Covitality moderates the relationship between victimisation and loneliness

ABSTRACT

Objective: Research suggests that low levels of school belonging and high levels of victimisation predict negative emotions, including loneliness. However, few studies have examined this relation among Chinese elementary school students. The protective role of covitality against victimisation and loneliness also remains unexplored. This study examines the relations between bullying victimisation, school belonging, covitality, and loneliness over 6 months, and whether covitality moderates the relations between victimisation and loneliness.

Method: Eight hundred students from five elementary schools in China completed self‐report surveys at two time points (6 months apart).

Results: Bullying victimisation, school belonging, and covitality predicted loneliness 6 months later. Students who experienced more bullying victimisation, lower levels of school belonging, and lower covitality reported more loneliness 6 months later. Covitality buffered the relation between verbal victimisation at Time 1 (T1) and loneliness at T1 but did not buffer the relation between victimisation T1 and loneliness at Time 2 (T2).

Conclusions: Schools should prevent bullying, foster school belonging, and promote covitality (positive psychological traits) to reduce Chinese youths' feelings of loneliness.

KEY POINTS

What is already known about this topic:

• (1) Victimisation is a risk factor for loneliness for children in the United States.

• (2) School belonging is a protective factor for loneliness for children in United States.

• (3) Covitality is a protective factor for youth.

What this topic adds:

• (1) Victimisation is a risk factor for loneliness for children in China.

• (2) School belonging and covitality are protective factors against loneliness for children in China.

• (3) Covitality moderates the relation between verbal victimisation and loneliness using cross‐sectional data only.

KEYWORDS:

• Bullying victimisation
• Chinese elementary school
• covitality
• loneliness
• school belonging

1. Covitality moderates the relationship between victimisation and loneliness

During middle childhood, the school context provides unique and rich opportunities for youth to develop foundational social relationships that will influence their sense of self and relatedness to the world. However, not all youth experience positive social connections with peers during these formative years. Peer victimisation tends to increase during middle childhood in the United States and is associated with the subjective and interpersonal experience of loneliness defined as sad or aching feelings of emotional and social isolation (MacEvoy, Weeks, & Asher, 2011). Despite loneliness being a growing concern among children and adolescents (Campbell, 2013; Perlman & Landolt, 1999), few researchers have explored risk and protective factors for loneliness in an Eastern collectivist context like China, where belonging to the social group is highly valued, and the social and relational factors involved in the development of loneliness may differ (Chen et al., 2004). Additional limitations in the literature include the lack of longitudinal research that examines students' experiences in school and their subsequent well‐being and growth (Tian, Zhang, Huebner, Zheng, & Liu, 2016) and the neglect of younger students (i.e., elementary school age) compared to adolescents. It is important to examine victimisation and loneliness among Chinese elementary school students because victimisation increases around middle childhood to early adolescence (Acquah, Topalli, Wilson, Junttila, & Niemi, 2016), and the majority of research on bullying and loneliness has been conducted in Western countries with limited generalizability to other cultural groups. Peer victimisation has become a growing concern possibly due to rapid changes occurring in Chinese society, such as globalisation and changes in family structures (Huang, Hong, & Espelage, 2013). As a result, more research is needed to identify the protective factors (e.g., individual characteristics/assets) that may buffer against loneliness. The goal of this current study is to examine (a) relations between bullying victimisation, school belonging, and loneliness among elementary school students over time, and (b) whether covitality (the co‐occurring and combinatorial effects of multiple positive psychological traits together; Furlong, You, Renshaw, O'Malley, & Rebelez, 2013) moderates the relations between victimisation and loneliness. As a highly social species, humans are intrinsically motivated to associate with others and be socially accepted. Understanding these relationships can help researchers and practitioners develop culture‐sensitive/specific programming for diverse students who experience loneliness.

1.1. Theoretical framework

This current study is grounded using Bronfenbrenner's Process-Person-Context-Time (PPCT) model (Bronfenbrenner, Morris, Lerner, & Damon, 2006), which accentuates the interplay of four components: (a) interactions of proximal environment often called proximal processes, (b) the characteristics of the individual, (c) social context of the person, and (d) change over time. This PPCT framework allows researchers to parse out the roles of both environmental/contextual characteristics and individual factors on developmental processes and outcomes. In our study, school belonging and peer victimisation are proximal process variables given the significant amount of time that students spend in school contexts with peers. We also explore personal characteristics, such as covitality, and whether covitality may interact with peer victimisation to predict youth outcomes. Finally, we situate the study in the larger social context of China and utilise a longitudinal design to examine the change of loneliness over time.

1.2. Loneliness

Loneliness reflects a subjective experience wherein an individual recognises their deficiency in social and personal relationships and experience subsequent negative feelings of sadness, emptiness, or longing (Asher & Paquette, 2003, p. 75). It is a relatively ubiquitous phenomenon. Based on the Australian Loneliness Report published by the Australian Psychological Association and Swinburne University of Technology (2018), about half (50.5%) of Australian adults felt lonely at least 1 day a week. Loneliness is also a problem among children and adolescents. Up to 10–20% of youth report that they have experienced continuous or recurrent loneliness (Perlman & Landolt, 1999). In Hong Kong, adolescents also reported intermediate levels of loneliness (average score of 18.7 on a scale with 36 as the maximum) and self‐reports of loneliness negatively correlated with self‐esteem, self‐efficacy, and life‐satisfaction (To, 2016).

1.3. School belonging

School belonging can be a protective factor against loneliness for students. This construct represents students’ feelings of respect, inclusion, and support within the school environment (Goodenow, 1993, p. 80) and encompasses three factors: caring relations or perceptions of caring adult relationships in school, acceptance or students' sense of acceptance or belongingness in school, and rejection which corresponds to perceptions of disrespect or rejection in school (You, Ritchey, Furlong, Shochet, & Boman, 2011). School belonging has been linked to positive psychosocial and educational outcomes, such as less bullying victimisation, better academic achievement (Fredricks, Blumenfeld, & Paris, 2004; Goodenow & Grady, 1993), and subjective well‐being (Tian et al., 2016). When individuals are satisfied with their social networks they experience less loneliness (Nazzal, Cruz, & Neto, 2017). In contrast, if a youth's needs for school belonging are not met, they become more vulnerable to negative affect such as feelings of loneliness (Shochet, Smith, Furlong, & Homel, 2011), depression (Millings, Buck, Montgomery, Spears, & Stallard, 2012; Osterman, 2000), and school drop‐out (Lee & Breen, 2007).

1.4. Victimisation and loneliness

Bullying or peer victimisation is characterised by aggressive physical (e.g., assault, destruction of property), verbal (e.g., taunts, threats, name‐calling), and social/relational (e.g., peer exclusion, gossiping, spreading rumours) behaviours (Nansel et al., 2001). The relationship between loneliness and victimisation may be bi‐directional (Campbell, 2013). On one hand, the experience of victimisation may contribute to negative feelings such as loneliness. On the other hand, loneliness may also lead to victimisation because loneliness is associated with negative feelings, and the expression of negative feelings may lead to negative peer interactions and victimisation (Liu et al., 2017). Individuals with perceived interpersonal difficulties and internalising symptoms may be especially vulnerable to prospective bullies who assume these victims will not retaliate given their weak social networks (Acquah et al., 2016).

However, few studies have examined this reciprocal relation between victimisation and loneliness, and none of the studies focused on younger Chinese students. Previous research showed that adolescents who experience a lack of peer social networks or affiliations (i.e., social loneliness) are more likely to be victimised at school 6 months later (Acquah et al., 2016). Among younger children, researchers suggested that peer victimisation precedes loneliness. One study showed that peer victimisation in kindergarten was predictive of changes in loneliness 4–6-months later but early reports of loneliness did not predict changes in the frequency of peer victimisation (Kochenderfer & Ladd, 1996). Children who moved from non‐victim to victim classification between kindergarten and third grade had increasing levels of loneliness and decreasing social satisfaction; however, when the victimisation experience stopped, there was no significant improvement in loneliness or social satisfaction (Kochenderfer‐Ladd & Wardrop, 2001). Given the inconsistency in previous research regarding the directionality of effects among victimisation and loneliness, we seek to examine this relation using longitudinal data among Chinese students.

1.5. Covitality

In recent years, an increased interest in resilience and positive psychology has led researchers to explore the co‐occurrence of positive psychological traits such as covitality. Covitality is the “synergistic effect of positive mental health resulting from the interplay among multiple positive psychological building blocks” (Renshaw et al., 2014, p. 14). Such positive building blocks include gratitude, zest, optimism, and persistence (Furlong et al., 2013). Furthermore, these psychological dispositions are indicators of a student's positive mental health. For example, covitality (measured by Social Emotional Health Survey‐Primary) has been significantly and positively associated with positive youth outcomes such as prosocial behaviour, feeling of school safety, and school belonging (Furlong et al., 2013). Also, Lenzi, Dougherty, Furlong, Sharkey, and Dowdy (2015) found that having more positive psychological traits (or greater number of psychological assets across different domains as measured by the Social Emotional Health Survey‐Secondary) were protective against behavioural and emotional difficulties (such as depressive symptoms) among youth. Among Chinese elementary school students, research showed that covitality was associated with fewer depressive symptoms, higher prosocial behaviours, and better academic achievement (Wang, Yang, Jiang, & Furlong, 2018). Therefore, covitality may serve as a protective factor for loneliness. In addition, researchers suggested that covitality may buffer against the negative effect of victimisation (Smith, Campain, & Stuck, 2013). However, based on our knowledge, no study has examined this moderation effect.

1.6. Current study

This study examines the following research questions:

1. Does school belonging and victimisation predict loneliness over time (6 months later)? We hypothesised that higher levels of school belonging and lower levels of victimisation (overall victimisation, as well as physical, verbal, and relational victimisation) would predict less loneliness over time.

2. Does covitality predict less loneliness over time above and beyond school belonging and victimisation? We hypothesised that covitality would predict less loneliness over time.

3. Does loneliness predict greater levels of victimisation over time? We also hypothesised that loneliness would predict greater victimisation over time.

4. Does covitality moderate the relation between victimisation and loneliness? We hypothesised that covitality would serve as a buffer and moderate the relation between victimisation and loneliness because students with high covitality have more positive psychological traits/dispositions and may possess more inner resources to cope with bullying. As a result, victimisation may lead to less negative outcomes among students with higher covitality compared to students with lower covitality.

2. Method

2.1. Participants

Survey data were collected from third‐ to sixth‐grade students from five elementary schools in Sichuan, China over two time points (November 2017 and May 2018). Students who only completed survey at T1 but not T2 were excluded in the analyses. The final data set included 800 students (56.2% boys; M_age = 10.5-years, SD = 1.21-years; range 8 to 14-years). Based on students' reports of family demographics, there was substantial variation in parents' level of education (fathers' education: 5.9% elementary or below, 15.1% middle school education, 22.2% high school education, 45.7% college graduates, and 11.1% with advanced degrees; mothers' education: 5.1% elementary or below, 15.5% middle school education, 24.4% high school education, 44.9% college graduates, and 10% with advanced degrees). Most fathers (82.0%) and mothers (75.3%) were employed.

2.2. Measures

Students responded to demographic questions (age, gender, grade level, parents' highest level of education, and parents' employment status) and completed the following measures in Chinese.

2.2.1. School belonging

The Psychological Sense of School Membership Chinese Version (PSSM; Goodenow, 1993) was used to measure school belonging. Participants rated whether they agreed with 18 statements (e.g., “Do you feel like you are a real part of your school?”) using a 6‐point response scale: 1 = no, never, to 6 = completely true. We used the mean score of all items as an indicator for school belonging. Prior studies have shown high internal consistencies between .78 and .95 for the PSSM across samples from different countries including China (Chan, Yang, Furlong, Dowdy, & Xie, 2019; Wagle et al., 2018; You et al., 2011,). The internal consistency for the complete scale was .83 at T1.

2.2.2. Covitality

We used the Social Emotional Health Survey‐Primary‐Chinese Version (SEHS‐P; Furlong et al., 2013; Wang, Yang, et al., 2018; Wang et al., 2018) to measure covitality at T1. SEHS‐P included 21 items and four subscales: gratitude (e.g., “Do you feel thankful to go to your school?”), optimism (“Do you believe that you can make new friends at school?”), zest (“Do you get really excited about your schoolwork?”), and persistence (“Do you keep working until you get your schoolwork correct?”), all of which loaded onto the latent construct of covitality. Students responded to each item on a 6‐point scale, ranging from 1 (no, never) to 6 (yes, always). We used the mean score of all 21 items as the indicator of overall covitality in this study. SEHS‐P showed a high internal consistency of .85 to .94 in previous studies in China (Chan et al., 2019; Wang, Boyanton, et al., 2018). The internal consistency was .95 at T1.

Victimisation

The Delaware Bullying Victimisation Scale‐Student‐Chinese Version (DBVS‐S; Bear et al., 2016; Xie et al., 2016) was used to assess students' perceived bullying victimisation during the current school year. Specifically, we measured physical victimisation (e.g., “I was pushed or shoved on purpose,” four items), verbal victimisation (e.g., “A student said mean things to me,” four items), and social/relational victimisation (e.g., “A student told/got others to not like me,” four items). Responses ranged from 1 (never), 2 (less than once a month), 3 (once a month or more), 4 (once a week), 5 (several times a week), to 6 (every day). The Chinese version of DBVS‐S has demonstrated high internal consistency (.70 to .95) and good construct validity (Wang, Boyanton, et al., 2018; Xie et al., 2016). The internal consistency was .92 at T1 and .95 at T2 for the entire scale, and 76, .83, and .83for physical, verbal, and relational victimisation subscales at T1, and .85, .89, and .90. for physical, verbal, and relational victimisation subscales at T2, respectively.

2.2.3. Loneliness

We used the Me and My School Questionnaire (Deighton et al., 2013) to measure emotional difficulties (10 items) and behaviour difficulties (6 items). For this study, we used one item (“I feel lonely”) to measure loneliness at both T1 and T2. Students responded by choosing three options: “never,” “sometimes,” or “always.”

2.3. Procedures

Principals from five elementary schools in Sichuan Province, China agreed to participate in this research study. Parents of children in Grades 3–6 in these schools were notified about this study and were given the option to withdraw their children from participation. No parents withdrew their children from participation. Students completed a 20‐ to 30‐minute online survey during class in the school's computer lab. School personnel used a script prepared by the first author to explain to students that the information they provided in the survey was confidential and they could exit the survey at any time without penalty. School staff also explained to students that there were no right or wrong answers and encouraged students to respond to questions truthfully. Students indicated their willingness to participate in the project by signing the electronic assent form. If students skipped any questions in the online survey, an automatic reminder prompted the students to answer any missing questions on that page before they were able to move on to the next page. Students were still able to skip any questions if they chose to do so after seeing the reminder. There were few missing data. We used mean sores for all subscales in the analyses. We used this procedure for data collection at both time points. The Institutional Review Board at the first author's university approved this project. Students did not receive any compensation for their participation.

3. Results

3.1. Descriptive statistics

The repeated measures t‐test showed that the score of loneliness did not change from T1 (M = 1.44, SD = 0.60) to T2 (M = 1.47, SD = 0.60), t = −1.17, p = .24. Loneliness T1 significantly correlated with loneliness T2, r = .26, p < .001. Specifically, 303 (37.88%) students reported feeling lonely “sometimes” or “always” at T1 and 331 (41.38%) students reported feeling lonely “sometimes” or “always” at T2. The repeated measures t‐test showed that overall peer victimisation did not change significantly from T1 (M = 1.29, SD = 0.64) to T2 (M = 1.27, SD = 0.66), t = −0.53, p = .59. Overall peer victimisation T1 correlated with victimisation T2, r = .38, p < .001. Specifically, 34.1% (n = 273) of participants at T1 and 26.3% (n = 210) of participants at T2 reported experiencing at least one type of peer victimisation, at least once a month. Loneliness correlated with higher levels of victimisation, lower school belonging, and lower covitality, which is consistent with our hypotheses (see Table 1). We found gender was related to loneliness (girls reported more loneliness than boys), but grade level and parent education were unrelated to any variables of interest. As a result, we controlled for gender in all regression analyses. We did not control for grade level and parent education because they were not correlated with loneliness or any other variables of interest in this study.

Table 1. Mean, SD, and correlations for variables of interest

Variables	1	2	3	4	5	6	7	8
1. Overall Vic T1	—
2. Verbal Vic T1	.93**	—
3. Physical Vic T1	.90**	.76**	—
4. Relational Vic T1	.92**	.80**	.74**	—
5. BelongingT1	−.28**	−.21**	−.22**	−.26**	—
6. Covitality T1	−.29**	−.25**	−.27**	−.28**	.65**	—
7. Loneliness T1	.42**	.38**	.34**	.44**	−.35**	−.34**	—
8. Loneliness T2	.26**	.21**	.25**	.26**	−.16**	−.17**	.26**	—
Mean	1.29	1.81	1.59	1.70	4.49	4.07	1.44	1.47
SD	0.64	1.03	0.81	0.94	0.85	0.69	0.60	0.60
Minimum	0.76	1.00	1.00	1.00	1.72	1.00	1.00	1.00
Maximum	4.59	6.00	6.00	6.00	6.00	4.85	3.00	3.00
Gender	−.07	−.02	−.14**	−.002	−.02	−.01	.04	.08*
Grade level	.03	−.07	−.01	.02	.05	.002	.002	−.03
Fathers' education	−.02	−.05	−.001	−.002	.06	.05	−.01	.05
Mothers' education	−.03	−.06	−.01	−.01	.05	.06	−.01	.07

Notes: Three decimal places look more rigorous and precise but are not needed given the level of measurement error. Vic = Victimisation; Belonging = School Belonging.

*p < .05; **p < .01.

3.2. Main effects

We ran several linear regression models using loneliness at T2 as the dependent variable while controlling for gender to answer our first two research questions. First, to answer research question 1 (“Does school belonging and victimisation predict less loneliness over time?”), we used school belonging and overall peer victimisation at T1 as the predictors for loneliness at T2. We found that overall victimisation (β = .222, p < .001) and school belonging (β = −.077, p = .025) at T1 significantly predicted loneliness at T2 when not controlling for loneliness at T1, F(3, 745) = 19.39, R² = .072, p < .001 (Table 2, Model 1). Students who experienced more peer victimisation and reported lower levels of school belonging at T1 tended to experience more loneliness at T2. Second, we added loneliness at T1 in the regression model to control for prior level of loneliness. After controlling for loneliness at T1, school belonging was no longer a significant predictor for loneliness at T2, β = −.040, p = .235, but overall victimisation continued to be a significant predictor (β = .175, p < .001), F(4, 743) = 18.073, R² = .089, p < .001 (Table 2, Model 2).

Table 2. Victimisation, school belonging, and covitality at Time 1 predicting loneliness at time 2

	Loneliness T2 (Model 1)		Loneliness T2 (Model 2)		Loneliness T2 (Model 3)
	b (SE)	β	b (SE)	β	b (SE)	β
Victimisation T1	.01(.002)	.22***	.01(.002)	.18***	.01(.002)	.16***
School belonging T1	−.05(.03)	−.08*	−.03(.03)	−.04	.01(.03)	.02
Loneliness T1			.15(.04)	.14***	.14(.04)	.13**
Covitality T1					−.08(.04)	−.10+
Gender	.11(.04)	.09**	.10(.04)	.08*	.10(.04)	.08*
F(df)	19.39(3, 745)***		18.07(4, 743)***		14.19(5, 724)***
R^2	.07		.09		.09
ΔR^2			.02*		.01+

+p = .05; *p < .05; **p < .01; ***p < .001.

Third, to answer research question 2 (“Does covitality predict loneliness over time above and beyond school belonging and victimisation?”), we added covitality as a predictor to the regression (Model 3). We found that covitality, β = −.095, p = .05, at T1 predicted loneliness at T2, F(5, 724) = 14.19, R² = .089, p < .001 (Table 2, Model 3). By adding covitality to the model, the model explained .005 (p = .05) more variance of loneliness at T2. Therefore, covitality appeared to be a protective factor for loneliness above and beyond victimisation and school belonging.

Fourth, to examine the unique effects of different types of victimisation at T1 on loneliness at T2, we analysed three types of peer victimisation (physical, verbal, and relational) separately by including all of them as predictors for loneliness in the regression analysis. Results suggested that physical victimisation, β = .183, p = .008, and relational victimisation, β = .128, p = .043, were significant positive predictors for loneliness at T2 after controlling for gender and loneliness at T1, F(5, 752) = 16.97, R² = 0.101, p < .001 (Table 3). Verbal victimisation was not a significant predictor for loneliness when the other two types of victimisation (i.e., physical and relational) were also included in the regression, β = −.092, p = .183. Because different types of victimisation had different relations with loneliness, we analysed them separately in the moderation analyses.

Table 3. Three types of bullying at Time 1 predicting loneliness at Time 2

	Loneliness T2
	b (SE)	β
Verbal bullying victimisation	−.05(.04)	−.09
Physical bullying victimisation	.14(.04)	.18**
Relational bullying victimisation	.08(.04)	.13*
Loneliness T1	.15(.04)	.15***
Gender	.12(.04)	.10**
F(df)	16.97 (5, 752)***
R^2	.10

* p < .05; **p < .01; ***p < .001.

To answer research question 3 (“Does loneliness predict greater levels of victimisation over time?”), we used four different victimisation (overall, physical, verbal, and relational) at T2 as the outcome variable and used loneliness at T1 as the predictor in our regression analyses. We found that loneliness at T1 predicted verbal victimisation at T2 after controlling for gender and verbal victimisation at T1, β = .083, p = .025, F(3, 753) = 33.99, R² = .119, p < .001; however, loneliness at T1 did not predict overall victimisation or physical, or relational victimisation at T2 (Table 4).

Table 4. Loneliness at Time 1 predicting different types of victimisation at Time 2

	Overall victimisation T2		Physical victimisation T2		Verbal victimisation T2		Relational victimisation T2
	b (SE)	β	b (SE)	β	b (SE)	β	b (SE)	β
Loneliness T1	.03(.04)	.03	.18(.20)	.03	.57(.25)	.08*	.21(.24)	.03
[vic]subscale T1	.38(.04)	.37***	.34(.04)	.33***	.30(.04)	.31***	.36(.04)	.35***
Gender	−.01(.05)	−.005	−.24(.23)	−.04	−.10(.28)	−.01	.27(.26)	.04
F(df)	41.24(3, 730)***		34.59(3, 748)***		33.99(3, 753)***		39.76(3, 752)***
R^2	.15		.12		.12		.14
Loneliness T1	.03(.04)	.03	.18(.20)	.03	.57(.25)	.08*	.21(.24)	.03
[vic]subscale T1	.38(.04)	.37***	.34(.04)	.33***	.30(.04)	.31***	.36(.04)	.35***
Gender	−.01(.05)	−.005	−.24(.23)	−.04	−.10(.28)	−.01	.27(.26)	.04
F(df)	41.24(3, 730) ***		34.59(3, 748) ***		33.99(3, 753) ***		39.76(3, 752)***
R^2	.15		.12		.12		.14

* p < .05; **p < .01; ***p < .001.

3.3. Moderation analysis

To examine the moderator effect of covitality (research question 4), we ran four regression models separately for victimisation (overall, physical, verbal, and relational) to avoid collinearity due to including all different victimisation and their interactions in the model simultaneously. First, we used loneliness at T1 as the outcome variable to examine the moderation effect cross‐sectionally. We found that the interaction between covitality at T1 and verbal victimisation at T1 was a significant predictor for loneliness at T2, β = −.074, p = .03 (Table 5). By adding the interaction term, the model explained .005 more variance of loneliness at T2, F(5, 737) = 40.65, R² = .216, p < .001. To further examine the moderation effect, we ran a simple slope test. The simple slope test showed that the slope for the low covitality group (b = 0.20) is significant, t(750) = 7.99, p < .001, as well as the slope for the high covitality group (b = 0.13), t(750) = 4.34, p < .001 (see Figure 1), but the relation between victimisation and loneliness is stronger for the low covitality group. This suggests that high covitality buffered the effect of verbal victimisation on loneliness cross‐sectionally at T1. Covitality did not interact with physical victimisation or relational victimisation at T1.

Table 5. Covitality moderating the relations between victimisation at Time 1 and loneliness at Time 1

	Overall victimisation		Physical victimisation		Verbal victimisation		Relational victimisation
	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)
	β	β	β	β	β	β	β	β
Covitality	−.17***	−.20***	−.18***	−.17***	−.18***	−.17***	−.17***	−.17***
Belonging	−.14**	−.13**	−.16***	−.17***	−.16***	−.17***	−.14**	−.14**
[vic] subscale	.30***	.30***	.24***	.23***	.28***	.26***	.33***	.32***
[vic]*covitality		−.04		−.05		−.07*		−.03
Gender	.05	.05	.07*	.07*	.03	.04	.04	.04
F(df)	52.73(4, 725) ***	42.37(5, 724)***	43.45(4,733)***	35.16(5, 732)***	49.37(4, 738) ***	40.65(5, 737)***	58.43(4, 735) ***	46.82(5, 734)***
R^2	.23	.23	.19	.19	.21	.22	.24	.24
ΔR^2		.002		.002		.005*		.001

*p < .05; **p < .01; ***p < .001.

Figure 1. Covitality at Time1 moderated the relation between verbal victimisation at Time1 and loneliness at Time1. Graph is needed… see https://media.wiley.com/assets/7323/92/electronic_artwork_guidelines.pdf

Second, we examined the moderation effect of covitality longitudinally using loneliness T2 as the outcome while controlling for loneliness T1. We did not find any significant moderation effects (Table 6).

Table 6. Covitality did not moderate the relations between victimisation at Time1 and loneliness at T2

	Victimisation		Physical victimisation		Verbal victimisation		Relational victimisation
	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)	Model 1 (main effect)	Model 2 (interaction)
	β	β	β	β	β	β	β	β
Covitality	−.10	−.09	−.07	−.07	−.09	−.10*	−.09	−.09
Belonging	.02	.02	.004	.002	.002	.01	.02	.01
[vic] subscale	.16***	.16***	.18***	.18***	.10**	.11**	.16***	.16***
[vic]*covitality		−.01		−.01		−.04		−.03
Loneliness T1	.13**	.13**	.15***	.15***	.18***	.18***	.14**	.14**
Gender	.08*	.08*	.10**	.10**	.07*	.04*	.07*	.07*
F(df)	14.19(5, 724) ***	11.82(6, 723) ***	15.79(5, 732) ***	13.16(6, 731) ***	13.50(5, 737) ***	11.49(6, 736) ***	14.67(5, 734) ***	12.33(6, 733) ***
R^2	.09	.09	.10	.10	.08	.09	.09	.09
ΔR^2		0		0		.002		.001

*p < .05; **p < .01; ***p < .001.

4. Discussion

Based on our knowledge, this is the first study to examine the longitudinal relations between peer victimisation, school belonging, covitality, and loneliness. It is also first to illustrate the moderation effect of covitality on the relation between victimisation and loneliness using cross‐sectional data among Chinese elementary school students. Consistent with prior research conducted in the United States (e.g., Furlong et al., 2013; Kochenderfer & Ladd, 1996) and Australia (Shochet et al., 2011), we found that the experience of peer victimisation predicted more loneliness at T2 after controlling for loneliness at T1. On the other hand, a sense of belonging at school, and the possession of positive psychological traits (e.g., gratitude, optimism, zest, and persistence) served as protective factors against loneliness for Chinese elementary school students. When different types of victimisation were examined, both physical victimisation and relational victimisation positively predicted loneliness over time. Although adults tend to pay more attention to physical victimisation, relational victimisation appears to be equally troublesome for Chinese elementary school students. It is important for adults to not only intervene when they witness physical bullying but also relational bullying.

The buffering effect of covitality at T1 is somewhat new and encouraging but should also be interpreted with caution because this effect was only observed with the cross‐sectional data and only for verbal victimisation (not for physical or relational victimisation). In our sample, students actually experienced slightly more verbal victimisation (M = 1.81) than physical (M = 1.59) and relational victimisation (M = 1.70). This may be because physical bullying is strongly discouraged in China due to strong cultural values for harmonious social relationships. On the other hand, verbal bullying may be tolerated or unnoticed by the teachers under the disguise of good‐natured teasing. However, the buffering effect of covitality disappeared when longitudinal data were examined. Future studies should examine other buffering variables (e.g., social support) over time. However, covitality still had a significant main effect and it negatively predicted loneliness over time after controlling for loneliness at T1. As a result, it is still important to promote covitality to reduce loneliness over time.

Some researchers suggested that the relation between victimisation and loneliness may be reciprocal (Campbell, 2013; Liu et al., 2017). In other words, victimisation contributes to more loneliness over time, and loneliness also contributes to later victimisation. Our findings provided partial support for this reciprocal hypothesis. We found that victimisation predicted more loneliness at T2, and loneliness at T1 predicted more verbal victimisation (but not overall victimisation, or physical or relational victimisation) at T2. Because the findings vary for different types of victimisation, it is important to examine the effects of victimisation subtypes separately. Considering that interdependence and group harmony are emphasised in China, schools greatly discourage aggressive behaviours and encourage students to work in groups. Our findings suggest that Chinese students who feel lonely may not be physically bullied and excluded from the peer group. However, it is still concerning that those students who felt lonely at T1 reported experiencing more verbal victimisation at T2. To stop such a vicious cycle, it is important to implement prevention and early intervention programs for students who feel lonely and who experience verbal victimisation. We found 26 to 34% of students experienced victimisation at least once a month. These prevalence rates of peer victimisation are consistent with reported rates in the United States (Cheng et al., 2010; Wang, Iannotti, & Nansel, 2009). However, such high prevalence also suggests that it is important for teachers and school administrators to implement schoolwide bullying prevention programs to stop students from being victimised.

Consistent with prior research showing that school climate is negatively related to mental health difficulties among Chinese elementary school students (Wang, Boyanton, et al., 2018), we also found that school belonging was a protective factor for loneliness. School staff can foster students' sense of school belonging by implementing schoolwide prevention programs, which are discussed further in the implication section.

4.1. Limitations

There are several limitations to this study. First, we only used one item from the PSSM to measure loneliness. Using only one item may not capture all unique aspects of loneliness. However, using a single‐item measure of loneliness is widely used, accepted, and perceived valid especially for survey data (see Rönkä, Rautio, Koiranen, Sunnari, & Taanila, 2014; Stickley et al., 2016). Future studies should use a more complex measure of loneliness to capture different aspects of loneliness, such as social loneliness and emotional loneliness. Second, data were only collected from five schools from urban areas in one province in China. The results may not be generalizable to other samples with different demographic backgrounds. Future researchers should collect data from a more diverse sample of Chinese students. Third, all data were based on student self‐report, and shared method variance is a concern. Future studies should collect data from peers, teachers, and parents to measure students' psychological traits and experience with bullying more comprehensively. Fourth, we only found covitality buffered the relation between victimisation and loneliness using cross‐sectional data at T1, and we did not find this moderation effect longitudinally, although covitality predicted less loneliness overtime. Future studies are needed to continue to test the moderation effect of covitality.

4.2. Implications

There are multiple strategies that teachers and school staff can use to combat loneliness. Teachers can promote students' sense of school belonging by fostering positive relationships with students and making sure that every student feels like they are a member of the school community. Research has repeatedly shown that having a positive relationship with teachers at school promotes youth academic and mental health outcomes (Wang, Boyanton, et al., 2018). It is also important to create an inclusive environment so that nobody feels rejected. For example, when designing activities (e.g., field trips, science fairs) at school, teachers should make every effort to ensure that students from different socioeconomic backgrounds can participate freely in these activities. In other words, students are not excluded from activities because of financial or other difficulties. Teachers can also facilitate inclusion and reduce students' feeling of loneliness by proactively making sure that students have peers to interact with during unstructured activities. For example, teachers can assign a “buddy” to a new student who just moved into the school or give students time to plan for activities in dyads or small groups before recess so that nobody feels left out.

It is also important to reduce the prevalence of bullying because of the reciprocal relation between victimisation and loneliness found in this study. Schools can promote a positive school climate to deter bullying. First, teachers and school staff (e.g., school psychologists) can work with school administrators to implement system‐wide programs to promote positive school climate and foster a welcoming school environment for every student. Second, classroom lead teachers in China can engage students in class meetings (usually once a week in Chinese elementary schools) to brainstorm the best ways to prevent bullying and make school a safe environment for each student. Teachers can then incorporate these students' ideas into the classroom and throughout the school. Peers should be encouraged to keep each other accountable to engage in prosocial behaviour and be positive bystanders during bullying incidents. Last, discussing difficult topics related to bullying, such as dominance and popularity during class meetings may be beneficial as well (Wang, Berry, & Swearer, 2013).

Finally, families and schools should foster student resilience by promoting covitality and other positive psychological traits. There are some existing programs targeting gratitude (Froh, Kashdan, Ozimkowski, & Miller, 2009), and persistence/grit (Bernard & Walton, 2011). To promote these positive psychological traits, teachers can incorporate a well‐being curriculum into classroom instruction by explicitly teaching students covitality related skills to enhance their positive emotions. Students can also be exposed to positive psychology by studying these topics in their academic curriculum. For example, students can study hope and gratitude in their language arts curriculum by analysing the characters and text and then reflect on it in a journal (Waters, 2011). By implementing these suggestions, we can develop effective culture‐sensitive interventions that foster social connectedness and reduce loneliness for all.