Factor Analysis and Development of a Short Version of the Bullying Participant Behaviors Questionnaire (BPBQ) in a German Sample of Adolescents

Abstract

The Bullying Participant Behavior Questionnaire (BPBQ) is widely used to assess bullying roles in different populations. This study aimed to investigate the structure, construct validity, and reliability of a German version of the BPBQ in a sample of 250 German-speaking adolescents aged 10–16 years (M = 12.92; SD = 1.36; 51.6% female). A second goal was to develop a short version that retains key behaviors but enhances usability. Results indicated that the five oblique factors model fit the long version best, consistent with the original BPBQ and prior research. The short version demonstrated a better model fit, strong internal consistency, and did not require error correlations or modifications. The shorter format improves the questionnaire’s practicality while preserving its accuracy. This study underscores the importance of refining tools like the BPBQ to better capture bullying dynamics and highlights the advantages of a concise version for varying purposes.

Keywords

Bullying Participant Behaviors Questionnaire adolescents confirmatory factor analysis short version

Introduction

Bullying is a multifaceted and dynamic phenomenon (Rose et al., 2015) that occurs in schools worldwide (Cosma et al., 2020; Eilts et al., 2022; Schütz et al., 2022). It is generally defined as aggressive behavior intended to harm others repeatedly over time, characterized by a perceived power imbalance between the bully and the victim (Olweus, 2013; Smith, 2016). The negative consequences of bullying, such as depression (Victim and Bully), academic difficulties (Victim and Bully), and delinquency (Bully), emphasize the importance of early detection and intervention of bullying (Moore et al., 2017; Zych et al., 2017). Owing to its social and dynamic nature, bullying often involves a range of participants beyond just the bully and the victim (Espelage & Swearer, 2003; Rodkin & Hodges, 2003; Thornberg, 2007). The social-ecological model posits that the entire school environment contributes to the persistence of bullying behaviors, involving every individual in some capacity, whether directly or indirectly (Swearer & Espelage, 2004, 2011). While research has traditionally concentrated on the roles of the bully and the victim (Bäker & Schütz-Wilke, 2023; Fischer et al., 2020; Jenkins et al., 2022), a broader perspective encompasses other roles such as assistants to the bully, reinforcers of the bully, defenders of the victim, and outsiders (Jenkins et al., 2022; Salmivalli et al., 1996). This comprehensive concept facilitates a deeper understanding of the various influences within schools and communities, underscoring the importance of examining the roles of all participants in bullying situations (Jenkins et al., 2022). To identify these roles, the Bullying Participant Behaviors Questionnaire (BPBQ), developed by Summers and Demaray (2008), can be used. The BPBQ has been extensively used in studies (e.g., Eilts & Bäker, 2024; Xie et al., 2023), has been psychometrically validated in both U.S. (e.g., Demaray et al., 2016) and Chinese samples (Qiu et al., 2021), and has demonstrated high validity and reliability in studies involving elementary school children (e.g., Jenkins et al., 2022) and middle school adolescents (e.g., Demaray et al., 2016; Jenkins & Canivez, 2021). However, studies analyzing the factor structure tested and identified different models and structures (e.g., five or four orthogonal or oblique factors, higher-order factor, and bifactor models, see Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021). To our knowledge, no research to date has yet examined the factor structure within a German sample. Thus, the primary aim of this study is to translate and evaluate the psychometric structure of the BPBQ in a community sample of German-speaking adolescents, contributing to the existing body of research on different factor structures across diverse samples and countries. Additionally, given the identification of different factor structures and the BPBQ’s length of 50 items, a secondary aim of this study is to develop and test a German short version of the BPBQ. A shorter version could facilitate more efficient screening for bullying behaviors in classroom settings, enabling quicker data-based interventions.

Theoretical Background

Bullying

Olweus (1993) defines bullying as a series of negative actions directed at a student repeatedly and over time. These actions can originate from another student or a group of students. Bullying manifests primarily in three forms: physical bullying (i.e., intentional infliction of injury such as pushing or hitting), verbal bullying (e.g., threats, mockery, teasing, or insults), and relational bullying (e.g., social exclusion or spreading rumors; Card & Little, 2006; Margraf & Pinquart, 2016). Olweus (1991, 1993) also emphasizes that bullying involves repetitive actions against an individual, but he acknowledges that a single severe incident can be classified as bullying under certain circumstances. A single severe incident can be classified as bullying if it meets the other key criteria (Imbalance of power: The aggressor holds more power, whether physical, social, or psychological, over the victim; Intent to cause harm: The incident is deliberate and aimed at causing distress or injury; Significant harm: The severity of the incident is so extreme that, despite being a one-time occurrence, it creates serious emotional, physical, or psychological damage to the victim). Salmivalli et al. (1996) identified different bullying roles which can be captured using the BPBQ, developed by Summers and Demaray (2008). The BPBQ comprises 50 self-report items aimed at identifying five specific roles in bullying contexts: (1) bully (those who consistently and intentionally exert aggression toward peers they overpower), (2) victim (the individual subjected to this aggression), (3) assistant to the bully (those who support the bully), (4) defender of the victim (those who support the victim), and (5) outsider (those who passively ignore or overlook the bullying).

Studies discuss the pros and cons of self-report versus third party assessments, such as teacher or peer nominations (Furlong et al., 2010). Self-reports better capture the student’s perspective and detect both overt and covert bullying (Furlong et al., 2010; Griffin & Gross, 2004), and are easier to implement compared to methods like behavioral observations (Thomas et al., 2015). Efficient measuring of bullying in schools is not only relevant for researchers but also for teachers, to identify problems in class and choose targeted interventions (Eilts et al., 2023).

Based on the 2022 Health Behaviour in School-aged Children study in Germany, 14% of 11-, 13-, and 15-year-olds reported experience with school bullying, showing a general decrease since 2009/10 but stability between 2017/18 and 2022 (Fischer & Bilz, 2024). Salmivalli et al. (1996) reported varying prevalence rates for different roles in bullying: 8.2% as bullies, 6.8% as assistants, 19.5% as reinforcers, 11.7% as victims, 17.3% as defenders, 23.7% as outsiders, and 12.8% without a clear role. The negative consequences of bullying affect not only the victims (e.g., PTSD, depression, and suicidal ideation; Moore et al., 2017) but also the bullies (e.g., externalizing behavior problems and academic difficulties; Schoeler et al., 2018), assistants or reinforcers (e.g., substance abuse; Quinn et al., 2016), and outsiders (e.g., depressive symptoms and emotional regulation issues; Camodeca & Nava, 2022; Midgett & Doumas, 2019). Additionally, positive effects are observed for defenders (e.g., increased empathy; Rizkyanti et al., 2021) offering opportunities for intervention approaches in school (Ruggieri et al., 2013).

Factor Structure of the Bullying Participant Behaviors Questionnaire

Originally developed as a five-dimensional measure, recent discussions opened the discourse to other factor structures regarding bullying. Aside from the “classical” roles of bully and victim, and the six roles identified by Salmivalli et al. (1996), literature also discusses pro-bullying and pro-victim roles (e.g., Jenkins & Canivez, 2021). These discussions thus lead to the examination of different factor structures in recent psychometric studies of the BPBQ (e.g., Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021). In total, the authors (e.g., Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021) tested 14 hypothesized models. These models were divided into two sets, with one set considering five group factors: bully, assistant, outsider, victim, and defender, and the other set considering four group factors: bully/assistant, outsider, victim, and defender. In exploring the factor structure, the central distinction among the models lies in how they conceptualize the relationships between the group factors.

Model 1 assumed five or four independent (orthogonal) factors, implying no correlation between them. In contrast, Model 2 allowed for correlations among the five or four factors, adopting an oblique approach that assumes potential interrelationships. Model 3 introduced a nuanced variation of the oblique model, clustering the factors into two correlated sets: one set comprising bully, assistant, and outsider (or a combined bully/assistant factor and outsider), and the other set consisting of victim and defender. Moving toward hierarchical structures, Model 4 extended the assumptions of Model 2 by proposing a higher-order model where a single general factor influenced the five or four group factors. Model 5 further refined Model 3 by introducing two higher-order dimensions: one termed Pro-Bully, which influences the bully, assistant, and outsider factors (or bully/assistant and outsider), and another termed Pro-Victim, influencing the victim and defender factors. Further extending these hierarchical approaches, Model 6 adopted a bifactor structure based on Model 4. This model retained a general factor but included additional specific group factors, tested with both five and four group factors. Finally, Model 7 paralleled Model 5 but took a bifactor approach with two general dimensions—Pro-Bully and Pro-Victim—in combination with specific group factors, also tested with five and four factors.

After comparing these models, the data generally supported the five-factor model (Jenkins et al., 2022; Qiu et al., 2021), which provided the clearest structure for understanding the relationships between the group factors. However, Jenkins and Canivez (2021) identified the two-general-factor model as the best-fitting model, indicating that the higher-order dimensions of Pro-Bully and Pro-Victim provided a more comprehensive explanation of the data. Thus, all identified models serve as the foundation for the current analysis.

Current Study

In the current study, we aim to explore the different models proposed by Jenkins and Canivez (2021) using a German translation of the BPBQ (Summers & Demaray, 2008). By applying these models to our dataset, we seek to validate and compare their applicability in the context of German-speaking populations. Furthermore, we will endeavor to develop a short version of the BPBQ that retains the essential components (according to Olweus, 1993 definition of bullying) of the full measure while offering a more concise tool for practical use in research and educational settings. This short version aims to provide an efficient yet comprehensive assessment of bullying roles and behaviors, making it accessible for broader application in schools.

Method

Participants and Procedure

The sample for this study comprised students from German secondary schools who were contacted via email and invited to participate. The schools were chosen at random from a list of schools in the relevant federal states. Once the schools agreed to participate, information letters were sent to both students and their parents, informing them about the study’s purpose, voluntary participation, procedures, and the measures taken to ensure data confidentiality. Participation required informed written consent from both the students and their parents, which was obtained through signed consent forms provided alongside the informational letters. A total of 252 adolescents provided consent to participate in the study; however, two chose not to take part after consent was obtained, resulting in a final sample of N = 250 German-speaking adolescents aged between 10 and 16 years (M = 12.92; SD = 1.36; 51.6% female).

Following the receipt of informed consent, data collection occurred during school hours in a controlled group setting. Students completed the paper-and-pencil questionnaire under the supervision of researchers or trained supervisors in a quiet classroom environment. This setup ensured that participants could ask for clarification if needed and that the integrity and confidentiality of their responses were maintained. The study received approval from the university’s ethics committee (Approval Number: Drs.Nr.EK/2020/047), which ensured that all procedures adhered to ethical standards for research involving human participants. Furthermore, the data protection authority provided the necessary approval, ensuring that data handling complied with privacy regulations.

Instrument

The BPBQ (Summers & Demaray, 2008) was used in this study. The BPBQ includes 50 items, with 10 items for each of the five roles: Bully (e.g., “I have made fun of another student”), Victim (e.g., “I have been ignored”), Assistant (e.g., “When someone was verbally threatening another student, I joined in”), Defender (e.g., “I defended someone who had things purposely taken from them”), and Outsider (e.g., “I ignored it when someone else pinched or poked another student”). This structure allows for a detailed assessment of each participant’s involvement in various bullying behaviors, providing a comprehensive understanding of their roles. The instructions of the questionnaire do not provide an explicit definition of bullying or label the roles at the beginning (Summers & Demaray, 2008). This is done to avoid priming the students (e.g., Kert et al., 2010; Vaillancourt et al., 2008; Vivolo-Kantor et al., 2014). Instead, it queries participants about situations that embody the essence of bullying and the corresponding roles. Participants were asked to rate their behavior over the past 30 days on a five-point scale, ranging from (0) never, (1) 1 to 2 times, (2) 3 to 4 times, (3) 5 to 6 times, to (4) 7 or more times. Therefore, higher scores reflect greater frequency of involvement in or experience with that particular role. Originally designed in English, the BPBQ was translated into German for this study using a back-translation method to ensure accuracy and cultural relevance. A fluent German speaker with proficiency in English conducted the initial translation. A second independent German speaker, proficient in English, then back-translated the items into English. Both translators compared the back-translation with the original questionnaire to identify discrepancies and refine the German version accordingly, ensuring conceptual equivalence.

Data Analysis

The aim of our study was to examine the factor structure of the BPBQ (Summers & Demaray, 2008) within a German sample of children using confirmatory factor analysis (CFA) in R (R Core Team, 2023; the R code is available at: https://osf.io/jxzub/). To achieve this, we tested various models informed by existing literature on the BPBQ. In total, 14 models outlined in the theoretical section were evaluated (e.g., Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021). The measured variables in this analysis range from 0 (never), 1 (1–2 times), 2 (3–4 times), 3 (5–6 times), to 4 (7 or more times). Given that the mean values of most items were below 1, indicating potential zero-inflation, we conducted an additional analysis comparing the observed proportion of zero responses to the expected proportion under a Poisson distribution. We included this step as an initial exploratory analysis to assess whether zero-inflation posed a systematic issue across the scale. The decision to compare the observed frequency of zero responses to an expected Poisson distribution was based on literature for evaluating potential deviations, such as zero-inflation, in ordinal or count data distributions (e.g., Feng et al., 2014; Min & Agresti, 2002). The results showed that while some items exhibited significant zero-inflation, this was not a systematic issue across the scale. Additionally, we have added a table to the supplemental material that presents the number of zero responses per item, providing a more detailed account of the zero-inflation patterns in the data (S1 Zero response). Since the BPBQ employs an ordinal Likert-type response format (0–4), we retained robust maximum likelihood estimation (MLM) as the most suitable method for our analysis, consistent with prior studies that have also used robust estimators (Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021). To address zero inflation, in addition to conducting confirmatory factor analyses (CFA), we also estimated a Two-Part Zero-Inflated CFA to account for the presence and absence of behaviors (Kim & Muthén, 2009). We tested the Two-Part Zero-Inflated CFA models using the weighted least square mean and variance adjusted (WLSMV) estimators. To keep consistency with previous research on the BPBQ, we report the results of both analyses.

The quality of these models was assessed using robust estimates of the Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), root mean square error of approximation (RMSEA), and standardized root mean squared residual (SRMR). A good model fit was characterized by RMSEA and SRMR values below .08, and CFI and TLI values above .90 (Hu & Bentler, 1999). Chi-square values were also reported, although the sensitivity of the Chi-square statistic in larger samples (N >200) should be considered (Vandenberg, 2006). Effect coding was used in the final reported models of the CFA (Little, 2013). Missing data, which comprised 15.2% of cases with at least one missing item, was excluded from the analysis. Factor loadings were assessed according to Stevens (2002) and Hair et al. (2009). A power analysis using semPower (Moshagen & Bader, 2024) was conducted to determine the necessary sample size for performing a CFA with five correlated latent factors. The expected correlation between the factors was set to .30, with each factor being measured by 10 indicators. The average factor loading was assumed to be .40. The analysis was performed using an a-priori approach with a significance level (α) of .05 and a desired statistical power of .80. The results indicated that a minimum sample size of 201 participants was required to detect the expected effects with sufficient power.

The final model with the best fit, including error correlations with similar overlaps in content (e.g., Item 25: I have been pushed around, punched, or slapped, and Item 26: I have been pushed or shoved; Cole et al., 2007; Urban & Mayerl, 2014), was then analyzed to identify potential items for omission in creating a short version of the BPBQ. Additionally, the internal consistencies of the questionnaire items within the identified factors were determined using Cronbach’s alpha and McDonald’s Omega.

Results

Descriptive Results

Prior to confirmatory testing, descriptive statistics and intercorrelations of the BPBQ items were calculated and presented in Table 1 and Table S2 in the supplementary material.

Table 1.

Descriptive Statistics of the Items.

Item		M	SD
1	I have called another student bad names	1.29	1.47
2	I have made fun of another student	0.85	1.07
3	I have purposely left out another student	0.26	0.66
4	I have pushed, punched or slapped another student	0.61	1.02
5	I have told lies about another student	0.09	0.40
6	I have tried to make people dislike another student	0.11	0.40
7	I have stolen things from another student	0.15	0.60
8	I have thrown things at another student	0.43	0.83
9	I have said bad things about another student	0.76	0.96
10	I have talked about someone behind their back	0.91	1.11
11	When someone was making fun of another student, I joined in	0.47	0.80
12	When someone was verbally threatening another student, I joined in	0.02	0.14
13	When someone bumped into another person, I have joined in	0.13	0.51
14	I have made fun of someone when they were pushed, punched, or slapped	0.21	0.63
15	I have made fun of someone who was being called mean names	0.32	0.71
16	When someone else broke something that belonged to another student, I stopped to watch	0.33	0.71
17	When someone else tripped another student on purpose, I laughed.	0.68	1.11
18	When someone else knocked books out of another student’s hands on purpose, I laughed.	0.09	0.41
19	When someone else pinched or poked another student, I joined in	0.18	0.59
20	When someone else threw something at another student, I joined in	0.19	0.53
21	I have been called mean names	1.71	1.53
22	I have been made fun of	1.23	1.32
23	I have been purposely left out of something	0.66	1.10
24	I have been ignored	0.94	1.22
25	I have been pushed around, punched or slapped	0.85	1.28
26	I have been pushed or shoved	0.73	1.01
27	People have told lies about me	0.85	1.19
28	People have tried to make others dislike me	0.58	1.07
29	I have been threatened by others	0.36	0.80
30	I have had things taken from me	0.54	0.94
31	I tried to become friends with someone after they were picked on	0.71	0.91
32	I encouraged someone to tell an adult after they were picked on	0.94	1.13
33	I defended someone who was being pushed, punched, or slapped	1.20	1.24
34	I defended someone who had things purposely taken from them	0.87	1.13
35	I defended someone who was being called mean names	1.15	1.19
36	I tried to include someone if they were being purposely left out	0.94	1.06
37	I helped someone who had their books knocked out of their hands on purpose	0.51	0.86
38	I helped someone who was purposely tripped	0.63	0.93
39	When I saw someone being physically harmed, I told an adult	0.87	1.15
40	I defended someone who I thought was being tricked on purpose	0.88	1.08
41	I pretended not to notice when things were taken or stolen from another student	0.47	0.89
42	I pretended not to notice when rumors were being spread about other students	0.46	0.84
43	I ignored it when I saw someone making fun of another student	0.53	0.89
44	I pretended not to notice a situation that purposely left someone out	0.34	0.81
45	I ignored it when I saw someone breaking or damaging another student’s things	0.34	0.73
46	I pretended not to notice when someone else tripped another student on purpose	0.45	0.98
47	I ignored it when someone else pinched or poked another student	0.62	1.16
48	I ignored it when someone else threw something at another student	0.59	1.07
49	I ignored it when someone else tricked another student	0.66	1.05
50	I pretended not to notice when someone was destroying another student’s property	0.26	0.72

Confirmatory Factor Analyses

In the first step, the results of the CFA, to examine the factor structure of the BPBQ, are presented. We tested 14 alternate CFA models with MLM estimator based on previous research studies (e.g., Jenkins & Canivez, 2021; Jenkins et al., 2022; Qiu et al., 2021). The model-fit indices for all models tested are summarized in Table 2. The results indicate that none of the confirmatory models show a good model fit. Additionally, we tested the same models in a Two-Part Zero-Inflated CFA using WLSMV estimator. The results are reported in Table 3.

Table 2.

Model-Fit Indices of CFA Models of the BPBQ With MLM Estimator.

Model	X ²	df	p	TLI	CFI	RMSEA	95% CI	SRMR
Four orthogonal factors (B/A, O, V, D)	1652.987	1179	<.001	.773	.781	.060	[.053; .067]	.177
Five orthogonal factors (B, A, O, V, D)	1746.884	1180	<.001	.731	.741	.066	[.059; .072]	.197
Four oblique factors (B/A, O, V, D)	1561.522	1173	<.001	.816	.823	.054	[.047; .061]	.088
Five oblique factors (B, A, O, V, D)	1555.227	1170	<.001	.818	.826	.054	[.047; .061]	.097
Two oblique (B/A, O)/two oblique (V, D)	1586.482	1177	<.001	.803	.811	.056	[.049; .063]	.136
Three oblique (B, A, O)/two oblique (V, D)	1585.854	1176	<.001	.804	.812	.056	[.049; .063]	.144
Four group factors, one higher order	1631.816	1175	<.001	.785	.793	.059	[.052; .065]	.119
Five group factors, one higher order	1727.008	1175	<.001	.746	.757	.064	[.057; .070]	.130
Four group factors, two higher order	1632.014	1174	<.001	.784	.793	.059	[.052; .065]	.119
Five group factors, two higher order	1730.078	1174	<.001	.745	.755	.064	[.057; .070]	.129
Four group factors, one general bifactor	1571.180	1129	<.001	.791	.808	.058	[.051; .064]	.144
Five group factors, one general bifactor	1694.727	1130	<.001	.743	.763	.064	[.058; .070]	.122
Four group factors, two general bifactor	1520.872	1119	<.001	.809	.825	.055	[.048; .062]	.137
Five group factors, two general bifactor	1530.077	1120	<.001	.801	.818	.056	[.049; .063]	.137

Note. X² = Chi-square statistic, df = degrees of freedom, TLI = Tucker–Lewis Index; CFI = Comparative Fit Index, RMSEA = root mean square error of approximation, SRMR = standardized root mean squared residual.

Table 3.

Model-Fit Indices of CFA Models of the BPBQ With WLSMV Estimator and With Two-Part-Modelling.

Model	X ²	df	p	TLI	CFI	RMSEA	95% CI	SRMR
Four orthogonal factors (B/A, O, V, D)	11,395.571	4834	<.001	.968	.969	.215	[.213; .217]	.321
Five orthogonal factors (B, A, O, V, D)	9605.040	4825	<.001	.978	.978	.190	[.179; .182]	.261
Four oblique factors (B/A, O, V, D)	10,238.705	4822	<.001	.976	.977	.187	[.186; .189]	.292
Five oblique factors (B, A, O, V, D)	7285.050	4805	<.001	.990	.990	.122	[.120; .123]	.191
Two oblique (B/A, O)/two oblique (V, D)	11,111.381	4830	<.001	.971	.971	.207	[.205; .209]	.312
Three oblique (B, A, O)/two oblique (V, D)	8572.536	4817	<.001	.983	.984	.156	[.154; .157]	.229
Four group factors, one higher order	Not converge
Five group factors, one higher order	Not converge
Four group factors, two higher order	Not converge
Five group factors, two higher order	Not converge
Four group factors, one general bifactor	Not converge
Five group factors, one general bifactor	Not converge
Four group factors, two general bifactor	Not converge
Five group factors, two general bifactor	Not converge

Modified Confirmatory Testing (Post-Hoc Analyses)

Post-hoc analyses were performed by calculating a modified CFA model. The five oblique factor model was chosen for this purpose as it showed one of the best fits in both analysis (CFA with MLM and Two-Part Zero-Inflated CFA with WLSMV) and was the model proposed by the authors of the BPBQ. To identify poorly discriminating items, item-total correlations across the BPBQ items were considered. Items with item-total correlations below .40 (Hair et al., 2009) were excluded in the subsequent confirmatory testing. The items deleted were Item 5 (Bully; λ = .327), Item 12 (Assistant; λ = .351), Item 16 (Assistant; λ = .235), and Item 18 (Assistant; λ = .344). However, this deletion did not significantly improve the model fit. Consequently, errors for items with overlapping content were allowed to correlate (Cole et al., 2007; Urban & Mayerl, 2014). The following error terms were correlated: 1 and 2, 3 and 6, 9 and 10, 4 and 8, 13 and 14, 11 and 15, 13 and 20, 21 and 22, 22 and 23, 23 and 24, 25 and 26, 27 and 28, 32 and 39, 33 and 39, 37 and 38, 31 and 40, 36 and 39, 42 and 43, 45 and 50, 43 and 49, and 47 and 48. Despite this, the error correlations did not improve the model fit to a satisfactory level. Therefore, modification indices were examined to identify further improvements. Changes were made iteratively until the model fit was deemed satisfactory, with suggested modifications incorporated only if they were theoretically justified. Correlations between error terms were allowed between the roles of the Bully and the Victim, given that the bully–victim role has been shown to exist. Additionally, correlations between the Bully and the Assistant were allowed due to their high correlation (Eilts et al., 2024). Specifically, based on these theoretical considerations, the following error term correlations were added: Items 1 and 21, 2 and 11, 4 and 25, 8 and 20, and 3 and 28. Although these additional correlations improved the model fit, it was still not satisfactory. As a final step, items with high cross-loadings were successively deleted, and the corresponding error term correlations were removed. The items deleted were Item 2 (Bully), which also loaded on the Victim scale; Item 14 (Assistant), which also loaded on the Outsider scale; and Item 31 (Defender), which also loaded on the Victim scale. The final model achieved a satisfactory fit with χ² = 994.904; df = 828; p < .001; CFI = .923; TLI = .916; RMSEA = .041; SRMR = .071.

Results of the Short Version

For the development of a short version of the BPBQ, the fourth model with the modifications was used. Items with the lowest loadings and those that had overlapping content (e.g., Item 9: I have said bad things about another student, and 10. Item: I have talked about someone behind their back, or Item 25: I have been pushed around, punched, or slapped, and Item 26: I have been pushed or shoved) with other items were deleted. In each case, we retained the item with the higher factor loading and better contribution to model fit, ensuring that the shortened measure maintained strong psychometric properties while minimizing redundancy. The final short version included the following items: Bully: 1, 3, 4, 8, and 9; Assistant: 11, 13, 15, 17, 19, and 20; Victim: 21, 22, 24, 26, 27, and 29; Defender: 33, 34, 35, 36, 39, and 40; and Outsider: 43, 44, 45, 46, 48, and 49. The final model with MLM estimator (without any error term correlations) showed a satisfactory fit with χ² = 459.003; df = 367; p < .001; CFI = .933; TLI = .926; RMSEA = .046; SRMR = .065. Factor loadings and Cronbach’s Alpha values are presented in Table 4. The final Two-Part Zero-Inflated CFA model with WLSMV also showed an acceptable fit (χ² = 2949.467; df = 15,500; p < .001; CFI = .989; TLI = .988; RMSEA = .064; SRMR = .161). Factor loadings for this model are presented in Table 5. The items of the short version are detailed in the supplemental material (S3).

Table 4.

Factor Loadings and Reliability of the Short Version (MLM Estimator).

Factor	Item	Loading	p-Value	α (ω)
Bully	BPBQ1	.66	<.001	.80 (.84)
	BPBQ3	.58	<.001
	BPBQ4	.72	<.001
	BPBQ8	.72	<.001
	BPBQ9	.66	<.001
Assistant	BPBQ13	.67	<.001	.83 (.90)
	BPBQ15	.71	<.001
	BPBQ17	.71	<.001
	BPBQ11	.62	<.001
	BPBQ19	.68	<.001
	BPBQ20	.73	<.001
Victim	BPBQ21	.82	<.001	.86 (.90)
	BPBQ22	.79	<.001
	BPBQ24	.71	<.001
	BPBQ26	.67	<.001
	BPBQ27	.72	<.001
	BPBQ29	.57	<.001
Defender	BPBQ33	.82	<.001	.88 (.93)
	BPBQ34	.83	<.001
	BPBQ35	.82	<.001
	BPBQ36	.66	<.001
	BPBQ39	.68	<.001
	BPBQ40	.70	<.001
Outsider	BPBQ43	.66	<.001	.83 (.88)
	BPBQ44	.55	<.001
	BPBQ45	.65	<.001
	BPBQ46	.70	<.001
	BPBQ48	.80	<.001
	BPBQ49	.77	<.001

Table 5.

Factor Loadings for the Two-Part Zero-Inflated CFA.

Factor	Item	Loading	p-Value
Bully	BPBQ1 cont	.73	<.001
	BPBQ1 zero	.71	<.001
	BPBQ3 cont	.74	<.001
	BPBQ3 zero	.69	<.001
	BPBQ4 cont	.73	<.001
	BPBQ4 zero	.70	<.001
	BPBQ8 cont	.75	<.001
	BPBQ8 zero	.68	<.001
	BPBQ9 cont	.73	<.001
	BPBQ9 zero	.70	<.001
Assistant	BPBQ13 cont	.80	<.001
	BPBQ13 zero	.74	<.001
	BPBQ15 cont	.78	<.001
	BPBQ15 zero	.76	<.001
	BPBQ17 cont	.78	<.001
	BPBQ17 zero	.76	<.001
	BPBQ11 cont	.77	<.001
	BPBQ11 zero	.76	<.001
	BPBQ19 cont	.80	<.001
	BPBQ19 zero	.74	<.001
	BPBQ20 cont	.81	<.001
	BPBQ20 zero	.73	<.001
Victim	BPBQ21 cont	.79	<.001
	BPBQ21 zero	.81	<.001
	BPBQ22 cont	.80	<.001
	BPBQ22 zero	.78	<.001
	BPBQ24 cont	.80	<.001
	BPBQ24 zero	.78	<.001
	BPBQ26 cont	.81	<.001
	BPBQ26 zero	.78	<.001
	BPBQ27 cont	.81	<.001
	BPBQ27	.79	<.001
	BPBQ29 cont	.81	<.001
	BPBQ29 zero	.79	<.001
Defender	BPBQ33 cont	.81	<.001
	BPBQ33 zero	.81	<.001
	BPBQ34 cont	.81	<.001
	BPBQ34 zero	.80	<.001
	BPBQ35 cont	.80	<.001
	BPBQ35 zero	.81	<.001
	BPBQ36 cont	.81	<.001
	BPBQ36 zero	.80	<.001
	BPBQ39 cont	.81	<.001
	BPBQ39 zero	.80	<.001
	BPBQ40 cont	.79	<.001
	BPBQ40 zero	.81	<.001
Outsider	BPBQ43 cont	.79	<.001
	BPBQ43 zero	.75	<.001
	BPBQ44 cont	.78	<.001
	BPBQ44 zero	.77	<.001
	BPBQ45 cont	.79	<.001
	BPBQ45 zero	.76	<.001
	BPBQ46 cont	.79	<.001
	BPBQ46 zero	.76	<.001
	BPBQ48 cont	.79	<.001
	BPBQ48 zero	.76	<.001
	BPBQ49 cont	.79	<.001
	BPBQ49 zero	.75	<.001

Note. Cont = loading on continuous variable, zero = loading on binary variable.

Discussion

The primary purpose of this study was to examine the factor structure of the BPBQ within a German sample. We explored the various factor structures proposed by Jenkins and Canivez (2021) and aimed to develop an economical short version of the BPBQ for use in schools and research settings.

In the following discussion, we first focus on the results of the factor structure analysis and critically evaluate these findings in light of previous research. We also provide implications for the use of the BPBQ in educational settings. This is followed by a discussion of the developed short version of the BPBQ, its potential benefits, and the next steps researchers should consider.

Factor Structure

On the basis of the model-fit indices, the five oblique factors model seems to be the model best fitting to the data without any additional modifications. This is the model that the authors originally intended for the questionnaire (Summers & Demaray, 2008) and this result is also in line with previous research (e.g., Jenkins & Nickerson, 2017; Salmivalli et al., 1996). Therefore, we would also recommend the use of the five distinct roles in further examinations of bullying in German-speaking samples. Our findings are only partly in line with the findings of Jenkins and Canivez (2021) and Jenkins et al. (2022) who also support the five oblique factors model but found a better model fit for the model with additional higher-order factors (pro-bully/pro-victim). This might be due to the use of a bullying definition prior to the questionnaire in these studies. However, the use of a bullying definition is not in line with the instructions for the BPBQ (Summers & Demaray, 2008) and can result in priming the students (Kert et al., 2010; Vaillancourt et al., 2008; Vivolo-Kantor et al., 2014).

To improve the model fit, four original items had to be deleted. These were one bully and three assistant items. This indicates that the role of the assistant does not seem to work as well in a German sample as it does in other samples. One possible explanation for the weaker differentiation of the assistant role in the German sample could be cultural differences in how bullying roles manifest (Scheithauer et al., 2016). In some contexts, bullying behaviors may be more structured, with clear participant roles, whereas in Germany, assistants might blend more into the bully or reinforcer roles. Additionally, the strong emphasis on inclusive education and cooperative learning environments in German schools may influence peer dynamics, potentially reducing the prevalence of clearly defined assistant behaviors. This is consistent with previous research suggesting a high overlap between bully and assistant roles (Jenkins & Canivez. 2021; Jenkins et al., 2022). The modification indices also show that there is a high correlation between the role of the bully and victim and the bully and assistant as these are the error terms that are recommended to be correlated. This is in line with Jenkins and Canivez (2021) who discussed that bullying roles are not exclusive, and adolescents can engage in multiple roles depending on the context. The high correlation between the bully and victims has also been demonstrated in other studies that report the presence of a bully-victim (e.g., Eilts et al., 2024; Vlachou et al., 2011). This is in line with the social-ecological model highlighting the social and dynamic nature of bullying (Swearer & Espelage, 2004, 2011; Thornberg, 2007). Therefore, a strength of the BPBQ is that it addresses the engagement of students in different roles. This also better represents the reality as students rarely occupy only one of the roles.

Teachers could use the BPBQ to inquire which students might occupy more than one role and in which situations this is the case. This could lead to improved data-based intervention strategies that the teachers can use. Additionally, teachers can infer who ignores bullying or who is assisting to see which intervention strategies could be useful for their class (Ruggieri et al., 2013). Nevertheless, this increases the complexity of factor analysis (Jenkins & Canivez, 2021). The Cronbach’s alpha values (ranging from .81 to .91) and MacDonald’s Omega (ranging from .87 to .93) also support the five-factor structure. The bully role has the lowest alpha value and the lowest factor loadings of all of the five roles. This might be due to the self-reporting nature of the instrument. The role of the defender shows the highest alpha value and the highest overall factor loadings of the five bullying roles. This is in line with previous research that also reports the highest alpha values for the defender role (e.g., Qiu et al., 2021).

Short Version

Additionally, we aimed to investigate whether developing a short version of the BPBQ would be feasible for more efficient use in future studies and schools. During the modification and shortening of the questionnaire, we focused on retaining the main behaviors described in the definition of bullying (Olweus, 1993), such as kicking and hitting for the bully role, and including the three forms of bullying (i.e., physical, verbal, and relational; Card & Little, 2006; Margraf & Pinquart, 2016). The short version demonstrates a better model fit than the original five oblique factors model and does not require error correlations or further modifications. Consistent with the final model of the CFA, the bully role has the lowest alpha value, while the defender role has the highest. The alpha values (ranging from .80 to .88) and McDonald’s omega (ranging from .84 to .93) for the short version indicate good reliability. The strength of the short version is economical and reliable application of the instrument in research and schools. Further studies should examine the factor structure of the short version. It should be evaluated in different national and international studies to see whether it is a reliable and economic instrument.

Limitations and Further Research

The instrument does not capture the sociometric dynamics of bullying. Teachers and researchers cannot infer the specific bully–victim relationships, making it impossible to determine whether victims exhibit bullying behaviors toward their own perpetrators. Such behavior would be considered aggressive rather than bullying, as it lacks the power imbalance characteristic of bullying (Smith, 2016). Another limitation is the variation in the use of the BPBQ. Jenkins and Canivez (2021) and Jenkins et al. (2022) provided a definition of bullying, which we did not, affecting the comparability of our results. Future studies should compare BPBQ results with and without a bullying definition to investigate any differences in bullying prevalence and check the measurement invariance. Additionally, as a self-report measure, the BPBQ does not capture classroom dynamics. Future research should include multiple informant perspectives to better understand bullying roles and reduce the likelihood of social desirability bias in students’ responses (Furlong et al., 2010). The lowest alpha values for the bully role also suggest that self-report measures might lead to a lower willingness to report perpetrator behavior.

Prior studies did not account for the high proportion of zero responses in the BPBQ, potentially leading to biased factor structures and inaccurate conclusions. The Two-Part Zero-Inflated CFA (ZICFA) addresses this limitation by distinguishing between the presence of a behavior (zero vs. nonzero responses) and its frequency, providing a more precise measurement. This approach improves model fit, reduces bias in factor loadings, and offers a more nuanced understanding of bullying dynamics, especially for less frequently endorsed roles. Future research should compare ZICFA to standard CFA in BPBQ validation studies.

For future research, the proposed short version should be psychometrically validated. It would also be valuable to examine cultural differences in bullying, such as those arising from differences in school environments (e.g., some countries do not use lockers, so students carry their books; e.g., see Items 18 and 37). Previous studies have also highlighted gender differences (e.g., Bäker et al., 2023; Eilts et al., 2022; Jenkins & Nickerson, 2017) and differences concerning students with disabilities (Eilts et al., 2022; Eilts & Koglin, 2022; Rose et al., 2011; Schütz et al., 2022). Additionally, there are diverse findings regarding age differences in bullying (Blake et al., 2012; Craig et al., 2009; Scheithauer et al., 2006; Swearer et al., 2012). Therefore, future studies should investigate the measurement invariance of the BPBQ across these subgroups.

Conclusion

In conclusion, the five oblique factors model best fits the data without additional modifications, aligning with the original intentions of the BPBQ authors (Summers & Demaray, 2008) and prior research (e.g., Jenkins & Nickerson, 2017; Salmivalli et al., 1996). This supports the use of five distinct roles in studying bullying in German-speaking samples. Since studies from other countries also support the five-factor solution (Jenkins et al., 2022; Qiu et al., 2021) the results seem to be generalizable for other countries as well. However, our findings diverge slightly from those of other studies, such as Jenkins and Canivez (2021) and Jenkins et al. (2022), which found a better fit with models incorporating additional higher-order factors. This discrepancy may be attributed to methodological differences, such as the inclusion of a bullying definition prior to administering the questionnaire. In contrast, our study adhered to the BPBQ’s original instructions, which could influence how respondents perceive and report their experiences.

Furthermore, Jenkins and Canivez (2021) and Jenkins et al. (2022) collected data from entire classes or schools, potentially providing a more comprehensive view of the bullying dynamics. Our study, however, only included a subset of students from each class. Given that bullying is a process that occurs within the classroom and involves various roles for each student (Salmivalli et al., 1996). To better capture the complexities of bullying, future research should aim to collect data from entire classes. By focusing on the factor structure of the questionnaire within a more representative sample, future studies can offer more nuanced insights into the prevalence and interplay of various bullying roles. This approach would likely enhance the development of more effective interventions and prevention strategies, tailored to the specific dynamics observed in real classroom settings.

The study also suggests that the role of the assistant may not function as effectively in a German context, evidenced by the need to remove certain items related to this role. High correlations between roles, such as bully and victim or bully and assistant, highlight the fluidity of these roles among adolescents, reinforcing the idea that students often engage in multiple roles depending on the situation. It could also be a question of the translation of the instrument. However, the results of Jenkins and Canivez (2021) also indicate a low discrimination between the role of the bully and the assistant for their sample. This complexity, while challenging for factor analysis, makes the BPBQ a valuable tool for teachers to understand and intervene in bullying dynamics in their classrooms.

The development of a short version of the BPBQ aimed at retaining key behaviors and bullying forms showed a better model fit than the original, with reliable alpha values, suggesting it as a viable option for more efficient use in future studies and educational settings.

Supplemental Material

Supplemental Material - Factor Analysis and Development of a Short Version of the Bullying Participant Behaviors Questionnaire (BPBQ) in a German Sample of Adolescents

Supplemental Material for Factor Analysis and Development of a Short Version of the Bullying Participant Behaviors Questionnaire (BPBQ) in a German Sample of Adolescents by Jule Eilts, Fabio Sticca, and Jessica Wilke in Journal of Psychoeducational Assessment

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Statement

ORCID iD

Jule Eilts

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.*

Supplemental Material

Supplemental material for this article is available online.

References

Bäker

Schütz-Wilke

(2023). Behavioral changes during the first year of the COVID-19 pandemic: A longitudinal comparison of bullying, cyberbullying, externalizing behavior problems and prosocial behavior in adolescents. COVID, 3(2), 289–300. https://doi.org/10.3390/covid3020022

Bäker

Wilke

Eilts

von Düring

(2023). Understanding the complexities of adolescent bullying: The interplay between peer relationships, emotion regulation, and victimization. New Directions for Child and Adolescent Development, 2023(1), Article 9916294. https://doi.org/10.1155/2023/9916294

Blake

Lund

Zhou

Kwok

Benz

(2012). National prevalence rates of bully victimization among students with disabilities in the United States. School Psychology Quarterly: The Official Journal of the Division of School Psychology, American Psychological Association, 27(4), 210–222. https://psycnet.apa.org/doi/10.1037/spq0000008

Camodeca

Nava

(2022). The long-term effects of bullying, victimization, and bystander behavior on emotion regulation and its physiological correlates. Journal of Interpersonal Violence, 37(3–4), 2056–2075. https://doi.org/10.1177/0886260520934438

Card

Little

(2006). Proactive and reactive aggression in childhood and adolescence: A meta-analysis of differential relations with psychosocial adjustment. International Journal of Behavioral Development, 30(5), 466–480. https://doi.org/10.1177/0165025406071904

Cole

Ciesla

Steiger

(2007). The insidious effects of failing to include design-driven correlated residuals in latent-variable covariance structure analysis. Psychological Methods, 12(4), 381–398. https://doi.org/10.1037/1082-989X.12.4.381

Cosma

Walsh

Chester

Callaghan

Molcho

Craig

Pickett

(2020). Bullying victimization: Time trends and the overlap between traditional and cyberbullying across countries in Europe and North America. International Journal of Public Health, 65(1), 75–85. https://doi.org/10.1007/s00038-019-01320-2

Craig

Harel-Fisch

Fogel-Grinvald

Dostaler

Hetland

Simons-Morton

de Mato

M. G.

Overpeck

Due

Pickett

HBSC Violence & Injuries Prevention Focus Group HBSC Bullying Writing Group . (2009). A cross-national profile of bullying and victimization among adolescents in 40 countries. International Journal of Public Health, 54(Suppl 2), 216–224. https://doi.org/10.1007/s00038-009-5413-9

Demaray

Summers

Jenkins

Becker

(2016). Bullying participant behaviors questionnaire (BPBQ): Establishing a reliable and valid measure. Journal of School Violence, 15(2), 158–188. https://doi.org/10.1080/15388220.2014.964801

10.

Eilts

Bäker

(2024). The role of CU traits, empathy and moral disengagement in bullying. Current Psychology, 43(7), 6136–6148. https://doi.org/10.1007/s12144-023-04807-7

11.

Eilts

Bäker

Rademacher

Wilke

(2024). Bullying in adolescents: Analyzing the co-occurrence of bullying roles and associations with social-emotional difficulties and learning disabilities. Empirische Sonderpädagogik.

12.

Eilts

Koglin

(2022). Bullying and victimization in students with emotional and behavioural disabilities: A systematic review and meta-analysis of prevalence rates, risk and protective factors. Emotional & Behavioural Difficulties, 27(2), 133–151. https://doi.org/10.1080/13632752.2022.2092055

13.

Eilts

Schipper-Bäker

Schütz-Wilke

Koglin

(2022). Bullying und Viktimisierung. Ein Vergleich zwischen Jugendlichen mit und ohne den Förderschwerpunkten Lernen und emotional-soziale Entwicklung. Zeitschrift für Padagogische Psychologie, 28(3), 211–222. https://doi.org/10.1024/1010-0652/a000351

14.

Eilts

Wilke

von Düring

Bäker

(2023). Bullying Perpetration: The role of attachment, emotion regulation and empathy. Emotional & Behavioural Difficulties, 28(4), 219–233. https://doi.org/10.1080/13632752.2023.2267230

15.

Espelage

Swearer

(2003). Research on school bullying and victimization: What have we learned and where do we go from here? School Psychology Review, 32(3), 365–383. https://doi.org/10.1080/02796015.2003.12086206

16.

Feng

Wang

Chen

X. M.

(2014). Log-transformation and its implications for data analysis. Shanghai Archives of Psychiatry, 26(2), 105–109. https://doi.org/10.3969/j.issn.1002-0829.2014.02.009

17.

Fischer

Bilz

HBSC Study Group Germany . (2024). Traditional bullying and cyberbullying at schools in Germany: Results of the HBSC study 2022 and trends from 2009/10 to 2022. Journal of Health Monitoring, 9(1), 42–61. https://doi.org/10.25646%2F11872

18.

Fischer

John

Melzer

Kaman

Winter

Bilz

(2020). Traditional bullying and cyberbullying among children and adolescents in Germany—Cross-sectional results of the 2017/18 HBSC study and trends. Journal of Health Monitoring, 5(3), 56–72. https://doi.org/10.25646/6894

19.

Furlong

Sharkey

Felix

Tanigawa

Green

(2010). Bullying assessment: A call for increased precision of self-reporting procedures. In Jimerson

S. R.

Swearer

Espelage

D. L.

(Eds.), Handbook of bullying in schools: An international perspective (pp. 329–345). Routledge.

20.

Griffin

Gross

(2004). Childhood bullying: Current empirical findings and future directions for research. Aggression and Violent Behavior, 9(4), 379–400. https://doi.org/10.1016/S1359-1789(03)00033-8

21.

Hair

Black

Babin

(2009). Multivariate data analysis: A global perspective (7th ed.). Prentice Hall.

22.

Bentler

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

23.

Jenkins

Canivez

(2021). Hierarchical factor structure of the bullying participant behavior questionnaire with a middle school sample. International Journal of School & Educational Psychology, 9(1), 55–72. https://doi.org/10.1080/21683603.2019.1636734

24.

Jenkins

Canivez

Braun

(2022). Structural validity of the bullying participant behavior questionnaire with an elementary school sample. International Journal of Bullying Prevention, 4(2), 115–129. https://doi.org/10.1007/s42380-021-00091-w

25.

Jenkins

Nickerson

(2017). Bullying participant roles and gender as predictors of bystander intervention. Aggressive Behavior, 43(3), 281–290. https://doi.org/10.1002/ab.21688

26.

Kert

Codding

Tryon

Shiyko

(2010). Impact of the word “bully” on the reported rate of bullying behavior. Psychology in the Schools, 47(2), 193–204. https://doi.org/10.1002/pits.20464

27.

Kim

Muthén

B. O.

(2009). Two-part factor mixture modeling: Application to an aggressive behavior measurement instrument. Structural Equation Modeling: A Multidisciplinary Journal, 16(4), 602–624. https://doi.org/10.1080/10705510903203516

28.

Little

(2013). Longitudinal structural equation modeling. Guilford Press.

29.

Margraf

Pinquart

(2016). Bullying and social support: Variation by school-type and emotional or behavioural disturbances. Emotional & Behavioural Difficulties, 21(3), 258–270. https://doi.org/10.1080/13632752.2016.1165970

30.

Midgett

Doumas

(2019). Witnessing bullying at school: The association between being a bystander and anxiety and depressive symptoms. School Mental Health, 11(3), 454–463. https://doi.org/10.1007/s12310-019-09312-6

31.

Min

Agresti

(2002). Modeling nonnegative data with clumping at zero: A survey. Journal of the Iranian Statistical Society, 1(1), 7–33.

32.

Moore

Norman

Suetani

Thomas

Sly

Scott

(2017). Consequences of bullying victimization in childhood and adolescence: A systematic review and meta-analysis. World Journal of Psychiatry, 7(1), 60–76. https://doi.org/10.5498/wjp.v7.i1.60

33.

Moshagen

Bader

(2024). semPower: General power analysis for structural equation models. Behavior Research Methods, 56(4), 2901–2922. https://doi.org/10.3758/s13428-023-02254-7

34.

Olweus

(1991). Bully/victim problems among school children: Basic facts and effects of a school based intervention program. In Pepler

Rubin

(Eds.), The development and treatment of childhood aggression. Erlbaum.

35.

Olweus

(1993). Bullying at school: What we know and what we can do. Blackwell Publishing.

36.

Olweus

(2013). School bullying: Development and some important challenges. Annual Review of Clinical Psychology, 9(1), 751–780. https://doi.org/10.1146/annurev-clinpsy-050212-185516

37.

Qiu

Zhang

Wang

Liu

Wang

(2021). Psychometric properties of the bullying participant behaviors Questionnaire (BPBQ) among Chinese middle school students. Journal of Psychopathology and Behavioral Assessment, 43(3), 545–558. https://doi.org/10.1007/s10862-021-09873-4

38.

Quinn

Fitzpatrick

Bussey

Hides

Chan

(2016). Associations between the group processes of bullying and adolescent substance use. Addictive Behaviors, 62(1), 6–13. https://doi.org/10.1016/j.addbeh.2016.06.007

39.

R Core Team . (2023). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

40.

Rizkyanti

Wahyuni

Alatas

(2021). Empathy and defender role in bullying at school: Student-teacher relationship as mediator. Electronic Journal of Research in Education Psychology, 19(54), 227–246. https://doi.org/10.25115/ejrep.v19i54.3564

41.

Rodkin

Hodges

(2003). Bullies and victims in the peer ecology: Four questions for psychologists and school professionals. School Psychology Review, 32(3), 384–400. https://doi.org/10.1080/02796015.2003.12086207

42.

Rose

Monda-Amaya

Espelage

(2011). Bullying perpetration and victimization in special education: A review of the literature. Remedial and Special Education, 32(2), 114–130. https://doi.org/10.1177/0741932510361247

43.

Rose

Nickerson

Stormont

(2015). Advancing bullying research from a social–ecological lens: An introduction to the special issue. School Psychology Review, 44(4), 339–352. https://doi.org/10.17105/15-0134.1

44.

Ruggieri

Friemel

Sticca

Perren

Alsaker

(2013). Selection and influence effects in defending a victim of bullying: The moderating effects of school context. Procedia - Social and Behavioral Sciences, 79(1), 117–126. https://doi.org/10.1016/j.sbspro.2013.05.060

45.

Salmivalli

Lagerspetz

Björkqvist

Österman

Kaukiainen

(1996). Bullying as a group process: Participant roles and their relations to social status within the group. Aggressive Behavior, 22(1), 1–15. https://doi.org/10.1002/(SICI)1098-2337(1996)22:1<1::AID-AB1>3.0.CO;2-T

46.

Scheithauer

Hayer

Petermann

Jugert

(2006). Physical, verbal, and relational forms of bullying among German students: Age trends, gender differences, and correlates. Aggressive Behavior, 32(3), 261–275. https://doi.org/10.1002/ab.20128

47.

Scheithauer

Smith

P. K.

Samara

(2016). Cultural issues in bullying and cyberbullying among children and adolescents: Methodological approaches for comparative research. International Journal of Developmental Science, 10(1–2), 3–8. https://doi.org/10.3233/DEV-16000085

48.

Schoeler

Duncan

Cecil

Ploubidis

Pingault

J.-B.

(2018). Quasi-experimental evidence on short- and long-term consequences of bullying victimization: A meta-analysis. Psychological Bulletin, 144(12), 1229–1246. https://doi.org/10.1037/bul0000171

49.

Schütz

Bäker

Koglin

(2022). Bullying in school and cyberbullying among adolescents without and with special educational needs in emotional-social development and in learning in Germany. Psychology in the Schools, 59(9), 1737–1754. https://doi.org/10.1002/pits.22722

50.

Smith

(2016). Bullying: Definition, types, causes, consequences and intervention. Social and Personality Psychology Compass, 10(9), 519–532. https://doi.org/10.1111/spc3.12266

51.

Stevens

(2002). Applied multivariant statistics for the social science. Erlbaum.

52.

Summers

Demaray

(2008). Bully participant role questionnaire. Northern Illinois University.

53.

Swearer

Espelage

(2004). Introduction: A social-ecological framework of bullying among youth. In Espelage

Swearer

(Eds.), Bullying in American schools: A social-ecological perspective on prevention and intervention (pp. 1–12). Lawrence Erlbaum Associates Publishers.

54.

Swearer

Espelage

(2011). Expanding the socialecological framework of bullying among youth: Lessons learned from the past and directions for the future. In Espelage

Swearer

(Eds.), Bullying in North American schools (2nd ed., pp. 3–10). Routledge.

55.

Swearer

Wang

Maag

Siebecker

Frerichs

(2012). Understanding the bullying dynamic among students in special and general education. Journal of School Psychology, 50(4), 503–520. https://doi.org/10.1016/j.jsp.2012.04.001

56.

Thomas

Connor

Scott

(2015). Integrating traditional bullying and cyberbullying: Challenges of definition and measurement in adolescents–a review. Educational Psychology Review, 27(1), 135–152. https://doi.org/10.1007/s10648-014-9261-7

57.

Thornberg

(2007). A classmate in distress: Schoolchildren as bystanders and their reasons for how they act. Social Psychology of Education, 10(1), 5–28. https://doi.org/10.1007/s11218-006-9009-4

58.

Urban

Mayerl

(2014). Strukturgleichungsmodellierung. Springer.

59.

Vaillancourt

McDougall

Hymel

Krygsman

Miller

Stiver

Davis

(2008). Bullying: Are researchers and children/youth talking about the same thing? International Journal of Behavioral Development, 32(6), 486–495. https://doi.org/10.1177/0165025408095553

60.

Vandenberg

(2006). Statistical and methodological myths and urban legends. Where, pray tell, did they get this idea? Organizational Research Methods, 9(2), 194–201. https://doi.org/10.1177/1094428110391814

61.

Vivolo-Kantor

Martell

Holland

Westby

(2014). A systematic review and content analysis of bullying and cyber-bullying measurement strategies. Aggression and Violent Behavior, 19(4), 423–434. https://doi.org/10.1016/j.avb.2014.06.008

62.

Vlachou

Andreou

Botsoglou

Didaskalou

(2011). Bully/victim problems among preschool children: A review of current research evidence. Educational Psychology Review, 23(3), 329–358. https://doi.org/10.1007/s10648-011-9153-z

63.

Xie

Man

Liu

(2023). A PRISMA-based systematic review of measurements for school bullying. Adolescent Research Review, 8(2), 219–259. https://doi.org/10.1007/s40894-022-00194-5

64.

Zych

Farrington

Llorent

Ttofi

(2017). Protecting children against bullying and its consequences. Springer International Publishing. https://doi.org/10.1007/978-3-319-53028-4

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.30 MB