







































 

 

 

 

 

The Mediating Role of Anti-Bullying 

Administrative Measures in the Relationship 

between Bullying and Students’ 

Core Competencies

I-Hua Chen,
1
 Ni Xie,

2
 Zhi-Yuan Meng

1
 

 
1. Qufu Normal University, Qufu 273165, Shandong, China 

2. Guizhou Normal University, Guiyang 550001, Guizhou, China 

Abstract. This research evaluates the role of school administrative 

measures (including creating a greater school belonging and paying 

attention to student attendance) as independent variables and their re-
sulting core competencies through the mediator of students’ experience 

with school bullying. This study adopted a multi-level mediation model 

to empirically analyze data from middle school students and school ad-
ministration in four Chinese provinces based on the 2015 Programme 

for International Student Assessment (PISA). Data were collected from a 

total of 9,060 students and 260 administrative staff. The results were: (i) 
Relational bullying was significantly and negatively correlated with the 

three core competencies, although no significant impact was found for 
either verbal or physical bullying; (ii) Schools which were successful in 

creating a more positive environment, including greater school belong-

ing and greater attention to students’ attendance, demonstrated lower 
levels of relational bullying; (iii) In terms of school-level variables, a 

greater sense of shared belonging had a direct effect on improving stu-
dent performance on math and science competencies, while greater at-

tention to students’ attendance was associated with higher student 

scores on all three core competencies; (iv) Furthermore, school-level 
variables, including the sense of shared belonging and greater attention 

to students’ attendance demonstrated a positive indirect effect on stu-

dents’ core competencies through the mediating effect of reduced rela-
tional bullying. 

Best Evid Chin Edu 2020; 5(2):681-701. 

Doi: 10.15354/bece.20.ar050. 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No.2, 2020 682 

How to Cite: Chen, I.H., Xie, N., Meng, Z.Y. (2020) The Mediating Role of Anti-

Bullying Administrative Measures on the Relationship between Bullying and Stu-

dents’ Core Competencies. Best Evid Chin Edu, 5(2):681-701. 

Keywords: Core Competencies; School Bullying; School Administration; Attend-

ance; Sense of Belonging; Multi-Level Mediation Modeling; 2015 PISA.

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

About the Authors: Ni Xie, D.Ed., School of Education, Guizhou Normal University, Guiyang 550001, Guizhou, 

China. Email: 364236069@qq.com. 
Zhi-Yuan Meng, Doctorate Candidate, School of Statistics, Qufu Normal University, Qufu 273165, Shandong, 

China. Email: 1360881683＠qq.com. 

Correspondence to: I-Hua Chen, D.Ed., Professor, Chinese Academy of Education Big Data, Qufu Normal Uni-
versity, Qufu 273165, Shandong, China. Email: aholechen@gmail.com. 

Funding: The 2016 General Education Project of the China Social Science Fund, “College Entrance Examination 
and Research on Transformation and Development of General High Schools" (BHA169000). 

Conflict of Interests: None. 

mailto:364236069@qq.com
mailto:aholechen@gmail.com


Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 683 

Introduction 

LOBALLY, educational administrators and policymakers hold the standard-

ized scores reported by the Programme for International Student Assessment 

(PISA) in high regard, placing a strong emphasis on PISA’s measurement of 

core competencies, holding these scores as indicators of school success (Wu, 2013). 

The emphasis of student performance on PISA, thus, reflects the role that school admin-

istrators must play in the cultivation of students’ core competencies. However, PISA 

not only evaluates the core competencies of 15-year-old middle school students across 

the globe (in terms of mathematics, reading, and science), but also collects a wide range 

of relevant background information, including student- and school-level factors which 

may influence student performance. As such, PISA data serves as a valuable source for 

educational researchers in empirically investigating the potential relationships between 

performance on core competencies and other relevant background factors, the results of 

which can reveal, to a certain degree, which factors related to school administration 

demonstrate the greatest influence on students’ academic performance in terms of the 

three core competencies. 

Given the availability of student- and school-level background data, some Chi-

nese researchers have already begun to conduct studies using a variety of items from 

PISA datasets. In the case of students’ science competencies, Zhao, Guo, and Jiao (2017) 

utilized multi-level analysis to evaluate PISA data, demonstrating the significant and 

positive effects of school-level variables, including the provision of creative extracur-

ricular activities, scientific resources, and greater cooperation among science teachers. 

In terms of reading, science, and mathematics competencies, Chen (2017) found signif-

icant effects for non-cognitive factors, including achievement motivation, parental emo-

tional support, reduced test anxiety, and a sense of belonging to the school. Although a 

sense of belonging to the school, as measured in Chen’s (2017) study, was based on 

individual students’ perceptions, the construct of “belongingness” can also serve as a 

higher-level (school-level) variable in multi-level modeling, as in the shared perceptions 

among students of the same school. In this manner, the use of school-level belonging-

ness can better reflect the degree to which school administration is effective in develop-

ing an appropriate environment for teaching and learning. 

While the aforementioned studies have evaluated the relationship between 

school measure variables, such as the provision of activities and resources, with stu-

dents’ performance on core competencies, the relationships among school bullying, 

school administration efforts and students’ core competencies have not yet been empiri-

cally evaluated. Therefore, this study utilized the PISA data released in 2015 to empiri-

cally explore the relationships among school administration factors, student bullying, 

and the three core competencies (mathematics, reading, and science). Variables belong-

ing to different levels, both school-level and student-level, were included in the multi-

level mediation modeling analysis of the present analysis. The aim of this analysis was 

to evaluate how school-level variables, including school administration, can influence 

student-level variables, such as performance on core competencies, by considering the 

G 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 684 

mediating role of school bullying. As such, this study contributes a new perspective on 

the prevention of school bullying that, in turn, has the potential for improving adoles-

cent students’ performance in terms of core competencies.  

Independent Variables: Creating a Sense of Shared Belonging 

and Attention to Student Attendance  

Based on the limitations facing existing studies, which were based on single-level mod-

els, this study adopted a multi-level analysis, including mediation, using aggregated 

data from the 2015 PISA datasets. This data was used to generate school-level inde-

pendent variables including a) creating a sense of shared belonging and b) attention to 

student attendance. Since these two school-level factors reflect the administrative ac-

tions or measures implemented by school staff, they are highly practical in nature and 

can be generalized to other contexts. Based on preliminary analytical results adopting 

single-level analysis, these two variables were supported as relevant to both students’ 

performance on core competencies and school bullying, serving as the basis for the de-

cision to further test these relationships using a multi-level mediation model. 

In terms of the sense of shared belonging to one’s school, Chen (2017) demon-

strated that higher levels for the sense of shared belonging were associated with higher 

core competencies for adolescent students. Furthermore, students’ perceived sense of 

belonging to the school was negatively associated with reported bullying (Chen & Zhi, 

2017; Didaskalou et al., 2017). In terms of student attendance, research has demonstrat-

ed that student truancy has negative impacts on academic achievement and is associated 

with reported bullying on campus (Gastic, 2008). Based on the aforementioned studies, 

administrative measures that both develop a sense of shared belonging and greater at-

tention to student attendance are relevant to the incidence of school bullying as well as 

student performance on core competencies. 

Mediator: School Bullying 

The results of a meta-analysis have shown that school bullying has a significant nega-

tive effect on academic achievement (Nakamoto & Schwartz, 2012). Empirical evi-

dence supporting the negative influence of bullying served as a basis for this study in 

selecting school bullying as a mediating variable. Moreover, the school bullying is both 

a new item for PISA, added for the first time in 2015 (OECD, 2017a), and an increas-

ingly prevalent issue in Chinese education in recent years (Yang et al., 2017; Li, 2017; 

Wang et al., 2015). According to the OECD (2017a), after controlling for the influence 

of the overall socio-economic status of a school’s families, students who report being 

more frequently bullied at school scored 47% lower on the science competency exam as 

compared with students who reported infrequent bullying. 

Given the fact that peer victimization has a profound impact on student learn-

ing performance, the PISA report (OECD, 2017a) urges schools to take concrete 

measures to curb school bullying, adding that school bullying directly and negatively 

impacts students’ performance on core competency measures. Furthermore, the report 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 685 

(OECD, 2017a) suggests that, in order to improve campus safety and enhance students’ 

potential learning success, schools must actively enact student management policies. 

Based on this logic, the mediating role of school management efforts is seen as a means 

to indirectly improve students’ performance on core competencies by directly reducing 

school bullying. 

Following the release of the 2015 PISA data and report, some studies in China 

have conducted secondary data analysis on school bullying-related topics (e.g., Chen & 

Zhi, 2017; Huang, 2017). However, thus far, no studies have adopted a multi-level me-

diation model including the school-level administration variables of (i) developing a 

sense of shared belonging and (ii) attention to student attendance. 

Research Questions  

The purpose of this study was to investigate the degree to which school-level independ-

ent variables indirectly influence students’ performance on core competencies through 

the mediation of the student-level variable of school bullying. This study utilized raw 

data from students and administrators in mainland China from the 2015 PISA datasets 

in the development of a multi-level mediation model. Based on recommended proce-

dures for examining mediation effects in multi-level models (Wen & Chiou, 2009), the 

research questions of this study are stated as follows: 

RQ1: Does school bullying significantly influence students’ performance in 

terms of reading, mathematics, and science competencies? 

RQ2: Does an administrative approach that seeks to develop a sense of shared 

belonging and pays greater attention to student attendance significantly reduce 

students’ experiences of school bullying? 

RQ3: Does the reported sense of shared belonging and student attendance sig-

nificantly explain differences among students in terms of performance on read-

ing, mathematics, and science competencies? 

RQ4: Does a sense of shared belonging and student attendance have a signifi-

cant indirect effect on students’ performance on reading, mathematics, and sci-

ence competencies through the mediating factor of school bullying? 

 

Multi-Level Modeling Procedures 

Assuming that school administrative policies and procedures will directly reduce school 

bullying which, in turn, will indirectly influence students’ performance on core compe-

tencies, a fundamental feature of the research model was the relationship between 

school administration factors and school bullying. In order to clarify the relationship 

between these two factors, related variables were first controlled. The following sec-

tions discuss the control variables included in the multi-level model and research hy-

potheses corresponding to the paths of the multi-level model. 

Control Variables 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 686 

Based on a review of school bullying research, certain student-level and school-level 

demographic variables have been found to influence the occurrence of school bullying. 

However, since many of these variables were not the focus of the model developed in 

this study, they are treated as control variables. In terms of student-level variables, gen-

der, grade, and family socioeconomic status were included as control variables. Based 

on the findings of Huang (2017), who evaluated 2015 PISA data from mainland China, 

boys, and lower grade students were more likely to be victims of bullying. Furthermore, 

Jansen et al. (2012) reported that adolescents in the Netherlands from families of lower 

social status reported a higher proportion of physical and psychological bullying symp-

toms and, as such, more strongly experienced the negative effects of school bullying.  

In terms of school-level variables, the school’s overall family socioeconomic 

status, school size, and location of the school district were included as control variables 

in the multi-level model. The school’s overall family socioeconomic status was neces-

sarily included based on the results of a multi-level study of high school students from 

various countries, which revealed that large gaps in the levels of economic purchasing 

power among families in a school district were associated with more frequent occur-

rences of school bullying (Due et al., 2009). Based on the recommendations of Due et 

al.’s study, a school’s overall family socioeconomic status, as a school-level variable, 

must be controlled. The size and location of schools were also evaluated by the model, 

in reference to Betts’ (2014) investigation into the relationship between the size and 

location of schools and differences in the occurrence of school bullying, with students 

from larger schools demonstrating greater vulnerability to physical bullying, and loca-

tion (urban vs. rural) showing no significant influence. Thus, although there no signifi-

cant impact on school location was noted in Betts’ (2014) study, due to the large educa-

tional gaps between China’s urban and rural areas, it was considered prudent to incor-

porate school location in the model as a school-level control variable. 

Research Hypotheses 

The research hypotheses proposed by this study are in reference to the limited findings 

of prior single-level studies. The expansion of the proposed model to a methodology 

adopting multi-level analysis was deemed prudent, given the potential contribution of 

both the student-level and school-level variables discussed in previous sections of this 

paper. 

Sense of Shared Belonging  

The first school-level variable included in the multi-level model was “sense of shared 

belonging.” A sense of shared belonging can be conceptualized as a student-level con-

structor, in the case of this study, computed as a school-level variable through the ag-

gregation of data from a school’s entire student population. Thus, as a school-level vari-

able, the interpretation of “sense of shared belonging” differs from the student-level 

construct in that, in addition to representing the shared and collective perceptions re-

garding the school environment by a school’s student population, it can also reflect the 

school’s overall administrative efforts towards the development of an appropriate envi-



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 687 

ronment for teaching and learning, including care for and acceptance of students (Free-

man et al., 2007). While past empirical studies have evaluated sense of belonging as a 

student-level variable (Chen, 2017; Chen & Zhi, 2017), in order to evaluate the efficacy 

of school administration in anti-bullying efforts (Stewart, 2008), a sense of shared be-

longing was set as a school-level construct in this study. Furthermore, it should be noted 

that Liu and Liu’s (2011) follow-up study found no significant correlation between stu-

dents’ sense of belonging (as a student-level variable) and academic achievement, after 

differentiating initial stage and linear growth values. As such, this study adopts students’ 

sense of shared belonging as an emotional factor, with the influence of emotional fac-

tors on academic achievement operating as an indirect, rather than direct, effect. 

Based on the findings of previous studies investigating students’ sense of be-

longing, it seems more likely that a sense of shared belonging, as a school-level variable, 

can best reflect the degree to which school administrators actively and effectively de-

velop an environment wherein students develop positive feelings towards the school, 

leading to better academic performance. However, although the school-level sense of 

shared belonging influences the core competencies of individual students across levels, 

the possible mediating effects have yet to be investigated. As Liu and Liu (2011) con-

clude, students’ sense of belonging serves as an emotional factor that indirectly influ-

ences academic achievement through other mediating variables (Dong & Yu, 2010). 

Thus, this study hypothesizes that a school-level sense of shared belonging will influ-

ence students’ performance in terms of PISA core competencies (RQ3). 

As such, this study proposed school bullying as a potential mediator for the in-

fluence of a sense of shared belonging on students’ core competencies. Specifically, 

victims of bullying often feel a lack of acceptance within a school’s environment and, 

as a result, fail to seek assistance, retreat from school life, and demonstrate lower levels 

of confidence, resulting in an increased likelihood of being bullied further, resulting in a 

long-term vicious cycle (Chen & Zhi, 2017; Didaskalou et al., 2017). Moreover, a sig-

nificant correlation exists between the experience of being bullied and low academic 

achievement, based on the meta-analysis of Nakamoto and Schwartz (2010). Thus, this 

study hypothesizes that bullying directly impacts students’ performance on core compe-

tencies (RQ1). 

From the previously noted single-level empirical studies, students’ sense of 

shared belonging and school bullying appear to be causally linked, making it difficult to 

determine which variables are independent and which are mediators. However, by ag-

gregating individual data in computing a school-level variable for a sense of shared be-

longing and adopting bullying as a student-level variable, the independent variable and 

mediating variable can be more clearly assessed. From the perspective of multi-level 

mediation analysis, higher-level (school-level) variables are generally adopted as inde-

pendent variables and are hypothesized to influence the lower-level (student-level) vari-

ables (Wen & Ye, 2014). Therefore, the “sense of shared belonging” as a school-level 

was adopted as the independent variable and school bullying as the mediating variable. 

After determining the possible paths among the three variables (students’ sense of 

shared belonging, school bullying, and student core competencies), we hypothesized 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 688 

that school administrators, by actively developing students’ sense of shared belonging 

would a) directly prevent school bullying (RQ2) while b) indirectly improving students’ 

core competencies (RQ4). 

Attention to Student Attendance 

Based on PISA data, the factor regarding the school administration’s efforts towards 

“attention to student attendance” is evaluating using three items evaluating: students’ 

lateness, skipped classes, and truancy. These are student-level variables in the original 

2015 PISA datasets but were aggregated as a school-level variable reflecting the admin-

istration’s attentiveness to school attendance. Students’ timeliness and attendance are 

undoubtedly necessary to ensure the quality of learning. If students are not willing to 

attend classes or are truant, they are not likely to perform well academically (Fang, 

2007). As such, paying closer attention to students’ attendance in one measure by which 

school administrators can promote students’ attendance classes and timeliness, factors 

assumed to have a direct effect on students’ performance on core competencies. Thus, 

this study hypothesizes that school-level sense of attention to student attendance will 

demonstrate an influence on students’ performance in terms of PISA core competencies 

(RQ3). 

Furthermore, the PISA report (OECD, 2017a) suggests that when students real-

ize that the school will attentively enforce school rules, bullying events will be reduced. 

Therefore, it is possible that more attentive student management by the administration, 

as a school-level variable, can deter the occurrence of school violence, reduce opportu-

nities for students to experience bullying, and have a positive indirect effect on students’ 

core competencies. Thus, in the multi-level model, school bullying likewise serves as a 

mediator between school administrators’ efforts to enforce student attendance and stu-

dents’ resulting performance on core competencies. 

Close relationships have been found among school bullying, student attendance, 

and academic achievement (Gastic, 2008). In order to avoid bullying, students may pro-

tect themselves by avoiding the school environment. However, by skipping classes, 

such students fail to keep up with lessons, resulting in poor academic performance. The 

research of Zhao and Zhu (2012), targeting both juvenile offenders and non-offending 

middle school students, found that skipping classes was closely associated with peer 

victimization, demonstrating that skipping classes weakened the relationship between 

students and the school, leaving them vulnerable to negative behaviors. As such, skip-

ping classes has a mutually causal relationship with school bullying, wherein bullied 

students attempt to skip classes to avoid peer victimization but, by skipping classes, lose 

their connection with the school, preventing victimized students from obtaining assis-

tance and continuing to suffer bullying. Thus, this study hypothesizes that bullying di-

rectly influences students’ performance on core competencies (RQ1). 

As previously noted, if the single-level analysis is adopted, it is difficult to de-

termine the independent variable from a bi-directional correlation. However, by aggre-

gating attendance as a higher-level (school-level) variable, the independent and mediat-

ing variable can be clearly distinguished. Since the school-level factor of attention to 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 689 

students’ attendance, as defined in this study, reflects the attentiveness of the school’s 

administration in the implementation and enforcement of school rules, this variable 

serves as a higher-level, school-level variable and an independent variable in evaluating 

the effectiveness of school bullying prevention. Therefore, this study hypothesizes that 

if the school administration pays attention to students’ attendance through a more atten-

tive administrative approach, they will demonstrate that the school attaches importance 

to students’ learning, resulting in a strengthened relationship between students and the 

school and more effective prevention of school bullying. In this way, an attentive school 

administration, which pays attention to students’ attendance, mediated by a decrease in 

school bullying, has the potential to improve students’ core competencies. After deter-

mining the possible paths among the three variables (attention to school attendance, 

school bullying, and student core competencies), we hypothesized that school adminis-

trators, by paying more attention to student’s attendance would a) directly prevent 

school bullying (RQ2) while b) indirectly improving students’ core competencies 

(RQ4). 

Research Methods 

Data Source 

Variables from the 2015 PISA dataset were selected for analysis, including background 

information on students and school administration. These samples represented the Chi-

nese provinces of Beijing, Shanghai, Jiangsu, and Guangdong. Data from a total of 

9,841students and 268 schools (with each school represented by one school administra-

tor) were analyzed. After excluding missing data, the number of valid subjects included 

9,060 students (52.4% male and 47.6% female) and 260 administrative staff. From 

among these schools, schools with a total student population of less than 1,600 account-

ed for 59.2% of schools, while schools with more than 1,600 students accounted for 

41.8% of total schools. Schools located in towns and small cities accounted for 50.8% 

of the total, with schools located in large cities accounting for 37.3% of the total, and 

schools in rural areas accounting for 11.9% of total schools. 

Variables in the Multi-Level Model and Corresponding 2015 

PISA Data 

This section lists the variables included in the research model and provides examples of 

items from the corresponding 2015 PISA dataset (see Table 1). Included in Table 1 is a 

column entitled “Variables” which includes the variables of the multi-level mediation 

model, including dependent and independent variables, a mediator, and control varia-

bles belonging to either the student- or school level. The column entitled “Correspond-

ing 2015 PISA data” includes items from the 2015 PISA dataset used in computing 

model variables (including both measurement indicators and the original items on 

which these measurement indicators were based). Derived variables included in the 

model include two generated from the PISA data, including the original PISA item of 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 690 

Table 1. The Variables of the Model and the Corresponding Data 
of PISA 2015. 
 Variables Corresponding data 

  Measurement Indicators Original PISA Items 

Dependent 
Variables 
(student-
level) 

Mathematics, reading, 
and science compe-
tencies 

Ten plausible values (PV) 
for each competency 

Mathematics: PV1MATH - 
PV10MATH 
Reading: PV1READ - 
PV10READ 
Science: PV1SCIE - 
PV10SCIE 

Independent 
Variables  
(school-
level)* 
 
 

Sense of shared be-
longing 

The aggregate of the varia-
ble “Belong” for students 
from the same school 

Belong: ST034Q01TA-
ST034Q06TA 

Attention to students’ 
attendance 

The aggregate of the num-
ber of days students are 
late for school each week 

Late for school: ST062Q03TA 

The aggregate of the num-
ber of days students skip 
class each week 

Skipping classes: 
ST062Q02TA 

The aggregate of the num-
ber of days students are 
truant each week 

Truancy: ST062Q01TA 

Mediator  
(student-

level) 

Being bullied  Verbal bullying: Mean PISA 
verbal bullying scores. 

Verbal bullying: ST038Q04NA 
and 05NA 

Relational bullying: Mean 
PISA relational bullying 
scores. 

Relational bullying: 
ST038Q03NA and 08NA 

Physical bullying: Mean 
PISA physical bullying 
scores. 

Physical bullying: 
ST038Q06NA and 07NA 

Control 
Variables 
(student-
level) 
 

Gender Male or female based on 
PISA codes 

Gender: ST004D01T 

Grade Grade based on PISA 
codes 

Grade: ST001D01T 

Family socioeconomic 
status 

“ESCS” Weighted “HOMEPOS”, 
“HISEI” and “PADER” values 

Control 
Variables 
(school-level) 
 

School size “SCHSIZE” Addition of the number of 
SC002Q01TA (boys) and 
SC002Q02TA (girls) 

Type of school Private v.s. public from 
PISA codes 

SC013Q01 (1 = private 
school, 3 = public school) 

School location Location-based on PISA 
codes 

SC001Q01 (1-5 representing 
the village, small town, town, 
city, and big city, respectively) 

School  family socio-
economic status 

The aggregate of the 
“ESCS” value for all stu-
dents from the same school 

 

* Averaging scores for students in the same school and aggregate them as a school-level variable 

Note 1: The variable names in the table are exactly the same as those in the PISA 2015 data file. 

Note 2: According to the PISA 2015 manual, in order to correctly analyze the PV values, a one-time analysis of the ten 
PVs in terms of the mean value is not recommended. Rather, each PV must first be analyzed and then combined with 
the results of the other 10 PV values in order to establish significance. This study utilized HLM software to perform the 
above-mentioned processes. 

Note 3: In addition to variables such as the size, type, and location of the school, other school-level independent varia-
bles and control variables were aggregated from student-level variables. 

Note 4: For the calculation of family socioeconomic status, the following items were included: “HOMEPOS” (home pos-

sessions), “HISEI” (highest parental occupation), and “PADER” (parental education). 

 

 

 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 691 

“Belong” (representing the sense of shared belonging) and “ESCS” (representing the 

overall school families’ socioeconomic status). The technical manual for PISA (OECD, 

2017b) stated that the derived variables are scale scores generated by computing origi-

nal items through item response theory, with these resulting derived scores being suita-

ble for using for direct comparisons among OECD member countries. 

Data Analysis 

Data analysis consists of two parts: descriptive statistics and multi-level mediation 

modeling. In terms of descriptive statistics, since the PISA data involves ten plausible 

values (PVs) for estimating students’ core competencies, these values cannot be directly 

processed using SPSS statistical software. Therefore, a syntax was first generated using 

IDB Analyzer 4.0 (IEA, 2018), and then executed using SPSS 22.0 to compute descrip-

tive statistics. For the multi-level mediation model, we used HLM 6.0 to analyze the 

data, following the suggestions provided by the PISA technical manual (OECD, 2017b) 

to weigh variables at the student-level and school-level, utilizing the values of 

“W_FSTUWT” (Final student weight) and “W_SCHGRN” (Final school weight). In 

terms of the steps involved in conducting multi-level mediation modeling, this study 

followed the recommended procedures of Wen and Chiou (2009) to address the four 

research questions. 

 Step 1 (RQ1). This study first evaluated the coefficient for the influence of the me-

diator on the dependent variables. The dependent variables included students’ 

mathematics, reading, and science competencies. The mediator included three types 

of school bullying (verbal, relational, and physical). Based on the results of the first 

step, we retained the bullying types with statistically significant coefficients before 

continuing to conduct a follow-up analysis. 

 Step 2 (RQ2). Next, we computed the coefficient for the influence of the school-

level independent variables on the mediator variable of school bullying. The 

school-level variables in this study were a) sense of shared belonging and b) atten-

tion to students’ attendance. In terms of attentive school administration efforts, 

which pay attention to students’ attendance, three items were included: “late for 

class,” “skipping classes,” and “truancy.” In Step 2, we tested the impact of these 

two school management measures on the mediator (school bullying) and retained 

those context variables with statistically significant influence. 

 Step 3 (RQ3). The third step was to test the coefficient for the influence of the 

school-level variables on the dependent variables (core competencies). From Step 2, 

only school-level variables with significant coefficients were used in testing their 

effects on the dependent variables. It should be noted that only school-level varia-

bles with coefficients reaching statistical significance in Step 3 were then retained 

for further analysis in Step 4.  

 Step 4 (RQ4). The final step was to confirm whether or not a mediating effect ex-

isted within the model. The criterion for a mediation effect is that when the school-

level variables and the mediator are included in the same model, the coefficient for 

the influence of the school-level variables on the dependent variables must be lower 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 692 

than the coefficient when the mediator is not included. In this case, if the coeffi-

cient for the influence of the school-level variables on the dependent variables (core 

competencies) was not significant, and the influence of the mediator on the depend-

ent variables was significant, a complete mediating effect would be indicated. 

However, if the coefficient for the influence of the school-level variables on the de-

pendent variables (core competencies) still reached a statistically significant level, a 

partial mediating effect would be indicated. 

Results 

Descriptive Statistics 

The descriptive statistics are provided in Table 2. In terms of school-level variables, the 

average school size was 1,476 students (SD = 1,445). The mean for overall school fami-

lies’ socio-economic status was -1.16 (SD = 0.76), while the mean shared sense of 

shared belonging was -0.30 (SD = 0.21). Mean values were also calculated for being 

late for school (Mean = 1.52, SD = 0.24), skipping classes (Mean = 1.11, SD = 0.11), 

and truancy (Mean = 1.03, SD = 0.05), which indicated that, on average, less than two 

absences or late arrivals were observed over the most recent two weeks. The average 

value for student-level family socioeconomic status was -1.04 (SD = 1.11). Since this 

value was negative, it suggests that the family socioeconomic status of students was 

lower than the average for students in other OECD countries. Among the three core 

competencies, the mathematics competence was the highest (Mean = 538.38, SD = 

104.19), followed by science competence (Mean = 524.38, SD = 101.84), and reading 

competence (Mean = 501.18, SD = 106.87). The mean value for verbal bullying was 

1.30 (SD = 0.56) and relational bullying was 1.30 (SD = 0.59). The mean for physical 

bullying was 1.35 (SD = 0.56). Given the above values and the variation among differ-

ent types of bullying, student-level bullying ranged widely among schools, from situa-

tions where students reported no bullied to schools where students reported experienc-

ing bullying several times a year. 

Mediation Effects 

The analytical results of multi-level mediation modeling are provided in Table 3. In 

order to confirm that the intraclass correlation coefficient (ICC) and the variation com-

ponents met the requirements of multi-level analysis, we first tested a null model in-

cluding the dependent variables of the three core competencies (mathematics, reading, 

and science). The results demonstrated that the ICC for mathematics, reading and sci-

ence competencies were 48%, 51%, and 49%, respectively, and that the various compo-

nents of these three variables were significantly different from zero, which indicated 

that the core competencies for students in the same school were similar, with significant 

differences between schools. Since the assumptions required for multi-level mediation 

modeling were met, this analysis was adopted to evaluate the nested data derived from 

the 2015 PISA through non-independent sampling. In addition, based on the recom- 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 693 

Table 2. Descriptive Statistics. 

 Mean SD 

School-level   

School size 1,476 1,445 

School-level family socio-economic status (ESCS) -1.16 0.76 

Sense of shared belonging -0.30 0.21 

Late for school 1.52 0.24 

Skipping classes 1.11 0.11 

Truancy 1.03 0.05 

Student-level   

Family socioeconomic status (ESCS) -1.04 1.11 

Mathematics competency 538.38 104.19 

Reading competency 501.18 106.87 

Science competency 524.38 101.84 

Verbal bullying 1.30 0.56 

Relational bullying 1.30 0.59 

Physical bullying 1.35 0.56 

 

 

 

mendations of Wen and Chiou (2009) school-level variables adopted a fixed slope and 

grand mean-centered when the multi-level mediation model was tested.  

The Relationship between School Bullying and the Core Competencies 

The multi-level model equations are shown as follows. The results demonstrate that 

when student- and school-level control variables were included, only relational bullying 

was shown to negatively influence all three core competencies (with a coefficient for 

mathematical competency of γ50 = - 8.53, p = 0.01; a coefficient form reading compe-

tency of γ50 = -6.21, p = 0.04; and a coefficient for science competency of γ50 = -6.93, p 

= 0.01). Verbal bullying and physical bullying showed no significant effects on the 

three core competencies. 

 

Student-level (Equation 1-1) 

Core Competency ij ＝β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socioeconomic status ij) 

＋ β4j(Verbal bullying ij) ＋ β5j(Relational bullying ij) ＋ β6j(Physical bullying ij) ＋ γij 

 

School level (Equation 1-2) 

β0j ＝ γ00 ＋ γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j)＋ 

γ04(School-level socioeconomic status j) ＋ U0j; β1j ＝ γ10; β2j ＝ γ20; β3j ＝ γ30; β4j ＝ γ40; 

β5j ＝ γ50; β6j ＝ γ60 

The Relationship between School Administrative Measures and School 

Bullying 

 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 694 

Table 3. Multi-Level Mediation Model. 

 
Null 
Model 

Step 1 Step 2 Step 3 Step 4 

Dependent 
Variable 

CC CC RB CC CC (Without Mediator) CC (With Mediator) 

  M R S  M R S M R S M R S 

Fixed Effects 

γ00 267.25** 229.48** 195.53** 1.83** 223.56** 203.88** 165.85** 
241.26**a 
246.92**b 

213.30** 
b 

172.71** 
a 
177.46**b 

246.09** 
a 
251.29**b 

216.19 
**b 

176.19** 
a 
180.63** 
b 

γ01 
School type 

-18.41 -15.35 -10.13 0.02 -12.78 -11.16 -5.12 
-11.22 a 
-17.98 b 

-14.86 b 
-3.49 a 
-9.53 b 

-10.94 a 
-17.44 b 

-14.52 b 
-3.31 a 
-9.16 b 

γ02 
Size 

-0.00 -0.00 -0.00 -0.00 0.00 0.00 0.00 
0.00 a 
0.00 b 

0.00 b 
0.00 a 
0.00 b 

0.00 a 
0.00 b 

0.00 b 
0.00 a 
0.00 b 

γ03 
Location 

-5.03 -0.05 -2.09 0.02 -3.35 1.35 -0.47 
-3.95 a 
-4.11 b 

0.81 b 
-1.07 a 
-1.11 b 

-3.76 a 
-3.90 b 

0.94 b 
-0.94 a 
-0.96 b 

γ04 
School-level ESCS 

62.03** 63.05** 62.47** 0.01 42.02** 46.89** 43.78** 
46.51** a 
51.76** b 

53.86** b 
48.31** a 
52.07**b 

46.49**a 
51.37**b 

53.61** b 
48.30** a 
51.79 **b 

γ05 
School-level shared 
belonging  

   -0.33** 62.82** 45.04 53.51* 90.85** a  82.06** a 87.59** a  79.76 ** a 

γ06 
Late for class 

   0.02          

γ07 
Skipping classes 

   0.27* -182.27** -164.70** -185.00** -210.34** b 
-184.86** 
b 

-208.91 ** 
b 

-206.01** b 
-182.05** 
b 

-205.85** 
b 

γ08 
Truancy 

   0.40          

γ10 
Gender 

9.28* -13.83** 13.02** 0.12** 8.24* -14.43**  
8.31* a 
8.26* b 

-14.41** b 
12.48* a 
12.43* b 

9.23*a 
9.19* b 

-13.81** b 
13.13** a 
13.09** b 

γ20 
Grade 

32.38** 34.26** 36.16** -0.08** 34.46** 35.73**  
33.95** a 
33.95** b 

35.37** b 
37.42** a 
37.38** b 

33.18** a 
33.19** b 

34.87** b 
36.87**a 
36.83**b 

γ30 
ESCS 

6.83* 8.48** 5.94** -0.01 6.89** 8.50**  
6.90**a 
6.91** b 

8.51** b 
5.97** a 
5.98** b 

6.80** a 
6.81** b 

8.44** b 
5.90**a 
5.91**b 

γ40 
Verbal bullying 

-1.67 -0.72 -1.19           

γ50 
Relational bullying 

-8.53* -6.21* -6.93*        
-8.26** a 
-8.30** b 

-5.38* b 
-5.81* a 
-5.86* b 

γ60 
Physical bullying 

2.29 2.25 3.30           

Random Effects 

σ2  
(Intragroup 
variation) 

M: 
5562.65 
R 
5481.30 
S:  
5091.36 

5282.33 5135.06 4757.59 0.36 5305.17 5144.49 4769.17 

SB: 
5304.63 
SC: 
5305.24 

SC: 
5144.62 

SB: 
4768.62 
SC: 
4769.43 

SB: 
5283.25 
SC: 
5283.85 

SC: 
5136.39 

SB: 
4759.05 
SC: 
4759.86 

τ00 
(Intergroup 
variation) 

M: 
5145.82* 
R: 
5667.24* 
S: 
4901.72* 

1849.56** 1648.24** 1466.85** 0.004* 1370.26** 1297.71** 1005.84** 

SB: 
1671.39** 
SC: 
1469.10** 

SC: 
1344.84** 

SB: 
1315.21**  
SC: 
1074.77** 

SB: 
1636.61** 
SC: 
1435.53** 

SC: 
1324.72** 

SB: 
1292.61** 
SC: 
1053.93** 

Note: CC: Core Competency; RB: Relational Bullying; M: Math; R: Reading; S: Science; SB: Shared belonging; SC: Skipping classes. 
 *p < 0.05, **p < 0.01; 

a
 coefficient for the influence from sense of shared belonging to the school (school-level); 

b
 coefficient for the influence of skipping classes (school-

level) 

 

 

 

Based on the results of Step 1, only relational bullying had a significant effect on stu-

dents’ core competencies. Thus, only relational bullying was utilized to test the associa-

tion between school administration measures and bullying. Based on Equations 2-1 and 

2-2, the results demonstrate a significant and negative coefficient for sense of shared 

belonging (γ05 = -0.33, p < 0.01) and a significant and positive coefficient for skipping 

classes (γ07 = 0.27, p = 0.03). However, the coefficients for late for school and truancy 

were not significant. These results imply that the higher the sense of shared belonging 

(as an aggregate measure) and the more attentive the student administration was to-

wards students’ attendance (as evidenced by lower rates of skipping classes), the lower 

the level of reported school bullying.  

 

Student-level (Equation 2-1) 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 695 

Relational bullying ij＝β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ γij 

 

School-level (Equation 2-2) 

β0j ＝ γ00 + γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ γ06(Late 

for school j) ＋ γ07(Skipping classes j) ＋ γ08(Truancy j) ＋ U0j；β1j ＝ γ10；β2j ＝ γ20；

β3j ＝ γ30 

The Relationship between School Administrative Measures and Stu-

dents’ Core Competencies 

Since the sense of shared belonging and attention to student attendance had a significant 

effect on relational bullying in Step 2, we tested the influence of these two school-level 

independent variables on students’ core competencies during Step 3. Adopting Equa-

tions 3-1 and 3-2, the analytical results are as follows. First, in terms of mathematical 

competency, the coefficients for sense of shared belonging (γ05 = 62.82, p < 0.01) and 

skipping classes (γ07 = -182.27, p < 0.01) reached statistical significance. Second, for 

reading competency, the explanatory effect for skipping classes (γ07 = -164.04, p < 0.01) 

was significant, while the sense of shared belonging was related at a statistically insig-

nificant level. Third, regarding science competency, both sense of shared belonging (γ05 

= 53.51, p = 0.02) and skipping classes (γ07 = -185.00, p < 0.01) demonstrated statisti-

cally significant effects. The results of Step 3 imply that higher degrees of sense of 

shared belonging are associated with higher scores in terms of students’ mathematics 

and science competencies. Furthermore, more attentive school administrative measures 

(by managing students’ attendance and lowering the number of skipped classes) were 

associated with higher scores for all three core competencies. 

 

Student-level (Equation 3-1) 

Core Competency ij ＝ β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ γij 

 

School-level (Equation 3-2) 

β0j ＝ γ00 + γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ 

γ07(Skipping classes j) ＋ U0j; β1j ＝ γ10; β2j ＝ γ20; β3j ＝ γ30 

The Indirect Effect of School Administrative Measures on Students’ 

Core Competencies through the Mediator of School Bullying 

After conducting Steps 1 through 3, the independent variables meeting the criteria for 

analysis by Step 4 of the multi-level mediation model were: a) sense of shared belong-

ing and b) skipping classes, with a mediator of relational bullying. In terms of depend-



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 696 

ent variables, because the sense of shared belonging and students’ reading competency 

was not significantly related, we only tested the mediation effect from the sense of 

shared belonging on mathematics and science competencies while examining the medi-

ation effect of skipping classes on all three core competencies. 

First, regarding a sense of shared belonging, we used Equations 4-1 and 4-2, 

including the dependent variables of mathematics and science competencies, to obtain 

the coefficients for their influence without including the mediator of school bullying. 

Then, following Equations 4-3 and 4-4, the coefficients for the sense of shared belong-

ing for mathematics and science competencies were computed while including school 

bullying as a mediator. The results demonstrated that the coefficient for the influence on 

mathematical competency changed from 90.85 to 87.59, and the coefficient for the in-

fluence on science competency changed from 82.06 to 79.76. Based on these findings, 

the influence of relational school bullying on mathematical (γ 50 = - 8.26, p < 0.01) and 

science (γ 50 = - 5.81, p = 0.02) competencies were significant. Given that when the var-

iable of relational school bullying was included as a mediator, the coefficient for the 

influence of sense of shared belonging on the dependent variables was lower as com-

pared to the model which did not include relational school bullying, we conclude that a 

mediation effect did exist. However, this mediation was only partial, since the coeffi-

cient for the sense of shared belonging (a school-level variable) was still significant. 

Second, in terms of the effects of the school-level variable of skipping classes, 

we adopted the same procedure described above. Equations 4-5, 4-6, 4-7, and 4-8 were 

used to test the effects on mathematics, reading, and science competencies. The results 

demonstrated that when the variable of relational bullying was included, the coefficients 

for the influence of relational bullying on the three core competencies were all signifi-

cant (mathematical competency: - 8.30, p < 0.01; reading competency: - 5.38, p = 0.02; 

science competency: - 5.86, p = 0.02), while the coefficients for the school-level varia-

ble of skipping classes changed from - 210.34 to - 206.01 for mathematics competency 

and from -184.86 to -182.05 for reading competency. Likewise, the coefficient for sci-

ence competency changed from -208.91 to -205.85. Thus, after relational bullying was 

included, the coefficient for the influence of the school-level variable of skipping clas-

ses on the three core competencies was lower than the model that did not include the 

mediator of relational bullying. Since the coefficient of the context variable of skipping 

classes reached a significant level, relational bullying was concluded to provide partial 

mediation. 

 

Student-level (Equation 4-1) 

Core Competency ij ＝ β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ γij 

 

School-level (Equation 4-2) 

β0j ＝ γ00 + γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ U0j; β1 

＝ γ10; β2j ＝ γ20; β3j ＝ γ30 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 697 

 

Student-level (Equation 4-3) 

Core Competency ij ＝β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ β5j(Relational bullying ij) ＋ γij 

 

School-level (Equation 4-4) 

β0j ＝ γ00 ＋ γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ U0j; β1j 

＝ γ10; β2j ＝ γ20; β3j ＝ γ30; β5j ＝ γ50 

 

Student-level (Equation 4-5) 

Core Competency ij ＝ β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ γij 

 

School-level (Equation 4-6) 

β0j ＝ γ00 ＋ γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ 

γ07(Skipping classes j) ＋ U0j; β1j ＝ γ10; β2j ＝ γ20; β3j ＝ γ30 

 

Student-level (Equation 4-7) 

Core Competency ij ＝β0j＋β1j(Gender ij)＋β2j(Grade ij)＋β3j(Socio-economic status ij) 

＋ β5j(Relational bullying ij) ＋ γij 

 

School-level (Equation 4-8) 

β0j ＝ γ00 ＋ γ01(School type j) ＋ γ02(School size j) ＋ γ03(School location j) ＋ 

γ04(School-level socio-economic status j) ＋ γ05(Sense of shared belonging j) ＋ 

γ07(Skipping classes j) ＋ U0j; β1j ＝ γ10; β2j ＝ γ20; β3j ＝ γ30; β5j ＝ γ50 

Discussion 

Based on the findings of previous single-level studies, our research expanded upon this 

framework in developing a multi-level mediation model using the 2015 PISA data from 

mainland China. The results demonstrate that two independent variables played indirect 

and positive roles in terms of influencing students’ core competencies by reducing 

school bullying. These two independent variables are a sense of shared belonging 

(which had a positive effect on students’ competencies) and skipping classes (which 

when lowered, is associated with better outcomes in terms of students’ competencies). 

These school-level variables reflect schools’ administrative efforts to develop a positive 

environment for teaching and learning (reflected in the aggregate score of students’ 

sense of shared belonging) and the schools’ attentive administrative efforts, wherein 

monitoring of students’ attendance was consistently adopted, as reflected by students’ 

attendance (with more attentive administrations deterring students from skipping class 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 698 

mitigating the relationship between skipping classes and experiencing bullying, as well 

as reducing the negative relationship between skipping classes and students’ core com-

petencies). Since these two independent variables are deemed to be highly practical and 

directly applicable to the field of educational management, the findings of this study 

provide a reference for school administrators to prevent and control school bullying and 

simultaneously improve students’ core competencies. In light of the impact of students 

being late for school or truant, which are in violation of school rules and may indicate 

the presence of school bullying, and the resulting impacts on students’ core competen-

cies, according to this study, school management should focus more closely on students’ 

attendance, particularly in terms of skipping classes. 

Application of Multi-level Mediation Model 

This study adopted a multi-level mediation model, which is rare in research regarding 

both school bullying and students’ core competencies. In contrast, most studies use re-

gression analysis including multiple predictor variables in the model simultaneously. 

However, a multi-level mediation model is a more appropriate method for further elabo-

rating on the specific paths of influence among factors that, in turn, can provide more 

specific feedback and recommendations for educational administrators. Studies adopt-

ing regression analysis on school bullying among adolescents in Guangdong Province 

(Wang et al., 2012), Ontario, Canada (Betts, 2014), and Quebec, Canada (Di Stasio, 

Savage, & Burgos, 2016) have been conducted, but with mixed or unclear results. 

Likewise, Zhao et al. (2017) and Huang (2017), in analyzing science competency and 

school bullying among students from four provinces of China, also adopted the 2015 

PISA data. However, although these studies found several significant explanatory vari-

ables at different levels, in terms of core competencies or school bullying, they were 

unable to further specify paths of influence for the factors included in their models, due 

to their use of multiple regression. 

Thus, our multi-level mediation model aims to evaluate the factors that have 

demonstrated significant explanatory power in the past, and further evaluating paths 

among variables and their relative influence, which has potential theoretical contribu-

tions for the related research topics. At present, only Wang and Meng (2017) have pub-

lished research using a multi-level mediation model, finding that the school atmosphere 

directly influences a number of student-level variables, which then has an indirect posi-

tive role in science competency. With the application of multilevel analysis in educa-

tional research becoming more and more popular, we expect that multilevel mediation 

model applications will be developed more extensively in the future. 

Attentive Measures for Preventing School Bullying 

The results of this study demonstrate, in terms of school administration, that it is bene-

ficial for schools to enact measures that combine attentive approaches to the enforce-

ment of school rules (in particular, students’ attendance) as well as measures that can 

develop a sense of shared belonging. Building students’ sense of shared belonging and 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 699 

paying greater attention to students’ attendance can reduce the occurrence of relational 

bullying among students, resulting in an indirect positive impact on their performance, 

in terms of core competencies. These two types of administrative approaches can jointly 

contribute to the development of a safe and positive learning environment for students. 

In this learning atmosphere, bullying can be reduced and, as a result, students may ex-

perience greater feelings of acceptance, resulting in greater willingness to seek help 

from others, even if they are bullying by their peers and avoidance of subsequent bully-

ing. At the same time, when school administrators and teachers are focused on monitor-

ing students’ attendance, they can identify students who may be suffering from school 

bullying (which may be the underlying cause of student lateness, skipping classes, or 

truancy), which is conducive to immediate interventions. The results of this study echo 

the recommendations of PISA’s school bullying report (OECD, 2017a) which calls on 

schools to create environments where students feel closer connections with teachers and 

clearly recognize that the school is an orderly place wherein school rules are followed. 

This recommendation is highlighted by the findings of this study that suggest school 

administration efforts can make students feel at ease and can prevent bullying events. 

Limitations 

The paths constructed by our multi-level mediation model can provide empirical evi-

dence that school bullying serves as a mediator between the school administrative 

measures and students’ core competencies. However, it is still unclear whether or not 

there are other mechanisms influencing the relationships among the variables evaluated 

in this study. Since some factors are beyond the scope of this study and are not ad-

dressed in this manuscript, we suggest this lack of data as a limitation and potential for 

future analysis. For example, if a school adopts a more attentive model of student at-

tendance management, this approach might first influence the overall sense of shared 

belongingness which, in turn, can decrease bullying and its influence on students’ core 

competencies. As such, future studies can evaluate multiple paths of mediation. Next, 

since this study was based on a multi-level mediation model, we were required to fol-

low corresponding steps for data analysis. During this process, although we found other 

interesting findings not directly related to the primary focus of this study there were not 

described in more detail in this manuscript. For example, the results of the first step of 

our multi-level mediation model demonstrated no significant correlations between ver-

bal and physical bullying and students’ core competencies; therefore, these two types of 

school bullying were not included in follow-up analyses, and their impact in terms of 

school management measures was not tested further. Finally, the data used in this study 

was limited to data reported in the 2015 PISA survey, which limits the explanatory 

power of some research results. For example, it is impossible to conclude why a sense 

of shared belonging at the school-level did not affect students’ reading competency but 

had a positive effect on mathematics and science competencies. Since this issue cannot 

be explained by existing 2015 PISA data, further data collection and analysis are rec-

ommended. 

 



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 700 

Acknowledgments: We highly appreciate Professor Gamble's help in translating articles. 

Professor Gamble has made a great contribution to the proofreading and semantic correc-

tion of the article, so that the Chinese version can be accurately translated into English. 

 

 

 

 

 

 

 

References

Betts, C. (2014) Multi-level analysis of stu-

dent and school-level characteristics asso-

ciated with bullying victimization in On-

tario high school students (Unpublished 

master’s thesis).University of Waterloo, 

Ontario. Canada. 

Chen, C. Z. & Zhi, T. X. (2017) The influenc-

ing factors of school bullying and the 

construction of long-term prevention 

mechanisms - Based on the analysis of 

2015 bullying behavior measurement data 

of youth campus. Edu Dev Res, 20:31-41. 

Chen, Z .Z. (2017) The impact of non-

cognitive factors on the comprehensive 

literacy of Chinese 15-year students - 

Based on PISA 2015 survey data. Edu 

Acad Monthly, 4:80-88. 

Di Stasio, M. R., Savage, R., & Burgos, G. 

(2016) Social comparison, competition, 

and teacher-student relationships in junior 

high school classrooms predict bullying 

and victimization. J Adoles, 53:207-216. 

Didaskalou, E., Roussivergou, C., & Andreou, 

E. (2017) School belongingness and cop-

ing with victimization in bullied and non-

bullied students: A Discriminant analysis 

approach. J Modern Greek Stud, Special 

Issue:110-124. 

Dong, W. & Yu, G. L. (2010) The Influence 

of adolescent academic emotions on aca-

demic achievements. Psychol Sci, 4:934-

937+945. 

Due, P., Merlo, J., Harel-Fisch, Y., 

Damsgaard, M. T., Holstein, B. E., 

Hetland, J., Lynch, J. (2009) Socioeco-

nomic inequality in exposure to bullying 

during adolescence: A comparative, cross-

sectional, multilevel study in 35 countries. 

Am J Pub Health, 99(5): 907-914. 

Fang, J. H. (2007) Psychological analysis and 

correction strategies of students’ absen-

teeism. Education Science, 23(6), 87-90. 

Freeman, T. M., Anderman, L. H., & Jensen, 

J. M. (2007) Sense of belonging in col-

lege freshmen at the classroom and cam-

pus levels. J Exp Edu, 75(3):203-220. 

Gastic, B. (2008) School truancy and the dis-

ciplinary problems of bullying victims. 

Edu Rev, 60(4):391-404. 

Huang, L. (2017) The situation of school 

bullying in China’s 15-year-old school 

students and its influencing factors - 

Based on the analysis of PISA 2015 data 

of four provinces and cities in China. Edu 

Sci Res, 11:36-42. 

IEA. (2018) Help Manual for the IEA IDB 

Analyzer (Version 4.0). Hamburg, Ger-

many. (Available from 

www.iea.nl/data.html) 

Jansen, P. W., Verlinden, M., Dommisse-van 

Berkel, A., Mieloo, C., van der Ende, J., 

Veenstra, R., Verhulst, F.C., Jansen, W., 

Tiemeier, H. (2012) Prevalence of bully-

ing and victimization among children in 

early elementary school: Do family and 

school neighborhood socioeconomic sta-

tus matter? BMC Pub Health, 12(1):494-

503.  



Chen et al. Anti-Bullying Measures and Students’ Core Competencies. 

Vol.5, No. 2, 2020 701 

Li, F. X. (2017) School bullying behavior 

survey and intervention strategy research. 

Ningxia Soc Sci, 3:133-136. 

Liu, Y., & Liu, Z. (2011) Trajectories of Chi-

nese students’ sense of school belonging 

and academic achievement over the high 

school transition period. Learn Indiv Diff, 

21(2):187-190.  

Nakamoto, J., & Schwartz, D. (2010) Is peer 

victimization associated with academic 

achievement? A meta-analytic review. 

Soc Dev, 19(2):221-242. 

OECD (2017a) How much of a problem is 

bullying at school? PISA in Focus, No. 74. 

Paris: OECD Publishing. 

https://doi.org/10.1787/728d6464-en. 

OECD (2017b) PISA 2015 Technical Report. 

Paris: OECD Publishing.  Retrieved from: 

https://www.oecd.org/pisa/sitedocument/P

ISA-2015-technical-report-final.pdf 

Stewart, E. B. (2008) School structural char-

acteristics, student effort, peer associa-

tions, and parental involvement: The in-

fluence of school-and individual-level 

factors on academic achievement. Edu 

Urban Society, 40(2):179-204. 

Wang J. Y. & Meng, H. Y. (2017) Multi-level 

analysis of the influence of school atmos-

phere on the scientific literacy of middle 

school students: based on the research of 

PISA 2012 Shanghai data. Foreign Prim 

Second Edu, 1:22-30. 

Wang, H., Zhou, X., Lu, C., Wu, J., Deng, X., 

Hong, L., Gao, X., & He, Y. (2012). Ado-

lescent bullying involvement and psycho-

social aspects of family and school life: A 

cross-sectional study from Guangdong 

Province in China. PLoS One, 

7(7):e38619. 

Wang,Y. J., Hong, Y. Q., & Zhang, W. (2015) 

Peer invasion phenomenon investigation 

and intervention strategy. Contemp Youth 

Res, 2:112-118. 

Wen, F. X. & Chiou, H. Z. (2009) Multi-level 

model methodology: key issues and trial 

of hierarchical linear models. NTU 

Manag Rev, 19(2):263-294. 

Wen, Z. L. &Ye, B. J. (2014) Analyses of 

mediating effects: The development of 

methods and models. Adv Psychol Sci, 

22(5):731-745. 

Wu, J. N. (2013) Key features and implica-

tions of successful schools – A review of 

PISA 2009 report “what makes school 

success”. Foreign Edu Res, 1:39-45. 

Yang, F., Yu, B., Zhu, Y. X., & Xu, Q. Y. 

(2017) Correlation between school bully-

ing and school belonging: Evidence from 

new educational experiments. Cours 

Textb Teach Meth, 5:113-120. 

Zhao, D. C, Guo, Y. G., & Jiao, L. Y. (2017) 

The scientific literacy performance of 15-

year-old students in four provinces (cities) 

and their influencing factors - Based on 

the analysis of PISA 2015 data. Edu Res, 

6:80-86. 

Zhao, J. & Zhu, P. Y. (2012) School connec-

tion tightness and juvenile causal experi-

ence research - Taking lessons, truancy, 

and dropout as indicators. Edu Res Exp, 

1:51-56. 

 

Received: 21 April 2020 

Revised: 02 June 2020 

Accepted: 12 June 2020 

 

 

The Chinese version of this article has been published in Journal of Sichuan Normal University (So-

cial Sciences Edition), 2019; 46(4):85-95. The English version has been authorized for being publica-

tion in BECE by the author(s) and the Chinese journal. 

陈奕桦，谢妮，孟志远., 校园欺凌防治与中学生核心素养关系实证研究. 四川师范大学学报 (社
会科学版) 2019; 46(4):85-95. 

 


	Article-YihuaChen-BECE_title_15July2020
	Article-YihuaChen-BECE_Maintext15July2020

