A General Strain Approach to Psychological and Physical Bullying

Abstract

Although studies have been conducted to examine the applicability of Agnew’s general strain theory (GST) to the explanation of school bullying, GST research on the phenomenon remains limited in number and scope. To fill this gap in research, using data from a sample of 296 middle school students in a southwestern state of the United States, this article examined whether different types of strain and negative emotions are positively associated with psychological, physical, and general bullying. Overall findings of negative binomial regression analyses tended to be consistent with our expectations, while some aspects of GST received more empirical support than others. Strains and negative emotions were mostly related positively either to psychological or physical bullying, with negative emotions, anger and depression, partly mediating the strain-bullying relationship. However, we found mixed patterns of interactions among strains, negative emotions, and conditioning factors. Theoretical and practical implications were discussed.

Keywords

school bullying general strain theory strain negative emotions conditioning factors

Introduction

Over the past two decades, there has been tremendous research interest in the prevalence and consequences of school bullying for victims and perpetrators in the United States and other countries. For example, using a nationally representative sample of 15,686 students in Grades 6 through 10 in American public and private schools, Nansel et al. (2001) found that about 30% of students were involved in school bullying, with 13% being bullies. More recently, Schneider, O’Donnell, Stueve, and Coulter (2012) reported that 26% of 20,406 high school students in Massachusetts were school bullying victims in the past 12 months, and that victims were more likely to suffer from psychological distress, depressive symptoms, and self-injury.

Due et al.’s (2005) study of school-aged adolescents from 28 countries in Europe and North America showed that the prevalence of bullying ranged from 11% in Sweden to 80% in Lithuania, with the United States in between at about 27%. Their study also revealed that exposure to school bullying was significantly related to physical and psychological symptoms such as morning tiredness, loneliness, helplessness, and sleeping difficulty in all 28 countries, providing further evidence that bullying has detrimental impact on victims, regardless of cultural differences.

Similarly, other studies (see Olweus, 1993; Schneider et al., 2012) indicate that school bullying victims are more likely to suffer physically, psychologically, and academically from depression, anxiety, school dropout, learning difficulties, and even suicide. In addition, research findings show that bullying is a significant predictor of later delinquency and adult criminal behaviors. For example, Ttofi, Farrington, Lösel, and Loeber (2011) performed a meta-analysis of 28 studies that assessed the relationship between bullying behavior and criminal offending in later life. Their results show that childhood bullies are more likely to engage in antisocial and criminal behaviors as adults. This long-term outcome of school bullying has led criminologists to examine causes as well as consequences of bullying by applying theories of delinquency.

This study intends to contribute to the literatures on school bullying, a ubiquitous problem with nontrivial consequences (see Bosworth, Espelage, & Simon, 1999; Moon, Hwang, & McCluskey, 2011; Nansel et al., 2004; Olweus, 1993; Wang, Iannotti, & Nansel, 2009), by examining its etiology based on Agnew’s (1992, 2006) general strain theory (GST). Specifically, we propose to test whether an adolescent’s strains (or stressors) and strain-generated negative emotions (anger and depression) are positively related to school bullying, focusing on negative emotions’ role as mediators in the strain-bullying relationship and conditioning effects involving strain and negative emotions. To this end, we applied negative binomial regression to analyze data from a sample of about 300 American middle school students. Before presenting our hypotheses, however, an overview of GST and its application to school bullying is in order.

GST: Key Propositions and Empirical Findings

First, GST proposes that strains are the main sources of delinquent behaviors, including school bullying. According to Agnew (2001), certain types of strains (e.g., physical/emotional punishment by significant others, criminal victimization, racial discrimination) are more likely to result in delinquency because they are more likely to be perceived as unjust and/or high in magnitude and associated with low self-control or to create incentives for delinquent coping. GST’s second proposition is that in response to strains, individuals are more likely to experience negative emotions, especially anger and depression, which in turn lead to delinquency. That is, negative emotions are expected to mediate, in part, the influence of strain on delinquency as strained individuals are pressured to engage in delinquent coping to correct undesirable situations or alleviate their negative emotions. A third proposition is that conditioning factors (e.g., relationship with parents, self-control, delinquent peer association) moderate the link between strains and negative emotions, on one hand, and deviant behaviors, on the other.

Over the past two decades, an increasing number of empirical studies with culturally and racially diverse samples were conducted, providing general support for the theory’s propositions. For example, they show that individuals who are criminally victimized, physically abused, or racially discriminated are more likely to commit deviant behaviors (see Agnew, Brezina, Wright, & Cullen, 2002; Baron, 2004; Moon, Hays, & Blurton, 2009). It has also been found that anger has a significant effect on delinquency and partly mediates the connection between strains and delinquency (see Aseltine, Gore, & Gordon, 2000; Mazerolle & Piquero, 1997; Moon et al., 2009). Research also provides some support for the proposition about conditioning factors. For example, strained individuals were found to be more likely to engage in deviant behaviors when they were associated with delinquent peers or had low self-control; whereas they were less likely to do so if they had a positive relationship with parents or were religiously involved (see Agnew et al., 2002; Jang & Johnson, 2003; Mazerolle & Maahs, 2000). However, in general, this proposition has received less empirical support relative to the other two (Agnew, 2006).

GST and School Bullying

In light of the research findings on negative outcomes for bullying victims and perpetrators, a growing number of criminological studies have been conducted to examine the applicability of GST as a potential cause of the phenomenon (see Lee, 2011; Moon, Morash, & McCluskey, 2012; Moon et al., 2011; Patchin & Hinduja, 2011). Empirical findings show that strains (i.e., test-related strain, victimization experience, or teachers’ physical and emotional punishment) and anger (a key negative emotion in GST) are significantly related to school bullying (see Bosworth et al., 1999; Espelage, Bosworth, & Simon, 2000; Moon et al., 2011; Moon et al., 2012; Patchin & Hinduja, 2011). These findings suggest that Agnew’s (1992, 2006) GST may be applicable to the etiology of school bullying.

In recent years, more studies (see Moon et al., 2011; Moon et al., 2012; Patchin & Hinduja, 2011) were conducted to examine the applicability of GST. For example, Moon et al. (2011), using longitudinal data with 655 South Korean middle school students, examined the effects of six strains (e.g., family conflict, parental punishment, examination-related strain, criminal victimization) and negative emotions (anger and depression) on school bullying. The results show that Korean youths who experienced higher level of test-related strain are more likely to report their involvement in school bullying as perpetrators. Interestingly, trait-based anger was found to have no significant relationship with bullying, whereas trait-based depression had a positive effect on bullying.

With a sample of approximately 2,000 middle school students in the United States, Patchin and Hinduja (2011) examined GST’s applicability in explaining traditional and cyber bullying. The findings indicated that general strain had a significant effect on traditional and cyber bullying in the expected direction. Anger and frustration was also positively related to school bullying in that angry and/or frustrated youth were more likely to engage in both types of school bullying. However, the results showed no mediating effect of strain on bullying via negative emotions. This was because strain continued to exert a significant effect on traditional and cyber bullying even after the inclusion of negative emotions in the final model.

Most recently, Moon et al. (2012) examined the GST’s applicability to school bullying in a nationally representative longitudinal sample of Korean adolescents. They measured seven key strains (e.g., part-time work, victimization experience, conflict with parents) and their effect on school bullying. These strains are the same factors that prior studies suggest are also related to juvenile delinquency. Four conditioning factors (e.g., low self-control, delinquent peer association) were also measured. The findings indicated that youth who were criminally victimized or had conflicting relationship with parents were more likely to bully others. The study unexpectedly found that trait-based anger had no significant direct effect on school bullying, which was contrary to previous findings (see Bosworth et al., 1999; Espelage et al., 2000; Patchin & Hinduja, 2011). In addition, the results showed no mediation of anger between strains and bullying. Regarding the interactions between the effects of strains on bullying and conditioning factors, the findings showed that only 4 of 28 interaction terms were significantly related to school bullying. Overall, these findings provided only partial support of GST’s applicability in explaining school bullying.

Despite the generally positive evidence from initial studies, GST research on school bullying remains limited in scope as well as in number. Specifically, previous studies (see Moon et al., 2011; Patchin & Hinduja, 2011) were confined to examining positive associations among strains, negative emotions, and bullying, but failed to test whether negative emotions mediated positive associations between strains and bullying, or whether the effects of strains and negative emotions on bullying were conditioned by other factors. Second, prior GST studies (see Moon et al., 2011; Moon et al., 2012) failed to examine the predictors of physical and psychological bullying separately, despite etiological differences between the two types of bullying behaviors.

In the current study, using a sample of 296 U.S. youths, we first tested whether the four “strains most likely to cause crime” (Agnew, 2006, p. 70) among juveniles were positively related to school bullying, separately for psychological and physical bullying. Second, we examined whether anger and depression were not only positively associated with school bullying but also mediated the relationship between strains and bullying. A caveat here is that temporal order cannot be fully established among strains, negative emotions, and school bullying using cross-sectional data, which previous GST researchers often used to test the mediation hypothesis (e.g., Jang & Johnson, 2003; Moon et al., 2009). Therefore, our examination of the hypothesized mediation is exploratory, and thus results should be interpreted with caution. Third, interactions involving strains, negative emotions, and conditioning factors were tested; first to see whether strained youths with low self-control or delinquent peer associations were more likely to bully others. The second interaction effect was that strained youth who have parents who control or support them were less likely to engage in bullying behavior.

Hypotheses

Taken together, we test the following hypotheses:

Hypothesis 1: Strain is likely to be positively associated with school bullying.

Hypothesis 2: The positive association between strain and school bullying is likely to be explained in part by anger and depression.

Hypothesis 3: Parental control and support are likely to ameliorate the effects of strain and negative emotions on school bullying, whereas low self-control and delinquent peer association are expected to aggravate the effects.

Method

Sample

To test the hypotheses, the present study analyzed data from a sample of 296 adolescents in two middle schools in a southwestern city of the United States. Each school had around 700 students who were predominately of Latino ethnicity (approximately 98%), and more than half (52%) were female. The schools were located in low-income neighborhoods, as around 94% of students in both school districts received free or reduced-price school lunch or qualified for public assistance.

In 2008, the school district and administrators approved the research and parents/legal guardians of all 6th and 7th graders in each school (a total of 620 students in both schools) were notified about the purpose and the administration of questionnaires at school. Approximately 1 week later, 360 students returned parental consent forms that indicated permission for their child to voluntarily participate in the research.

As compensation, each student received two pens after the completion of the questionnaire. Students whose parents allowed them to participate in the research were asked to complete the questionnaire in a classroom or a school cafeteria. A total of 320 students completed the survey, but 24 questionnaires were incomplete and thus were discarded. Of the 296 adolescents who completed the survey (158 and 138 students from each school), 168 students (57%) were girls, and a large majority of students in the sample were Hispanic (94%), which was consistent with the school district’s race/ethnic composition.

Independent Variables

Strains

In the current study, four measures of strains were constructed: family conflict, teachers’ emotional punishment, racial discrimination, and criminal victimization. All the strain indexes were coded so that a higher score indicated a higher level of strain. Table 1 presents descriptive statistics, including means, of all independent and dependent variables.

Table 1.

Descriptive statistics of independent and dependent variables.

Variables	Mean	SD	Minimum	Maximum
Family conflict	3.64	2.09	0	9
Teachers’ emotional punishment	2.77	3.96	0	20
Racial discrimination	3.79	4.46	0	20
Criminal victimization	0.88	2.07	0	16
Anger	1.83	2.27	0	9
Depression	2.50	2.44	0	9
Low self-control	47.05	13.63	24	86
Delinquent peer association	5.09	7.64	0	41
Parental control	5.80	2.12	2	8
Parental support	6.17	2.12	2	8
Psychological bullying	3.89	4.23	0	21
Physical bullying	1.83	3.41	0	18
General bullying	5.72	7.06	0	36

Total Number = 296

The family conflict scale consists of three items, adapted from Aseltine et al. (2000), asking about “frequent arguments between parents,” family members arguing/disagreeing with each other, or a respondent arguing/disagreeing with parents during the last 12 months prior to survey. The response options for each item ranged from 0 (never) to 3 (always), and their Cronbach’s alpha was .65. The teachers’ emotional punishment scale was created by summing five items, adapted from Moon and Morash (2004). The five items (α = .84) asked respondents whether their teachers “isolated,” “embarrassed,” “negatively compared to others,” “ignored,” or “made negative comments about” them. The items’ response options ranged from 0 (never) to 4 (10 or more times). The racial discrimination index was adapted from Landrine and Klonoff (1996) and created by summing seven items that measured the degree to which a respondent experienced various racial discrimination during the previous year, such as being treated unfairly or called a derogatory name or slur because of his or her race (see the appendix). The response options for the items (α = .78) ranged from 0 (none) to 4 (10 or more times). The criminal victimization scale (α = .71) consisted of five items (partially adapted from Baron, 2004), and measured whether a respondent had been victimized by theft, robbery, burglary, sexual assault, or physical assault. The items’ response options ranged from 0 (never) to 4 (4 or more times).

Negative emotions: Anger and depression

The state anger (henceforth, anger) scale (α = .81) was created by summing three items adapted from Derogatis (1977) and measured whether a respondent had “uncontrollable outbursts of temper,” “urges to beat or harm someone,” or “urges to break things” during the last 12 months. The state depression (henceforth, depression) scale (α = .80) consisted of three items adapted from Piquero and Sealock (2000), that measured the degree to which a respondent “felt sad and depressed,” had “no interest in things,” or “preferred to be alone” over the last year. Their response options for anger and depression ranged from 0 (never) to 3 (always). Both scales were coded so that a higher score indicated high levels of anger or depression.

Conditioning factors

Agnew (2006) suggested that various conditioning factors influenced the likelihood of criminal coping in response to strains and negative emotions. We focused on four of those factors: low self-control, delinquent peer association, parental control, and parental support. First, an index of low self-control measured the concept’s six dimensions identified by Gottfredson and Hirschi (1990), such as impulsivity, volatile temper, preference for simple tasks, and physical activities. The scale consisted of 24 items derived from Grasmick, Tittle, Bursik, and Arneklev (1993), and the response options for the items ranged from 1 (strongly disagree) to 4 (strongly agree). A principal component factor analysis indicated a unidimensional solution for the 24 items because only one component (eigenvalue = 3.3) had an eigenvalue greater than 1.0. Therefore, we combined 24 items to create a low self-control scale, coded so that a higher score indicated lower self-control. The scale’s alpha level was .90.

Second, an index of delinquent peer association was created by summing 11 items, and measured whether a respondent’s close friends engaged in various types of deviant behaviors during the last 12 months (e.g., using alcohol, smoking cigarettes, stealing something of worth, attacking another student, taking other student’s money with force). Next, a parental control index consisted of two items, and measured whether a respondent’s parents/guardians recognized whether their child was in trouble or engaged in delinquent behaviors, with their response options ranging from 1 (strongly disagree) to 4 (strongly agree). The index’s alpha was .84, and it was coded so a higher score indicated a high level of parental control. Finally, a parental support index (α = .85) was created by combining two items that measured whether a respondent’s parents/guardians loved and showed interest in a respondent during the last 12 months. The index was coded so that a high score indicated high parental support for a respondent; response options ranged from 1 (strongly disagree) to 4 (strongly agree).

Dependent Variables

In the current study, school bullying was operationalized as physical and psychological aggression to other students by a single student or a group of students occurring on or off school grounds (see Bosworth et al., 1999; Espelage & Swearer, 2003; Kim, Koh, & Leventhal, 2004). To measure various dimensions of school bullying, we used 15 items partially adapted from Kim et al. (2004), asking respondents how often they engaged in physical and psychological aggression toward others during the previous year. To create a psychological bullying index, we combined seven items such as “making fun of other student(s),” “leaving other student(s) out during recess,” “ignoring other student(s),” “calling other student(s) names,” and “speaking ill of other student(s).” The items’ Cronbach’s alpha was .88, which indicated high reliability. Physical bullying (α = .87) was measured by combining eight items tapping the extent of physical aggression toward other students. Some of these items were: “taking or damaging others’ belongs,” “hitting and pushing other students,” and “physically attacking other students.” Finally, a general bullying scale (α = .92) was constructed by combining physical and psychological bullying. Response options for each of these 15 items ranged from 0 (never) to 3 (6 times or more) and the indexes were coded so that a higher score indicated more involvement in physical and psychological aggression toward others during the previous year.

Control Variables

Gender, age, parental income, and academic performance were controlled for because they tend to be associated with school bullying and deviant behaviors (see Bao et al., 2004; Moon et al., 2012). Gender was a dichotomous variable (male = 1, female = 0). Age was a continuous variable that ranged from 11 to 14 years old. Parental income was measured by asking parents’ yearly household income. Academic performance measured a student’s average grade during the previous year and was coded so that a higher score indicated lower academic grades.

Findings

Table 2 presents the zero-order correlations among strains, negative emotions, conditioning factors, and bullying measures. The results show that all four strains were significantly related to school bullying in the expected positive direction. Consistent with the GST proposition, anger and depression were both positively correlated with all three measures of school bullying.

Table 2.

Correlation matrix among major independent and dependent variables.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
(1)	1
(2)	.36*	1
(3)	.42*	.46*	1
(4)	.30*	.33*	.56*	1
(5)	.34*	.40*	.46*	.42*	1
(6)	.29*	.24*	.29*	.25*	.60*	1
(7)	.42*	.27*	.31*	.33*	.48*	.41*	1
(8)	.43*	.44*	.45*	.51*	.53*	.36*	.49*	1
(9)	−.22*	−.20*	−.18*	−.08	−.06	.06	−.01	−.09	1
(10)	−.20*	−.13*	−.17*	−.08	−.07	.07	.02	−.10	.78*	1
(11)	.36*	.38*	.43*	.37*	.56*	.38*	.47*	.58*	−.13*	−.10	1
(12)	.32*	.43*	.38*	.36*	.46*	.24*	.38*	.62*	−.21*	−.18*	.71*	1
(13)	.37*	.44*	.44*	.40*	.55*	.35*	.46*	.64*	−.18*	−.15*	.94*	.91*	1

Note¹: * = p < .05

Note²: (1) Family conflicts, (2) Teachers’ emotional punishment, (3) Racial discrimination, (4) Criminal victimization, (5) Anger, (6) Depression, (7) Low self-control, (8) Delinquent peer association, (9) Parental control, (10) Parental support, (11) Psychological bullying, (12) Physical bullying, (13) General bullying

Correlations with anger (.56, .46, and .55) were larger than those for depression (.38, .24, and .35). The table also showed that youths who had low self-control or who associated with delinquent peers were more likely to bully other students, whereas those who reported higher levels of parental control and support were less likely to engage in school bullying.

Tables 3 through 5 show the results from estimating five models with each of three bullying behaviors regressed on strains, negative emotions, and conditioning factors. The first model included all four strains as well as sociodemographic variables; and then anger and depression were separately added to the baseline model in the second and third model. In the next model (Model 4), anger and depression were included together, whereas the full model (Model 5) added all four conditioning factors as well. These various models allowed us to explore whether negative emotions mediate the effects of strains on school bullying before adding conditioning variables to the model. Given that a large majority of youths in the sample indicated no involvement in school bullying behaviors, negative binomial regression was applied to estimate the models (see Long, 1997). To test hypothesized relationships, one-tailed tests were used. For any other relationships, including those whose direction is opposite to what was expected, two-tailed tests were conducted.

Table 3.

Negative Binomial Regression Models of the Psychological Bullying (N=296).

	Model 1	Model 2	Model 3	Model 4	Model 5
Strains
Family conflict	0.10* (0.03)	0.07* (0.03)	0.07* (0.03)	0.06* (0.03)	0.01 (0.03)
Teachers’ emotional punishment	0.04* (0.02)	0.01 (0.02)	0.03* (0.02)	0.01 (0.02)	0.01 (0.01)
Racial discrimination	0.04* (0.02)	0.02 (0.02)	0.04* (0.02)	0.03* (0.02)	0.02* (0.01)
Criminal victimization	0.04 (0.03)	0.01 (0.03)	0.02 (0.03)	0.01 (0.03)	−0.03 (0.03)
Negative emotions
Anger		0.16* (0.03)		0.13* (0.03)	0.08* (0.03)
Depression			0.11* (0.02)	0.05* (0.03)	0.05* (0.03)
Conditioning factors
Low self-control					0.02* (0.00)
Delinquent peer association					0.02* (0.01)
Parental control					0.01 (0.04)
Relationship with parents					−0.05 (0.04)
Control factors
Gender (Male=1)	−0.04 (0.12)	0.01 (0.11)	0.05 (0.12)	0.04 (0.11)	0.06 (0.11)
Age	0.16* (0.07)	0.19* (0.07)	0.20* (0.07)	0.20* (0.07)	0.18* (0.06)
Parental income	0.03 (0.03)	0.05* (0.03)	0.05 (0.03)	0.06* (0.03)	0.03 (0.03)
Academic performance	−0.03 (0.04)	−0.02 (0.04)	−0.03 (0.04)	−0.02 (0.04)	−0.05 (0.04)
Constant	−1.40* (0.86)	−1.93* (0.81)	−2.06* (0.84)	−2.17* (0.81)	−2.23* (0.83)
Nagelkerke R²	0.05	0.07	0.06	0.08	0.09

Note 1: Robust Standard errors in parentheses

Note 2: *p < .05 (One tailed-test)

Table 4.

Negative Binomial Regression Models of the Physical Bullying (N=296).

	Model 1	Model 2	Model 3	Model 4	Model 5
Strains
Family conflict	0.09 (0.06)	0.04 (0.06)	0.07 (0.06)	0.04 (0.06)	−0.07 (0.06)
Teachers’ emotional punishment	0.10* (0.03)	0.08* (0.03)	0.09* (0.03)	0.09* (0.03)	0.07* (0.03)
Racial discrimination	0.06* (0.03)	0.03 (0.03)	0.06* (0.03)	0.03 (0.03)	0.01 (0.03)
Criminal victimization	0.10* (0.06)	0.06 (0.06)	0.09 (0.06)	0.06 (0.06)	0.01 (0.06)
Negative emotions
Anger		0.21* (0.05)		0.22* (0.06)	0.09 (0.06)
Depression			0.07 (0.05)	−0.02 (0.05)	−0.01 (0.05)
Conditioning factors
Low self-control					0.03* (0.01)
Delinquent peer association					0.06* (0.02)
Parental control					−0.01 (0.08)
Relationship with parents					−0.13* (0.08)
Control factors
Gender (Male=1)	0.11 (0.22)	0.22 (0.22)	0.19 (0.23)	0.20 (0.22)	0.24 (0.21)
Age	0.17 (0.12)	0.22* (0.12)	0.20 (0.12)	0.22* (0.12)	0.14 (0.11)
Parental income	−0.06 (0.06)	−0.03 (0.06)	−0.06 (0.06)	−0.03 (0.06)	−0.06 (0.06)
Academic performance	−0.01 (0.08)	0.01 (0.08)	−0.01 (0.08)	0.01 (0.08)	−0.03 (0.07)
Constant	−2.60* (1.45)	−3.48* (1.42)	−3.06* (1.47)	−3.50* (1.45)	−2.44* (1.46)
Nagelkerke R²	0.06	0.07	0.06	0.07	0.10

Note 1: Robust Standard errors in parentheses

Note 2: *p < .05 (One tailed-test)

Table 5.

Negative Binomial Regression Models of the General Bullying (N=296).

	Model 1	Model 2	Model 3	Model 4	Model 5
Strains
Family conflict	0.11* (0.04)	0.07* (0.03)	0.08* (0.04)	0.07* (0.03)	−0.01 (0.04)
Teachers’ emotional punishment	0.06* (0.02)	0.03* (0.02)	0.04* (0.02)	0.03* (0.02)	0.03* (0.02)
Racial discrimination	0.04* (0.02)	0.02 (0.02)	0.04* (0.02)	0.03 (0.02)	0.02 (0.02)
Criminal victimization	0.06 (0.04)	0.03 (0.04)	0.05 (0.04)	0.02 (0.04)	−0.02 (0.03)
Negative emotions
Anger		0.17* (0.03)		0.16* (0.03)	0.08* (0.03)
Depression			0.10* (0.03)	0.04 (0.03)	0.04 (0.03)
Conditioning factors
Low self-control					0.02* (0.01)
Delinquent peer association					0.03* (0.01)
Parental control					0.01 (0.04)
Relationship with parents					−0.09* (0.04)
Control factors
Gender (Male=1)	−0.01 (0.13)	0.05 (0.13)	0.08 (0.13)	0.07 (0.13)	0.10 (0.11)
Age	0.16* (0.08)	0.19* (0.07)	0.19* (0.08)	0.21* (0.07)	0.18* (0.07)
Parental income	0.01 (0.04)	0.04 (0.03)	0.02 (0.04)	0.04 (0.03)	0.01 (0.03)
Academic performance	−0.03 (0.04)	−0.01 (0.04)	−0.03 (0.04)	−0.01 (0.04)	−0.05 (0.04)
Constant	−1.08 (0.93)	−1.80* (0.89)	−1.72* (0.93)	−1.99* (0.90)	−1.83* (0.90)
Nagelkerke R²	0.05	0.07	0.05	0.07	0.09

Note 1: Robust Standard errors in parentheses

Note 2: *p < .05 (One tailed-test)

Table 3 presents the results of negative binomial regression of psychological bullying on strains, negative emotions, and conditioning factors. The results in Model 1 show partial support for the first hypothesis about the strain-bullying relationship. Specifically, three of the four strains were positively related to psychological bullying as hypothesized: family conflict (.10), teachers’ emotional punishment (.04), and racial discrimination (.04). In the second model, anger had a positive effect on psychological bullying (.16), being consistent with GST’s prediction. More importantly, the expected mediation of anger between strains and psychological bullying (Hypothesis 2) was indicated by two of the three strains’ significant effects in Model 1 becoming nonsignificant (i.e., teacher’s emotional punishment and racial discrimination) as well as a 30% reduction in the coefficient of family conflict (.07). That is, strained students were found to be at risk of psychological bullying partly because they were likely to experience high levels of anger, which in turn increased psychological bullying.

Anger was replaced by depression in Model 3. The results show that depression was a significant predictor of psychological bullying (.11). However, depression did not mediate the effects of strains on bullying as much as anger, given that strains continued to exert significant effects on bullying. When anger and depression were added together to the baseline model (Model 4), they were both not only positively related to psychological bullying (.13 and .05) but also jointly mediated the effects of strains on bullying. In the final model, only racial discrimination was found to be significantly related to psychological bullying (.02). Consistent with the findings in the previous models, anger (.08) and depression (.05) have significant effects on psychological bullying. Among the four conditioning factors, low self-control (.02) and delinquent peer association (.02) were significant predictors of psychological bullying. In all models, age was positively associated with psychological bullying, that is, older respondents were more likely to engage in the behavior than their younger counterparts.

In Table 4, three strains—teachers’ emotional punishment, racial discrimination, and criminal victimization—were found to have positive associations with physical bullying, which provided partial support for Hypothesis 1, while the effect of criminal victimization on physical bullying became nonsignificant when anger was added (see Models 2 and 4), indicating anger’s mediation of the strain effect. Unlike anger, however, depression was not significantly related to physical bullying and had a limited role in mediating between strains and physical bullying.

In Model 4, anger remained a significant predictor of physical bullying, whereas depression continued to have no significant association with the bullying behavior. Model 5 showed that three conditioning factors were significantly related to physical bullying in the expected direction. Youths with low self-control or association with delinquent peers were more likely to report their involvement in physical bullying, whereas juveniles who had parental support were less likely to bully other youths. Teachers’ emotional punishment was the only strain that had a significant impact on physical bullying in the final model, whereas neither anger nor depression was found to have positive association with physical bullying. These findings failed to provide support for Hypothesis 2.

Family conflict, racial discrimination, and teachers’ emotional punishment had significant effects on general bullying (see Table 5). When entered individually, anger (Model 2) and depression (Model 3) were significant predictors of the bullying behavior in the expected direction, .17 and .10, respectively. However, the findings showed that these strains (two strains in Model 2 and three strains in Model 3) were positively associated with general bullying in Model 1, and that they continued to exert significant effects on bullying behavior even after the inclusion of anger or depression variables (Hypothesis 2). Consistent with the findings in Table 4, anger was a significant predictor of general bullying, whereas depression was no longer significantly related to bullying, once anger was added to the baseline model. Family conflict and teachers’ emotional punishment continued to have positive impact on general bullying as shown in Model 4. Regarding the effects of conditioning factors on general bullying, low self-control, delinquent peer association, and parental support were significantly related to general bullying in the expected direction. Teachers’ emotional punishment and anger continued to have a significant impact on general bullying in the final model.

To test Hypothesis 3 about conditioning factors, a total of 12 interaction terms were constructed. To decrease the complexity of the analysis, we first constructed a composite measure of strain by combining the four strains (family conflict, teachers’ emotional punishment, racial discrimination, and criminal victimization). Second, all constituent variables were centered to reduce the potential for multicollinearity problem (see Aiken & West, 1991). Third, four interaction terms of strain and each conditioning factor (e.g., Strain × Low self-control) were created and the effects of these interaction terms on the different types of school bullying are presented in Table 6, specifically, for Models 1, 4, and 7. In addition, eight more interaction terms were constructed by replacing strain with anger and depression and their effects are shown in Models 2, 5, 7 and Models 3, 6, 9, respectively. Given the difficulty of interpreting the results of interaction effects in nonlinear models, like negative binomial regression (Ai & Norton, 2003), we used a series of ordinary least square regressions to test the hypothesized interactions.

Table 6.

OLS regression models predicting various bullying behaviors with interactions (N=296).

	Psychological Bullying			Physical Bullying			General Bullying
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9
Gender (Male=1)	0.04 (0.38)	0.09 (0.37)	0.03 (0.37)	−0.10 (0.31)	−0.16 (0.31)	−0.24 (0.31)	−0.03 (0.60)	−0.08 (0.59)	−0.21 (0.59)
Age	0.60* (0.22)	0.60* (0.21)	0.63* (0.21)	0.17 (0.17)	0.16 (0.17)	0.15 (0.18)	0.69* (0.35)	0.75* (0.34)	0.78* (0.34)
Parental income	0.12 (0.10)	0.13 (0.10)	0.12 (0.10)	−0.00 (0.08)	0.03 (0.08)	0.01 (0.08)	0.15 (0.17)	0.16 (0.16)	0.13 (0.16)
Academic performance	−0.10 (0.13)	−0.09 (0.13)	−0.10 (0.13)	0.03 (0.10)	0.04 (0.11)	0.02 (0.11)	−0.05 (0.21)	−0.05 (0.20)	−0.09 (0.20)
Strain	0.06* (0.03)	0.04 (0.03)	0.04* (0.03)	0.03 (0.02)	0.04* (0.02)	0.05* (0.02)	0.10* (0.05)	0.08* (0.04)	0.09* (0.04)
Anger	0.46* (0.12)	0.46* (0.12)	0.49* (0.12)	0.24* (0.10)	0.23* (0.10)	0.23* (0.10)	0.68* (0.19)	0.69* (0.19)	0.71* (0.19)
Depression	0.10 (0.10)	0.12 (0.10)	0.07 (0.10)	−0.06 (0.08)	−0.08 (0.08)	−0.10 (0.08)	0.04 (0.16)	0.04 (0.16)	−0.03 (0.16)
Low self-control	0.04* (0.02)	0.04* (0.02)	0.05* (0.02)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	0.04* (0.03)	0.05* (0.03)	0.06* (0.03)
Delinquent peer association	0.16* (0.04)	0.14* (0.04)	0.15* (0.04)	0.15* (0.03)	0.17* (0.03)	0.19* (0.03)	0.30* (0.06)	0.30* (0.06)	0.34* (0.06)
Parental control	−0.08 (0.13)	−0.05 (0.13)	−0.00 (0.13)	−0.11 (0.10)	−0.12 (0.11)	−0.08 (0.11)	−0.23 (0.21)	−0.17 (0.20)	−0.08 (0.21)
Relationship with parents	−0.01 (0.13)	0.03 (0.13)	0.00 (0.13)	−0.08 (0.11)	−0.04 (0.11)	−0.10 (0.11)	−0.08 (0.21)	−0.01 (0.20)	−0.10 (0.21)
Strain x Low self-control	−0.00 (0.00)			−0.00 (0.00)			−0.00 (0.00)
Strain x Delinquent peers	−0.00 (0.00)			0.01* (0.00)			0.01 (0.00)
Strain x Parental control	−0.01 (0.01)			−0.03* (0.01)			−0.04* (0.02)
Strain x Relationship with parents	0.02 (0.01)			0.02 (0.01)			0.04 (0.02)
Anger x Low self-control		−0.01 (0.01)			−0.01 (0.01)			−0.02 (0.01)
Anger x Delinquent peers		0.02* (0.01)			0.03* (0.01)			0.05* (0.02)
Anger x Parental control		−0.03 (0.05)			−0.09* (0.04)			−0.12 (0.08)
Anger x Relationship with parents		0.12⁺ (0.05)			0.10⁺ (0.04)			0.22⁺ (0.08)
Depression x Low self-control			−0.00 (0.01)			−0.01 (0.01)			−0.01 (0.01)
Depression x Delinquent peers			0.01 (0.01)			0.01 (0.01)			0.02 (0.02)
Depression x Parental control			−0.02 (0.05)			−0.08* (0.04)			−0.10 (0.07)
Depression x Relationship with parents			0.14⁺ (0.05)			0.08⁺ (0.04)			0.22⁺ (0.07)
Constant	−4.51 (2.70)	−3.50 (2.63)	−3.88 (2.64)	−0.84 (2.19)	−0.27 (2.15)	0.16 (2.20)	−5.35 (4.28)	−3.77 (4.15)	−3.72 (4.21)
R²	0.48	0.49	0.49	0.48	0.47	0.46	0.53	0.54	0.54

Note 1: Standard errors in parentheses

Note 2: *p < .05 (One tailed-test), ⁺p < .05 (Two tailed-test)

The results for Model 1 showed that none of the four interaction terms involving strain and conditioning factors had a significant effect on psychological bullying, whereas two of the four were found to be significant for physical bullying. Specifically, Model 4 shows that strained youths were more likely to engage in physical bullying when they associated with delinquent peers (.01), while students were less likely to do so when they were supervised closely by their parents (−.03). Also, parental control’s buffering effect (−.04) was found for the general bullying model (see Model 7).

Next, Models 2, 5, and 8 showed that the two significant conditioning factors—delinquent peer association and parental control—were found to be significant in physical and general bullying, although not significant with psychological bullying (i.e., Models 5 and 8). For example, delinquent peer association increased the influence of anger on physical bullying (.03), whereas parental control tended to decrease it (−.09). The same was observed for the general bullying model. However, parental support was found to interact with anger on school bullying in the opposite direction (.12, .10, and .22): that is, parental support was likely to aggravate the influence of anger on bullying instead of ameliorating it, just as we anticipated it would. This pattern was found for interactions involving depression as well (.14, .08, and .22; see Models 3, 6, and 9). Besides this anomaly, we had only one significant interaction involving depression in Model 6, where parental control was found to weaken the positive association between depression and physical bullying (−.08).

Discussion and Conclusion

This article intended to empirically examine the generality of Agnew’s (1992, 2006) general theory of strain to explain adolescent bullying behaviors, to which GST has less often been applied relative to other behaviors. Overall findings tended to be consistent with our expectations, although some aspects of GST received more empirical support than others. Below we discuss the key findings of the current research and policy implications.

All four measures of strain were found to be positively related to psychological, physical, or general bullying, which supported the first hypothesis. All of them were strains that Agnew (2006) proposed were “most likely to cause crime,” thereby being in line with Agnew’s (2006) proposed distinction between most and least criminogenic strains among juveniles (p. 70).

The second hypothesis about negative emotions as an explanation of strain on bullying received support, with just one exception. Specifically, the significant effects of strain on bullying became nonsignificant or reduced in size when the hypothesized explanatory variables—anger and, to a lesser extent, depression—were added individually or jointly to a baseline model. So, our study provides empirical evidence for the GST proposition that strained adolescents are at risk of committing delinquency because they are likely to feel angry and/or depressed in response to strain, which in turn pressures the adolescent to take a corrective action, including deviant coping behavior like bullying. The exception for the second hypothesis was the effect of teacher’s emotional punishment on physical bullying, which remained significant across models. This might suggest that teachers, a potential source of strain experienced by student at school, more directly influenced student behaviors, like bullying, than other types of strains.

Results from our interactive models provided some support for GST, which made a key contribution that conditioning factors explained why only some, but not all, strained individuals turn to crime and delinquency. First, as expected, delinquent peer association was found to increase the antisocial influence of strain on physical bullying, whereas parental control decreased it; however none of the conditioning factors significantly changed the effect size of strain on psychological bullying. Delinquent peer association and parental control had the same pattern of moderation for anger and, to a lesser extent, depression. That is, the former factor aggravated the effect of anger on physical bullying, and the latter ameliorated the effect. The antisocial influence of depression on physical bullying was buffered by parental control, but not delinquent peer associations. Second, low self-control failed to moderate the effects of strain or negative emotions on bullying. Third, parental support that was negatively associated with physical bullying (although not with psychological bullying) was found to moderate the effects of negative emotions on all measures of bullying in the opposite direction. Parental support increased the antisocial influence of anger and depression. To further examine this counterintuitive finding, we conducted supplemental analysis by re-estimating all nine interactive models with only parental support included in each model, because the unexpected interaction effects might have been a methodological anomaly. For example, including all four multiplicative interaction terms simultaneously might have caused multicollinearity problems for this particular conditioning factor, despite the fact that these terms were constructed after their constituent terms were centered to avoid the problem. Supplemental analysis results were generally consistent with our expectations. That is, the interaction involving parental support was found to be nonsignificant in eight of the nine models with Model 2 being an exception (complete results are available on request).

A closer observation of our interaction results revealed a potentially interesting pattern: parental control was found to have conditioning effects, but had no significant direct effect on bullying. This might have had something to do with the obvious fact that parental control is a variable tapping into the domain of family, whereas bullying is a deviant behavior committed at school, and thus the former is unlikely to directly explain the latter. So, although the null finding could have been simply a data-related artifact or random oddity, it might illustrate that some variables help little to explain crime and deviance directly, but provide explanation primarily indirectly by either mediating or moderating other variables’ influence on the dependent variable. Conversely, some variables might be likely to behave the other way round, having a direct influence, but no significant mediating or moderating effect (e.g., low self-control in the present study). If this is the case, the present finding suggests that GST researchers should be creative and not confine their search for significant conditioning factors to the “predictors” of crime and deviance. The limited success in identifying such factors (Agnew, 2006) might be a blind spot in research on GST.

Finally, we need to acknowledge methodological limitations of our study so future research may address and further contribute to GST research on bullying. First of all, we analyzed data collected from a nonprobability sample of predominantly Hispanic students. This ethnic composition is, in fact, a potential strength of this study given that Latino youth have less often been studied in criminology compared with Caucasian and African American students. However, caution is warranted in interpreting our results given the lack of representativeness and statistical basis for confirmatory research. As recognized earlier, another limitation concerns the cross-sectional nature of our data, which does not allow us to establish causal ordering among our key variables. Despite suggestions by Agnew (1992) that GST be examined based on concurrent rather than lagged associations (unless panel data were collected with relatively short intervals of less than 3 months), it is not ideal to test GST using variables that have the same measurement period (i.e., during the last 12 months prior to survey). In addition, although we operationalized anger and depression as state rather than trait emotions as Agnew (2006) suggested, more research is needed to examine situational emotions such as anger or depression generated by specific strain, which is included in the same analysis (see Jang & Johnson, 2003; Mazerolle, Piquero, & Capowich, 2003).

We believe that the current research provides empirical evidence of the importance of developing programs to reduce students’ stress to prevent school bullying. Especially, as teachers’ emotional punishment was found to be one of key strains leading to various types of school bullying, schools need to adequately address this issue by promoting a positive relationship between teachers and students and through teacher training (e.g., focusing on alternative disciplinary methods not involving emotional punishment) and supervision. Also, schools should assess the degree to which students experience criminal victimization and racial discrimination within or outside school boundaries to formulate a plan. If students typically experience these strains within school, school administrators and teachers need to develop a plan to help students cope with them.

In conclusion, our study contributes to the criminological literature by applying GST to explain an adolescent deviant behavior, which is widespread in the United States but has not often been studied from the strain perspective. This research also provides empirical evidence of the generality of GST, including its applicability to Hispanic adolescents, a group that is relatively understudied in criminology as well as GST research. Although the types of strain we found to be positively associated with our dependent variable are not equally amenable to bullying prevention program, the most consistent correlate of the deviant behavior—teachers’ emotional punishment—implies a significant negative influence of teacher behavior on student behaviors at school.

Footnotes

Appendix

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.

Author Biographies

Byongook Moon, is Associate Professor in the Department of Criminal Justice at the University of Texas, San Antonio. His research focuses on the application of criminology theories to explaining delinquency. He also has been involved in research on bullying and cyber deviant behaviors. Recent publications appear in Crime & Delinquency, Journal of Research in Crime and Delinquency, and Youth Violence and Juvenile Justice.

Sung Joon Jang, is Research Professor of Sociology at the Institute for Studies of Religion at Baylor University. His research focuses on the effects of strain, religion, and spirituality on criminal offending and desistance. Recent publications have appeared in Criminology, Journal of Criminal Justice, and Sociological Focus. He is co-principal investigator on a study examining the long-term effectiveness of seminary programs in maximum security prisons.

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-87.

Agnew

(2001). Building on the foundation of general strain theory: Specifying the types of strains most likely to lead to crime and delinquency. Journal of Research in Crime & Delinquency, 38, 319-361.

Agnew

(2006). General strain theory: Current status and directions for further research. In Cullen

F. T.

Wright

J. P.

Blevins

(Eds.), Taking stock: The status of criminological theory: Advances in criminological theory (pp. 101-123). New Brunswick, NJ: Transaction.

Agnew

Brezina

Wright

J. P.

Cullen

F. T.

(2002). Strain, personality traits, and delinquency: Extending general strain theory. Criminology, 40, 43-71.

Norton

E. C.

(2003). Interaction terms in logit and probit models. Economics letters, 80, 123-129.

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Aseltine

R. H.

Gore

Gordon

(2000). Life stress, anger and anxiety, and delinquency: An empirical test of general strain theory. Journal of Health and Social Behavior, 41, 256-275.

Bao

Haas

(2004). Life strain, negative emotions, and delinquency: An empirical test of general strain theory in the People’s Republic of China. International Journal of Offender Therapy and Comparative Criminology, 48, 281-297.

Baron

S. W.

(2004). General strain, street youth and crime: A test of Agnew’s revised theory. Criminology, 42, 457-483.

10.

Bosworth

Espelage

D. L.

Simon

T. R.

(1999). Factors associated with bullying behavior in middle school students. Journal of Early Adolescence, 19, 341-362.

11.

Derogatis

L. R.

(1977). SCL 90: Administration, scoring, and procedures manual for the revised version. Baltimore, MD: Johns Hopkins University.

12.

Due

Holstem

B. E.

Lynch

Diderichsen

NicGabhain

Scheidt

, . . . The Health Behaviour in School-Aged Children Bullying Working Group. (2005). Bullying and symptoms among school aged children: International comparative cross sectional study in 28 countries. European Journal of Public Health, 15, 128-132.

13.

Espelage

D. L.

Bosworth

Simon

T. R.

(2000). Examining the social context of bullying behaviors in early adolescence. Journal of Counseling & Development, 78, 326-333.

14.

Espelage

D. L.

Swearer

S. M.

(2003). Research on school bullying and victimization: What have we learned and where do we go from here?. School Psychology Review, 32, 365-383.

15.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

16.

Grasmick

Tittle

Bursik

R. J.

Arneklev

B. J.

(1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research in Crime & Delinquency, 30, 5-29.

17.

Jang

S. J.

Johnson

B. R.

(2003). Strain, negative emotions, and deviant coping among African Americans: A test of general strain theory. Journal of Quantitative Criminology, 19, 79-105.

18.

Kim

Y. S.

Koh

Leventhal

B. L.

(2004). Prevalence of school bullying in Korean middle school students. Archives of Pediatrics & Adolescent Medicine, 158, 737-741.

19.

Landrine

Klonoff

E. A.

(1996). The schedule of racist events: A measure of racial discrimination and a study of its negative physical and mental health consequences. Journal of Black Psychology, 22, 144-168.

20.

Lee

(2011). An ecological systems approach to bullying behaviors among middle school students in the United States. Journal of Interpersonal Violence, 26, 1664-1693.

21.

Long

S. J.

(1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage.

22.

Mazerolle

Maahs

(2000). General strain and delinquency: An alternative examination of conditioning influences. Justice Quarterly, 17, 753-778.

23.

Mazerolle

Piquero

(1997). Violent responses to strain: An examination of conditioning influences. Violence and Victims, 12, 323-343.

24.

Mazerolle

Piquero

Capowich

G. E.

(2003). Examining the links between strain, situational and dispositional anger, and crime: Further specifying and testing general strain theory. Youth & Society, 35, 131-157.

25.

Moon

Hays

Blurton

(2009). General strain theory, key strains, and deviance. Journal of Criminal Justice, 37, 98-106.

26.

Moon

Hwang

McCluskey

(2011). Causes of school bullying: Empirical test of a general theory of crime, differential association theory, and general stain theory. Crime & Delinquency, 57, 849-877.

27.

Moon

Morash

(2004). Adaptation of theory for alternative cultural contexts: Agnew’s general strain theory in South Korea. Journal of International and Comparative Criminal Justice, 28, 77-104.

28.

Moon

Morash

McCluskey

(2012). General strain theory and school bullying: An empirical test. Crime & Delinquency, 58, 827-855.

29.

Nansel

T. R.

Craig

Overpeck

M. D.

Saluja

Ruan

W. J.

, & the Health Behavior in School-aged Children Bullying Analyses Working Group. (2004). Cross-national consistency in the relationship between bullying behaviors and psychosocial adjustment. Archives of Pediatrics and Adolescent Medicine, 158, 730-736.

30.

Nansel

T. R.

Overpeck

Pilla

R. S.

Ruan

W. J.

Simons-Morton

Scheidt

(2001). Bullying behaviors among US youth: Prevalence and association with psychosocial adjustment. Journal of the American Medical Association, 285, 2094-2100.

31.

Olweus

(1993). Bully/victim problems among school-children: Long-term consequences and an effective intervention program. In Hodgins

(Ed.), Mental disorder and crime (pp. 317-349). Thousand Oaks, CA: Sage.

32.

Patchin

J. W.

Hinduja

(2011). Traditional and nontraditional bullying among youth: A test of general strain theory. Youth & Society, 43, 727-775.

33.

Piquero

Sealock

M. D.

(2000). Generalizing general strain theory: An examination of an offending population. Justice Quarterly, 17, 449-484.

34.

Schneider

S. K.

O’Donnell

Stueve

Coulter

R. W.

(2012). Cyberbullying, school, and psychological distress: A regional census of high school students. American Journal of Public Health, 102, 171-177.

35.

Ttofi

M. M.

Farrington

D. P.

Lösel

Loeber

(2011). Do the victims of school bullies tend to become depressed later in life? A systematic review and meta-analysis of longitudinal studies. Journal of Aggression, Conflict and Peace Research, 3, 63-73.

36.

Wang

Iannotti

R. J.

Nansel

T. R.

(2009). School bullying among adolescents in the United States: Physical, verbal, relational, and cyber. Journal of Adolescent Health, 45, 368-375.