nonlinear growth across four time points. More importantly, significant variances were found for the intercept and quadratic slope of bullying and victimization (ps < .005), suggesting that there were significant individual differences in the initial level and change in bullying and victimization. These results indicated that parallel-process LCGAs are necessary.

Joint Developmental Trajectories of Bullying and Victimization

To determine the most appropriate number of trajectory classes, two to six classes were conducted using parallel-process LCGAs. Based on the fit indices and substantive interpretability of the trajectory classes, the four-class model was determined to be the final model. The fit indices are listed in Table 1. Entropy values do not differ much among the three to six classes, indicating that these classes have high classification accuracy. The five-class and six-class models were not chosen due to nonsignificant p values on the LMR-LRT and the small size of the smallest class (n < 5%). Although the LMR-LRT did not favor the four-class model, the relative improvements in AIC, BIC, and a-BIC were much smaller after four classes. More importantly, the model chosen was based not only on statistical indices but also on the interpretability and meaningfulness of each class. After carefully checking each class model, we found that compared with the three-class model, the “new class” found in the four-class model was meaningful and significantly different from the previous classes. Furthermore, compared with the four-class model, the five-class model simply subdivided one of the classes in the four-class model into two similar small classes. Therefore, the four-class model was chosen.

Table 1.

Fit Statistics for Parallel Latent Class Growth Analyses.

Number of Profiles	AIC	BIC	a-BIC	Entropy	LMR-LRT	Proportions Min
2	9,483.46	9,581.17	9,514.48	0.92	<.001	0.20
3	8,772.20	8,902.48	8,813.56	0.95	0.002	0.08
4	8,455.14	8,617.99	8,506.85	0.95	0.178	0.06
5	8,238.14	8,433.57	8,300.20	0.96	0.112	0.02
6	8,068.97	8,296.96	8,141.36	0.96	0.359	0.01

Note. Boldface indicates the selected model. AIC = Akaike information criterion; BIC = Bayesian information criterion; a-BIC = sample-size-adjusted Bayesian information criterion; LMR-LRT = Lo–Mendell–Rubin adjusted likelihood ratio test.

The four joint trajectory classes of bullying and victimization are depicted in Figure 1. Specifically, Class 1 accounted for 7.6% of the total sample (n = 58). Children in this class were characterized by low levels of bullying and victimization at the initial point and a steep increase during the last year. Thus, this class was labeled the involvement trajectory group. Class 2 accounted for 6.1% of the total sample (n = 45). Contrary to children in Class 1, those in Class 2 were characterized by high levels of bullying and victimization initially that slowly declined over the next 3 years. Thus, this class was labeled the desisted trajectory group. Class 3 accounted for 13.2% of the total sample (n = 103). Children in this class were characterized by stable high levels of victimization and stable low levels of bullying. Thus, this class was labeled the victimization trajectory group. Finally, Class 4 had the largest number of children and accounted for 73.1% of the total sample (n = 569). Children in this class were characterized by extremely low levels of bullying and victimization; thus, this class was labeled the noninvolved trajectory group.

Figure 1.

Latent trajectory classes of bullying and victimization: (A) Class 1, (B) Class 2, (C) Class 3, (D) Class 4.

Sex and Grade Differences in Trajectory Class Membership

The chi-square test was used to examine whether there were significant sex and grade differences in class membership. For sex, the results of chi-square analysis were significant (χ² = 53.04, df = 3, p < .001). We further conducted post hoc analyses using standardized adjusted residual scores. The results revealed that boys were more likely than girls to be in the involvement trajectory group (z = 4.3), the desisted trajectory group (z = 2.9), and the victimization trajectory group (z = 4.0). However, girls were more likely than boys to be in the noninvolved trajectory group (z = 7.2). For grade, chi-square analysis was nonsignificant (χ² = 7.46, df = 6, p = .281).

Forms of Bullying and Victimization Over Time

Finally, means were used to describe how different forms of bullying/victimization change over time in different subgroups (Figure 2). In general, for the three classes involving bullying or victimization (Classes 1, 2, and 3), three forms of bullying and victimization were consistent with the general developmental trend of bullying and victimization, showing an increasing trend, a decreasing trend, or stable levels. Each form of bullying/victimization exhibited similar levels at the same time points.

Figure 2.

Changes in forms of bullying/victimization over time by trajectory class: (A) Class 1 bullying, (B) Class 1 victimization, (C) Class 2 bullying, (D) Class 2 victimization, (E) Class 3 bullying, (F) Class 3 victimization, (G) Class 4 bullying, (H) Class 4 victimization.

Discussion

School bullying has attracted wide attention in the fields of psychology, pedagogy, criminology, and public health (Piquero et al., 2013). However, the joint development process of bullying and victimization is not yet clear. This study is the first to use a parallel-process LCGA approach to better understand the developmental relationship between bullying and victimization. In a longitudinal study involving four annual waves of data, we found four distinct joint development trajectories of bullying and victimization, revealing a new horizon for understanding the developmental process of the two. In addition, we also investigated the changes in bullying/victimization forms in each class over time and found that all forms of bullying/victimization were consistent with the overall trend and showed similar levels.

Joint Developmental Trajectories of Bullying and Victimization

This study found four different joint developmental trajectories of bullying and victimization: involvement trajectory group (7.6%), desisted trajectory group (6.1%), victimization trajectory group (13.2%), and noninvolved trajectory group (73.1%).

We found a group with both bullying and victimization increased over time (involvement trajectory group, 7.6%). A similar group was found in Barker et al. (2008) but not in the other two studies (Cho & Lee, 2019; Haltigan & Vaillancourt, 2014). Why did bullying and victimization increase at almost the same time? Aggression, on one hand, can exacerbate potential bullying perpetration, and those children who attack others may become victims (Brendgen et al., 2016). On the other hand, aggressive children can create a hostile social atmosphere, which leads to poor peer relationships and subsequent peer victimization experiences (Ettekal & Ladd, 2017). In turn, victimization experiences promote patterns of hostile attribution and aggressive responses, which further exacerbate and sustain children’s aggression (Ettekal & Ladd, 2017). As a result, bullying and victimization reinforce each other over time.

We found a group characterized by higher initial levels of bullying and victimization, both of which decreased over time (desisted trajectory group, 6.1%). Similar joint trajectories were found in Cho and Lee (2019) but not in the other two studies (Barker et al., 2008; Haltigan & Vaillancourt, 2014). Decreasing trajectories of bullying/victimization were common in previous studies (Brendgen et al., 2016; Espelage et al., 2018; Goldbaum et al., 2003; Pabian & Vandebosch, 2016). In fact, patterns of desistance were ubiquitous. That is, many children initially at risk can escape from bullying and victimization over time (Barker & Maughan, 2009; Burt, 2012). Expanding the view of the interactional continuity perspective, a reduction in aggression co-occurs with a reduction in victimization. Therefore, once children show less aggressive behavior, their risk of being bullied decreases accordingly, and then they become part of the noninvolvement group (Ettekal & Ladd, 2017).

Unfortunately, our findings showed that some children were persistently victims from childhood to adolescence (victimization trajectory group, 13.2%). This trajectory class was common in previous studies (Biggs et al., 2010; Cho & Lee, 2019; Goldbaum et al., 2003), indicating that victimization shows a stable trend (Haltigan & Vaillancourt, 2018; Salmivalli, 2018; Zych et al., 2020). For example, Zych et al. (2020) examined the transition of different bullying roles among 916 children from age 11 to 17. Results indicated that the role of victimization had high stability. Most of the victims at age 11 remained victims until age 17. Compared with bully-victims (children in the involvement trajectory group and desisted trajectory group), greater stability in victimhood is observed perhaps because once a child becomes a victim, his or her reputation as a victim will be established, making it difficult for the child to establish new friendships (Sugimura et al., 2017). Furthermore, it is difficult for such children to establish a peer support system that may help them escape from being bullied. Therefore, being a victim is stable in the long term. Why did victims not become bullying perpetrators or bullying victims? One possible reason is that the children who exhibit passive-submissive and withdrawal behavior styles often become targets of bullying. Their victimization experiences in turn strengthen their passive-submissive and withdrawal behavior styles, thus promoting continuity in victimization (Ettekal & Ladd, 2017).

Fortunately, we found that most children in our sample belonged to the noninvolved trajectory group (73.1%). This finding was consistent with previous studies, which indicated that most children were not involved in any bullying or victimization for a long time (Brendgen et al., 2016; Cho & Lee, 2019; Ettekal & Ladd, 2017). However, this result does not support our hypothesis that a larger proportion of children will not be involved in bullying and victimization because of Chinese collectivist culture. The proportion (73.1%) we found is close to or even lower than that of Western samples (75% in Barker et al., 2008; 73% in Haltigan and Vaillancourt, 2014). One possible reason for these findings is that our sample is in the transition period from primary school to junior high school, which relates to the higher levels of bullying and victimization (Schäfer et al., 2005).

Consistent with some studies (Cho & Lee, 2019; Davis et al., 2020) but not with other studies (Barker et al., 2008; Haltigan & Vaillancourt, 2014; Zych et al., 2020), we did not find a trajectory of a “pure” bullying class. On one hand, this finding was consistent with the possible theoretical groups elaborated by the interactional continuity model: children who were bully perpetrators and became bullying victims slowly, and those who were victims initially and remained victims in all periods (for more details, see Ettekal & Ladd, 2017). That is, over time, pure bully perpetrators develop multiple identities and are no longer pure bully perpetrators. For example, research has found that pure bully perpetrators only exist among children in fourth grade of primary school and then transition into other roles (Williford et al., 2011). This theoretical viewpoint of interactional continuity also explains why our study found that bullying is always accompanied by victimization but that victimization is not necessarily accompanied by bullying. However, we need to be cautious about these results. Our finding does not mean that there are no pure bully perpetrators. In fact, previous research has found that pure bully perpetrators exist and have a certain stability (Zych et al., 2020). The reasons why this study and other studies have not detected pure bully perpetrators may include the following. First, some children who engage in bullying do not think that their behavior is bullying (Hill et al., 2017). Interviews and open surveys also found that children often had difficulty distinguishing among jokes, teasing, and bullying (Guerra et al., 2011). In their view, bullying behaviors are often merely part of a joke (Randa & Hayes, 2018). In addition, bully perpetrators often think highly of themselves, feel good about themselves, and underestimate the harm they cause to others (Juvonen & Graham, 2014). Moreover, social expectations may also impact the self-reporting of bullying (Hill et al., 2017). Finally, the consideration of generalizability may limit the identification of pure bully perpetrators. As mentioned before, in the framework of group-based modeling, each subgroup is defined to contain at least 5% of the sample to increase the generalizability (Nylund, Asparouhov, et al., 2007). However, according to the prior results, which are focused only on bullying, persistent perpetrators comprise a small group, and their prevalence is often lower than 10% (Pabian & Vandebosch, 2016; Pepler et al., 2008). Among them, when bully-victims are excluded, pure perpetrators tend to comprise less than 5% of the total sample, resulting in their being more rare and less representative. Prior research supports the viewpoint that the pure aggression subgroup comprises a small proportion of the total number of students, even lower than 1% (Orpinas et al., 2014).

Sex and Grade Differences in Trajectory Class Membership

Regarding sex differences, this study found that boys were more likely to be involved in bullying and victimization than girls. Specifically, boys were more involved than girls in the involvement trajectory group, desisted trajectory group and victimization trajectory group, whereas girls were more involved than boys in the noninvolved trajectory group. In general, our findings were consistent with most previous findings that boys were more involved than girls in bullying (Kim et al., 2009; Pepler et al., 2008), victimization (Geoffroy et al., 2018), and bully victimization (Nylund, Bellmore, et al., 2007). For example, in a sample of rural children in North America, both perpetration and victimization were more frequent in boys than in girls (Pellegrini & Long, 2002). However, inconsistent with the findings of our study, a small number of studies have found that girls report more victimization than boys (Zych et al., 2020). Considering that comparisons of sex differences need to be carried out in the same cultural context, we recommend more research to explore this issue. Consistent with previous studies (Barker et al., 2008), we did not find any significant grade difference in class membership, possibly because these children have similar cognitive levels and behavioral patterns.

Forms of Bullying and Victimization Over Time

In general, we found that in the three classes involved in bullying/victimization (Classes 1, 2, and 3), the three forms of bullying/victimization were either increasing or decreasing, maintaining a trend consistent with the overall trend. That is, children used all forms of bullying equally, and no single form of bullying dominated in a particular period of time. Our finding was inconsistent with the theoretical view of heterotypic continuity (Björkqvist, 1994; Björkqvist et al., 1992; Underwood et al., 2009) but consistent with previous meta-analysis and a large number of empirical studies (Bellmore & Cillessen, 2006; Card et al., 2008) suggesting a moderate to high correlation between different forms of bullying/victimization. Our finding was also consistent with the view that bullying should be classified according to severity rather than form (Bettencourt et al., 2013; Haltigan & Vaillancourt, 2018; Nylund, Bellmore, et al., 2007; Williford et al., 2011). A possible explanation is that physical forms of bullying can generate not only punishments but also rewards, such as favorable social feedback (Juvonen & Graham, 2014). Therefore, older children still use physical forms of bullying as they grow older.

This does not mean that our findings were contrary to those of previous theoretical and empirical studies indicating that relational bullying increases and physical bullying decreases with age (Björkqvist, 1994; Björkqvist et al., 1992; Underwood et al., 2009). The age of our sample did not cover late adolescence. Physical bullying may become less acceptable until late adolescence, but it is relatively common in early and mid-adolescence. However, previous studies have also found more physical and verbal bullying than theoretical relational bullying in late adolescence (Ettekal & Ladd, 2017; Giang & Graham, 2008). More research extending to a longer life span is needed to investigate how forms of bullying and victimization change over time.

In addition, our results also need to be interpreted in the context of Chinese culture. We found that among Chinese rural children, relational aggression remains at almost the same level as physical and verbal aggression. Our results supported the view that the influence of culture on indirect aggression is the same as that on direct aggression (Forbes et al., 2009). However, we cannot conclude that our results are inconsistent with the two views that Chinese children use more indirect aggression than direct aggression (Lansford et al., 2012) or that they use more direct aggression than indirect aggression (Zhang et al., 2001), as only means were used to describe the developmental trend of the three forms of bullying/victimization over time in each subgroup, without comparisons of the differences among them because of the small sample size in each subgroup. However, our findings provide a perspective for this issue. Namely, the three forms of bullying/victimization remain at similar levels.

Strengths and Limitations

Our study has some strengths. This study had a relatively large sample size and 3 years of follow-up. In addition, we adopted a promising new method to understand the developmental process of bullying and victimization, filling a gap in knowledge about the number and shape of the joint trajectories of bullying and victimization in one model.

Despite the above strengths, our study also has several limitations. First, the data in this study were self-reported by children, which may lead to systematic underreporting or overreporting (Jetelina et al., 2019). Future research should use multiple measurement methods, such as a combination of questionnaires, interviews, and observations (Berger, 2007). However, self-reporting is still the main method for measuring bullying and victimization. Victims in particular are the most aware of the harm they have suffered (Berger, 2007). Second, although our research spanned 4 years, it was not long enough to determine when children “enter in” or “escape from” bullying/victimization. Although this study provided some evidence, future research should span a longer period of development, including preschool, late adolescence, and even adulthood. Third, our research did not include cyberbullying. In fact, with the increasing availability of internet, children and adolescents may use cyberbullying, which is closely related to traditional bullying (Raskauskas & Stoltz, 2007), and it can be as harmful or even more harmful than traditional bullying. It is very important for future research to include cyberbullying. Fourth, the study sample was not sufficiently representative of the population. The children in our study were from rural areas, and the sample consisted of more boys than girls. This may be related to the traditional preference for boys in many rural areas of China; thus, boys are more likely than girls to obtain better educational resources. Therefore, the conclusions obtained from this sample cannot be extended to urban samples. Future research should include both urban and rural samples. Finally, given the high attrition rate (more than 50%), the data for the present study are based on the 775 children who reported bullying and victimization on four measurements, rather than the 1,587 children who participated in the initial survey. Although we conducted supplementary analysis using the data of the 1,587 children and found that the results were similar to those of the data of the 775 children, there may be problems with the excluded cases, which should be avoided in future research.

Implications for Practice

This study provides some insight for anti-bullying intervention projects. First, the joint development process of bullying and victimization should be fully considered. Particularly for those involved in both bullying and victimization, it may be more effective to intervene in both. As bullying and victimization can reinforce each other over time, if we intervene only in victimization, a child may be involved in victimization again because of bullying others.

Second, individual differences should be fully considered. Tailored anti-bullying interventions for different individuals may be more effective. However, we cannot deny the role of general intervention programs in creating a good school climate (Juvonen & Graham, 2014). It is more effective to combine general interventions for all groups with tailored interventions for specific groups.

Third, boys should be given special attention in anti-bullying interventions, especially victimization. In Chinese culture, girls are considered to be vulnerable groups and targets of bullying, whereas victimization among boys is often ignored. The study reminds us of the need to pay special attention to bullying among boys. Families and educational institutions need to take more action, such as encouraging children, especially boys, to seek help and providing them with counseling and safe havens (Leadbeater et al., 2014).

Finally, we found that three forms of bullying/victimization remain similar at all time points, which provides a perspective for detecting bullying/victimization that is not accessible to direct observation. Physical bullying/victimization can be easily observed, whereas other forms are difficult to find (Zych et al., 2020). Hence, when clinicians or teachers find that a child is being bullied in physical ways, they should ascertain whether the child was experiencing other forms of bullying.

Footnotes

Acknowledgements

We thank the school and the children for their participation.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this study was provided by the programs of Pioneer Initiative of the Chinese Academy of Sciences, Feature Institutes Program (TSS-2015-06), and Consulting and Appraising Project of Chinese Academy of Sciences (Y7CX134003).

ORCID iDs

Yueyue Zhou

Yiming Liang

Author Biographies

Yueyue Zhou is a doctoral candidate at the Institute of Psychology, Chinese Academy of Sciences. Her research focuses on multiple aspects of child and adolescent development, especially those in adversity. She is particularly interested in the relationship between the quality of early parent–child interaction and development outcomes in adolescence.

Hao Zheng is a master’s candidate in the applied psychology program at the Institute of Psychology, Chinese Academy of Sciences. His research focuses on the developmental outcomes of parent–child attachment relationships. In particular, he concentrates on the effects of attachment-based intervention programs for left-behind children in China.

Yiming Liang is a doctoral candidate in the clinical psychology program at the Institute of Psychology, Chinese Academy of Sciences. His research focuses on the development of children and adolescents in the face of adversity. He is particularly interested in the long-term development and diagnosis of post-traumatic stress disorder (PTSD) among children after disasters.

Jiazhou Wang is a master’s candidate in the applied psychology program at the Institute of Psychology, Chinese Academy of Sciences. Her research focuses on the growth and development of children and adolescents, including attachment and psychological trauma.

Ru Han, PhD, is an assistant professor at the Institute of Psychology, Chinese Academy of Sciences. Her research focuses on the social behavior and development of disadvantaged children.

Zhengkui Liu, PhD, is a professor at the Institute of Psychology, Chinese Academy of Sciences. His research focuses on major traumatic events and mental health. He and his team are devoted to developing effective and sustainable psychological assistance systems and establishing a model and tool for evaluating psychological trauma after major stress events suitable for the Chinese population.

References

Archer

Coyne

S. M.

(2005). An integrated review of indirect, relational, and social aggression. Personality and Social Psychology Review, 9, 212–230.

Barker

E. D.

Arseneault

Brendgen

Fontaine

Maughan

(2008). Joint development of bullying and victimization in adolescence: Relations to delinquency and self-harm. Journal of the American Academy of Child and Adolescent Psychiatry, 47, 1030–1038.

Barker

E. D.

Maughan

(2009). Differentiating early-onset persistent versus childhood-limited conduct problem youth. American Journal of Psychiatry, 166, 900–908.

Bellmore

A. D.

Cillessen

A. H.

(2006). Reciprocal influences of victimization, perceived social preference, and self-concept in adolescence. Self and Identity, 5, 209–229.

Berger

K. S.

(2007). Update on bullying at school: Science forgotten? Developmental Review, 27, 90–126.

Bettencourt

Farrell

Liu

Sullivan

(2013). Stability and change in patterns of peer victimization and aggression during adolescence. Journal of Clinical Child & Adolescent Psychology, 42, 429–441.

Betts

K. S.

Williams

G. M.

Najman

J. M.

Alati

(2014). Maternal depressive, anxious, and stress symptoms during pregnancy predict internalizing problems in adolescence. Depression and Anxiety, 31, 9–18.

Biggs

B. K.

Vernberg

Little

T. D.

Dill

E. J.

Fonagy

Twemlow

S. W.

(2010). Peer victimization trajectories and their association with children’s affect in late elementary school. International Journal of Behavioral Development, 34, 136–146.

Björkqvist

(1994). Sex differences in physical, verbal, and indirect aggression: A review of recent research. Sex Roles, 30, 177–188.

10.

Björkqvist

Lagerspetz

K. M.

Kaukiainen

(1992). Do girls manipulate and boys fight? Developmental trends in regard to direct and indirect aggression. Aggressive Behavior, 18, 117–127.

11.

Brendgen

Girard

Vitaro

Dionne

Boivin

(2016). Personal and familial predictors of peer victimization trajectories from primary to secondary school. Developmental Psychology, 52, 1103–1114.

12.

Burt

S. A.

(2012). How do we optimally conceptualize the heterogeneity within antisocial behavior? An argument for aggressive versus non-aggressive behavioral dimensions. Clinical Psychology Review, 32, 263–279.

13.

Card

N. A.

Stucky

B. D.

Sawalani

G. M.

Little

T. D.

(2008). Direct and indirect aggression during childhood and adolescence: A meta-analytic review of gender differences, intercorrelations, and relations to maladjustment. Child Development, 79, 1185–1229.

14.

Chen

Zhang

Deaterdeckard

(2019). Developmental trajectories of Chinese adolescents’ relational aggression: Associations with changes in social-psychological adjustment. Child Development, 90, 2153–2170.

15.

Cho

Lee

J. R.

(2018). Impacts of low self-control and delinquent peer associations on bullying growth trajectories among Korean youth: A latent growth mixture modeling approach. Journal of Interpersonal Violence. Advance online publication. https://doi.org/10.1177/0886260518786495

16.

Cho

Lee

J. R.

(2019). Joint growth trajectories of bullying perpetration and victimization among Korean adolescents: Estimating a second-order growth mixture model-factor-of-curves with low self-control and opportunity correlates. Crime & Delinquency. Advance online publication. https://doi.org/10.1177/0011128719890271

17.

Cruz

R. A.

King

K. M.

Cauce

A. M.

Conger

R. D.

Robins

R. W.

(2017). Cultural orientation trajectories and substance use: Findings from a longitudinal study of Mexican-origin youth. Child Development, 88, 555–572.

18.

Davis

J. P.

Ingram

K. M.

Merrin

G. J.

Espelage

D. L.

(2020). Exposure to parental and community violence and the relationship to bullying perpetration and victimization among early adolescents: A parallel process growth mixture latent transition analysis. Scandinavian Journal of Psychology, 61, 77–89.

19.

Enders

C. K.

(2010). Applied missing data analysis. Guilford Press.

20.

Espelage

D. L.

Van Ryzin

M. J.

Holt

M. K.

(2018). Trajectories of bully perpetration across early adolescence: Static risk factors, dynamic covariates, and longitudinal outcomes. Psychology of Violence, 8, 141–150.

21.

Ettekal

Ladd

G. W.

(2017). Developmental continuity and change in physical, verbal, and relational aggression and peer victimization from childhood to adolescence. Developmental Psychology, 53, 1709–1721.

22.

Forbes

Zhang

Doroszewicz

Haas

(2009). Relationships between individualism-collectivism, gender, and direct or indirect aggression: A study in China, Poland, and the US. Aggressive Behavior, 35, 24–30.

23.

Forbes

G. B.

Collinsworth

L. L.

Zhao

Kohlman

LeClaire

(2011). Relationships among individualism-collectivism, gender, and ingroup/outgroup status, and responses to conflict: A study in China and the United States. Aggressive Behavior, 37, 302–314.

24.

Geoffroy

M. C.

Boivin

Arseneault

Renaud

Perret

L. C.

Turecki

. . .Tremblay

R. E.

(2018). Childhood trajectories of peer victimization and prediction of mental health outcomes in midadolescence: A longitudinal population-based study. Canadian Medical Association Journal, 190, E37–E43.

25.

Giang

M. T.

Graham

(2008). Using latent class analysis to identify aggressors and victims of peer harassment. Aggressive Behavior, 34, 203–213.

26.

Goldbaum

Craig

W. M.

Pepler

Connolly

(2003). Developmental trajectories of victimization: Identifying risk and protective factors. Journal of Applied School Psychology, 19, 139–156.

27.

Guerra

N. G.

Williams

K. R.

Sadek

(2011). Understanding bullying and victimization during childhood and adolescence: A mixed methods study. Child Development, 82, 295–310.

28.

Haltigan

J. D.

Vaillancourt

(2014). Joint trajectories of bullying and peer victimization across elementary and middle school and associations with symptoms of psychopathology. Developmental Psychology, 50, 2426–2436.

29.

Haltigan

J. D.

Vaillancourt

(2018). The influence of static and dynamic intrapersonal factors on longitudinal patterns of peer victimization through mid-adolescence: A latent transition analysis. Journal of Abnormal Child Psychology, 46, 11–26.

30.

Hill

R. M.

Mellick

Temple

J. R.

Sharp

(2017). The role of bullying in depressive symptoms from adolescence to emerging adulthood: A growth mixture model. Journal of Affective Disorders, 207, 1–8.

31.

Hysing

Askeland

K. G.

La Greca

A. M.

Solberg

M. E.

Breivik

Sivertsen

(2019). Bullying involvement in adolescence: Implications for sleep, mental health, and academic outcomes. Journal of Interpersonal Violence. Advance online publication. https://doi.org/10.1177/0886260519853409

32.

Isernhagen

Harris

(2004). A comparison of bullying in four rural middle and high schools. Rural Educator, 25, 5–13.

33.

Jetelina

K. K.

Gonzalez

J. M. R.

Cuccaro

P. M.

Peskin

M. F.

Pompeii

Atem

. . . Schuster

M. A.

(2019). Mechanisms and frequency of violent injuries among victims and perpetrators of bullying. Journal of Adolescent Health, 64, 664–670.

34.

Zhang

Jones

Smith

(2004). Gender differences of physical, verbal, and indirect bullying among primary and secondary school children: A cross-cultural comparison between China and Britain. Journal of Shandong Normal University (Humanities and Social Sciences), 49, 21–24.

35.

Juvonen

Graham

(2014). Bullying in schools: The power of bullies and the plight of victims. Annual Review of Psychology, 65, 159–185.

36.

Kawabata

Tseng

W. L.

Murray-Close

Crick

N. R.

(2012). Developmental trajectories of Chinese children’s relational and physical aggression: Associations with social-psychological adjustment problems. Journal of Abnormal Child Psychology, 40, 1087–1097.

37.

Kim

Y. S.

Boyce

W. T.

Koh

Y. J.

Leventhal

B. L.

(2009). Time trends, trajectories, and demographic predictors of bullying: A prospective study in Korean adolescents. Journal of Adolescent Health, 45, 360–367.

38.

Lansford

J. E.

Skinner

A. T.

Sorbring

Giunta

L. D.

Deater-Deckard

Dodge

K. A.

. . . Uribe Tirado

L. M.

(2012). Boys’ and girls’ relational and physical aggression in nine countries. Aggressive Behavior, 38, 298–308.

39.

Leadbeater

B. J.

Thompson

Sukhawathanakul

(2014). It gets better or does it? Peer victimization and internalizing problems in the transition to young adulthood. Development and Psychopathology, 26, 675–688.

40.

Modecki

K. L.

Minchin

Harbaugh

A. G.

Guerra

N. G.

Runions

K. C.

(2014). Bullying prevalence across contexts: A meta-analysis measuring cyber and traditional bullying. Journal of Adolescent Health, 55, 602–611.

41.

Muthén

L. K.

Muthén

B. O.

(1998–2015). Mplus user’s guide (7th ed.).

42.

Nylund

Bellmore

Nishina

Graham

(2007). Subtypes, severity, and structural stability of peer victimization: What does latent class analysis say? Child Development, 78, 1706–1722.

43.

Nylund

K. L.

Asparouhov

Muthen

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535–569.

44.

Olweus

(1996). The Revised Olweus Bully/Victim Questionnaire. Research Center for Health Promotion (HEMIL Center), University of Bergen.

45.

Orpinas

Mcnicholas

Nahapetyan

(2014). Gender differences in trajectories of relational aggression perpetration and victimization from middle to high school. Aggressive Behavior, 41, 401–412.

46.

Pabian

Vandebosch

(2016). Developmental trajectories of (cyber) bullying perpetration and social intelligence during early adolescence. Journal of Early Adolescence, 36, 145–170.

47.

Pellegrini

A. D.

Long

J. D.

(2002). A longitudinal study of bullying, dominance, and victimization during the transition from primary to secondary school. British Journal of Developmental Psychology, 20, 259–280.

48.

Pepler

Jiang

Craig

Connolly

(2008). Developmental trajectories of bullying and associated factors. Child Development, 79, 325–338.

49.

Piquero

A. R.

Connell

N. M.

Piquero

N. L.

Farrington

D. P.

Jennings

W. G.

(2013). Does adolescent bullying distinguish between male offending trajectories in late middle age? Journal of Youth and Adolescence, 42, 444–453.

50.

Randa

Hayes

B. E.

(2018). Addressing onset and desistance of bullying behavior: Surveying perpetrators. Violence and Gender, 5, 93–102.

51.

Raskauskas

Stoltz

A. D.

(2007). Involvement in traditional and electronic bullying among adolescents. Developmental Psychology, 43, 564–575.

52.

Reijntjes

Vermande

Goossens

F. A.

Olthof

van de Schoot

Aleva

van der Meulen

(2013). Developmental trajectories of bullying and social dominance in youth. Child Abuse & Neglect, 37, 224–234.

53.

Salmivalli

(2018). Peer victimization and adjustment in young adulthood: Commentary on the special section. Journal of Abnormal Child Psychology, 46, 67–72.

54.

Schäfer

Korn

Brodbeck

F. C.

Wolke

Schulz

(2005). Bullying roles in changing contexts: The stability of victim and bully roles from primary to secondary school. International Journal of Behavioral Development, 29, 323–335.

55.

Sugimura

Berry

Troop-Gordon

Rudolph

K. D.

(2017). Early social behaviors and the trajectory of peer victimization across the school years. Developmental Psychology, 53, 1447–1461.

56.

Sumter

S. R.

Baumgartner

S. E.

Valkenburg

P. M.

Peter

(2012). Developmental trajectories of peer victimization: Off-line and online experiences during adolescence. Journal of Adolescent Health, 50, 607–613.

57.

Underwood

M. K.

Beron

K. J.

Rosen

L. H.

(2009). Continuity and change in social and physical aggression from middle childhood through early adolescence. Aggressive Behavior, 35, 357–375.

58.

Vaillancourt

Miller

J. L.

Fagbemi

Côté

Tremblay

R. E.

(2007). Trajectories and predictors of indirect aggression: Results from a nationally representative longitudinal study of Canadian children aged 2–10. Aggressive Behavior, 33, 314–326.

59.

Williford

A. P.

Brisson

Bender

K. A.

Jenson

J. M.

Forrest-Bank

(2011). Patterns of aggressive behavior and peer victimization from childhood to early adolescence: A latent class analysis. Journal of Youth and Adolescence, 40, 644–655.

60.

Zhang

Wang

Lin

(2001). Types of bullying behavior and its correlates. Psychological Development and Education, 17, 12–17.

61.

Zhang

Jones

(1999). Revision of Chinese version of Olweus child bullying questionnaire. Psychological Development and Education, 15, 7–11.

62.

Zych

Ttofi

M. M.

Llorent

V. J.

Farrington

D. P.

Ribeaud

Eisner

M. P.

(2020). A longitudinal study on stability and transitions among bullying roles. Child Development, 91, 527–545.