Contextualizing Adolescent Bullying: The Overlap Between Victimization and Perpetration

Abstract

This study used data from 2019 Minnesota Student Survey to explore patterns of traditional bullying perpetration. Using conjunctive analysis of case configurations (CACC), results from a sample of 166,351 students show that (1) incidents of traditional bullying perpetration cluster significantly among dominant situational profiles; (2) students most likely to be bullies experience victimization, but students least likely to bully their peers rarely experience bullying victimization; and (3) being a victim of traditional bullying can increase the chances of traditional bullying perpetration by as much as 100% in some situational contexts, but can have almost no influence on perpetration in other contexts. Current findings are discussed considering existing bullying scholarship and recommendations for policy and future areas of research are presented.

Keywords

bullying perpetration bullying victimization conjunctive analysis of case configurations

It is widely recognized that school is an important setting for the social development of adolescents. Broadly speaking, the social interactions that occur in a school setting ought to be pleasurable and supportive of building relationships. Within the school environment, students learn to interact with their peers as well as face conflict both of which are inevitable parts of adolescent growth and development. However, at times some students bully their peers, which can have severe consequences for victims as well as perpetrators. As a result, various risk factors that contribute to the onset of traditional bullying perpetration among school-age children have emerged in the literature over the past several years (see Álvarez-García et al., 2015).

Past scholarship consistently demonstrates that traditional bullying victimization is a significant predictor of traditional bullying perpetration (Chan & Wong, 2015; Lebrun-Harris et al., 2019; Shetgiri et al., 2012; Yang & Salmivalli, 2013) and growing concern regarding this relationship has fostered the development of school programs and policies aimed at preventing its occurrence. Although traditional bullying victimization is a predictor of perpetration, it is unclear whether this risk factor has a consistent affect across a multitude of situational contexts that define traditional bullying perpetration, or whether its influence is context dependent. Therefore, improving our understanding of the interaction between traditional bullying victimization and other relevant predictors of traditional bullying perpetration could increase the effectiveness of prevention and intervention initiatives.

To date, no known study has attempted to disentangle the influence of traditional bullying victimization on perpetration. In response, the current study was conducted to fill this gap in contemporary bullying literature. This study aims to identify situational antecedents that will put adolescents most at risk for bullying perpetration. Specifically, through a conjunctive analysis of case configurations (Miethe et al., 2008), the current study assesses patterns of bullying perpetration, with an emphasis on the role that traditional bullying victimization plays throughout various distinctive situational contexts that characterize bullying perpetration.

Literature Review

The Nature of Bullying and its Impact on Adolescents

In contemporary literature, the various ways in which bullying has been conceptually defined reflect the various ways in which the behavior can manifest. For instance, traditional bullying is recognized in the literature as intentionally aggressive behavior that is repetitive and involves a power imbalance where one student maintains a level of dominance over their victim (Juvonen & Graham, 2014; Menesini & Salmivalli, 2017; Olweus, 2013). Bullying occurs in physical, verbal, and social forms that are broadly referred to as traditional bullying. Physical bullying includes violence such as hitting, kicking, pushing, and taking another peer’s property. Verbal bullying involves name-calling, taunting, teasing, and threatening harm. These forms of bullying are considered overt since they usually occur directly between the bully and their victim. Social bullying includes hurting an individual’s social status in school through rumor spreading, social exclusion, or public humiliation. This form of bullying is usually deemed as indirect since it occurs by more covert means (Olweus, 2013).

Although bullying impacts each student differently, those who are victimized generally experience poor academic achievement, fear at school and avoidance, negative health outcomes, and suicidal ideation. Research has demonstrated that victims of bullying are also more likely to suffer from fear at school and avoidance behaviors; however, results have been mixed across studies (Hutzell & Payne, 2018). Previous studies also indicate that victims of bullying suffer from negative health outcomes such as depression, anxiety, and suicide ideation (see Hong et al., 2015).

Previous research also demonstrates that expansions in technology have increased the capacity to communicate and offer adolescents a greater range to harass, intimidate, or embarrass their peers (Hinduja & Patchin, 2017; Paez, 2018). These behaviors are regarded as cyberbullying from which research has conceptualized as intentional and repetitive harm perpetrated through computers, cell phones, and other electronic devices (Peter & Petermann, 2018). These technologies provide adolescents with anonymous methods to impose harm on victims regardless of time and setting (Hinduja & Patchin, 2017), making cyberbullying opportunistic by virtue of the concealment this approach affords to perpetrators. Like the outcomes of studies that explored the consequences of traditional bullying on victims, the impacts of cyberbullying on adolescents include the development of school avoidance, depression, suicidal ideation, and poor academic performance (Keith, 2018).

Predictors of Adolescent Traditional Bullying Perpetration

Demographic and behavioral predictors

Previous studies demonstrate that bullying perpetration is linked to gender (Álvarez-García et al., 2015; Lebrun-Harris et al., 2019; Merrin et al., 2018, Smith et al., 2019). For example, Álvarez-García et al.’s (2015) systematic review of 85 studies on predictors of school bullying perpetration among adolescents consistently demonstrates that females were less likely to bullying their peers compared to their male counterparts. Research by Lebrun-Harris et al. (2019) on correlates of bullying perpetration from 50,212 responses to the 2016 National Survey of Children’s Health, support the findings of Álvarez-García et al. (2015) and Merrin et al. (2018) wherein females were less likely to bully their peers compared to males. Evidence also suggests that bullying perpetration is associated with race/ethnicity, but the nature of this relationship remains inconsistent across studies (Álvarez-García et al., 2015; Juvonen & Graham, 2014; Merrin et al., 2018). Juvonen and Graham (2014) suggest that varied outcomes among these studies might be influenced by the setting or a function of racial diversity among the data used.

As indicated in contemporary literature, analyses that explored the relationship between age and traditional bullying perpetration demonstrate that older students generally desist from this behavior as they mature (Juvonen & Graham, 2014; Lebrun-Harris et al., 2019; Merrin et al., 2018). However, the exact age that students desist from bullying remains unclear (see Álvarez-García et al., 2015). For example, Merrin et al.’s (2018) assessment of a national representative sample of U.S. youth (N = 12,185) indicated that older students reported less bullying perpetration compared to their younger peers. Research by Lebrun-Harris et al. (2019) that examined correlates of bullying perpetration supports the findings of Merrin et al. (2018) in which older adolescents were less likely to bully others.

Other individual characteristics have received academic attention and been linked to bullying perpetration includes fighting at school, running away from home, and academic performance (Álvarez-García et al., 2015; Low & Espelage, 2014; Shetgiri et al., 2012; Wang et al., 2012). For instance, Wang et al.’s (2012) analysis of survey responses from 8,342 Chinese middle school students revealed that students who engaged in physical fights at school exhibited a higher probability of bullying their peers. Research by Low and Espelage (2014) from data on bullying involvement, fighting, and deviance among 1,232 U.S. students supports the findings of Shetgiri et al. (2012) and Wang et al. (2012) that students who reported fighting in school were more likely to bully their peers. Finding from Wang et al.’s (2012) study also demonstrate that students who ran away from home exhibited a greater probability of bullying their peers. However, it is important to note that research between these factors is sparse.

Research conducted in the United States and abroad that examined predictors of bullying behaviors demonstrate that academic achievement was a significant factor in bullying perpetration, but the results are mixed (see Álvarez-García et al., 2015). For example, Chang et al.’s (2013) study of predictors of bullying perpetration among Taiwanese youth supports the findings of Shetgiri et al.’s (2012) that youth who reported lower academic performance were more likely to be bullies. Conversely, Yang et al. (2013) examined survey responses from 1,344 Korean youth to identify predictors of bullying behaviors and found no significant relationship between academic performance and being a bully. Their findings also support the work of Wang et al. (2012) that also indicated no association between students’ academic performance and in bullying.

Contextual predictors

Prior studies conducted in the U.S. and abroad have indicated that contextual factors such as school climate (i.e., students’ perception of fair treatment and overall safety), physical or verbal abuse from a parent or adult in the home (Bayraktar, 2012; Casas et al., 2013). For instance, Bayraktar’s (2012) study on bullying among 1,052 adolescents in North Cyprus and Turkey indicated that students’ who perceived teachers who promote a positive school climate less often reported being a bully. Similar work by Casas et al. (2013) on bullying utilizing survey responses from 893 Spanish students revealed that the respondents with positive perceptions of safety were negatively associated with being a bully. Finally, maltreated children commonly have issues with self-regulation, resulting in negative external behaviors like bullying (see Lucas et al., 2016). For example, Hong et al. (2017) study on bullying using 2,219 survey responses from the 2011 to 2013 wave of the Korea Children and Youth Panel Study demonstrated a positive association between punitive parenting (that included physical abuse) and bullying perpetration.

Finally, prior victimization has been identified as a significant risk factor of bullying involvement among adolescents (Chan & Wong, 2015; Lebrun-Harris et al., 2019; Shetgiri et al., 2012; Yang & Salmivalli, 2013). These adolescents are often referred to as a bully-victim. For example, Chan and Wong (2015) used survey data from a sample of Hong Kong secondary school adolescents (N = 1,880) to examine the overlap between school bullying perpetration and victimization and found that both were positively correlated. Research by Lebrun-Harris et al. (2019) supports the work of Chan and Wong (2015), Yang and Salmivalli (2013), and Shetgiri et al. (2012) that also demonstrated children who were bullied by their peers were more likely to bully others.

Variable-Oriented Approaches to Examining Bullying

Variable-oriented approaches used to understand bullying usually involve examining a small number of independent variables (e.g., demographic, behavioral, and contextual predictors) across many observations and identifying a limited set of causal variables that explain as much variation as possible in the dependent variable (e.g., perpetration). This is achieved by arranging a representation of relationships, predicated on patterns observed across various cases that demonstrate a linear view of causation which, in turn, makes this approach insensitive to causal complexity. The variable-oriented approach utilized in various studies on bullying views causation as a challenge between independent variables to explain variation in an outcome.

However, a case-oriented approach aims to explore various features of each observed case profile and to design a representation of individuals from the links among the aspects of these configurations. A case-oriented approach regards cases as configurations—combinations of features and situations—and purports to comprehend them at the level of specific instance. This approach generates causal conditions that may combine in different and occasionally inconsistent ways to produce the same outcome. In contrast to a variable-oriented approach, a case-oriented approach regards causation as often both conjunctural and varied.

Although much of the extant literature on bullying is based on a variable-oriented approach, our knowledge and understanding of the situational context of bullying perpetration can be enhanced through case-oriented approaches, for example, a conjunctive analysis of case configurations (CACC) that narrows the gap between the variable-oriented and other case-oriented approaches (Miethe et al., 2008). “Situational context” is distinct from “contextual predictors.” Situational context denotes the unit of analysis used in CACC and contextual predictors refers to a single predictor of an outcome used in variable-oriented analytic techniques like regression. The phrase “situational profiles” and “case configurations” are used interchangeably with “situational context” throughout the remainder of the paper. While traditional analytical methods generally rely on main-effect models, CACC centers on identifying the unique combinations of variable attributes that yield “causal conditions” that result in specific outcomes. Thus, in the context of bullying research, a qualitative case-oriented approach could be used to develop a better empirical understanding of the context in which perpetration occurs.

The Current Study

Using a sample of adolescents, the current study is designed to extend previous research by examining the situational contexts for traditional bullying perpetration to answer three research questions: Is traditional bullying perpetration context dependent; and if so, to what extent? Does the influence of traditional bullying victimization vary across the distinct situational profiles that characterize traditional bullying perpetration? Which dominant situational contexts that define traditional bullying perpetration among adolescents are most (and least) influenced by traditional bullying victimization? The data and methods employed to answer these questions are presented in the following sections.

Methods

Data

The Minnesota Student Survey (MSS) was designed to collect data on the health and well-being of students and is managed by the MSS Interagency Team. From 1989 through 2010, students in grades six, nine, and twelve participated in the survey ranging in age from roughly 10 to 18 years old. In 2013, the grade levels changed to grades five, eight, nine, and eleven. The MSS is a population-based, cross-sectional survey administered every 3 years to students attending public, charter, and tribal schools. Since its inception, the MSS has expanded its collection of health-related data that includes information on health and nutrition, school climate, bullying, illegal substance use, and relationships. The current study is based on anonymous data collected from students during the 2019 wave of the MSS (N = 166,351). However, regarding this study, the data utilized originally consist of 170,128 records but missing or blank data were discarded to improve the validity of the dataset.

Measures

Dependent variable

Students reported on traditional types of bullying perpetration (i.e., physical, verbal, and social) in the past 30 days based on six questions embedded in the 2019 MSS. Two questions assessed the frequency of physical bullying perpetration: “During the last 30 days, how many times have you pushed, shoved, slapped, hit or kicked someone when you weren’t kidding around?” or “During the last 30 days, how many times have you threatened to beat someone up?.” One question assessed the frequency of verbal bullying perpetration: “During the last 30 days, how many times have you made sexual jokes, comments or gestures towards someone else?” Two questions assessed the frequency of social bullying perpetration: “During the last 30 days, how many times have you spread mean rumors or lies about someone else?” or “During the last 30 days, how many times have you excluded someone from friends, other students or activities?” All six questions used a 5-point frequency scale to capture students’ involvement in traditional bullying behaviors (1 = Never to 5 = Every day).

Responses to these questions were dichotomized to distinguish between no involvement (i.e., never) and involvement (i.e., once or twice to every day). This approach was based on past studies that dichotomized categorical measures in varying approaches to examine student bullying (Leemis et al., 2018; Taliaferro et al., 2020). Finally, the responses to these six questions were aggregated and recoded to 0 (No) for students that responded to never being involved in these acts while students that responded to involvement in any of the forms of traditional bullying at least once were recoded to 1 (Yes) to create traditional bully. It is noteworthy to mention that a definition of traditional bullying was not included in the MSS, various forms through which this harm can be inflicted on victims were included in the questions utilized to capture traditional bullying perpetration.

Independent variables

A total of 11 independent variables of traditional bullying perpetration were used in this analysis. Six of the 11 variables correspond to individual characteristics, whereas five are related to the contextual characteristics of traditional bullies. Demographic characteristics of students include dichotomized measures of sex (0 = Female, 1 = Male), race (0 = White, 1 = Non-White) and grade. Grade denotes the student’s class standing that spanned multiple grades (0 = Middle school, 1 = High school). Middle school students were composed of adolescents who identified as being in the 5th or 8th grade, while students who identified as being in the 9th or 11th grade at the time of the survey were classified as being in high school.

Students who engage in violent behavior, like fighting, were measured by the question: “During the last 12 months, how often have you hit or beat up another person?” A 5-point frequency scale captured students who reportedly harmed their peers (1 = Never to 5 = 10 or more times). Responses were dichotomized to distinguish between no involvement (i.e., never) and involvement in violent behavior (i.e., once or twice to 10 or more times) to create fighting by recoding the original responses to 0 (No) for students that responded never to this question and 1 (Yes) for students that reported fighting at least once or twice. Academics was developed regarding students’ overall academic achievement and later aggregated mostly A’s and mostly B’s as an indicator of above-average performance while mostly C’s through F’s were noted at or below-average performance (0 = Above average, 1 = At or below average). Finally, students who run away from home were assessed by the question: “During the last 12 months, how often have you run away from home?” A 5-point frequency scale captured students who reportedly ran away from home (1 = Never to 5 = 10 or more times). Responses were dichotomized to distinguish whether respondents did not run away (i.e., never) and those that reported running away from home (i.e., once or twice to 10 or more times) to create runaway by recoding the original responses to 0 (No) for students that responded never and 1 (Yes) for all other responses.

Contextual correlates of traditional bullying perpetration were also examined. For example, students’ perception of the fairness of school rules was measured by the question: “The school rules are fair?” A 4-point scale captured students’ responses to the above stated question (1 = Strongly agree, 2 = Agree, 3 = Disagree, 4 = Strongly disagree). The responses were aggregated to create fairness by recoding the original responses to 0 (Agree) for students that responded strongly agree and agree to this question and 1 (Disagree) for the remaining responses. Students’ perception of their well-being while in school was assessed by the question: “I feel safe at school” A 4-point scale captured students’ responses to the preceding question (1 = Strongly agree, 2 = Agree, 3 = Disagree, 4 = Strongly disagree). Responses were aggregated to create safety by recoding the original responses to 0 (Agree) for students that responded strongly agree and agree to this question and 1 (Disagree) for the remaining responses. Lastly, students were asked if they were verbally (i.e., Does a parent or other adult in your home regularly swear at you, insult you or put you down?) or physically (i.e., Has a parent or other adult in your household ever hit, beat, kicked, or physically hurt you in any way?) abused by a parent or adult in the home. The dichotomous responses (0 = No, 1 = Yes) to these questions were aggregated to develop abuse by recoding the original responses to 0 (No) for students who did not report abuse and 1 (Yes) for students that reported verbal or physical maltreatment.

Cyberbullying victimization was measured by the question: “During the last 30 days, how often have you been bullied through e-mail, chat rooms, instant messaging, websites, or texting?” This question used a 5-point frequency scale to capture cyberbullying victimization (1 = Never to 5 = Every day). Responses were dichotomized and recoded to 0 (No) if a student did not report being a victim of cyberbullying (i.e., never) and 1 (Yes) if a student indicated being a victim of cyberbullying at least once or twice to create cyberbullied.

Students reported on traditional types of bullying victimization (i.e., physical, verbal, and social) in the past 30 days based on six questions embedded in the 2019 MSS. Two questions assessed the frequency of physical bullying victimization: “During the last 30 days, how often have other students at school pushed, shoved, slapped, hit or kicked you when they weren’t kidding around?” or “During the last 30 days, how often have other students at school threatened to beat you up?”. One question assessed the frequency of verbal bullying victimization: “During the last 30 days, how often have other students at school made sexual jokes, comments or gestures towards you?”. Two questions assessed the frequency of social bullying victimization: “During the last 30 days, how often have other students at school spread mean rumors or lies about you?” or “During the last 30 days, how often have other students at school excluded you from friends, other students or activities?”. All six questions used a 5-point frequency scale to capture traditional bullying victimization (1 = Never to 5 = Every day).

Responses to these questions were dichotomized to distinguish between non-victimization (i.e., never) and victimization (i.e., once or twice to every day). This approach was based on past studies that dichotomized categorical measures in varying approaches to examine student bullying victimization (Fredkove et al., 2019; Taliaferro et al., 2020). Finally, the responses to these six questions were aggregated and recoded to 0 (No) for students that responded never to all the questions regarding traditional bullying victimization and students that reported being a victim of the various forms of traditional bullying at least once were recoded to 1 (Yes) to create bullied. It is important to note that the MSS did not provide students with a definition of the forms of bullying prior to the questions related to bullying behaviors.

Table 1 presents the descriptive statistics of the measures included in the current study and shows that roughly 27% of the students in the 2019 MSS sample reported bullying their peers. The typical respondent in the current sample is male (50%) and in middle school (52%). A considerable number of students in the current sample is White (68%) and roughly two-thirds (75%) of students perform “above average” academically. Finally, neither fighting nor running away from home was generally identified among most of the students surveyed (87% and 95%, respectively).

Table 1.

Descriptive Statistics for Study Measures (N = 166,351).

Variable (coded label)	Total	Traditional bully (%)
Variable (coded label)	Total	No	Yes
All students (traditional bully)	166,351	73.3	26.7
Individual predictors of traditional bullying perpetration
Demographics
Sex (sex)
Female	83,091	75.6	24.4
Male	83,260	70.9	29.1
Race (race)
White	113,545	74.1	25.9
Non-white	52,806	71.4	28.6
Grade (grade)
Middle school	86,562	70.2	29.8
High school	79,789	76.5	23.5
Behavioral characteristics
Fight at school (fighting)
No	145,033	77.9	22.1
Yes	21,318	41.4	58.6
Academic performance (academics)
Above average	125,240	75.3	24.7
At or below average	41,111	66.8	33.2
Runaway from home (runaway)
No	157,806	74.7	25.3
Yes	8,545	46.3	53.7
Contextual predictors of traditional bullying perpetration
School climate
Fairness of school rules (fairness)
Agree	121,440	76.9	23.1
Disagree	44,911	63.4	36.6
Safety at school (safety)
Agree	145,134	74.2	25.8
Disagree	21,217	66.3	33.7
Parenting style
Verbal or physical abuse (abuse)
No	142,483	76.0	24.0
Yes	23,868	56.8	43.2
Victimization
CyberBullying victimization (cyberbullied)
No	142,151	77.0	23.0
Yes	24,200	51.1	48.9
Traditional bullying victimization (bullied)
No	84,258	90.5	9.5
Yes	82,093	55.5	44.5

Analytical Strategy

The current study applies a conjunctive analysis of case configurations (CACC) to answer the above-mentioned research questions that guide this analysis. CACC is a qualitative case-oriented method developed as an exploratory and comparative data analysis technique (Hart, 2020). For example, this technique has been used to explore the situational context that defines traditionally bullying victimization (Hart et al., 2013). CACC can be outlined by a few procedures, which begin with creating a “truth table” that includes an outcome variable (e.g., bullying perpetration) and independent variables (e.g., individual and contextual factors) that are contained in an existing data file (e.g., 2019 MSS). The truth table consists of all theoretically observable situational contexts created from the independent variables’ attributes. Each row in the truth table constitutes all possible combinations of variable attributes interconnected with the outcome (e.g., bullying perpetration) and denote all possible outcomes; columns represent the specific variables used in the analysis. Next, each observation is aggregated into one of the rows that delineate the truth table that has been created. The truth table is then examined to determine the frequency with which a specific profile is observed and how often the outcome variable is observed for each case profile.

Decision rules for defining dominant case configurations are applied to the truth table after it is constructed and populated. Hart (2020) points out that the threshold for defining a dominant case configuration is typically 10 or more observations for larger samples. The subsequent truth table encompasses one row for each dominant case configuration that is empirically observed in the original data utilized in the analysis, the number of times it was observed, and the proportion of times it produced the outcome of interest (i.e., the dependent variable). In addition, recent methods have been advanced to determine whether there is significant clustering of the data aggregated to a truth table’s case configurations; and if there is, quantifying the magnitude clustering (Esteve et al., 2019; Hart, 2020). Methods for exploring the main effects that each independent variable has on the outcome can also be established utilizing these recently developed techniques (Esteve et al., 2019).

In the current study, attention is placed on “causal conditions” or situational profiles that produce traditional bullying perpetration among adolescents. It is important to note that “causal conditions” does not imply that CACC can establish causation. This phrase is frequently used in case-oriented research like CACC to denote that causation can be conjunctural and varied. In other words, causal conditions may combine in different and sometimes conflicting ways to produce the same outcome. Unlike traditional variable-oriented approaches that primarily aim at establishing generalized relationships between variables, CACC can be used to highlight where some contexts matter in some instances. Finally, Hart’s (2020) test of situational clustering is used to assess whether sample data cluster significantly among dominant situational profiles that define traditional bullying perpetration; and the Situational Clustering Index (SCI) is applied to quantify the magnitude of clustering when it is observed (see Hart, 2020).

Results

CACC was applied to a sample of students participating in the 2019 MSS survey (N = 166,351) to answer our first research question: Is traditional bullying perpetration context dependent; and if so, to what extent? The CACC matrix was constructed from 11 dichotomous demographic, behavior, and contextual factors associated with bullying perpetration. Although a total of 2,048 unique situational profiles could have been observed in the data (i.e., 2¹¹ = 2,048), only 1,864 case configurations were observed. Furthermore, only 1,039 of the observed profiles were determined to be dominant case configurations, based on the application of established minimum frequency rules (see Miethe et al., 2008). Table 2 shows that the likelihood of victimization across all dominant profiles ranges from 0% to 100% (M = .47, SD = .26). The average likelihood of bullying perpetration for all dominant case configurations is somewhat higher than for the sample, as a whole (see Table 2), which may indicate that bullying perpetration is context dependent.

Table 2.

Situational Factors and the Likelihood of Traditional Bullying Perpetration (N = 166,351).

Situation ID	Sex	Race	Grade	Fighting	Academics	Runaway	Fairness	Safety	Abuse	Cyberbullied	Bullied	P (TB)	N
Dominant profiles most likely to result in traditional bullying perpetration
1	Female	Non-white	Middle School	Yes	At or below average	Yes	Disagree	Agree	No	Yes	Yes	1.00	18
2	Male	Non-white	Middle School	Yes	Above Average	No	Disagree	Agree	Yes	Yes	Yes	1.00	16
3	Male	White	High School	Yes	At or below average	No	Disagree	Disagree	Yes	Yes	Yes	1.00	14
4	Female	Non-white	High School	Yes	Above Average	No	Disagree	Disagree	Yes	Yes	Yes	1.00	13
5	Male	Non-white	Middle School	Yes	Above Average	Yes	Disagree	Agree	Yes	Yes	Yes	1.00	12
6	Female	White	Middle School	Yes	At or below average	Yes	Disagree	Disagree	No	Yes	Yes	1.00	11
7	Male	Non-white	Middle School	Yes	Above Average	Yes	Agree	Agree	Yes	Yes	Yes	1.00	11
8	Male	Non-white	Middle School	Yes	At or below average	Yes	Agree	Disagree	No	No	Yes	1.00	10
9	Male	Non-white	Middle School	Yes	At or below average	Yes	Disagree	Agree	No	Yes	Yes	1.00	10
10	Male	Non-white	Middle School	Yes	Above Average	Yes	Disagree	Agree	No	Yes	Yes	.95	20
Dominant profiles least likely to result in traditional bullying perpetration
1030	Female	Non-white	High School	No	At or below average	No	Disagree	Disagree	Yes	No	No	.00	17
1031	Female	White	Middle School	No	Above Average	No	Agree	Disagree	No	Yes	No	.00	16
1032	Female	White	High School	No	At or below average	No	Agree	Disagree	No	Yes	No	.00	16
1033	Male	Non-white	Middle School	No	Above Average	No	Agree	Disagree	Yes	No	No	.00	14
1034	Female	White	High School	No	Above Average	No	Disagree	Disagree	No	Yes	No	.00	12
1035	Female	White	High School	No	At or below average	No	Disagree	Agree	No	Yes	No	.00	12
1036	Male	Non-white	High School	No	Above Average	No	Disagree	Disagree	No	Yes	No	.00	12
1037	Female	Non-white	Middle School	No	At or below average	Yes	Agree	Agree	No	No	No	.00	11
1038	Female	White	Middle School	No	Above Average	No	Disagree	Disagree	No	Yes	No	.00	10
1039	Female	Non-white	High School	No	At or below average	Yes	Agree	Agree	Yes	No	No	.00	10
Mean=												.47	19
SD=												.26	23

Note. p(TB) denotes the probability of traditional bullying perpetration among the profiles.

Results of a goodness-of-fit test showed that the MSS data clustered significantly among a small sub-set of dominant situational profiles that define traditional bullying perpetration (X²(1,038, N = 1,864) = 2,809,007.0, P < .001). These findings provide empirical evidence that bullying perpetration is context dependent. Simply put, bullying incidents are not randomly distributed across causal conditions that result in perpetration; instead, they cluster among certain distinct combinations of variable attributes, which the existing literature has identified as correlates to bullying perpetration. The magnitude of clustering observed in the data was exceedingly strong (SCI = .86) and Figure 1 provides a visual representation of this pattern.

Figure 1.

A graphical depiction of the situational clustering index of traditional bullying perpetration from Minnesota student survey (MSS) data.

Results of our CACC also provide answers to the second research question: Does the influence of traditional bullying victimization vary across the unique situational profiles that characterize traditional bullying perpetration? The 10 causal conditions that produced the greatest and least likelihood of traditional bullying perpetration are presented in Table 2. For the top 10 profiles, students who are more likely to bully their peers always report fighting, being victims of some form of traditional bullying during the previous 30 days. Students also never report fighting and being a victim of traditional bullying in situations that characterize the causal conditions where a student is least likely to bully a peer. Thus, when adolescents are the most and least likely to bully others, previous victimization in a traditional form is the only factor that remains constant regardless of context.

To answer the third and final research question, the sample of MSS data used in the current study were split into two groups: students who reported being a victim of traditional bullying and those who were not. Two additional CACC matrixes assessing traditional bullying perpetration were created from both subsets of data, and identical profiles from each matrix were matched. The probability of perpetration for each matched situational profile was used to create a relative-risk ratio (RRR). The RRR represents the likelihood of traditional bullying perpetration, given the presence of traditional bullying victimization, relative to the likelihood of perpetration, given the lack of traditional bullying victimization. Creating this ratio assisted in isolating the main effect that traditional bullying victimization has on traditional bullying perpetration, controlling for all other variables simultaneously because all other factors in the situational profile were held constant through the matching process. Table 3 shows the matched profiles with the highest and lowest RRRs, which enable us to answer the third research question: Which dominant situational contexts that define traditional bullying perpetration among students are most (and least) influenced by traditional bullying victimization?

Table 3.

Matched Situational Profiles of Bullying Perpetration Where Students Experienced Traditional Bullying Versus When They Did Not Experience Traditional Bullying.

Situation ID #	Sex	Race	Grade	Fighting	Academics	Runaway	Fairness	Safety	Abuse	Cyberbullied	Traditionally bullied	Not traditionally bullied	RRR
Matched profiles that demonstrate the greatest effect that traditional victimization has on perpetration
654	Female	Non-white	High school	No	At or below average	No	Agree	Disagree	No	No	0.38	0.02	24.23
782	Female	White	Middle school	No	At or below average	No	Agree	Disagree	No	No	0.26	0.01	23.72
659	Female	White	High school	No	Above average	No	Agree	Disagree	No	Yes	0.38	0.02	17.23
Matched profiles that demonstrate the smallest effect that traditional victimization has on on perpetration
551	Female	Non-white	High school	Yes	Above average	Yes	Disagree	Agree	Yes	No	0.47	0.40	1.16
328	Male	White	Middle school	Yes	At or below average	Yes	Agree	Agree	No	No	0.63	0.61	1.04
302	Male	White	Middle school	Yes	Above average	Yes	Disagree	Disagree	No	No	0.65	0.64	1.03

Note: Traditionally bullied figures denote the probability of bullying perpetration in cases that students experienced traditional bullying victimization. Not traditionally bullied figures denote the probability bullying perpetration in cases that students did not experience traditional bullying victimization. The probability of bullying perpetration for each matched situational profile was used to create a relative-risk ratio (RRR). The relative-risk ratio (RRR) is calculated by dividing the traditionally bullied figure by the not traditionally bullied figure. The situation ID number represents the rank-order of situations based on their relative prevalence of bullying perpetration found in Table 1. All the figures in the last three columns are rounded to the nearest hundredth.

The top three matched situational profiles of traditional bullying shown in Table 3 demonstrate that being a victim of traditional bullying when all other factors are held constant, results in an approximately 17–24 times greater chance of a student engaging in bullying. A closer look at these profiles shows that race lacks variability, whereas the other ten factors change across contexts. These patterns suggest that when traditional bullying victims are identified, certain demographic, behavioral, and contextual factors significantly increase the likelihood of adolescent bullying perpetration when found in combination with one another.

Table 3 also shows that there are several situational profiles where the likelihood of being a bully is nearly unaffected (i.e., RRR = 1.00) by a student’s traditional bullying victim status. For example, the RRR for the bottom profile in Table 3 is 1.03, because the P(TB) for the traditionally bullying profile is 0.65, and 0.64 for the identical profile that involves students who are not traditional bully-victims. Again, these patterns suggest that when traditional bullying victims are identified, certain demographic, behavioral, and contextual factors do not affect the likelihood of bullying perpetration significantly, when in the context of a specific combination of other factors.

Discussion

Findings from the current study add to the growing literature on bullying and offer new insights into this significant social issue that could inform policies and programs, and direct future research. Through the application of CACC, the specific causal conditions that could lead to bullying perpetration, based on these established correlates, were examined and several patterns related to bullying perpetration were identified.

First, bullying occurrences clustered in an extremely strong pattern among a few dominant situational profiles; but the likelihood of perpetration varied considerably among them, regarding the unique combination of variable attributes. Non-white, females in middle school who reported fighting, performed at or below average academically, reportedly run away from home, disagreed with the fairness of school rules, felt safe while in school, did not report verbal or physical abuse from a parent or adult in the home, and reported both cyberbullying and traditional bullying victimization within the previous 30 days were the students most at risk of being a bully (Profile #1 in Table 2). This specific student profile is associated with a 100% chance of traditional bullying perpetration, based on the 2019 MSS data. Unlike past research, which suggested that males are more likely to engage in traditional bullying (Álvarez-García et al., 2015; Lebrun-Harris et al., 2019; Merrin et al., 2018), the present study findings show that females were most at risk of being a bully. This outcome supports previous studies that demonstrate gender differences exist among the types of traditional bullying, as females are more likely to engage in social bullying and less likely to engage in physical bullying (see Smith et al., 2019).

In addition, contrary to Hong et al.s’ (2017) findings that demonstrated a link between punitive parenting and bullying perpetration, the current study findings demonstrate that adolescents most at risk of being a bully did not report verbal or physical abuse from a parent or adult in the home. This outcome supports the results of Lucas et al.s’ (2016) study on the associations between reported violence at home and bullying among 3,202 Swedish children. Their results demonstrated a strong association between adolescents who reported abuse at home and being a victim of bullying in contrast to reporting bullying perpetration.

Second, the literature on bullying demonstrates that traditional victimization is a significant predictor of perpetration (Chan & Wong, 2015; Lebrun-Harris et al., 2019; Shetgiri et al., 2012; Yang & Salmivalli, 2013). However, these studies apply variable-oriented approaches to their analyses, which generally aim to determine the effect that an independent variable has on the variation in a dependent variable (see Hart, 2020). Although statistical models that apply these variable-oriented approaches have produced important insight into bullying perpetration, more empirical knowledge regarding bullying can be produced from case-oriented approaches like CACC.

Specifically, the current findings indicate that students most likely to be bully their peers have always experienced traditional bullying victimization, but students least likely to engage in traditional bullying behaviors almost never experience traditional bullying victimization. Not only are these findings analogous to the extant literature that highlights the influence that traditional bullying victimization has on perpetration, they also reveal the role that traditional bullying victimization plays among the distinctive situational profiles that delineate bullying perpetration. Understanding the context of traditional bullying perpetration where victimization exists or is lacking from a student profile may allow school administrators to modify prevention efforts around categorizations of students, notably when those students report being victims of traditional forms of bullying.

Finally, the affect that traditional bullying victimization has on bullying perpetration is highly contextual. For example, non-white, females in high school who were not involved in a fight, perform at or below average academically, do not run away from home, agree that the rules established in their school were fair, disagree with feeling safe while at school, do not report verbal or physical abuse by a parent or adult in the home and report cyberbullying victimization (Profile # 654 in Table 3) have an approximately 24% greater likelihood of bullying perpetration if they have also been a victim of traditional bullying, compared to the same student who has not been a victim of traditional bullying. However, being a traditional bullying victim has almost no effect on the likelihood of bullying perpetration for White males in middle school, who have been involved in a fight, who perform well academically, report running away from home, disagree with the fairness of the established rules in school, disagree with feeling safe while at school, do not report verbal or physical abuse by a parent or adult in the home and do not report cyberbullying victimization (Profile # 302 in Table 3). Thus, once a victim of traditional bullying victim has been recognized, intervention policies might have a greater impact on mitigating and preventing bullying by focusing on those profiles where being a victim of traditional bullying increases the likelihood of perpetration the most.

Limitations

Like most research, the current study is not without certain limitations. First, the MSS data do not contain information on all school-age children. The data are limited to adolescents in grades 5, 8, 9, and 12 ranging in age from roughly 10 to 18 years old. Therefore, only the situational contexts of cyberbullying among some middle and high school students were analyzed. Additionally, the racial/ethnic make of the sample utilized in this study was predominately white students (see Table 1). Thus, the disproportionate racial/ethnic make of the sample in this study may have distorted the influence that this factor has on bullying.

Second, the current study included several factors associated with bullying perpetration identified in the existing literature. However, other relevant factors related to school climate (e.g., peer and teacher relationships), indicators of socioeconomic status, and parental or peer attachments have also been shown to influence bullying perpetration were not measured by the MSS and could not be included in the analysis. Next, since the data from the MSS is derived from self-reported surveys, it is possible for students to over- or underreport bullying. For example, students who bully their peers may be reluctant to respond honestly, in fear of the consequences or may have may have objected to participate, and thus skewing the frequency of its occurrence.

Fourth, respondents were not given a definition that outlined the core components that construct an instance of bullying. Specifically, the questions embedded in the MSS to delineate responses to bullying behaviors did not note a power imbalance or the repetitiveness of the act. Thus, lacking a clear definition of bullying could have affected how the students responded to the bullying-related questions. It is plausible that a lack of congruence may exist between a student’s experience and perception of bullying and might affect the prevalence of its occurrence or inaccurately identify victims or perpetrators in the MSS data utilized in the current study. For example, previous analyses that examined the impact of how bullying is operationalized in research on the prevalence of its occurrence among adolescents revealed inconsistencies in cases as well as disparities between students’ experiences and perceptions according to definition applied (Connell et al., 2019).

Finally, the cross-sectional nature of the MSS data limits the ability to determine the causal direction or temporal ordering of the relationships identified in this study, because data for the MSS were collected at a single point in time. In addition, students may report bullying a peer during the survey that occurred outside of the survey reference period which may distort the frequency of reported bullying.

Conclusion

Considering the current findings, a few themes emerged that warrant elaboration and have implications for practice and prevention. First, future research on bullying should continue to explore the connection between harm that is caused by traditional bullying as well as cyber victimization and perpetration. Furthermore, it would be interesting to apply the techniques in the current study in assessing the relationship between cyberbullying victimization and cyberbullying perpetration. This approach might offer unique insight into the situational contexts that exist among instances where adolescents bully their peers by going beyond traditional variable-oriented analyses.

Regarding anti-bullying policies, the results of the current study demonstrate the broad variability in the occurrence of bullying throughout its contexts. Specifically, the findings of this study indicate that the likeliness of bullying perpetration is rooted in the complex social conditions that are not merely defined in a limited range of variables that remain sustained across all contexts.

Therefore, context matters in exploring how and why bullying occurs which is beneficial to the developing or improving prevention efforts. For example, a CACC method to identifying bullying perpetration offers school administrators and staff unique profiles of students that are more likely to engage in bullying along with experiencing victimization. Thus, practitioners working with adolescents should consider modifying their intervention strategies or programming according to the type of adolescent’s bullying status (e.g., bully-victim) or type of bullying perpetrated (Cho & Lee, 2018) rather than focusing on broad prevention efforts that are commonly used in schools. More importantly, these programming efforts should target children at higher risk of bullying (e.g., bully-victim) and promote prosocial behaviors and foster positive peer relationships (Hong et al., 2017; Yang & Salmivalli, 2013). Though, it is possible that some aspects may be neglected by school administrators or scholars and thus requires a further examination or the application of new methods to understand the unique conditions that encompass instances of bullying to prevent its occurrence.

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.

ORCID iD

G. R. Paez

Author Biographies

G. R. Paez is a faculty member in the Department of Criminology and Criminal Justice at the University of Tampa, where he teaches undergraduate and graduate courses in law enforcement, research methods, and criminal justice policy. His primary research interests include adolescent bullying, school violence, and policing.

O. L. Richmond is a recent graduate of the University of Tampa’s Master of Science program in Criminology and Criminal Justice. His primary research interests include school violence and policing.

References

Álvarez-García

García

Núñez

J. C.

(2015). Predictors of school bullying perpetration in adolescent: A systemic review. Aggression and Violent Behavior, 23, 126–136.

Bayraktar

(2012). Bullying among adolescents in North Cyprus and Turkey: Testing a multifactor model. Journal of Interpersonal Violence, 27(6), 1040–1065.

Casas

J. A.

Del Rey

Ortega-Ruiz

(2013). Bullying and cyberbullying: Convergent and divergent predictor variables. Computers in Human Behavior, 29, 580–587.

Chan

H. C. O.

Wong

D. S. W.

(2015). The overlap between school bullying perpetration and victimization: Assessing the psychological, familial, and school factors of Chinese adolescents in Hong Kong. Journal of Child and Family Studies, 24, 3224–3234.

Chang

Lee

Chiu

Hsi

Huang

Pan

(2013). Relationships among cyberbullying, school bullying, and mental health in Taiwanese adolescents. Journal of School Health, 83, 454–462.

Cho

Lee

J. M.

(2018). Explaining physical, verbal, and social bullying among bullies, victims of bullying, and bully-victims: Assessing the integrated approach between social control and lifestyles-routine activities theories. Children and Youth Services Review, 91, 372–382.

Connell

N. M.

Schell-Busey

N. M.

Hernandez

(2019). Experiences versus perceptions: Do students agree that they have been bullied? Youth & Society, 51(3), 394–416.

Esteve

Moneva

Hart

T. C.

(2019). CACC R package. Conjunctive analysis of case configurations (Version 1.0.3). CACC.

Fredkove

W. M.

Gower

A. L.

Sieving

R. E.

(2019). Association among internal assets, bullying and emotional distress in eight grade students. Journal of School Health, 89(11), 883–889.

10.

Hart

T. C.

(2020). Identifying situational clustering and quantifying its magnitude in dominant case configurations: New methods for conjunctive analysis. Crime & Delinquency, 66, 143–159.

11.

Hart

T. C.

Hart

J. L.

Miethe

T. D.

(2013). Situational context of student bullying victimization and reporting behavior: A conjunctive analysis of case configurations. Justice Research and Policy, 15(2), 43–73.

12.

Hinduja

Patchin

J. W.

(2017). Cultivating youth resilience to prevent bullying and cyberbullying victimization. Child Abuse & Neglect, 73, 51–62.

13.

Hong

J. S.

Kim

D. H.

Piquero

A. R.

(2017). Assessing the links between punitive parenting, peer deviance, social isolation and bullying perpetration and victimization in South Korean adolescents. Child Abuse & Neglect, 73, 63–70.

14.

Hong

J. S.

Kral

M. J.

Sterzing

P. R.

(2015). Pathways from bullying perpetration, victimization, and bully victimization to suicidality among school-aged youth: A review of the potential mediators and a call for further investigation. Trauma, Violence, & Abuse, 16(4), 379–390.

15.

Hutzell

K. L.

Payne

A. A.

(2018). The relationship between bullying victimization and school avoidance: An examination of direct associations, protective influences, and aggravating factors. Journal of School Violence, 17(2), 210–226.

16.

Juvonen

Graham

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

17.

Keith

(2018). How do traditional bullying and cyberbullying victimization affect fear and coping among students? An application of general strain theory. American Journal of Criminal Justice, 43, 67–84.

18.

Lebrun-Harris

L. A.

Sherman

L. J.

Limber

S. P.

Miller

B. D.

Edgerton

E. A.

(2019). Bullying victimization and perpetration among U.S. children and adolescents: 2016 national survey of children’s health. Journal of Child and Family Studies, 28, 2543–2557.

19.

Leemis

R. W.

Espelage

D. L.

Basile

K. C.

Kollar

L. M

Davis

J. P.

(2018). Traditional and cyber bullying and sexual harassment: A longitudinal assessment of risk and protective factors. Aggressive Behavior, 45(2), 181–192.

20.

Low

Espelage

(2014). Conduits from community violence exposure to peer aggression and victimization: Contributions of parental Monitoring, impulsivity, and deviancy. Journal of Counseling Psychology, 61(2), 221–231.

21.

Lucas

Jernbro

Tindberg

Janson

(2016). Bully, bullied and abused. Associations between violence at home and bullying in childhood. Scandinavian Journal of Public Health, 44, 27–35.

22.

Menesini

Salmivalli

(2017). Bullying in schools: The state of knowledge and effective interventions. Psychology, Health & Medicine, 22(1), 240–253.

23.

Merrin

G. J.

Espelage

D. L.

Hong

J. S.

(2018). Applying the social-ecological framework to understand the associations of bullying perpetration among high school students: A Multilevel Analysis. Psychology of Violence, 8(1), 43–56.

24.

Miethe

T. D.

Hart

T. C.

Regoeczi

W. C.

(2008). The conjunctive analysis of case configurations: An exploratory method for discrete multivariate analysis of crime data. Journal of Quantitative Criminology, 24(2), 227–241.

25.

Olweus

(2013). School bullying: Development and some important challenges. Annual Review of Clinical Psychology, 9, 751–780.

26.

Paez

G. R.

(2018). Cyberbullying among adolescents: A general strain theory perspective. Journal of School Violence, 17(1), 74–85.

27.

Peter

I.-K.

Petermann

(2018). Cyberbullying: A concept analysis of defining attributes and additional influencing factors. Computers in Human Behavior, 86, 350–366.

28.

Shetgiri

Lin

Flores

(2012). Identifying children at risk for being bullies in the United States. Academic Pediatrics, 12(6), 509–522.

29.

Smith

P. K.

López-Castro

Robinson

Görzig

(2019). Consistency of gender differences in bullying in cross-cultural surveys. Aggression and Violent Behavior, 45, 33–40.

30.

Taliaferro

L. A.

Doty

J. L.

Gower

A. L.

Querna

Rovito

M. J.

(2020). Profiles of risk and protection from violence and bullying perpetration among adolescent boys. Journal of School Health, 90(3), 212–223.

31.

Wang

Zhou

Deng

Hong

Gao

(2012). Adolescent bullying involvement and psychosocial aspects of family and school life: A cross-sectional study from Guangdong Province in China. PLoS One, 7(7), 1–10.

32.

Yang

Salmivalli

(2013). Different forms of bullying and victimization: Bully-victims versus bullies and victims. European Journal of Developmental Psychology, 10(6), 723–738.

33.

Yang

Stewart

Kim

Shin

Dewey

M. E.

Maskey

Yoon

(2013). Differences in predictors of traditional and cyber-bullying: A 2-year longitudinal study in Korean school children. European Child & Adolescent Psychiatry, 22, 309–318.