Stressful Life Events and Negative Emotions on Delinquency Among Korean Youth: An Empirical Test of General Strain Theory Assessing Longitudinal Mediation Analysis

Abstract

The current study examines the extent to which the argument of Agnew’s general strain theory extends to a sample of South Korean youth to be able to explain the hypothesized direct and indirect/mediated effects of negative emotions derived from various negative life experiences of strain on delinquent behaviors. These experiences can include bullying victimization, parental abuse, poor friend quality, and poor physical functioning. Data for this study were compiled from the Korean Welfare Panel Study (KOWEPS), which is a longitudinal study among 7,027 Korean households surveyed from 2006 through 2012. This study utilizes the longitudinal mediation models for temporal precedence. Results reveal that adolescents with poor-quality friends during the previous year are more likely to engage in later delinquency, and adolescents who experience bullying victimization, child abuse, and poor friendship quality during the previous year are more likely to later feel negative emotions. Furthermore, early feelings of negative emotions are significantly related to later delinquent behaviors and partially mediate the link between poor friend quality and later delinquent behaviors. Finally, bullying victimization, child abuse, and poor friend quality generated negative emotions, which led to a greater likelihood of delinquent behavior later.

Keywords

bullying victimization parental abuse poor friendship quality delinquent behaviors negative affective states longitudinal mediation models

Adolescents who engage in delinquency may have experienced early life events such as victimization (Arseneault, Bowes & Shakoor, 2010; Cullen, Unnever, Hartman, Turner, & Agnew, 2008; Hay & Meldrum, 2010), parental abuse (Lansford et al., 2007; Moylan et al., 2010), and negative peer interactions (Chapple, Tyler, & Bersani, 2005; Espelage & Holt, 2001). In recent years, many studies have examined the effect of victimization on offending behavior using Agnew’s (1992) general strain theory (GST). The literature found a positive effect, assuming that delinquency may be a coping technique that some individuals use when they perceive that there are no other legitimate alternatives to reduce the strain in their life (Agnew, 2001; Arseneault et al., 2010; Cullen et al., 2008). Thus, victimization experiences may lead to a negative emotional response, such as anger, frustration, or depression (Aseltine, Gore, & Gordon, 2000). To counter the strain produced from victimizations, an individual may turn to delinquent behaviors, such as acting aggressively, bullying, self-harm, and substance use (Bauman, Toomey, & Walker, 2013; Ford, 2014).

To date, many studies have contributed to the social sciences research literature via mediation modeling by which some variables exert an impact on others through intervening/mediating variables. For instance, it is even more informative that negative life experiences affect delinquency indirectly through negative affective states (Bao, Haas, Chen & Pi, 2014; Posick, Farrell, & Swatt, 2013). Individuals who experience negative life events such as victimization are more likely to respond with negative emotions, which, in turn, can increase the likelihood of engaging in delinquent behaviors and may be the mediating variable that leads a person to delinquent behavior.

The purpose of the current study is to discuss the application of statistical methods of the hypothesized indirect/mediated effects. Theoretically, the study used GST to extend the generalizability to a broader population by using longitudinal data from South Korea. This study first examined how negative life experiences, such as bullying victimization, parents’ physical abuse and psychological aggression, and poor friend quality, correspond to delinquency and negative emotions, respectively. Next, it estimated the relationship between negative emotions and delinquency. Finally, it examined whether negative emotions fully or partially mediate the link between negative life experiences and delinquency. In doing so, the study employed the longitudinal mediation analysis using 3-year data for describing the temporal order of events that helps in identifying the sequential association between negative life experiences, negative emotions, and delinquency. Examining delinquency in this way is important because it could provide evidence for future policy implications that target the victimization experience, negative emotional responses to these events and positive coping skills, and not just the delinquent act.

Mechanisms Underlying Multiple Sources of Negative Affective States and Delinquency

Agnew (1992) states that strain theory consists of social and psychological elements as well, which focus on individual and environmental components. Unlike other social theories, strain theory only focuses on an individual’s negative relations with others and was modified to include personal relations that contain negative or noxious stimuli (Agnew, 1985, 1992) Agnew suggests that negative affective states, such as anger, frustration, depression, and adverse emotions, contribute directly to delinquency, and those emotions are produced by a wide variety of sources of strain: failure to achieve positively valued goals, disjunction between expectations and achievement, removal of positively valued stimuli, and presentation of negative stimuli. Negative stimuli range from life experiences, such as criminal victimization, parents’ physical abuse and psychological aggression, and transfers to new schools, to poor friendship quality or dissatisfaction with peers. Those multiple sources of negative stimuli can bring on negative emotions and subsequent delinquency, such as (a) attempts to achieve goals through illegitimate means, (b) targeting the source of adversity, or (c) engaging in substance use to manage or self-medicate the negative feelings (Agnew, 1992).

Not all individuals react to negative states or noxious stimuli through deviant or delinquent behavior. Agnew (2001), therefore, enhances GST by determining when strains are most likely to lead to criminal or delinquent behavior and include events or conditions that are seen as unjust by the individual, are high in magnitude, contain an element of low self-control, and create a situation in which the individual feels pressured into using crime or delinquency to cope. The individuals who respond to strainful life events with negative emotions such as anger or frustration are more likely to engage in more aggressive or violent forms of delinquency (Aseltine et al., 2000). Although this is well documented in the literature, there is substantial literature on how negative life experiences leading to delinquency are mediated by the negative affective states that strain causes, though the findings have not been consistent.

Bullying Victimization and Delinquency

Olweus (1993) defines bullying as a series of repeated behaviors over time that are intended to inflict negative actions on the victim that may cause injury or discomfort. Bullying victimization can increase the likelihood of suicidal ideation and self-harm (Barzilay et al., 2017; Hay & Meldrum, 2010; Hong, Kral, & Sterzing, 2015), cause physical and emotional problems in adolescents (Hinduja & Patchin, 2007; Kim & Leventhal, 2008; Reijntjes, Kamphuis, Prinzie, & Telch, 2010), and lead to adolescents engaging in delinquent behaviors (Arseneault, Bowes, & Shakoor, 2010; Liang et al., 2007). Victims of bullying demonstrate an increased rate of aggressive and violent behavior and are more likely to carry a weapon than individuals who have not been bullied (Aseltine et al., 2000; Cullen et al., 2008). Other research suggests that bullying victims have an increased rate of internalizing delinquent behavior, such as substance abuse; however, this may be more prominent among males who are bullying victims. Therefore, following the GST, delinquency was found to be more prominent in individuals who “have little to lose,” (Cullen et al., 2008) such as those with weaker school social bonds from peer rejection, negative school experiences, and victimization experiences (Cullen et al., 2008; Hay & Meldrum, 2010).

Parents’ Physical Abuse and Psychological Aggression and Delinquency

Parental abuse on a child can be physical or nonphysical and can include acts of physical and verbal aggression, sexual abuse, neglect, and family violence. Abuse inflicted on a child by a parent can cause negative emotional responses by the child (Heim, Shugart, Craighead, & Nemeroff, 2010; Suliman, Mkabile, Fincham, Ahmed, Stein, & Seedat, 2009), reduce physical health status (Leeb, Lewis, & Zolotor, 2011; Shin & Miller, 2012), increase mental health issues (Norman et al., 2012), and lead to delinquent and aggressive behavior (Vissing, Straus, Gelles & Harrop, 1991). Lansford and colleagues (2002) examined child maltreatment in a longitudinal study that ranged from 1987 to 1999 and found that the consequences of the negative responses that children have from parental abuse and aggression include increased rates of aggression, social problems, social withdrawal, and increased mental illness. Similarly, a later longitudinal study conducted by Lansford et al. (2007) found that physically abused children were more likely to face additional consequences, such as failure to complete high school, higher rates of teen pregnancy, and being fired from a job, which could all cause increased rates of stress and strain in the individual. Moylan and associates (2010) examined the effects of child abuse and domestic violence on internalizing (being withdrawn, anxious, depressed) and externalizing (delinquent and aggressive) behavior and found that exposure to both child abuse and domestic violence increased the child’s risk for internal and external behaviors later in adolescence.

Poor Friendship Quality and Delinquency

Peer relationships may be a cause of strain and lead to delinquent behaviors under certain conditions, especially when an individual is receiving social support from a delinquent peer group (Brezina & Azimi, 2018). Research suggests that some individuals are rejected by prosocial peers because of individual characteristics (like low self-control) and are therefore more likely to select into delinquent groups of individuals (Chapple et al., 2005). Consequently, individuals may select into peer groups who are more likely to engage in aggressive or bullying types of behaviors, increasing the chance that the individual will be victimized (Espelage & Holt, 2001). Keenan, Loeber, Zhang, Stouthamer-Loeber, and van Kammen (1995), in a longitudinal study, indicated that boys who were associated with deviant peers were more likely to engage in disruptive or delinquent behaviors (Keenan et al., 1995). Consequently, those who engaged in these behaviors were more likely to associate and gain social status with other delinquent or aggressive individuals (Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988).

Mediated Effects of Multiple Sources of Strain on Delinquency Through Negative Emotions

Agnew (1992) states that there are intervening mechanisms that can help determine delinquency as a person’s response to strain. GST states that negative emotions can be caused by strain that an individual may face daily and that delinquency may be used as a coping strategy to mitigate the strainful feelings (Agnew, 1992). Delinquency, therefore, is a corrective action that a person utilizes to counter the negative emotions they are feeling from negative life events, such as bullying victimization, parental abuse, and conflicting peer relations. Numerous studies have demonstrated that these life events can lead to a negative emotional state, which mediates the relationship between the victimization experience and delinquent or criminal behavior (Carson, Sullivan, Cochran & Lersch, 2008; Posick et al., 2013; Slocum, Rengifo & Carbone-Lopez, 2011).

Bao and his colleagues (2012) indicated that higher levels of interpersonal strain, negative emotions, and reduced means of legitimate coping mechanisms increased the likelihood that a person engaged in delinquent behavior. Adolescents who were subject to multiple victimization often experience negative emotions, such as anger and depression, and frequently cope with the ensuing negative affective state through acts of juvenile delinquency. Similarly, Carson and colleagues (2008) tested whether early victimization affected an individual’s rate of drug use later in life and found that drug use was partially mediated by negative emotions and the strength of social bonds. Posick and associates (2013) also tested GST using negative affective states and found that negative emotions mediated the impact of strain on delinquent behavior; however, there were differences in how males and females respond to such strain. Namely, males were more likely to engage in external forms of deviance such as fighting, whereas females were more likely to demonstrate internalized deviance through avenues such as self-harm.

Korean Literature

Numerous studies have now been conducted outside of Western populations and are beginning to test GST and delinquency in countries such as South Korea, where cultural factors may influence how strain affects a person and whether there are differences in the negative affective states individuals experience from the strain. Moon and Morash (2004) were among the earliest scholars to examine GST in differing cultural contexts. In an early study of students from high schools in South Korea, two additional measures of strain were tested; these include “examination related strain” and “emotional and physical abuse by teachers,” which the authors note are not commonly tested in studies from the United States. Moon and Morash (2004) conclude that incorporating culturally specific variables into GST increases the theory’s generalizability and furthers the development for research in other regions.

Moon, Hwang, and McCluskey (2008) further examined GST on delinquency in South Korea and tested key aspects of strain that often lead to delinquency, including being bullied or victimized, and found some support for GST. However, in a finding that is inconsistent with the predictions of GST, bullying had a negative relationship with delinquency. Although the findings were not entirely as predicted, the authors demonstrated the importance of incorporating specific forms of strain that may be culturally specific, which is developed further by incorporating gender into cultural tests of GST among Korean youth (Morash & Moon, 2007). Moon, Morash, and McCluskey (2012) Stressful Life Events and Negative Emotions on Delinquency Among Korean Youth further tested the general theory of crime, GST, and differential association theory on a longitudinal sample of Korean youth and found limited support for all three theories on the applicability of each to explain bullying behavior. There were, however, significant effects of certain strains, such as physical and emotional punishment from teachers and stress/strain from examinations, on bullying.

Scholars are testing the mediating effects of emotions on delinquency in other cultural contexts as well. Moon, Morash, McCluskey, and Hwang (2009) sought to further GST by testing eight types of strain with situational and trait-based negative emotions, and delinquency, and found that situational and trait-based negative emotions had different mediating functions. Furthermore, in a longitudinal study of Korean adolescents, Jun and Choi (2015) demonstrated that academic stress is positively associated with negative emotions and addiction to the Internet, indicating that adolescents who are experiencing negative emotions may be more prone to Internet addiction. Jang and Song (2015) further tested a coping-process model originally proposed by Agnew (2013) and found that negative emotions (specifically anger) mostly mediated the effects of strain on delinquent coping (drug use and nondrug-related delinquencies).

Present Study

Many studies have examined mediation analysis testing if a mediating variable transmits the effect of an independent variable on a dependent variable. Little, however, is known about the temporal precedence from an independent variable to a mediator to a dependent variable. The current study uses longitudinal data that allow us to examine the temporal precedence inference of a mediating effect on the link between an independent variable and a dependent variable that are unavailable in cross-sectional data. Utilizing longitudinal data with a sample of South Korean adolescents, we propose the following research hypotheses:

Hypothesis 1 (H1): Adolescents who have negative life experiences such as bullying victimization, child abuse, and poor friendship quality (Wave 1) will be at a high risk of delinquent behaviors (Wave 3).

Hypothesis 2 (H2): Adolescents who have negative life experiences such as bullying victimization, child abuse, and poor friendship quality (Wave 1) are more likely to feel negative emotions (Wave 2).

Hypothesis 3 (H3): Adolescents who have negative emotions (Wave 2) will have a higher likelihood of engaging in delinquent behaviors (Wave 3).

Hypothesis 4 (H4): The tendency to experience negative emotions (Wave 2) will mediate the relationship between negative life experiences (Wave 1) and delinquent behaviors (Wave 3).

Method

Data and Sample

Data used in this study came from the on-going Korean Welfare Panel Study (KOWEPS) which was conducted by the Korean Institute of Social and Health Affairs in conjunction with the Social Welfare Research Institute of Seoul National University. The KOWEPS was designed to obtain a nationally representative, longitudinal study of 18,856 participants from 7,027 Korean households, which was surveyed annually since 2006 (www.koweps.re.kr). The KOWEPS survey population represented 90% of the census conducted in 2005, which was selected by a two-stage stratified cluster sampling design. Data were collected through face-to-face interviews conducted by trained interviewers at participants’ homes during the first year of the study (2006). A comprehensive data set provides a variety of information on families and individuals about demographic background, sources of income, social service needs, health care usage patterns, and subjective emotional and behavioral health status. The constituent questionnaires in the study consisted of the following surveys: (a) a survey of household members aged 15 years and above, (b) a household survey, and (c) a supplementary survey for specific topics. Multiple individuals were selected from the same household and were repeatedly measured at each time point of data collection.

The data for the present study were taken from the first (2006), fourth (2009), and seventh (2012) waves of KOWEPS to provide an overall picture on the temporal-ordered association among negative life experiences, negative affective states, and delinquency. To maximize the power of the analysis, the study merged two sets of data: (a) Study Population I, including participants who remained unemployed or became employed at the third wave (2008), and (b) Study Population II, including those who remained unemployed or became employed at the fifth wave (2010). The second (2007) and fourth (2009) waves were the baseline surveys, and the third (2008) and the fifth (2010) waves were the follow-up survey. Among the 7,072 households surveyed in the first wave, 18,856 subjects completed the survey questionnaire, and among the 6,207 households in the fourth wave, 16,255 subjects completed it. The final seventh wave sample for the analysis included 14,604 subjects from 5,735 households with a follow-up rate of 73.6% in 2012 compared with the first wave. Of the subjects who completed the survey, the present analyses included youth between Grades 4 and 6. Approximately 28% of the original sample dropped out of the study at some point during the 6-year period. The attrition rate was addressed in Mplus by specifying full information maximum likelihood (FIML), indicating all data contribution.

Measures

Dependent variable

Delinquency was measured at the third wave of survey year 2012 using a three-item measure, asking participants about the number of times they engaged in each of these following behaviors during the past year: “stealing,” “beating severely,” and “robbing” (Lin, Cochran & Mieczkowski, 2011). Those items were heavily and positively skewed. Each item was recoded into a dichotomous variable (1 for “yes” and 0 for “no”) and summed ranging from 0 to 3. Thus, delinquency was used as a count dependent variable. Each scale indicates how many different behaviors of that type were committed at least once during the past year. Three variety scales corresponding to several major classes of criminal behaviors, excluding status offenses, were used for two major reasons. First, such variety scores are less skewed than frequency scores. Second, they give equal weight to all delinquent behaviors, compared with frequency scores, which give more weight to minor, more frequent delinquencies and give less weight to serious, less frequent delinquencies (Lee, 2018). All variables for the analysis are described in Appendix A.

Independent variables

Bullying victimization was generated at the first wave of survey year 2006 using a six-item measure on a 4-point Likert-type scale, ranging from 1 = never to 4 = more than 4 times. This measure was created by asking students about the number of times they engaged in each of the following during the previous year: “Others severely teased me or made fun of me,” “Others completely ignored or excluded me from their group of friends and left me out of things on purpose,” “Others told lies or spread false rumors about me or sent mean notes and tried to make other students dislike me and other hurtful things like that,” “Others threatened me,” “Others robbed me,” and “Others hit, kicked, pushed, shoved around, or locked me” (Hinduja & Patchin, 2010). The Cronbach’s alpha for this measure was satisfactory (α = .71). We factor-analyzed these items to derive a single variable, with a higher value reflecting more frequent bullying victimization experiences.

Child abuse was assessed at the first wave of survey year 2006 utilizing seven items. The following questions were asked to parents about their behaviors to assess severe assault (physical abuse): “Hit him or her on some other part of the body besides the bottom/the wall,” “Threw something that could hurt him or her,” “Hit him or her with a fist or kicked him or her hand,” and “Hit him or her on some other part of the body with something like a stick or some other hard object” (Straus, Hamby, Finkelhor, Moore, & Runyan, 1998). Example items involved the following three questions to assess psychological aggression: “Scolded him or her severely as he or she felt insulted/ashamed,” “Said you would send him or her away or kick him or her out of the house,” and “Swore or cursed at him or her” (Straus et al., 1998). Those items were measured using a 5-point Likert-type scale, ranging from 1 = never to 5 = once or twice a week. The child abuse measure demonstrated satisfactory internal reliability at Wave 1 (α = .70). We conducted a confirmatory factor analysis to derive a single construct. A higher value on this latent variable reflects more frequent child abuse.

Poor friendship quality was measured at the first wave of survey year 2006, using a four-item latent variable measure, involving several questions to assess the level of students’ friendship quality. Example items included were as follows: “My friend helps me/ stays with me when I am having trouble with someone,” “My friend helps me if I needed it,” “My friend likes me,” and “My friend and I get along well” (Bukowski, Hoza, & Boivin, 1994). Those items were measured using a 4-point Likert-type scale, ranging from 1 = never to 4 = almost always. The Cronbach’s alpha for this measure was satisfactory at Wave 1 (α = .80). We attributed an answer of almost always with a score of 1, sometimes = 2, seldom = 3, and never = 4 for each question, giving participants with poor friendship quality a high score. Then, we factor-analyzed to create a single latent construct, with a higher value reflecting a lower level of friendship quality.

Negative emotion was generated at the second wave of survey year 2009, using 14 items, asking students about their mental health condition such as anxiety and depression, distinguished from poor physical functioning (Stewart, Hays, & Ware, 1988). Example items for anxiety included the following: “I am not worth much,” “I am inferior to others,” “I am tense/strained,” “I am frightened,” “I feel ashamed/embarrassed with my self-consciousness,” “I worry over misfortunes/feeling blue,” “I am worried/ feeling fearful or worried,” and I have difficulty concentrating/focusing on others” (α = .83). Example items for depression are as follows: “I felt that I was not just as good as other people,” “I felt lonely,” “I had crying spells,” “I felt fearful,” “I felt that people dislike me,” and “People were unfriendly” (α = .91; Radloff, 1977). Those items were measured using a 3-point Likert-type scale, ranging from 1 = hardly ever to 3 = often. Each measure of anxiety and depression demonstrated satisfactory internal reliability at Wave 2 (α = .83 and .91, respectively). We conducted the second-order confirmatory factor analysis to derive a single latent construct of the global domain, negative emotions, by using the two subdomains, anxiety and depression. A higher value reflected worse negative emotions.

Sociodemographic characteristics (sex, grade, lower class) and poor physical functioning were included as a control variable in the analyses. The sample consisted of 46.5% males and 52.5% females. The fourth graders were 34.8%, the fifth graders were 30.8%, and the sixth graders were 34.4%. Among all participants, 33.5% were youth in a lower class. Poor physical functioning was measured at the first wave of survey year 2006, asking students about their health condition (i.e., physical function) and about the extent to which health interferes with a variety of activities (e.g., sports, carrying groceries, climbing stairs, and walking) (Stewart et al., 1988). The response option was a 5-point Likert-type scale, ranging from 1 = very good to 5 = very bad. A higher value reflects a poorer health condition/poorer physical functioning.

Analysis

Negative binomial regression models were employed to assess delinquency, which led to unbiased estimates compared with models with a continuous variable utilizing Mplus 7.4 (Muthen & Muthen, 2011). These models make it difficult to interpret individual coefficients of variables, which requires exponentiation of the estimate. That is, taking the antilog of the coefficient allows one to get a multiplicative estimate of the change in the expected counts of delinquency for each unit increase in x (Pires & Jenkins, 2007). The likelihood analysis in negative binomial regression models is written as follows:

\log [L (β | y)] = - \sum_{i = 1}^{n} μ (x i, β) + \sum_{i = 1}^{n} y i \log [μ (x i, β)] + const .

The log link function is used for the negative binomial regression in the form of our model equation: Log (U, delinquency) = β₀ + β₁ (negative life experiences) + β₂ (negative emotions), where U is the count modeled and log(U) is the link function (i.e., how to transform U as a linear combination of the predictor variables). This implies the following: Numbers (delinquency) = exp (β₀ + β₁ (negative life experiences) + β₂ (negative emotions) = exp (β₀) × exp (β₁(negative life experiences)) × exp (β₂ (negative emotions)). The coefficients have an additive effect in the log scale. Longitudinal mediation models also were conducted to estimate the temporal precedence by quantifying coefficients relating variables to variables over time. The current study consisted of a series of stages. The first stage of the analysis was to examine zero-order relationships between both delinquency and each predictor. The second stage involved estimating longitudinal mediation models depicting the direct effect of multiple sources of negative life experiences on delinquency as well as the indirect/mediated effect of those sources on delinquency through negative emotions using the causal steps approach via the difference in coefficients approach.¹ Finally, this study estimated the magnitude of strength of indirect/mediated effects via the product of coefficients approach.² Figure 1 presents the single-mediator model.

Figure 1.

The single-mediator model.

Results

Correlations

The first stage of this analysis was the examination of zero-order relationships between each predictor and delinquency (see Table 1). The negative emotions variable at Wave 2 was significantly related to delinquency at Wave 3. Each of the predictors (child abuse, poor friend quality, poor physical functioning, and negative emotions) was significantly related to bullying victimization in the predicted (positive) directions, while grade was significantly and inversely related to bullying victimization. Poor physical functioning, negative emotions, and sex were correlated to child abuse and poor friend quality, respectively. The similar finding of the zero-order relationships to the higher order relationships in the multivariate model indicated there were no biased coefficients. Thus, it was decided that all intercorrelated predictors were included in the full multivariate model.

Table 1.

Correlations Among the Study Variables.

	1	2	3	4	5	6	7	8	9
1. Delinquency 2012	1.000
2. Bullying victimization 2006	.056	1.000
3. Child abuse 2006	.090	.286**	1.000
4. Poor friendship quality 2006	.073	.150**	.055	1.000
5. Negative emotions 2009	.146**	.203**	.166**	.128**	1.000
6. Sex 2006	.090*	.053	.161**	.085*	−.063	1.000
7. Grade 2006	.004	−.131**	−.023	−.022	.018	−.039	1.000
8. Lower class 2006	−.012	.037	−.066	.016	.024	−.051	.030	1.000
9. Poor physical functioning 2006	−.004	.125**	.083*	.228**	.087*	.087*	−.005	−.001	1.000

p < .05. **p < .01. ***p < .001.

Direct and Indirect Effects on Delinquency

We first estimated the longitudinal mediation model depicting only direct effects of multiple sources of strain (negative life experiences) at Wave 1 (2006) on delinquency at Wave 3 (2012) to estimate the coefficients relating the independent variables to the dependent variable (X → Y). Results are shown in Table 2 (see Model 1 of Table 2). The findings revealed that poor friend quality at Wave 1 (2006) had the direct and significant effect on delinquency (β = 0.31, p < .05). Regarding the model selection, we found that the coefficient of the natural log of the overdispersion was different from zero (β = 6.714, p < .001), indicating that the model is better estimated running a negative binomial regression model. It showed that our model fit well to our data.

Table 2.

Four Causal Steps of Mediation Analysis (X → M → Y).

	Model 1		Model 2		Model 3		Model 4		Model 5
	X → Y(Delinquency 2012)		X → M(Negative emotions 2009)		M → Y(Delinquency 2012)		X → Y(Delinquency 2012), adjusted for M		Indirect effect through negative emotions 2009 on delinquency 2012
	b	(SE)	b	(SE)	b	(SE)	b	(SE)	b	(SE)	95% CL
Sex 2006	0.308	0.353	−0.202**	0.081	0.72 *	0.329	0.564	0.329
Grade 2006	−0.238	0.207	0.053	0.048	−0.196	0.208	−0.237	0.203
Low class 2006	−0.142	0.375	0.069	0.086	−0.166	0.35	−0.271	0.358
Strain
Poor physical functioning 2006	−0.385	0.23	0.061	0.06	−0.297	0.209	−0.42	0.227
Bullying victimization 2006	0.128	0.125	0.145 **	0.06			0.062	0.121	0.067 *	0.035	−0.001	0.136
Child abuse 2006	0.187	0.121	0.127 *	0.057			0.106	0.132	0.059 *	0.030	0.000	0.118
Poor friendship quality 2006	0.31 *	0.148	0.106 *	0.055			0.275 *	0.143	0.049 *	0.028	−0.006	0.105
Mediator
Negative emotions 2009					0.546 ***	0.123	0.465 ***	0.134
Information criteria
Dispersion (Delinquency)	6.714*** (2.016)				6.732***(1.764)		5.901*** (1.708)
Model fit information
AIC	361.841		1,551.186		354.709		1,907.728		1,907.728
BIC	399.348		1,590.121		383.881		1,989.925		1,989.925
Sample-size-adjusted BIC(n* = (n + 2) / 24)	370.784		1,561.551		361.664		1,929.61		1,929.61

Note. AIC = Akaike information criterion; BIC = Bayesian information criterion.

Bold values are significant at *p < .05. **p < .01. ***p < .001.

Next, we estimated direct effects of multiple sources of strain on negative emotions to estimate the coefficients relating the independent variables to the mediator (X → M; see Model 2 of Table 2). The findings revealed that students who were bullied by others (β = 0.145, p < .01), who experienced child abuse (β = 0.127, p < .05), and who had poor-quality friendship (β = 0.106, p < .05) during the previous year were more likely to later feel negative emotions. Females were more likely to feel negative emotion (β = −0.202, p < .01).

In Model 3, it was estimated whether the mediating negative emotions variable at Wave 2 (2009) was significantly related to delinquency at Wave 3 (2012) (M → Y) after adjusting for control variables. Negative emotions significantly predicted delinquency at Wave 3 (β = 0.546, p < .001). Students who felt negative emotions during the previous year were generally at a high risk of engaging in delinquent behaviors later. Males were at a heightened risk of delinquency (β = 0.72, p < .05).

Finally, we tested whether the multiple sources of strain are still significant after adjusting for the mediating negative emotion variable in the full model (X → Y, adjusted for M; see Model 4 of Table 2). Poor friend quality at Wave 1 remained significant of delinquency at Wave 3 (β = 0.275, p < .05). Students with poor-quality friends during the last year (2006) were generally at a high likelihood of engaging in delinquent behaviors. Also, negative emotion at Wave 2 was significantly related to delinquency at Wave 3 (β = 0.465, p < .001). Students who felt negative emotions during the last year were generally at a higher risk of later delinquency.

Notably, it was observed that the coefficient relating the multiple sources of strain (Xs) on delinquency must be larger (in the absolute value) than the coefficient relating the multiple sources (Xs) on delinquency after accounting for the mediating negative emotions variable using the difference in the coefficients (c–c’) approach. By doing so, we could check whether the finding satisfied the fourth assumption of the causal steps approach, identifying whether the mediation is different from nonzero. The finding revealed that the coefficient relating poor friend quality to delinquency in the full model (β = 0.275) was smaller than the coefficient in the first model before adjusting for the mediator (β = 0.310). This means that the mediator did partially mediate the link between poor friend quality and delinquency. Regarding the model selection, we found the overdispersion, indicating that the coefficient was different from zero (β = 5.901, p < .001). It means the model is better estimated using a negative binomial regression model. Results of the path model for delinquency are depicted in Figure 2.

Figure 2.

Pathways on delinquency.

Finally, Model 5 estimated the extent to which the mediating negative emotions variable (M) transmits the effect of sources of strain (Xs) to outcome variables, delinquency (Y). The causal steps approach tests whether the hypothesized indirect/mediated effect is nonzero, but does not provide any numerical value of its strength. For this reason, this study turned to a discussion of an alternative approach, the product of coefficients to estimate the magnitude of strength of mediation. Consistent with the previous findings, we found significance on the magnitude of strength of specific indirect effects of all the negative life events on delinquency. Students who had the prior experience of bullying victimization, child abuse, and poor friend quality were more likely to feel negative emotions that, in turn, demonstrated the high odds of engaging in delinquent behaviors later (β = 0.067, β = 0.059, β = 0.049; p < .05).

Discussion

The goal of the current study was to examine the hypothesized direct and indirect/mediated effects by examining the mechanism of negative affective states derived from various negative life experiences of strain on delinquency by using GST. The study also conducted a statistical significance test of mediation via the longitudinal mediation models (i.e., the difference in the coefficients approach) providing information about the temporal precedence among variables over times. Furthermore, it provided a formal test of mediation (i.e., the product of coefficients approach) estimating the magnitude of strength of the specific indirect/mediated effect and confidence limits obtained via the negative binomial regression. Thus, the study was able to provide the more accurate estimate of the mediating negative emotion effect.

Findings from the longitudinal mediation model partially support the first hypothesis. There was a clear direct relationship between poor friend quality and delinquent behaviors. Adolescents with poor friend quality during the previous year were more likely to engage in later delinquency. This indicates that the exposure to poor-quality friends increased the child’s risk for delinquent behaviors later in adolescence. In other words, adolescents who have at least one friend who provides support, protection, and intimacy is less likely to engage in delinquent behaviors. This finding replicates prior research concluding that poor friend quality (Waldrip, Malcolm, & Jensen-Campbell, 2008) and peer conflict (Boman & Mowen, 2019; Brezina & Azimi, 2018; Espelage & Holt, 2001) carry substantively meaningful importance for delinquent behaviors. Thus, it supports Agnew’s (1992) argument that poor friend quality is a more consequential type of strain, meaning that this study conducted the full test of GST in a more direct way. However, the other negative life experiences (e.g., bullying victimization, and child abuse) did not show significant relationships. Regarding specific forms of strain, South Korean research has inconsistency found that bullying was significantly and inversely related to general delinquency (Moon, Hwang, & McCluskey, 2008), but criminal victimization was found to be a strong predictor of delinquency (Moon, Hwang, & McCluskey, 2012). The hypothesized differential effects of negative life experiences on delinquency according to age are based on developmental explanatory frameworks on delinquency (e.g., Benda & Corwyn, 2002). For instance, the effect of prior experiences (e.g., child abuse) on later delinquency is more likely to be mediated by other study factors among older adolescents than younger youth due to the temporal distance (Baron & Kenny, 1986). These two groups (older vs. younger) are congruent with developmental stages, identified in significant differences in social, moral, and cognitive development as well as in various types of delinquency (Moffitt, 1993). For these reasons, incorporating cross-cultural variables (e.g., failure of school and getting a low grade, parent–child relationship, excessive study pressure for college entrance) into GST plays a significant role in generalizing the theory and further developing research in other countries (Moon & Morash, 2004; Morash & Moon, 2007).

The second hypothesis, which premised that negative life experiences would relate to higher level of negative emotions, is fully supported. Adolescents who had negative life experiences such as bullying victimization, child abuse, and poor friend quality during the previous year were more likely to later feel negative emotions. This finding has broad implications for life stress research in predicting vulnerability for negative emotions and suggests that prior life events can have a high negative impact and may represent a vulnerability that can help account for variability in depression and anxiety following negative events. This finding is consistent with those from previous studies, supporting the notion that negative emotions can be caused by the stress or strain of negative life events (Cullen et al., 2008; Heim et al., 2010; Suliman et al., 2009). There is a substantial amount of support that GST focuses on an individual’s negative relations with others and was modified to include personal relations that contain negative or noxious stimuli (Agnew, 1985, 1992) Given the considerable importance of stressful life events on negative emotions to criminologists, future research should continue to examine the effect of interpersonal vulnerabilities on increasing the occurrence of stressful life events.

The next important finding shows that negative emotions in the previous year increased delinquent behaviors later. Adolescents who revealed feelings of anxiety and depression were more likely to engage in delinquent behaviors. These findings are consistent with previous research that delinquent behaviors are one adaptation that individuals may use in response to negative emotion to cope with strain (Arseneault et al., 2010). It supports GST stating that negative emotions can be caused by the stress or strain that an individual may face daily and that delinquency may be used as a coping strategy to mitigate the strainful feelings (Agnew, 1992).

The final finding of interest is the mediating/intervening effects of negative emotions on the link between negative life events and delinquent behaviors. According to the finding via the causal steps approach to mediation (i.e., the difference in the coefficients approach), only poor friend quality was significant even after adjusting for a mediating negative emotions variable. The difference of the coefficients of significant poor friend quality effects on delinquency between before and after the adjustment of the mediator was nonzero; specifically, the coefficient of poor friend quality before the adjustment of the mediator was greater than the coefficient of poor friend quality after the adjustment of the mediator. This is the result indicating that the mediated effect was nonzero (i.e., negative emotions partially mediated the link between poor friend quality and delinquency), but not any numerical value/the extent of mediated effect. Instead, this utilized the alternative method, the product of coefficients approach and found significant estimates (numerical value) of the mediating negative emotions variable that transmitted the effect of all the negative life events, bullying victimization, child abuse, and poor friend quality to delinquency. The mediating effect of negative emotions was strongest on the link between bullying victimization and delinquency. The overall findings confirm the conclusions from other recent studies (Bao, Haas, Chen & Pi, 2012; Jang & Song, 2015; Posick et al., 2013; Slocum et al., 2011) that strains generate negative emotion, which can lead to a greater likelihood of delinquent behavior. It provides strong evidence that not all adolescents who experience strain engage in delinquent behaviors—only those who feel anxiety and depression as a result of the strain. It supports Agnew’s (1992) argument that feeling negative emotions generates pressure for corrective action in the form of wrongdoing. In the finding support for significant antecedent strain variables and the mediating effect of negative emotions, we believe that both mediation analyses are reasonable for future research. Thus, it should continue to focus on understanding the nature, antecedents, and sequences of negative emotions and why it is channeled into delinquency. Results from this study imply several policy and practice recommendations. Schools should provide intervention programs for teaching children and adolescents skills to strengthen emotional self-regulation/management as well as problem-solving skills (Hampel, Meier, & Kümmel, 2008). Schools should also include physical and mental extracurricular activities that provide a break from self-consuming thoughts related to any stressful life experiences and thus help students find self-worth in explaining personal interests (Hampel & Petermann, 2005).

Limitation

Despite the strengths of the current study, some limitations need to be acknowledged. First, traditional criminologists have developed for more serious delinquent behaviors, especially male misbehaviors. This should not constrain future research from examining the gender difference in delinquency. Next, previous Korean literature suggests that culturally specific factors, such as examination/academic stress, are significantly predictive of delinquency. The data set used in the analyses does not include items related to academic stress. For future research, it would be better to incorporate these culturally specific factors into the strain theory to increase the generalizability. Last, the present analyses assessed only individual-level factors of personal characteristics. The statistical significance, in particular the Type I error rates, would be too high. Korean research has found the significant effects of school-generated strains on general delinquency. Thus, future studies might consider various types of strains students experience in the context of school environment using the multilevel mediation models and estimate whether a mediator clustered at the school level might mediate the link between predictors at the individual level and delinquency.

Conclusion

The current study advances the empirical development and applicability of Agnew’s (1992) GST by comprehensively examining the hypothesized direct and indirect/mediated effects of negative emotions derived from various negative life events on delinquency in a culturally unique setting. Results were consistent with much of the previous studies applying GST, with modest support for the direct effect of multiple sources of strain such as poor friend quality on delinquent behaviors in expected directions outside of the U.S. context. Despite no robust evidence of an intervening/mediating influence of negative emotions on the link between strain and delinquent behaviors, the results emphasize the importance of incorporating peer-related variables for GST to be applicable cross-culturally. For future research, interpersonal aggression remains a significant issue among adolescents who are trying to navigate the complexities of differing environments and social interactions during their formative years. If strain and negative emotions exacerbate the problems that adolescents face in daily life, then it is necessary for researchers to identify these contributing factors so that school staff and teachers can develop and/or modify intervention programs to reduce the incidence and intensity of juvenile delinquency.

Footnotes

Appendix

Appendix A

Description of the South Korean Youth Sample.

Variable	Scale	Wave 1	Wave 2	Wave 3
Variable	Scale	M (SD)	M (SD)	M (SD)
Dependent variable
Delinquency (n = 512)
Stealing during the past year (number of times)	Count			0.10 (0.49)
Beating severely the past year (number of times)				0.08 (0.44)
Robbing during the past year (number of times)				0.08 (0.46)
Control variables
Sex (n = 759)	1 (male) = 46.5%0 (female) = 52.5%
Grade (n = 759)	Fourth (n = 264)Fifth (n = 234)Sixth (n = 261)	5.12 (3.51)
Lower class (n = 759)	1 (low) = 33.5%0 (general) = 66.5%
Poor physical functioning 2006 (n = 605)
Extent to which health interferes with a variety of activities (e.g., sports, carrying groceries, climbing stairs, and walking)	1 (very good)2 (good)3 (so so)4 (bad)5 (very bad)	1.72 (0.81)
Independent variables
Bullying victimization 2006 (latent variable of . . .) (n = 747)
Others severely teased me or made fun of me	1 (never)2 (once)3 (twice or 3 times)	1.79 (1.05)
Others completely ignored or excluded me from their group of friends and left me out of things on purpose	1 (never)2 (once)3 (twice or 3 times)	1.31 (0.70)
Others told lies or spread false rumors about me or sent mean notes and tried to make other students dislike me and other hurtful things like that	4 (more than 4 times)	1.24 (0.59)
Others threatened me		1.10 (0.41)
Others robbed me		1.08 (0.45)
Others hit, kicked, pushed, shoved around, or locked me		1.24 (0.67)
Child abuse 2006 (latent variable of . . .) (n = 747)
Hit him or her on some other part of the body besides the bottom/ the wall (physical abuse)		1.06 (0.36)
Threw something that could hurt him or her (physical abuse)	1 (never)2 (1-2 times a year)3 (1-2 times in 2-3 months)4 (1-2 times a month)5 (1-2 times a week)	1.04 (0.24)
Hit him or her with a fist or kicked him or her hand (physical abuse)		1.07 (0.32)
Hit him or her on some other part of the body with something like a stick or some other hard object (physical abuse)		1.12 (0.42)
Scolded him or her severely as he or she felt insulted/ashamed (Psychological Aggression)		1.16 (0.57)
Said you would send him or her away or kick him or her out of the house (Psychological Aggression)		1.21 (0.59)
Swore or cursed at him or her (Psychological Aggression)		1.18 (0.56)
Poor friendship quality 2006 (latent variable of . . . ) (n = 747)
My friends spend their time with me when I feel lonely	1 (never)2 (seldom)3 (sometimes)4 (almost always)	1.83 (0.85)
My friends help me when I need their hands		1.71 (0.76)
My friends like me		1.68 (0.73)
I get along well with friends at school		1.47 (0.65)
Mediating variables
Negative emotions 2009 (latent variable of . . .) (n = 605)
I felt lonely (depression)			1.22 (0.48)
I had crying spells (depression)			1.40 (0.63)
I felt fearful (depression)			1.28 (0.51)
I felt that people dislike me (depression)			1.14 (0.40)
People were unfriendly (depression)			1.08 (0.33)
I feel of worthlessness/I feel like I am not worth much/I am inferior to others (state anxiety)			1.19 (0.46)
I am tense/strained (state anxiety)			1.45 (0.63)
I am frightened (state anxiety)			1.26 (0.54)
I feel ashamed/embarrassed with my self-consciousness (state anxiety)			1.33 (0.57)
I worry over misfortunes/feeling blue (state anxiety)			1.23 (0.51)
I am worried/feeling fearful or worried (state anxiety)			1.57 (0.66)
I have difficulty concentrating/focusing on others (state anxiety)			1.67 (0.74)

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

Sujung Cho

Notes

References

Agnew

(1985). A revised strain theory of delinquency. Social Forces, 64, 151-167.

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-88.

Agnew

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

Agnew

(2013). When criminal coping is likely: An extension of general strain theory. Deviant Behavior, 34(8), 653-670.

Arseneault

Bowes

Shakoor

(2010). Bullying victimization in youths and mental health problems: “Much ado about nothing”? Psychological Medicine, 40, 717-729.

Aseltine

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

Chen

(2012). Repeated strains, social control, social learning, and Delinquency. Youth & Society, 46(3), 402-424.

Bao

Haas

Chen

(2014). Repeated strains, social control, social learning, and Delinquency. Youth & Society, 46, 402-424.

Baron

R. M.

Kenny

D. A.

(1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

10.

Barzilay

Klomek

A. B.

Apter

Carli

Wasserman

Hadlaczky

Brunner

(2017). Bullying victimization and suicide ideation and behavior among adolescents in Europe: A 10-country study. Journal of Adolescent Health, 61, 179-186.

11.

Bauman

Toomey

R. B.

Walker

J. L.

(2013). Associations among bullying, cyberbullying, and suicide in high school students. Journal of Adolescence, 36, 341-350.

12.

Benda

B. B.

Corwyn

R. F.

(2002). The effect of abuse in childhood and in adolescence on violence among adolescents. Youth & Society, 33, 339-365.

13.

Boman

J. H.

Mowen

T. J.

(2019). Unpacking the role of conflict in peer relationships: Implications for peer deviance and crime. Deviant Behavior, 40, 882-895.

14.

Brezina

Azimi

(2018). Social support, loyalty to delinquent peers, and offending: An elaboration and test of the differential social support hypothesis. Deviant Behavior, 39, 648-663.

15.

Bukowski

W. M.

Hoza

Boivin

(1994). Measuring friendship quality during pre-and early adolescence: The development and psychometric properties of the friendship qualities scale. Journal of Social and Personal Relationships, 11, 471-484.

16.

Cairns

Neckerman

Gest

Gariepy

(1988). Social networks and aggressive behavior: Peer support or peer rejection? Developmental Psychology, 24, 815-823.

17.

Chapple

Tyler

Bersani

(2005). Child neglect and adolescent violence: Examining the effects of self-control and peer rejection. Violence and Victims, 20, 39-53.

18.

Carson

Sullivan

Cochran

Lersch

(2008). General strain theory and the relationship between early victimization and drug use. Deviant Behavior, 30(1), 54-88.

19.

Cullen

Unnever

Hartman

Turner

Agnew

(2008). Gender, bullying victimization, and juvenile delinquency: A test of general strain theory. Victims & Offenders, 3, 346-364.

20.

Espelage

Holt

(2001). Bullying and victimization during early adolescence. Journal of Emotional Abuse, 2, 123-142.

21.

Ford

(2014). Poor health, strain, and substance use. Deviant Behavior, 35, 654-667.

22.

Hampel

Meier

Kümmel

(2008). School-based stress management training for adolescents: Longitudinal results from an experimental study. Journal of Youth and Adolescence, 37, 1009-1024.

23.

Hampel

Petermann

(2005). Age and gender effects on coping in children and adolescents. Journal of Youth and Adolescence, 34, 73-83.

24.

Hay

Meldrum

(2010). Bullying victimization and adolescent self-harm: Testing hypotheses from general strain theory. Journal of Youth and Adolescence, 39, 446-459.

25.

Heim

Shugart

Craighead

Nemeroff

(2010). Neurobiological and psychiatric consequences of child abuse and neglect. Developmental Psychobiology, 52, 671-690.

26.

Hinduja

Patchin

(2007). Offline consequences of online victimization. Journal of School Violence, 6, 89-112.

27.

Hinduja

Patchin

(2010). Bullying, cyberbullying, and suicide. Archives of Suicide Research, 14, 206-221.

28.

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, 379-390.

29.

Jang

Song

(2015). A “rough test” of a delinquent coping process model of general strain theory. Journal of Criminal Justice, 43, 419-430.

30.

Jun

Choi

(2015). Academic stress and Internet addiction from general strain theory framework. Computers in Human Behavior, 49, 282-287.

31.

Keenan

Loeber

Zhang

Stouthamer-Loeber

van Kammen

(1995). The influence of deviant peers on the development of boys’ disruptive and delinquent behavior: A temporal analysis. Development and Psychopathology, 7, 715-726.

32.

Kim

Y. S.

Leventhal

(2008). Bullying and suicide: A review. International Journal of Adolescent Medicine and Health, 20, 133-154.

33.

Lansford

Dodge

Pettit

Bates

Crozier

Kaplow

(2002). A 12-Year Prospective Study of the Long-term Effects of Early Child Physical Maltreatment on Psychological, Behavioral, and Academic Problems in Adolescence. Archives of Pediatrics & Adolescent Medicine, 156(8), 824.

34.

Lansford

Miller-Johnson

Berlin

Dodge

Bates

Pettit

(2007). Early physical abuse and later violent delinquency: A prospective longitudinal study. Child Maltreatment, 12, 233-245.

35.

Lee

(2018). Contextualizing informal labeling effect on adolescent recidivism in South Korea. International Journal of Offender Therapy and Comparative Criminology, 62, 3117-3134.

36.

Leeb

R. T.

Lewis

Zolotor

A. J.

(2011). A review of physical and mental health consequences of child abuse and neglect and implications for practice. American Journal of Lifestyle Medicine, 5, 454-468.

37.

Liang

Flisher

Lombard

(2007). Bullying, violence, and risk behavior in South African school students. Child Abuse & Neglect, 31(2), 161-171.

38.

Lin

Cochran

Mieczkowski

(2011). Direct and vicarious violent victimization and juvenile delinquency: An application of general strain theory. Sociological Inquiry, 81, 195-222.

39.

Moffitt

T. E.

(1993). Adolescent-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701.

40.

Moon

Blurton

McCluskey

(2008). General strain theory and delinquency. Crime & Delinquency, 54, 582-613.

41.

Moon

Hwang

McCluskey

(2008). Causes of school bullying. Crime & Delinquency, 57, 849-877.

42.

Moon

Morash

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

43.

Moon

Morash

McCluskey

Hwang

(2009). A comprehensive test of general strain theory. Journal of Research in Crime and Delinquency, 46, 182-212.

44.

Moon

Morash

McCluskey

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

45.

Morash

Moon

(2007). Gender differences in the effects of strain on the delinquency of South Korean youth. Youth & Society, 38, 300-321.

46.

Moylan

C. A.

Herrenkohl

T. I.

Sousa

Tajima

E. A.

Herrenkohl

R. C.

Russo

M. J.

(2010). The effects of child abuse and exposure to domestic violence on adolescent internalizing and externalizing behavior problems. Journal of Family Violence, 25, 53-63.

47.

Muthén

(2011). Mplus user’s guide (6th ed.). Los Angeles, CA: Muthén & Muthén.

48.

Norman

Byambaa

Butchart

Scott

Vos

(2012). The long-term health consequences of child physical abuse, emotional abuse, and neglect: A systematic review and meta-analysis. PLoS Medicine, 9(11), e1001349.

49.

Olweus

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

50.

Pires

Jenkins

(2007). A growth curve analysis of the joint influences of parenting affect, child characteristics and deviant peers on adolescent illicit drug use. Journal Of Youth And Adolescence, 36(2), 169-183. doi:10.1007/s10964-006-9127-5

51.

Posick

Farrell

Swatt

(2013). Do boys fight and girls cut? A general strain theory approach to gender and deviance. Deviant Behavior, 34, 685-705.

52.

Radloff

L. S.

(1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401.

53.

Reijntjes

Kamphuis

J. H.

Prinzie

Telch

M. J.

(2010). Peer victimization and internalizing problems in children: A meta-analysis of longitudinal studies. Child Abuse & Neglect, 34, 244-252.

54.

Shin

S. H.

Miller

D. P.

(2012). A longitudinal examination of childhood maltreatment and adolescent obesity: Results from the national longitudinal study of adolescent health (add health) study. Child Abuse & Neglect, 36, 84-94.

55.

Slocum

Rengifo

Carbone-Lopez

(2011). Specifying the strain-violence link: The role of emotions in women’s descriptions of violent incidents. Victims & Offenders, 7(1), 1-29.

56.

Stewart

A. L.

Hays

R. D.

Ware

J. E.

(1988). The MOS short-form general health survey: Reliability and validity in a patient population. Medical Care, 26, 724-735.

57.

Straus

M. A.

Hamby

S. L.

Finkelhor

Moore

D. W.

Runyan

(1998). Identification of child maltreatment with the parent-child conflict tactics scales: Development and psychometric data for a national sample of American parents. Child Abuse & Neglect, 22, 249-270.

58.

Suliman

Mkabile

Fincham

Ahmed

Stein

Seedat

(2009). Cumulative effect of multiple trauma on symptoms of posttraumatic stress disorder, anxiety, and depression in adolescents. Comprehensive Psychiatry, 50, 121-127.

59.

Vissing

Straus

Gelles

Harrop

(1991). Verbal aggression by parents and psychosocial problems of children. Child Abuse & Neglect, 15(3), 223-238.

60.

Waldrip

A. M.

Malcolm

K. T.

Jensen-Campbell

L. A.

(2008). With a little help from your friends: The importance of high-quality friendships on early adolescent adjustment. Social Development, 17, 832-852.