Gendered Pathways to Delinquency

Abstract

This study examines whether gendered pathways to delinquency posited by general strain theory (GST) operate equivalently for contemporary Korean youth as compared to past studies focused on American youth. The study analyzed longitudinal data from 3,125 South Korean teenagers and revealed some support for GST. The findings suggest that gender equalization transition affected reports of alcohol use but not for aggressive behaviors. The findings also indicate that GST is gender-specific for Korean youth and depends on the type of delinquency, strains, negative emotionality, and conditioning factors. Unlike males, female strains tend to be more psychological and subtle, leading toward deviance through mediation effects of anger and various conditioning factors such as peer delinquency and gender role socialization. The findings also support GST as being applicable to Korean social and cultural contexts, thereby strengthening the theory’s generalizability. However, gendered pathways to delinquency in the GST framework may be conditioned by the contemporary Korean cultural context of asymmetric gender equalization transition.

Keywords

general strain theory South Korean youth gender deviance aggression

Introduction

Two key issues await resolution in criminology regarding the nexus between gender and crime. The first revolves around basic differences in criminality between males and females. Feminist criminologists argue that gender equalization must be considered in understanding such differences. As females become more socially active and gender differences become more equalized in society, disparities in criminality between males and females should deteriorate, with females becoming more likely to commit deviant acts. Still, research continues to suggest that male criminality exceeds that of females, at least for traditional forms of crime and delinquency (Broidy & Agnew, 1997).

When further classifying crime into subcategories (e.g., serious violent crimes and self-destructive deviance such as drug use, eating disorder, and alcohol use), an interesting dynamic appears as a society becomes increasingly equalized in terms of gender-specific social roles. For serious violent crimes, males continue to dominate prevalence rates. In contrast, the gender gap for deviance involving self-destructive behaviors has narrowed with increases in prevalence rates among females (Canter, 1982; Chesney-Lind & Shelden, 1992).

The second and more important issue pertains to criminogenic pathways leading to crime and deviance and how they differ across genders. Feminist criminologists suggest that mainstream criminological theories do not adequately explain female criminality and that this shortcoming is due to their heavily skewed focus toward male-oriented criminality (Messerschmidt, 1993). They argue that a causal pathway toward crime and deviance for females is qualitatively different compared to males. Thus, criminological theories should explain not only males’ pathway toward crime but also females’.

General strain theory (GST), a recent revision of classic strain theory (see Agnew, 1992, 2001, 2006), was subsequently extended to explain the gender differences in crime and female criminality. Specifically, Broidy and Agnew (1997) hypothesized that men and women experience gendered strain and negative emotionality and thus males experience strain and negative emotionality more conducive to “street” crimes, while females may be more prone to internalized behaviors (e.g., eating disorders, drug use, etc.).

In terms of empirical support, GST has fared reasonably well as key propositions of the theory continue to receive support in studies of crime and delinquency. Nonetheless, within the GST framework, gender remains a relatively less examined topic (Hay, 2003). Furthermore, the propositions of Broidy and Agnew (1997) are yet to be thoroughly examined beyond the U.S. social context. In order to establish the generalizability of GST, its principles should be validated in non-U.S. settings. This issue is particularly important in terms of understanding gender differences in the context of GST’s ability to explain criminality.

In the United States, society experienced the transition from gender inequality to more gender equality from 1950–1960s through 1970–1980s to today. Particularly during the 1970s, substantial gender equalization was witnessed across the nation. This gender equalization may have impacted gender roles, eventually leading to narrowing the gender gap in crime (Adler, 1975). This experience may be related to the rapid increase in female criminality in some areas of deviance such as shoplifting (Datesman & Scarpitti, 1980). GST propositions of gendered pathways toward crime were proposed, and subsequently tested within this peculiar experience of the U.S. settings. Research has yet to adequately explore such nuances with international samples.

Contemporary Korean society has experienced a rapid social and cultural transition from the traditional patriarchic society into the more gender-equality society. Since 1980s and 1990s, Korean society has undergone a dramatic change with regard to perceptions of gender roles. Women have experienced increases in power and professional responsibility, and the society is generally more gender equal in terms of familial, social, and political status (Lee, 2012, p. 904). For example, the average education level of South Korean women is now the highest in the world, with 82.4% of South Korean women attending a university in 2009, compared with 81.6% of men. The dramatic transformation of the female role has challenged the patriarchal history of South Korea. However, societal changes in Korea are still incomplete and show a distinctive pattern including an increase in the importance of females’ physical appearance, leading to females’ appearance-related stress (Lee, 2012, pp. 904–905). Since it is possible that Korea’s transition to gender equality could be quite different from that of the U.S. experience, it seems meaningful to examine the impact of gender equalization transition on crime and deviance and the applicability of gendered pathways toward crime of GST toward Korean context. This approach will serve to enrich the understanding of gendered nature of crime and deviance.

The current study serves that purpose. In an attempt to provide an initial examination of these issues within the South Korean context, we assess how GST operates in terms of explaining crime and deviance across genders among South Korean youth. The findings provide for a meaningful contribution to theoretical debate regarding gender and crime due to the cultural and societal difference between South Korean and the United States.

Theoretical Framework: GST and Gender

Responding to feminist criminologists’ argument that a gender-specific theory is needed to explain female criminality, proponents of GST contend that GST can explain both male and female crime and articulate the gendered difference in male and female criminality by paying attention to the way in which gender conditions the processes described by GST (Broidy & Agnew, 1997, p. 296). GST suggests that certain criminogenic strains lead to crime and deviance through negative emotionality and conditioning factors, such as social control and social learning variables (Cullen & Agnew, 2006, p. 207). According to GST, gender difference in crime and deviance is largely due to gendered pathways in that causal process from strains through negative emotionality and conditioning factors to crime.

GST (Agnew, 1992, 2001) also posits that there are three major sources of strains that increase the likelihood of crime and deviance: (1) the failure to achieve positively valued goals; (2) the removal of positively valued stimuli; and (3) the presentation of noxious/negative stimuli. Agnew (2001, 2006) subsequently suggested four major criteria of key strain and stressors increasing the possibility of crime and delinquency if strains are “seen as unjust, high in magnitude, associated with low social control, and related to some pressure or incentive to engage in criminal copying” (Agnew, 2001, p. 326). Empirical research has demonstrated support for GST in that certain strains and stressors meeting the four criteria tend to increase or decrease the possibility of crime and delinquency. Such strains and stressors include child abuse, criminal victimization, unemployment, and experience with prejudice and discrimination (Agnew, 2006).

Also, according to the theory, criminogenic strains are mediated through negative emotionality and conditioning factors. People respond to such strains through negative emotionality (e.g., anger, depression, fear, and frustrations), thus mediating the impact of strain on illegitimate forms of coping (e.g., crime). Empirical studies have found that anger tends to increase the likelihood of other directed deviance, such as interpersonal violence, while depression increases the likelihood of self-destructive behavior such as drug and substance abuse (Agnew, 2006; Jang, 2007). Agnew has also argued that conditioning factors mediate the causal process of strains and negative emotionality toward deviance. These conditioning factors include self-esteem, social support, individual coping strategy, social bonding variables, and social learning variables. Whether strain leads to crime is determined by these conditioning factors associated with the individual’s personality, resources, and interpersonal relationship, and so on (Agnew 1992, 2006; Cullen & Agnew, 2006, pp. 206–208).

A unique component of GST is that its scope has been extended to explain gender differences in crime and delinquency. Broidy and Agnew (1997) proposed three hypotheses to explain how and why women and men differ in crime types and rates in the context of GST. Their first hypothesis was that men and women experience gendered strain. That is, males experience strain more conducive to crime than females do because males tend to have greater exposure to financial strain, concern for extrinsic achievement, and experiences of victimization and interpersonal conflicts with their peers. In contrast, females experience more gender-specific strains, such as procedural justice, network relationship problems, and gender-based discrimination, due to a tendency toward a greater emphasis on relationships with others. This component can be argued in explaining attenuated crime rates for females and relatively confined to the engagement in self-directed behaviors, such as substance abuse (Agnew, 2006; Sharp, Brewster, & Love, 2005).

Broidy and Agnew’s (1997) second hypothesis suggested that one’s response to negative emotions differs across gender. Males are more likely to externalize anger, perhaps leading to crime and delinquency. By contrast, females are more likely to internalize the anger, and such internalization may manifest into other negative emotions such as depression, fear, guilt, and shame due to females’ cultural socialization. These differences may reduce the tendency toward aggressive type of crime for females (Broidy, 2001; Mirowsky & Ross, 1995).

Finally, their third hypothesis argued that males respond to strain with coping more conducive to crime, than females, due to the gendered difference in available conditioning resources. Those mediating factors include opportunity, coping skills, self-esteem, self-efficacy, social bonding, social learning, dispositions (i.e., hot temper), and other variables pointed out by other major criminological theories, particularly social learning and social bonding theories. Such factors condition an individual’s response to negative emotions and may result in different behavioral outcomes between genders. Females experience, respond to, and cope with strain differently from males.

Empirical Research on Gender and GST

To date, the major components of GST have received a fair amount of scholarly attention and have garnered considerable empirical support. However, the interest of most studies has largely been limited in examining the impact of serious strain and stressors on a set of delinquent and criminal behaviors, often neglecting an examination of the gender disparities (i.e., Broidy & Agnew’s 1997 hypotheses).

Early studies testing gender differences in the context of GST generally focused on examining the applicability of GST for both males’ and females’ delinquency and substance use. However, they tend to find no significant gender difference in the strain process (Agnew & Brezina, 1997; Hoffman & Cerbone, 1999; Hoffman & Su, 1997).

Some studies have found mixed support for the hypotheses; arguing gender differences in types of strains lead to deviance (Eitle, 2001; Hay, 2003; Mazerolle, 1998; Sharp et al., 2005). Mazerolle’s (1998) study found no evidence that the effects of strain on delinquency differed by gender. In his crime-specific analysis, however, he found some evidence that strain leads to different types (e.g., violent and property-related delinquency) of crime across gender. Studies have also found evidence relevant to the relationship between types of strains and emotional responses to strain (Baron, 2007; Broidy, 2001; Hay, 2003).

More recent studies have explicitly tested the hypotheses of GST with a full range of variables including more serious and gender-related strains, negative effects, legitimate and illegitimate coping strategies, and various constraints and dispositions conditioning the strain process (e.g., religiosity—see Johnson & Morris, 2008). Those studies found some evidence of a link between the two types of negative emotions (anger and depression) and legitimate or illegitimate copings to strain; yet, some of the findings are still mixed (Baron, 2007; Jennings, Piquero, Gover, & Perez, 2009; Kaufman, 2009; Piquero & Sealock, 2004).

Hay (2003) focused on the role of family-related strain in explaining the gender gap in delinquency. He collected self-reported data from a sample of 182 high school students (95 females and 87 males) and predicted that males would experience more family-related strain than females. He found that males tend to report more physical responses by their parents to their rule-violating behaviors than females. This physical punishment type of strain was associated with significant and positive effects on male delinquent behavior but not for females.

Piquero and Sealock (2004) examined gender differences in experiencing strain, negative effects, and legitimate and illegitimate coping techniques. Their findings were mixed in that there was no difference in the amount of strain experienced by both genders and females experienced higher levels of anger and depression, contrary to the propositions. Kaufman (2009) focused on serious strains that are relevant for males and females and found support for the theory’s proposition that serious strains (suicidal behavior by friends and family and violent criminal victimization) have a strong impact on both types of negative emotions (depressive symptom and bad temper) and different types of delinquency outcomes for males and females (suicidal thoughts, running away, weekly drinking alcohol, and violence). Jennings et al. (2009) replicated Piquero and Sealock’s (2004) study with a sample of Mexican American youth, finding mixed support for the hypotheses.

Using two waves of the National Youth Survey (NYS) data, Manasse and Ganem (2009) also tested Broidy and Angew’s hypothesis with a focus on examination of the link between victimization, depress, and gender. They found supportive evidence that victimization has a positive effect on depression and later delinquency for both genders. Depressed males were more likely to respond to victimization with delinquency than males without depressive symptoms. Also, this study found that trait depression does not fully mediate, rather it moderates the relationship between victimization and delinquency, particularly for males, in which implies that there is gender difference in the role of depression.

In summary, empirical research findings on gender and GST has generally supported principal GST propositions. However, a particular gendered pathway of strain-negative emotionality delinquency showed mixed results and depended on the types of strain, negative emotionality, and delinquency.

GST, Gender, Deviance, and South Korean Youth

South Korean society has experienced a rapid transition of gender equalization in the 1990s in every aspect of society and culture. Like the arguments of feminist criminologists, this changed gender relationship has impacted on the rapid increase in females’ participation in deviance. Substantially more number of young females consumed alcohol and committed interpersonal aggression compared to the past. However, it is also true that, just like the United States, young males’ deviance in general is still far greater than that of young females (Kim, 2002).

Empirical studies have been conducted to understand criminogenic factors of South Korean youth (Kim, 2002, 2005; Kim & Kim, 2012; Lee, 2012; Lee & Uh, 1999; Nam, 2012). Such findings have suggested that differential physical strength was identified as a significant causal factor on interpersonal violence that provides a differential opportunity toward one’s engagement of violence (Kim & Kim, 2012). Family dynamics were also identified (Kim, 2005). The perception of socialized gender role (equalization of gender role) positively affected female youths’ deviance but did not affect that of male youths (Nam, 2012). With the advancement of gender equalization, females’ physical appearance became highly emphasized, and this distinctive Korean context was leading to females’ appearance-related stress (Lee, 2012, pp. 904–905). This body of findings calls for further examination of gendered differences in the causal processes of criminality.

Some studies have tested for gender differences in the strain process among the South Korean youth (e.g., Lee, 2003; Moon & Morash, 2005; Morash & Moon, 2007). Morash and Moon (2007) examined the proposed GST and gender hypothesis in a sample of high school students from one city in South Korea and found some support for GST. They found that two stressors (abuse by teachers and exam-related stress) might have a differential impact on delinquency across genders. Similarly, Lee (2003) examined the GST propositions with a sample of 800 male high school students. Lee also found support for GST showing that the boys’ experiences of strain and stress from parents, school, and friends are positively associated with delinquency and that situational anger might mediate the effects of some strain and stresses.

However, there are several limitations within existing tests of GST and gender among Korean youth. First, each of the above-mentioned studies relied upon cross-sectional and nonrandom samples and was not representative of South Korean youth in general. Next, it appears that the findings of existing Korean studies do not fully support Broidy and Agnew’s (1997) hypotheses, rather the findings have been mixed. As the theory proposed, some found that certain types of strain lead to different types of negative emotions by gender, and there are gender differences in the coping mechanisms used by males compared to females. Less is known about the role of conditioning factors such as self-efficacy, self-esteem, negative self-image, and other social learning and social control variables because the studies lacked their focus on whether conditioning factors have different effect on the relationship between strain and delinquency by gender (Jennings et al., 2009; Morash & Moon, 2007). Also, previous studies have yet to account for negative effects (or only anger) and have included only one or two strain measures and neglected constraints and disposition variables conditioning males’ and females’ adoption of coping strategies—but see Kaufman (2009). Owing to those reasons, further research is needed to understand gender differences in the strain process among the South Korean youth.

The Current Study

The current study examines gender differences in alcohol use and interpersonal aggression among the South Korean adolescence within the GST context and also feminist criminology perspectives. Two delinquency variables were selected to compare external deviant responses (interpersonal aggression/violence) with internal deviant responses (alcohol use). Feminist criminologists argue that as gender equalization progresses in society and culture in general, the gender gap in crime and deviance will be narrowed. According to the U.S. experience and Broidy and Agnew’s proposition (1997), the feminist criminologists’ argument is partially valid. In self-directed illegitimate behaviors such as drug and alcohol use, the gender gap has been substantially closed. By contrast, in other directed deviance, such as interpersonal aggression, the substantial gender gap persisted, and males are far more likely to engage in interpersonal aggression. With the experience of substantial gender equalization transition, this pattern of gender difference regarding types of deviance is hypothesized to appear among South Korean youth.

The above outlined approach extends the literature surrounding GST and gender in several ways. First, the sample used in this study is a longitudinal and representative random sample collected from the entire South Korean youth. Thus, the existing study may produce more representative and general results for the Korean youth case. Second, this study utilizes a more complete set of variables in testing the proposed hypotheses: various types of individual strain and stress measures, two types of negative effect (anger, depression), and various types of conditioning factors drawn from previous studies and major criminological theories (social learning and social bonding theories). These variables were drawn from GST (Agnew, 1992, 2001) propositions and various existing literature. More specifically, in a more recent specification of GST, Agnew (2006) summarized the results of a large body of empirical literature carried out in recent decades, which examined GST propositions. His review presented an overview of those strains and stressors that have consistently been shown to be associated with crime and delinquency. These include subjective and/or objective stressful life events, such as “events and conditions that are disliked by individuals”; a wide range of abusive behavior (i.e., physical and verbal abuse, parental rejection); criminal victimization; a set of various negative secondary experiences occurring in school (e.g., negative relations with peers and teachers); abusive peer relations; the failure to achieve selected goals (e.g., the desire for monetary wealth); experience/exposure with/to prejudice; and other “vicarious” and/or “anticipated” strains (Agnew, 2006, pp. 102–116).

In line with GST, as outlined by Agnew (2006) and prior study results, the present study explores the following key strains: a series of personal problems (i.e., health, psychological problems, parental abuse); family conflicts and family financial problems (vicarious strains), school suspensions, school-related stress, and peer stress; victimization; and a range of stress measures (parental stress, monetary stress, and physical appearance stress). The strain variables of the current study were used in the previous literature on GST and gender (Agnew, 2006; Hay, 2003; Jang, 2007; Kaufman, 2009; Manasse & Ganem, 2009; Moon, Morash, & McCluskey, 2010; Sharp et al., 2005).

These strains can be divided into male, female, gender-neutral strains. Male strain is hypothesized to include parental physical punishment, parental stress, and peer stress. It has been argued that males are more likely than their female counterparts to react to deviance with physical punishment (Hay, 2003) and that males are more likely than females to have negative relations with adults (parents and teachers) and experience conflicts with peers because of the competitive nature of their interpersonal relations (Jang, 2007, pp. 529–530). Female strains are hypothesized to include psychological/mental problems, parental verbal abuse, family dysfunction, health problems, and appearance stress. Various literatures on deviance and gender indicated that females are more reactive to psychological, verbal, indirect, and covert impulse or aggression, the so-called relational aggression, such as psychological problems and parental verbal abuse (Bartol & Bartol, 2005, p. 258). Females are also more likely concerned with family interpersonal relationship, such as family dysfunction (Broidy & Agnew, 1997), individual health problems (Jang, 2007), and appearance stress (Lee, 2012). The remaining strain variables are hypothesized as gender-neutral strains including victimization, family financial problem, monetary stress, school stress, and suspension from school. Males are more reactive to physical victimization, while females are more reactive to emotional, verbal, and sexual victimization. Traditionally, it has been argued that monetary- and job-performance- or achievement-related stress was more likely related to male-type strains. However, with the advance of gender equalization in the general society, gendered nature of this type of strains was narrowed. Thus, here financial strains such as family financial problem and monetary stress and school performance strains such as school stress and suspension from school are set as gender-neutral strains (Broidy & Agnew, 1997).¹

Third, this study incorporates measures central to tests of gender and criminality in the GST context but have rarely been used in the past (e.g., appearance, attitudes toward gender roles, etc.). In the Korean context, appearance stress is a major source of strain for females due to the heavy emphasis on females’ appearance in the contemporary Korean society and culture (Lee, 2012). Previous research suggests that perceptions of masculinity/femininity may conflate the likelihood of deviance across genders (Broidy and Agnew, 1997; Nam, 2012).

Considering Broidy and Agnew’s (1997) major propositions and feminist criminologists’ perspectives combined with findings from previous studies, the current study tests the following hypotheses:

Hyopthesis 1: Males are far more likely to commit other-directed deviance, interpersonal aggression, than females, but the gender difference in self-destructive deviance, alcohol use, is insignificant and at best minimal due to the recent gender equalization transition in the current Korean society and culture.

Hypothesis 2: Males are subject to different strains than females, with male strains being more conducive to other-directed deviance such as interpersonal aggression and female strains being more conducive to self-destructive deviance, such as alcohol use.

Hypothesis 3: There are gender differences in the emotional response to strain. Although the level of anger and depression is similar for both genders, the causal process from types of strains to negative emotionality, anger and depression, differs across gender.

Hypothesis 4: Gendered pathways toward deviance exist. Females’ pathway from strains through negative emotionality and various conditioning factors toward deviance are distinct from males’.

Method

Data

The current study used the second and third waves of junior high school students’ survey data collected by the National Youth Policy Institution (NYPI) as a part of the Korea Youth Panel Survey (KYPS) projects (2003–2007) in South Korea. The NYPI² developed this project to construct a longitudinal database providing resources for academic research by supplying wide range of information about South Korean youths, such as participants’ educational achievements, leisure activities, and problematic behaviors. The KYPS data consisted of six waves of longitudinal data that studied a nationally representative sample of adolescents and children using a stratified multistage cluster sampling method. The KYPS project originally followed 8-year-old elementary school students and 14-year-old junior high school students for 4 and 6 years, respectively. The data collection was conducted in three stages. First, trained interviewers visited randomly selected schools and conducted face-to-face interviews with participants to fill out questionnaires asking about various aspects of participants’ experiences, social relationships, and learning environments. Second, participants took part in a self-administrated survey to respond participants’ delinquent behaviors such as smoking, drinking, and using violence. Third, telephone interviews with participants’ parents were conducted to obtain information such as family monthly income, parental educational level, and family structure (i.e., living with both or single parents). Confidentiality was assured to all participants before participants took part in the survey (Lee & Baek, 2007). In 2003, among initially selected students, a total of 3,449 students and their parents took part in the Wave 1 data collection with a response rate of 93.2%. In 2004, a total of 3,188 of those took part in the first wave survey responded to the survey and interview with a response rate of 92.4%. In 2005, at the third wave survey, a total of 3,125 of them return back to the third wave survey with a 90.6% follow-up rate from adolescents and their parents.

Measures

Dependent Variables

The current study measured two types of delinquent behaviors as the dependent variables, both drawn from Wave 3: drinking alcohol and interpersonal aggression. The classification is based on the theoretical assumption that the type and significance of strain measures and conditioning factors may predict the types of delinquency occurring differently by gender. Females, for example, may respond to strain with self-destructive behaviors, such as purging behaviors, eating disorders, and alcohol use, while males are more likely to express negative emotions involving other-directed illegitimate behaviors such as interpersonal aggressions (Jennings et al., 2009; Sharp et al., 2005; Sharp, Terling-Watt, Atkins, Giliam, & Sanders, 2001). To measure alcohol use, participants were asked whether they had imbibed any type of alcoholic beverage during the past year (yes = 1, no = 0). Interpersonal aggression was operationalized by participants’ involvement in (1) violent fighting, (2) joining gang activities, (3) money extortion, (4) verbal abuse, (5) threatening others, and/or (6) malicious behaviors against classmates during the last year. Responses for each behavior were measured on a binary response (never = 0, yes = 1) and summed as a scale and recoded as a dichotomous variable (0 = never; 1 = one or more).

Independent Variables

Independent variables are drawn from the Wave 2 data.

Strain Measures

Strain measures were operationalized, as the presentation of noxious stimuli and stain/stress measures were developed on the basis of theory and previous studies. Agnew (2006) suggested that the failures to find supportive evidence on the GST propositions regarding gender and crime in previous studies are because of using a composite strain scale masking the potential effects of the individual types of strain on crime. Thus, he recommended utilizing individual measures of strain for future studies. This study includes various strain measures (i.e., financial, interpersonal problems) identified as important measures in prior studies, including serious types of strain that more likely leading to crime and deviance (i.e., victimization, abuse; Agnew, 2006). Agnew (2001, 2006) also distinguished subjective strain from objective strain and suggested the importance using both types of strain measures. This current study includes both subjective (i.e., perceived psychological/ mental problems) and objective measures (i.e., victimization). The KSPY survey asked the participants to answer a series of personal problems in their life. Those problems include (1) health problems, (2) perceived psychological problems, (3) parental divorce and/or conflict, (4) family financial problem, (5) experience of parental verbal abuse, (6) parental psychical punishment, and (7) suspension from school. Each single item was used as an individual strain measure in analyses, and all of the items used for constructing those five additive scales were measured using a 5-choice Likert-type scale ranging from never (0) to always (5). Internal consistency for the above-mentioned measures was reasonable (α > .7).

Five composite measures reflecting various types of stressors that participants are currently experiencing were operationalized as strain measures. In the KSPY survey, a series of questions were asked to participants to determine whether they have the following stressors: (1) parental stress (a summed index of 4 items); (2) school-related stress (a summed index of 4 items), (3) stress from friends (a summed index of 3 items), (4) appearance stress (a summed index of 4 items), and (5) monetary stress (a summed index of 3 items). All items used for constructing those five additive scales were measured using a 5-choice Likert-type scale ranging from never (0) to always (5) and were reasonable in internal consistency (α > .7).

Victimization was measured with five questions about the participants’ prior experience of victimization incidences during the last year. Participants were asked to answer whether they have been a victim (yes = 1 or; no = 0) in any of the following incidences during the last year: (1) threatened, (2) money extortion, (3) aggressively beaten, (4) sexual abuse or rape, and (5) being bullied. It is one type of key strain reflecting “seen as unjust, high in magnitude” and is more likely to lead to criminal and deviant behaviors (Agnew, 2006). Participants’ victimization responses were summed up to construct a single index (ranging from 0 to 9; α > .7). The index was summed and coded as two binary responses, never (0) and once or more than once (1).

Negative Emotionality

Two types of trait-based negative emotions, anger and depression, were used in this study. Although situational negative emotions are generally preferred by the theory (Agnew, 2006), situational negative emotions were not available in the KYPS data. Mazerolle, Piquero, and Capowich (2003), however, found evidence that trait-based negative emotions are still relevant in testing GST and gender in which individuals reporting higher level of trait-based negative emotions, for example, anger, are more likely to demonstrate higher level of situational anger given exposure to strain and stress (Baron, 2007). Moreover, Agnew (2006) also argues that trait-based anger than situational-based anger is more conducive to criminal outcomes as it relates to chronic strains (Jennings et al., 2009, p. 416). With regard to the using of two emotional measures, anger and depression were selected on the basis of the findings from the prior literature examining GST. According to Agnew (2006), anger is the most influential negative emotion, arguing that “anger occupies a central role in GST for these reasons and it is not surprising that the GST research has focused on this emotion” (Agnew, 2006, p. 105).

Here, anger was measured through 4 items asking respondents if they agreed with the following statements: “I have a very hot temper,” “I feel that I am a very volatile person like explosives about to be blown up,” “I, sometimes, cannot control my temper that want to through anything when I upset,” and “I become very aggressive when I am upset” (1 = strongly disagree; 5 = strongly agree; α = .73). Depression was a composite scale consists of 4 items asking whether participants agree with the survey questions; “I always worry for everything,” “I am, sometimes, very lonely,” “sad,” and “fearful with no reason” (1 = strongly disagree; 5 = strongly agree; α = .84).

Conditioning Factors and Control Variables

Agnew (1992, 2006) stated that strain would be conditioned by various factors influencing the strain process to crime and deviant acts. The current study includes various conditioning factors, both constraints and dispositions, proposed by other major criminological theories. Agnew (2006) pointed out that conditioning factors such as social learning and social control variables tended to mediate the causal process from strain into deviance. In the current model, social learning measures include peer delinquency, while social bonding measures include parental attachment and parental supervision.

Peer delinquency was operationalized through a composite measure stemming from 5 items reflecting each respondent’s delinquent peer association. Respondents were asked how many of your close friends have ever engaged in (1) truancy, (2) being bully at school, (3) arrested by police, (4) fighting, and (5) stealing over the last year. Response for each item by participants was added as one scale (range from 0 to 62; α = .90). Parental attachment and parental supervision measures were used, as constraints would reduce stress and strain effects on delinquency among participants. The two scales were measured by 5 Likert-type scaled items ranging from strongly disagree (1) to strongly agree (5). Parental attachment is a composite scale of 5 items referring participants’ perception of a positive relationship with their parents (α = .85). Parental supervision variable was also indexed by 2 Likert-type scales measuring participants’ perception of their parents’ monitoring of their acts (α = .86).

This study also included two types of coping strategy measures, cognitive and physical (i.e., behavioral) coping strategies. Coping as a conditioning factor refers to noncriminal coping responses that tend to decrease the chance of criminal or deviant behavior that represents illegitimate coping strategies (McGivern, 2010). Agnew (2006, p. 109) suggested that the impact of strain on individual responses to crime and deviance may be ameliorated by either lowering standards used to assess outcome events or altering their perceptions of the outcomes, such as exaggerating their achievements or inflating their perceptions of their potential for future success. This cognitive coping strategy is called self-efficacy, which refers to beliefs in one’s own capacity to organize and execute the courses of action required to manage prospective situations (Grau, Salanova, & Peiro, 2001). Other coping resources that individuals may use to alleviate the adverse effects of strain include self-esteem, as a cognitive coping, and physical activities, as a physical coping (Jennings et al., 2009).

Three types of cognitive coping techniques were used in the current study. Self-efficacy and self-esteem (high on self-esteem) reflect individuals’ cognitive ability to do things well, while negative self-image (low on self-esteem) indicates individuals’ perceived self-image toward their inability to handle problems and the likelihood of self-blame. Self-efficacy is represented by a composite measure of 3 Likert-type scale items asking whether participants “trust their decision,” “believe can handle their own problems,” and “can manage their own life and issues” by themselves (1 = strongly disagree; 5 = strongly agree; α = .86). Self-esteem (high on self-esteem) was measured by 3 Likert-type scale items asking whether participants perceive themselves as “having a good personality,” consider as “an able man,” and “precious person” (1 = strongly disagree; 5 = strongly agree; α = .83). Negative self-image (low on self-esteem) was measured by 5 Likert-type scale items asking whether participants consider themselves as “worthless,” “a bad person,” “a loser,” “a troublemaker,” and “a delinquent” (1 = strongly disagree; 5 = strongly agree; α = .76). Physical coping was measured via an item asking whether participants involved in athletic activities over the past year (1 = yes, 0 = no).

Attitudes toward femininity and masculinity were measured by responses from participants to the statements reflecting their attitudes and perceptions toward gendered socialized role of women and men. Feminine social role identity construct was an additive measure by 3 items asking respondents 3 statements whether they believe that women should “have greater value in marriage than social success,” “pay more attention to men’s opinion than their own idea,” and “comply with men” (1 = strongly disagree; 5 = strongly agree; α = .72). Masculine social role identity was measured by 3 items containing statements “The most important thing for men is social success,” “men should assert his own opinion rather than comply with others,” and “men need to have power controlling others” (1 = strongly disagree; 5 = strongly agree; α = .76).

Two control variables were accounted for. They included age and family income. Age was used as a continuous measure at the time of interview (in years). Family income measured by a 3-item categorical variable represented by high- (>US$5,000 per month), medium-, and low-income (<US$1,500 per month) level, with medium income serving as the reference category. Gender moderation was accounted for through stratification. The survey items used in this study are provided in the appendix.

Analytic Strategy

In order to test the proposed hypotheses, gender-specific models were estimated for all analyses. First, mean difference tests of all variables were conducted by gender. Second, two ordinary regression analyses (Ordinary Least Squares [OLS]) were estimated (one for males, one for females) to examine the effects of all strain measures on two types of negative emotions (anger and depression). Third, two logistic regression analyses were carried out in order to examine the effects of strain/stress measures on adolescence alcohol use and interpersonal aggression. The regression analyses consisted of three steps: First, baseline models (Model 1) were developed to estimate individual strain effects on two dependent variables, alcohol use and interpersonal aggression. Second, anger and depression were added, respectively, to the baseline models to examine mediating effects of negative emotions on individual dependent variables (Model 2 and Model 3)—for similar approaches in tests of GST (see Kaufman, 2009; Swatt, Gibson, & Piquero, 2007). Third, full models were estimated to examine whether the strain process to deviance was mediated by negative emotions, and whether the condition factors had a different impact on the relationship across gender as the GST hypotheses proposed (Model 4).³ The stratification by gender allows for a determination to be made as to whether the above-noted relationships (i.e., mediating effects) are moderated by gender (correlation matrix presented in Appendix B).

Results

Table 1 presents descriptive statistics and t-test results between males and females across each measure. Table 2 presents the direct effects of individual strain measures for each negative emotion (anger and depression) across genders. Several strains exerted a significant and positive impact on depression, but this effect did not vary between genders and thus no gender difference (i.e., moderating effect) was found. Both males and females experienced depression because of their victimization, perceived psychological/mental problems, school stress, peer stress, and appearance stress. Z-tests for coefficient equality showed no statistically significant difference of effect size across gender for all variables with the significant effects on depression (Z < ±1.64).

Table 1.

Descriptive Statistics and Mean Difference Test for Variables.

Variables	Males (n = 1,509)		Females (n = 1,508)		Min	Max	p Value
Variables	Mean	SD	Mean	SD	Min	Max	p Value
Strain
Victimization***	0.14	0.34	0.39	0.48	0	1	.000
Health problem*	1.82	0.98	1.89	0.97	1	5	.032
Psychological/mental problem	1.63	0.87	1.63	0.86	1	5	.977
Family financial problem	1.95	1.03	1.94	0.97	1	5	.813
Family dysfunction	1.54	0.89	1.54	0.85	1	5	.96
Parental verbal abuse	1.82	1.00	1.66	0.91	1	5	.451
Parental physical punishment***	1.82	1.04	1.67	0.99	1	5	.000
Parental stress***	2.80	0.91	2.70	0.87	1	5	.000
School stress	2.97	0.86	3.03	0.82	1	5	.060
Peer stress***	2.23	0.85	2.10	0.82	1	5	.000
Appearance stress***	2.47	0.87	2.84	0.88	1	5	.000
Monetary stress***	2.60	1.04	2.73	1.04	1	5	.000
Suspended (school)	1.35	0.85	1.29	0.74	1	5	.077
Negative emotions
Anger**	2.60	0.82	2.70	0.84	1	5	.005
Depression***	2.72	0.87	3.03	0.88	1	5	.000
Conditioning factors
Physical coping	1.29	0.45	1.27	0.45	1	0	.317
Peer delinquency**	0.60	1.90	0.43	1.48	0	62	.008
Parental attachment***	3.32	0.75	3.52	0.76	1	5	.000
Parental supervision***	3.30	0.86	3.60	0.87	1	5	.000
Self-esteem (CC)	3.28	0.80	3.25	0.73	1	5	.187
Self-efficacy (CC) **	3.51	0.77	3.43	70	1	5	.002
Negative self-image (CC)	2.36	0.68	2.35	0.65	1	5	.587
Feminine social identity***	2.60	0.80	2.10	0.74	1	5	.000
Masculine social identity***	3.30	0.83	3.13	0.89	1	5	.000
Control
Age	15.21	0.42	15.21	0.41	14	16	.751
Low income (1 = < US$ 1500)	0.11	0.31	0.09	0.29	0	1	.339
High income (1 = ≥ US$5000)***	0.13	0.33	0.04	0.19	0	1	.000
Deviant outcome measures
Interpersonal aggression (1 = yes) ***	0.12	0.32	0.06	0.23	0	1	.000
Alcohol use (1 = yes)	0.36	0.48	0.38	0.49	0	1	.338

Note. CC = cognitive coping; SD = standard deviation.

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

Table 2.

Effects of Strain on Negative Emotions by Gender (OLS Regression).

	Anger		Depression
Variable	Males	Females	Males	Female
	B (SE)	B (SE)	B (SE)	B (SE)
Victimization	−.03 (.06)	.06 (.08)	.15 (.06)**	.20 (.08)*
Health problem	−.02 (.03)	−.05 (.03)	−.00 (.02)	.02 (.02)
Psychological/mental problem	.04 (.03)	.14 (.03) ***	.10 (.03)***	.08 (.03)**
Family financial problem	−.02 (.03)	−.01 (.02)	−.02 (.02)	−.01 (.03)
Family dysfunction	.04 (.03)	−.01 (.02)	.01 (.03)	.03 (.03)
Parental verbal abuse	−.01 (.03)	.08 (.04)*	.03 (.03)	−.02 (.03)
Parental physical punishment	.09 (.03) **	.02 (.03)	.00 (.03)	−.01 (.03)
Parental stress	.06 (.03)	.07 (.03)*	.05 (.03)	−.01 (.03)
School stress	.03 (.04)	.03 (.03)	.16 (.03)***	.24 (.04)***
Peer stress	.04 (.04)	−.01 (.03)	.11 (.03)**	.11 (.03)**
Appearance stress	.06 (.04)	.13 (.03)***	.12 (.03)**	.17 (.03)***
Monetary stress	.10 (.03)***	.09 (.03)**	.00 (.03)	.01 (.03)
Suspended (school)	.06 (.02)*	.15 (.03)***	.02 (.02)	.03 (.02)
R ²	.11	.16	.17	.20
n	1498	149	1494	1498

Note. OLS = Ordinary Least Squares; SE = standard error.

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

Regarding anger, a gender-specific pattern was observed. Only two strain measures, monetary stress and being suspended, were commonly influential on respondents’ anger in males and females. However, a gender differential was found for suspension from school (Z > ±1.64). Upon this result, females were more strongly affected by suspension from school than males do. Two factors of parental verbal abuse and parental stress, appearance stress, and psychological/mental problem were significantly related to anger among females only. By contrast, parental physical punishment was significantly related to anger among males only. This result suggested an interesting finding that gendered pathway appeared in anger. Males responded with anger on more overt and physical strain such as physical punishment. By contrast, females’ reaction to anger is a more covert and mental process. They mostly responded to mental or psychological strains. Even if suspension from school, which is not a direct and overt physical strain, significantly affected both males and females, females were more substantially affected. However, monetary stress, which is not a physical strain but can be considered as direct (not psychological but material) strain, equally affected both males and females on anger.

Logistic Regression Models

The result in Table 3 presents a series of models on alcohol use for males and females. In all the models, only two of the strain measures, monetary strain and suspension from school, were significant and positive stressors for males and females leading to alcohol use. Other strains were not associated with alcohol use. Suspension experience exerted the strongest impact on both male and female alcohol use; however, no statistically significant difference was found in this effect between genders (Z < ±1.64). Monetary strain is also related to alcohol use for both males and females. No statistical significant effect size difference was observed as well.

Table 3.

Logistic Regression of Alcohol Use on Measures of GST by Gender—Odds Ratios and Standard Errors.

	Model 1		Model 2		Model 3		Model 4
	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)
Victimization	1.22 (.19)	1.12 (.24)	1.23 (.20)	1.17 (.23)	1.20 (.19)	1.12 (.24)	1.14 (.19)	1.15 (.25)
Health problem	0.92 (.06)	1.07 (.06)	0.92 (.06)	1.07 (.07)	0.91 (.06)	1.07 (.07)	0.97 (.07)	1.10 (.07)
Psychological/mental problem	0.94 (.07)	1.00 (.07)	0.94 (.07)	0.99 (.07)	0.93 (.07)	1.01 (.08)	0.97 (.09)	0.95 (.08)
Family financial problem	1.03 (.06)	1.05 (.07)	1.03 (.06)	1.05 (.07)	1.03 (.06)	1.05 (.07)	1.02 (.07)	1.05 (.08)
Family dysfunction	1.12 (.08)	1.15 (.08)	1.11 (.08)	1.15 (.09)	1.11 (.08)	1.16 (.09)	1.06 (.08)	1.17 (.09)
Parental verbal abuse	1.06 (.08)	1.12 (.09)	1.06 (.08)	1.09 (.09)	1.06 (.08)	1.10 (.09)	0.95 (.08)	1.03 (.09)
Parental physical punishment	1.00 (.07)	1.06 (.08)	0.99 (.07)	1.07 (.08)	1.01 (.07)	1.06 (.08)	1.04 (.10)	1.00 (.08)
Parental stress	1.06 (08)	1.16 (.09)	1.05 (.09)	1.15 (.10)	1.05 (.08)	1.17 (.10)	1.07 (.10)	1.15 (.10)
School stress	1.16 (.10)	0.87 (.08)	1.16 (.10)	0.86 (.08)	1.14 (.10)	0.86 (.08)	1.10 (.08)	0.85 (.08)
Peer stress	0.93 (.07)	0.89 (.07)	0.92 (.07)	0.89 (.07)	0.91 (.07)	0.90 (.07)	0.96 (.09)	0.89 (.08)
Appearance stress	0.92 (.08)	0.99 (.08)	0.91 (.08)	0.98 (.08)	0.90 (.08)	0.97 (.09)	0.87 (.08)	0.97 (.09)
Monetary stress	1.20 (.08) **	1.23 (.08) **	1.17 (.08) *	1.21 (.08) **	1.20 (.08) **	1.23 (.08) **	1.19 (.08) *	1.10 (.08)
Suspended (school)	1.41 (.09) ***	1.69 (.13) ***	1.36 (.09) ***	1.66 (.13) ***	1.41 (.09) ***	1.68 (.14) ***	1.28 (.09) **	1.41 (.13) ***
Anger			1.15 (.08) *	1.15 (.08) *			1.07 (.08)	1.04 (.08)
Depression					1.14 (.08) *	1.05 (.08)	1.10 (.09)	0.95 (.08)
Physical coping							1.04 (.13)	0.94 (.07)
Peer delinquency							1.28 (.06) ***	1.45 (.12) ***
Parental attachment							0.97 (.09)	0.93 (.09)
Parental supervision							0.94 (.07)	0.89 (.07)
Self-esteem (CC)							1.01 (.08)	1.03 (.09)
Self-efficacy (CC)							1.06 (.09)	1.07 (.10)
Negative self-image (CC)							1.14 (.12)	1.47 (.17) **
Feminine social identity							1.10 (.08)	1.00 (.09)
Masculine social identity							1.02 (.08)	1.18 (.09) *
Age							0.87 (.12)	0.95 (.14)
Low income (1 ≤US$1,500)							1.09 (.21)	0.96 (.20)
High income (1 ≥ US$5,000)							0.77 (.14)	1.21 (.35)
Pseudo R²	.05	.06	.05	.06	.05	.06	.07	.10
n	1,494	1,494	1,494	1,491	1,494	1,494	1,455	1,474

Note. CC = cognitive coping; SE = standard error.

†p < 0.1. *p < .05. **p < .01. ***p < .001.

Results from Models 2 and 3 suggest that anger had a positive and significant effect on alcohol use for both the genders with no statistical difference of effect size (Z < ± 1.64). At the same time, the direct effects of two strain measures, monetary stress and suspension from school, still remain. This finding tells that anger partially mediates the two strain measures for both males and females. Depression was significantly related to alcohol use among males only. No significant effect of depression on alcohol use was found among females. For males, no mediation effect appeared and only direct effect of depression on alcohol use was observed. Also for females, the findings suggested that depression does not mediate strain for female alcohol use.

Model 4 presents the regression results of a full set of strain measures, negative emotions, and conditioning and control variables on alcohol use for males and females. For males, two strains, monetary stress and suspension from school, maintained their direct effects, but the direct effect of monetary stress was not observed in the full model for females’ drinking, although the direct effect of suspension was sustained. For conditioning factors, having delinquent friends was a significant and equivalently positive factor for both males and females (Z < ± 1.64). Negative self-image and the masculine social role had significant and positive effects on alcohol use only for females. Noticeably, anger and depression were not significant in the full model when conditioning factors were taken into account. This finding suggests an interesting mediation effect of conditioning factors and gendered pathways toward alcohol use. For males, two strain measures leading to alcohol use were partially mediated by two intervening factors, anger and peer delinquency. The first mediation effect of anger was completely mediated by the second mediation effect of peer delinquency. The direct effect of depression was also completely mediated by peer delinquency. However, this peer delinquency only partially mediated strain measures, whose direct effect sustained. For females, the mediation effects of three conditioning factors, not only peer delinquency but also negative self-image and masculine social identity, were observed. Monetary strain was partially mediated by anger but completely mediated by three conditioning factors that are the second mediation effect. The anger was also completely mediated by the conditioning factors. Suspension effect was partially mediated by conditioning factors, and its direct effect was persisted.

Table 4 demonstrates the results of gender-specific logistic regression models for interpersonal aggression. Victimization and school suspension had a positive and significant impact on interpersonal aggression for both males and females with no significant gender difference of effect size (Z < ± 1.64), while psychological/mental problems had positive influence only on female engagement in interpersonal aggression (Model 1). Anger was a significant predictor of both male and female interpersonal aggression with no significant gender difference in effect size (Z < ± 1.64). The findings indicated that anger only partially mediated strain measures, suspension for genders, victimization, and psychological/mental problem for females. The partial mediating effect of anger was observed for male’s victimization strain. The direct effects of all strains, which were significant in the Model 1, sustained in Model 2. Depression was a positive and significant predictor for males’ involvement in interpersonal aggression but not for females. This means that females’ depression was not related to interpersonal aggression unlike males. Depression had the mediation effect for males’ victimization, but no such effect was observed in the case of males’ suspension (Model 3).

Table 4.

Logistic Regression of Interpersonal Aggression on Measures of GST by Gender—Odds Ratios and Standard Errors.

	Model 1		Model2		Model3		Model4
	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)	Males Odds (SE)	Females Odds (SE)
Victimization	1.91 (.39) **	2.44 (.79) **	1.94 (.41) **	2.41 (.79) **	1.86 (.38) **	2.47 (.81) **	2.27 (.49) ***	2.32 (.89) *
Health problem	1.15 (.11)	0.97 (.12)	1.16 (.11)	0.97 (.13)	1.15 (.11)	0.95 (.12)	1.14 (.12)	0.98 (.15)
Psychological/mental problem	1.10 (.12)	1.52 (.21) **	1.09 (.12)	1.44 (.20) *	1.07 (.11)	1.54 (.21) **	1.10 (.13)	1.42 (.22) *
Family financial problem	0.95 (.09)	1.28 (.17)	0.97 (.09)	1.30 (.17)	0.97 (.09)	1.30 (.17)	1.03 (.11)	1.29 (.19)
Family dysfunction	0.96 (.10)	0.93 (.13)	0.95 (.10)	0.92 (.13)	0.95 (.10)	0.93 (.13)	0.95 (.11)	0.90 (.14)
Parental verbal abuse	0.83 (.10)	1.08 (.17)	0.83 (.09)	1.05 (.17)	0.83 (.09)	1.08 (.18)	0.76 (.10)	1.05 (.20)
Parental physical punishment	1.22 (.13)	0.98 (.15)	1.20 (.12)	0.97 (.15)	1.22 (.13)	0.94 (.14)	1.23 (.14)	0.88 (.14)
Parental stress	1.10 (.14)	0.88 (.15)	1.05 (.13)	0.88 (.15)	1.07 (.13)	0.92 (.16)	1.05 (.15)	0.91 (.18)
School stress	1.14 (.15)	0.88 (.15)	1.13 (.14)	0.88 (.17)	1.10 (.14)	0.85 (.18)	1.04 (.14)	0.82 (.16)
Peer stress	1.09 (.13)	0.93 (.15)	1.09 (.13)	0.95 (.15)	1.05 (.12)	0.93 (.15)	1.08 (.14)	1.02 (.18)
Appearance stress	0.90 (.13)	1.00 (.18)	0.88 (.13)	0.93 (.16)	0.87 (.11)	0.98 (.18)	0.92 (.13)	0.86 (.17)
Monetary stress	1.17 (.12)	1.24 (.18)	1.14 (.11)	1.18 (.17)	1.17 (.12)	1.22 (.17)	1.14 (.12)	1.11 (.18)
Suspended (school)	1.50 (.12) ***	2.05 (.21) ***	1.48 (.12) ***	1.92 (.21) ***	1.51 (.13) ***	2.09 (.23) ***	1.52 (.15) ***	1.77 (.23) ***
Anger			1.37 (.15) **	1.66 (.26) **			1.27 (.13) *	1.44 (.25) †
Depression					1.31 (.13) *	1.08 (.16)	1.12 (.15)	1.02 (.17)
Physical coping							1.12 (.22)	0.88 (.26)
Peer delinquency							1.05 (.04)	1.23 (.09) **
Parental attachment							1.10 (.16)	1.03 (.22)
Parental supervision							0.90 (.10)	0.79 (.14)
Self-esteem (CC)							1.00 (.12)	1.25 (24)
Self-efficacy (CC)							1.05 (.14)	0.94 (.19)
Negative self-image (CC)							1.05 (.16)	1.39 (.32)
Feminine social identity							1.01 (.13)	0.95 (.18)
Masculine social identity							1.01 (.13)	1.47 (.26) *
Age							0.68 (.16)	2.06 (.62) *
Low income (1 = < $1,500)							1.27 (.36)	1.34 (.52)
High income (1 = ≥ $5,000)							1.38 (.34)	0.70 (.53)
Pseudo R ²	.05	.15	.07	.16	.07	.15	.10	.21
n	1490	1494	1490	1491	1490	1489	1450	1474

Note. CC = cognitive coping.

† p < 0.1. * p < .05. ** p < .01. ***p < .001.

In the full model (Model 4), a very interesting gender-specific pattern appeared. The mediation effects of conditioning factors were conspicuous for females, but no mediation effects of both negative emotionality and conditioning factors were observed for males. Rather, the effect of anger was present for males. For females, direct effects of all strains were present in Models 1, 2, and 3, but their impacts were partially mediated by several conditioning factors (peer delinquency, masculine social identity, and age). Anger was completely mediated by those conditioning factors in Model 2, suggesting that females who are older, having more delinquent friends, and have positive attitude toward masculine social role identity were more likely to be involved in interpersonal aggression. By contrast, in case of males, no such mediation effects of conditioning factors were observed. The direct effect of anger was sustained but that of depression disappeared. The direct effects of two strain measures that were significant in the previous models remained, but the effect sizes increased. This meant that anger had mediated the effect for those two strains. This finding suggests that, for males, those who have strains of victimization or suspension from school may directly engage in interpersonal aggression unlike females. Males with anger are directly involved in interpersonal aggression. Finally, males with victimization and/or suspension have higher chance of interpersonal aggression if they reported higher levels of anger.

Discussion

The purpose of this study was to explore and explain the gendered nature of crime in the Korean social and cultural context through the lens of GST. In doing so, the study also examined whether recent gender equalization in Korean society has had any change in other-directed deviance and/or self-destructive deviance. The current study also contributes to developing the literature on gender, deviance, and GST by examining the generalizability of the theory to South Korean adolescents and extends the GST literature by exploring the impact of various types of strain and negative emotionality on different deviance outcome measures (internal and external deviant responses) with a more complete set of conditioning variables.

The results of a series of analyses generated some support for the proposed hypotheses (1–4). Further, the gendered pathways to deviance (alcohol use and interpersonal aggression) through GST seem to reflect a distinctive Korean cultural and social context. First, the results showed empirical supports for Hypothesis 1. According to the t-test results, the significant gender difference was observed in interpersonal aggression but not in alcohol use. This means that, in the contemporary Korean society, males tend to be more likely to commit other-directed deviance such as interpersonal aggression than females do, but the gender gap was almost nonexistent in self-destructive deviance such as alcohol use. Thus, the gender equalization trend in general society might have narrowed the gender gap in alcohol use but not for interpersonal aggression. This result may describe the dual standards of the gender role compared to the past Korean society where female alcohol use was considered deviant and unacceptable, unlike male alcohol use in the current Korean society. Due to the gender equalization transition, females’ alcohol use is socially and culturally perceived acceptable on a level comparable to males’. Gender differences within interpersonal aggression among Korean youth tend to mirror that of the United States (i.e., it is more acceptable among males and more taboo for females, generally speaking)

The results found mixed support for Hypothesis 2. Whether males are subject to different strains compared to females depends on types of negative emotionality and deviance. Korean males were found to be significantly more likely to report some strains, such as parental physical punishment, parental stress, and peer stress, while females are significantly more likely report other strains, such as victimization, health problems, and appearance stress. However, not every strain was related to negative emotionality and deviance. Regarding strains toward negative emotionality, gender-specific pathways emerged for the strain-anger dynamic, but for the pathway of strain-depression, both genders were represented by equivalent strain, reflecting a gender-neutral pattern. Still, not all strains leading toward negative emotionality had an impact on deviance. The findings that a substantive proportion of the gender differences exist within GST depend on the type of deviance. For alcohol use, suspension and monetary stress are gender neutral and were predictive of alcohol use through anger for both the genders and of depression in males. For interpersonal aggression, victimization and suspensions appear to be gender neutral and thought their impact on aggression was positive, operating through anger for males and females and depression in males.

We found that only psychological/mental problems can be considered a gender-specific strain variable that appears to be mediated by anger and several conditioning factors in explaining interpersonal aggression. In sum, the findings partially supported the proposition that there are some gender-specific strains to which only males or females are subjected on the ground that the gender-specific strains are highly conditioned by types of negative emotionality and types of deviance.

Turning to Hypothesis 3, the findings partially support the idea that there are gender differences in the emotional response to strain. Such gender differences appear to be conditioned by types of negative emotionality and types of strain. No exclusively gender-specific negative emotionality was found (e.g., anger for males and depression for females). Both males and females feel anger and depression. Interestingly, for both anger and depression, Korean females tend to have a significantly higher level of negative emotionality than males. This finding is interesting and not expected from the theoretical propositions and empirical studies for the U.S. subjects. This might explain the Korean cultural context that female youths are subjected to greater level of negative emotionality due to the relatively greater level of school and family control and social pressure for females compared to male youths. In case of anger, the Hypothesis 3 of the gender differences is supported, but the very same Hypothesis is rejected in case of depression, where no gender differences were found. These findings are consistent with past work (Broidy, 2001).

In short, Hypothesis 3 is supported under certain conditions. When females felt psychological/mental problem, parental verbal abuse, parental stress, or appearance stress, they responded with anger. Males did not show this causal process. By contrast, when males felt parental physical punishment, they responded with anger. This causal process did not occur in females. These findings are consistent with Hay’s study (2003). These findings indicate that males tend to respond with anger to physical and direct stain, while females tend to respond with anger to psychological and indirect strain.

Finally, Hypothesis 4 was supported. We found that pathways toward deviance for both outcomes differed between males and females. Depending on gender differences, causal processes of strains to types of deviance were differentially intermediated by different types of negative emotionality and conditioning factors. Among males, for example, alcohol use, criminogenic strains, monetary stress, and suspension were partially mediated through anger and only one conditioning factor (peer delinquency). The direct effects of anger and depression were completely mediated by peer delinquency. However, for females’ alcohol use, other conditioning factors, negative self-image, and masculine social identity as well as peer delinquency were important for criminogenic causal process. Depression was not related to females’ alcohol use. Females’ monetary stress was completely mediated by the presence of peer delinquency, low self-esteem (low self-image), and masculine gender role (masculine social identity).

Regarding interpersonal aggression, gendered pathways were also present, supporting Hypothesis 4. Males’ interpersonal aggression was affected by victimization and suspension only, which are more obvious and direct strains. Both negative emotions, anger and depression, were stepped in the causal process of strains to interpersonal aggression. Both anger and depression had mediation effects and were moderated by gender, depending on the interplay of types of strains and negative emotions. Eventually, with the inclusion of conditioning factors, the significance of depression effect evaporated, and the effect of anger sustained and showed gender moderation effects for strain measures. By contrast, females’ interpersonal aggression was more complicated. Based on this inclusion of psychological strain, the results showed that female interpersonal aggression includes not only direct and physical dynamics but also mental, subtle, and psychological dynamics. Anger had partial mediation effect, and it was again completely mediated by conditioning factors, peer delinquency, masculine social identity, and age. These conditioning factors also partially mediated significant strains. From this, a gendered pattern was apparent.

Female interpersonal aggression was affected by not only physical but also psychological strains unlike males. Female interpersonal aggression was the outcome of important interplay of strains, anger, and their peer association, gendered social role perception, and maturity. In case of male interpersonal aggression, strains and gender were directly related to the violent outcome without the intermediation of such conditioning factors.

Based on the current findings, it can be argued that feminist criminologists’ argument of gender equalization in crime and deviance is conditionally valid depending on the types of crime and deviance. This result also reflects the transitional stage of gender equalization of contemporary South Korean society

For Korean youth, gendered pathways appear to develop through the interplay of varying types of strains, negative emotionality, conditioning factors, and across different contexts of deviance. Female criminality was represented more so by psychological/mental, subtle, relational, and cultural strains and conditioning factors, while male criminality was the outcome of more direct and physical strains and conditioning factors.

Regarding negative emotionality, several strains were positively and significantly associated with anger and depression, which are consistent with the findings of the previous studies and Broidy and Agnew’s theory (1997). However, not all strains leading to negative emotions materialize into deviance. Appearance stress was responded to by anger for females and by depression for males. But the causal process did not necessarily lead to a deviant response, at least for alcohol use or interpersonal aggression. More importantly, gender-specific effects and also types of effect (mediation) of negative emotions were dependent on the type of strains and deviant behavior. These findings are consistent with previous studies (Baron, 2007; Hay, 2003; Jennings et al., 2009; Kaufman, 2009; Lee, 2003; Morash & Moon, 2007).

Taken together, the current study supports Broidy and Agnew’s (1997) hypothesis of gender and GST by showing some generalizability of the theory for the South Korean context. However, the findings of this study also suggested some culture-specific aspects, which means that gendered GST pathways to delinquency have some general applicability, but specific gendered pathways from strain through negative emotionality and conditioning factors toward deviance depend not only on specific combination of types of strains, negative emotions, conditioning factors, and deviance but also on specific types of cultural settings. In contemporary Korean society, various strains are not related to delinquency in general, because overall delinquency levels among general youth population are relatively low compared to youths in Western societies (particularly the United States). Most youths in Korea continue to be relatively controlled by school and family, compared to the United States, which may explain why many strains were found to lead toward negative emotions but did not manifest into delinquency.

At present, Korean society is in transition from patriarchal society into more gender-equalized society. Thus, there is asymmetry of gender equalization transition. In some aspects, gender equalization is more complete, and similar standard applies to both males and females. One such example is monetary success and school achievement. Scholarly achievement, quality job and social success, and monetary success are equally and highly emphasized equally for both males and females. Thus, males and females are equally subjected to severe strains of monetary matters and school performance. As a result, strains of this kind have no gendered difference in the findings. Certainly, these strains became criminogenic factors leading toward deviance relatively equally, affecting both genders. By contrast, there still remain patriarchal aspects in today’s Korea. Females are more controlled by family, school, and society compared to males. They are also subjected to more stress regarding physical appearance standards.

Also, the Korean cultural context of asymmetry in gender equalization transition also plays an important role in the difference between less gendered difference in alcohol use and more gendered difference in interpersonal aggression. Culturally, alcohol use among youths, especially for older youths, is not considered a serious delinquency but rather more socially tolerable as a growing-up process of socialization. Due to this rather generous attitude to alcohol use, very few strain variables had direct and conditioning effect on alcohol use. However, regarding interpersonal aggression, a gendered pathway was more conspicuous in the contemporary Korean society, relatively speaking. Thus, this kind of behavior is more common among young males in South Korea. No conditioning factors may be needed to commit such masculine behavior. Direct effects of victimization, suspension from school, and anger on interpersonal aggression appeared as illegitimate but mainly coping strategy. Victimization may be related to fighting back and suspension from school and anger may be related to letting out stress. By contrast, female involvement in interpersonal aggression is more subtle and complex, because female aggression is still less unacceptable and more sanctioned by family, school, and society. Thus, interpersonal aggression by South Korean female youth may require more interplay between strain variables and conditioning factors. The findings here point out that female interpersonal aggression is more psychological/mental coping strategy in association with peer socialization. This gendered pathway toward interpersonal aggression seems to illustrate double standard of aggression or violence applied to different genders in the current Korean society.

In closing, these findings must also be tempered by several limitations. Measures representing negative emotions such as fear, anxiety, and guilt were not available. Furthermore, although this study attempts to utilize various types of strains and stressors including gender-specific strain measures (e.g., appearance stress), other more gender-specific strain measures not available to the current study may need to be identified and accounted for. Also, this study examined only two types of deviance, alcohol use and interpersonal aggression. Perhaps other strains can be related to other types of deviance. In the criminogenic causal process, gendered pathways may or may not occur. The general applicability of the current findings to other societies or other time period in the Korean society also remains as a subject of the future study as does a more complete assessment of GST on South Korean youth (i.e., interactions between conditioning factors, strains, and negative emotions).

Footnotes

Appendix A

Appendix B

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.

Notes

References

Adler

(1975). Sisters in crime: The rise of the new female criminal. New York, NY: McGraw-Hill.

Agnew

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

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 and Delinquency, 38, 319–361.

Agnew

(2006). Pressured into crime: An overview of general strain theory. Los Angeles, CA: Roxbury.

Agnew

Brezina

(1997). Relational problems with peers, gender and delinquency. Youth and Society, 29, 84–111.

Baron

S. W.

(2007). Street youth, gender, financial strain, and crime: Exploring Broidy and Agnew’s extension to general strain theory. Deviant Behavior, 28, 273–302.

Bartol

C. R.

Bartol

A. M.

(2005). Criminal behavior: A psychosocial approach. 7th ed. Upper Saddle River, NJ: Pearson Prentice Hall.

Broidy

(2001). A test of general strain theory. Criminology, 39, 9–36.

Broidy

Agnew

(1997). Gender and crime: A general strain theory perspective. Journal of Research in Crime and Delinquency, 34, 275–306.

10.

Canter

(1982). Sex differences in self-reported delinquency. Criminology, 20, 372–393.

11.

Chesney-Lind

Shelden

R. G.

(1992) Girls, delinquency, and juvenile justice. Pacific Grove, CA: Brooks/Cole.

12.

Cullen

F. T.

Agnew

(2006). Criminological theory: Past to present. 3rd ed. Los Angeles, CA: Roxbury Publishing Company.

13.

Datesman

S. K.

Scarpitti

F. R.

(1998). Women, crime, and justice. New York, NY: Oxford University Press.

14.

Eitle

D. J.

(2001). Exploring a source of deviance-producing strain for females: Perceived discrimination and general strain theory. Journal of Criminal Justice, 30, 429–442.

15.

Grau

Salanova

Peiro

J. M.

(2001) Moderator effects of self-efficacy on occupational stress. Psychology in Spain, 5, 63–74.

16.

Hay

(2003). Family strain, gender, and delinquency. Sociological Perspectives, 46, 107–135.

17.

Hoffman

J. P.

Cerbone

F. G.

(1999). Stressful life events and delinquency escalation in early adolescence. Criminology, 37, 343–373.

18.

Hoffman

J. P.

S. S.

(1997). The conditional effects of stress on delinquency and drug use: A strain theory assessment of sex differences. Journal of Research in Crime and Delinquency, 34, 46–78.

19.

Jang

S. J.

(2007). Gender differences in strain, negative emotions, and coping behaviors: A general strain theory approach. Justice Quarterly, 24, 523–553.

20.

Jennings

W. G.

Piquero

N. L.

Gover

A. R.

Perez

D. M.

(2009). Gender and general strain theory: A replication and exploration of Broidy and Agnew’s gender/strain hypothesis among a sample of southwestern Mexican American adolescents. Journal of Criminal Justice, 37, 404–417.

21.

Johnson

M. C.

Morris

R. G.

(2008). The moderating effects of religiosity on the relationship between stressful life events and delinquent behavior. Journal of Criminal Justice, 36, 486–493.

22.

Kaufman

J. M.

(2009). Gendered responses to serious strain: The argument for a general strain theory of deviance. Justice Quarterly, 26, 410–444.

23.

Kim

(2002). An analysis of gender difference in delinquency among the Korean youths. Daehan Ganho Hakhoe Jee, 32, 492–505.

24.

Kim

(2005). The effect of stress and attachment on offending and victimization experience among adolescents: A focus on dual experience group and general experience group. Studies on Korean Youth, 16, 5–36.

25.

Kim

(2012). Effects of parent and peer attachment and body image on violent misbehaviors of youths. Sonyeonboho Yeongoo, 20, 37–58.

26.

Lee

(2012). The relationship between appearance-related stress and internalizing problems in South Korean adolescent girls. Social Behavior and Personality, 40, 903–918.

27.

Lee

(2003). A test of general strain theory in explaining violent delinquency: A situational approach. Hyungsajungchaek, 15, 85–105.

28.

Lee

K. S.

Baek

H. J.

(2007). Korea Youth Panel Survey (KYPS) III: A Summary report (Korean). Seoul, Korea: Korea Institute for Youth Development.

29.

Lee

1999. Analysis on psychological and environmental variables correlated with early adolescents’ violent behaviors. Studies on Korean Youth, 10, 119–141.

30.

Manasse

M. E.

Ganem

N. M.

(2009). Victimization as a cause of delinquency: The role of depression and gender. Journal of Criminal Justice, 37, 371–378.

31.

Mazerolle

(1998). Gender, general strain, and delinquency: An empirical examination. Justice Quarterly, 15, 65–91.

32.

Mazerolle

Piquero

A. R.

Capowich

G. E.

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

33.

McGivern

M. S.

(2010). The impact of cognitive coping on the strain-delinquency relationship: A test of general strain theory. (Unpublished doctoral dissertation). Graduate College The University of Iowa, Iowa City, Iowa.

34.

Messerschmidt

J. W.

1993. Masculinities and crime: Critique and reconceptualization of theory. Lanharm, MD: Rowman and Littlefield.

35.

Mirowsky

Ross

C. E.

(1995). Sex differences in distress: Real or artifact? American Sociological Review, 60, 449–468.

36.

Moon

Morash

(2005). 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.

37.

Moon

Morash

McCluskey

J. D.

(2010). General strain theory and school bullying: An empirical test in South Korea. Crime & Delinquency, 58, 827–855.

38.

Morash

Moon

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

39.

Nam

(2012). Gender and juvenile delinquency. Social Science Research, 29, 1–21.

40.

Paternoster

Brame

Mazerolle

Piquero

(1998). Using the correct statistical test for the equality of regression coefficients. Criminology, 36, 859–866.

41.

Piquero

N. L.

Sealock

M. D.

(2004). Gender and general strain theory: A preliminary test of Broidy and Agnew's gender/GST hypotheses. Justice Quarterly, 21, 125–158.

42.

Sharp

S. F.

Brewster

D. R.

Love

S. R.

(2005). Disentangling strain, personal attributes, affective response and deviance: A gendered analysis. Deviant Behavior, 26, 133–157.

43.

Sharp

S. F.

Terling-Watt

T. L.

Atkins

L. A.

Giliam

J. T.

Sanders

(2001) Purging behaviors in a sample of college females: A research note on general strain theory and female deviance. Deviant Behavior, 22, 171–188.

44.

Swatt

Gibson

Piquero

(2007). Exploring the utility of general strain theory in explaining problematic alcohol consumption by police officers. Journal of Criminal Justice, 35, 596–611.