Gender and General Strain Theory

Abstract

The present study of 659 Korean adolescents tests General Strain Theory’s (GST) utility in explaining gender differences in delinquency causation. It models the effects of key strains, negative emotions, and a composite measure of several conditioning factors separately for boys and girls and for delinquency. Consistent with the theory, males and females experience different strains and different emotions in response, and they vary in influences hypothesized to alter the connections of strains or emotions to delinquency. Strains that males experience more than females are significantly related to their violent and property delinquency, and those concentrated among females explain their status offending. For boys, family conflict influences different types of delinquency and examination-related strain predicts violent and status offending. The empirical research suggested that GST falls short in explaining boys’ and girls’ property and status offending, and in showing how a composite measure of conditioning factors act as a moderator in explaining their delinquency.

Keywords

General Strain Theory gender Korean youth delinquency

Introduction

The General Strain Theory (GST) has received considerable research attention for nearly two decades and has developed into a complex explanation of youths’ illegal behavior (Agnew & Brezina, 1997; Hoffmann & Su, 1997; Jennings, Piquero, Gover, & Perez, 2009; Kaufman, 2009; Mazerolle & Maahs, 2000; Moon & Morash, 2004; Moon, Morash, McCluskey, & Hwang, 2009; Piquero & Sealock, 2004). According to the theory (Agnew 1992, 2006), various negative social relations/situations (e.g., failure to achieve positively valued goals, removal of positively valued stimuli, and experience of negative stimuli) directly influence delinquency or indirectly influence it by increasing negative emotions. To adapt to strain, individuals use coping strategies that are cognitive (i.e., minimizing the importance of strain and negative outcomes, or accepting responsibility for adversity), emotional (i.e., alleviating negative emotions through physical exercise or meditation), and behavioral (i.e., eliminating the source of strain or engaging in vengeful behavior). Individuals who lack non-delinquent coping strategies and who have a strong disposition to engage in deviance are more likely than others to alleviate strain and negative emotions through delinquent behaviors (Agnew, 1992).

In a seminal piece that sets forth but does not present a new test of the relevance of GST to how gender affects delinquency, Broidy and Agnew (1997) proposed that the theory might explain gender differences in both the etiology and the amount of delinquency. Drawing on research conducted in the traditions of GST, feminist criminology/sociology, and the social-psychological literature on gender and stress, they proposed gender differences in the types of strain experienced, emotional reactions to strain, conditioning factors that moderate the effects of strain, and level and type of delinquency.

Suggesting the need for further study, the many empirical studies to test hypotheses derived from GST’s key propositions regarding gender have yielded inconsistent findings (Agnew & Brezina, 1997; Hoffmann & Su, 1997; Jang & Johnson, 2005; Kaufman, 2009; Mazerolle, 1998). Supporting the theory, some studies (Kaufman, 2009; Mazerolle, 1998; Piquero & Sealock, 2004) found significant gender differences in strains and negative emotions and in their effects on delinquency. However, other studies (Agnew & Brezina, 1997; Hoffmann & Su, 1997) revealed limited or no significant gender differences in strains, negative emotions, and conditioning factors.

Research to further understanding of GST’s contribution to understanding gender differences in delinquency requires consideration of several general issues relevant to testing the theory. First, empirical tests must include measures of key strains hypothesized to be related to the deviance of both boys and girls (Agnew, 2006). Second, a majority of prior studies (except Lin, Cochran, & Mieczkowski, 2011; Mazerolle & Maahs, 2000) assessed the interaction effect of each of several conditioning factors with strains on the strain-delinquency relationship. However, Agnew (2006) indicates that this approach has a methodological limitation, because it may not matter that a particular conditioning factor interacts with strains to increase or decrease delinquency, as long as some conditioning factor has an effect. To overcome this limitation, he suggested that the “overall standing” of these variables should be studied as a moderator of the relationship between strains and delinquency (Agnew, 2006, pp. 109-110). Limited research (Lin et al., 2011; Mazerolle & Maahs, 2000) has examined combined effects of conditioning factors on the strain-delinquency relationship, and we are unaware of any research to examine gender differences in the moderating influence of a composite measure of conditioning factors. Third, most prior studies examined the moderating effects of conditioning factors on the strain-delinquency relationship, but ignored the possible moderating effects of conditioning factors on the negative emotions-delinquency relationship. Fourth, previous empirical findings (Mazerolle, 1998; Morash & Moon, 2007) suggest that gender differences in the effects of strains, negative emotions, and conditioning factors depend on the type of delinquency. Supporting this view, some tests of GST have shown that girls tend to internalize reactions, which leads to status offenses, but boys tend to externalize their reactions by becoming aggressive or committing property offenses (Leadbeater, Blatt, & Quinlan, 1995). Therefore, research to understand gender differences in the effects of strains should separate violent, property, and status offending.

Gender, General Strain Theory, and Empirical Findings

Broidy and Agnew (1997) argue that gender differences in the types of strains may explain gender differences in delinquency. As males highly value material success and extrinsic achievement, they are more likely to experience financial strain resulting from the inability to achieve monetary goals. Also, males are more likely to experience peer conflict and criminal victimization because their peer relationships are based on competition and conflict. However, females experience more interpersonal strain due to their greater concern with maintaining close relationships with significant others. They also are expected to report more strain due to gender discrimination.

Broidy and Agnew (1997) also suggested that gender differences in negative emotional reactions to strains produce gender differences in deviant behaviors. In their view, although boys and girls report similar levels of anger in response to strain, girls more often respond to strain with a mix of negative emotions. Because boys are more likely to report that their anger results from moral outrage at being challenged or treated unjustly, they tend to externalize their reactions by becoming aggressive or committing property offenses (Leadbeater et al., 1995). In contrast, angry, frustrated, and depressed girls tend to internalize reactions to strains that may lead to substance abuse, truancy, or running away from home.

Finally, Broidy and Agnew (1997) wrote that girls less than boys respond to strain with delinquent behavior because of moderating influences that affect the connection of strains to delinquency. Parents and friends often expect girls to conform to gender-related expectations that limit their opportunities to break the law, provide them with more supervision and control, and restrict association with delinquent peers, all of which limit serious and aggressive delinquent behaviors. On the contrary, conducive to serious and aggressive delinquent behaviors, boys are more likely to have lower social/parental control and to associate with delinquent peers.

Over the last decade, a number of empirical studies have tested Broidy and Agnew’s extension of GST to explaining gender differences in the etiology of delinquency, and some but not all support the theory’s key propositions. Studies (Agnew & Brezina, 1997; Jennings et al., 2009; Kaufman, 2009; Mazerolle, 1998) showed that males more often report negative relations with peers, conflict with significant others, and physical punishment and criminal victimization, but females report more problems forming and maintaining positive relationships with family and friends and strain from expectations that they do housework, which suggests perceived gender discrimination. Contradicting these results, Morash and Moon (2007) showed no gender differences for adolescents in financial strain, academic performance related strain, negative life events, and parental physical/emotional punishment.

Researchers (Mazerolle, 1998; Morash & Moon, 2007; Piquero & Sealock, 2004) have also documented significant gender differences in the effects of various strains on delinquent behaviors. Family strains, negative life events, or a composite measure of strain had stronger effects on violent delinquency for males than females. Morash and Moon (2007) found the somewhat different pattern, that negative life events, parental punishment, and financial strains, had significant effects on female violent and status delinquency but not on male delinquent behaviors. Contrary to these findings, Hoffmann and Su (1997), Kaufman (2009), and Jennings et al. (2009) discovered no significant gender differences in the effects of some strains on delinquent behaviors.

Research on gender differences in the interaction effects between conditioning factors and strains on delinquency is especially limited and inconsistent in its findings (see Jang, 2007; Jang & Johnson, 2005; Jennings et al., 2009; Morash & Moon, 2007). Specific to gender differences, Jang (2007) and Jang and Johnson (2005) found that due to their being more religious, strained females were less likely than their male counterparts to commit interpersonal violence. Jennings et al. (2009) similarly showed that females with higher levels of physical/spiritual coping resources and/or peer social support were less likely to engage in interpersonal aggression or property delinquent behaviors. However, Morash and Moon (2007) found that the interaction between teachers’ punishment and delinquent peer association significantly, positively affected delinquency regardless of gender.

General Strain Theory in the Context of South Korea

The Korean context provides opportunities to investigate culture-specific childhood strains and the generalizability of GST outside the United States. Strains that are uncommon in the United States but which are pervasive in South Korea and other parts of the world include stressful, highly competitive examinations that strongly affect future success and physical and verbal punishment by teachers and parents. Also unique to Korea and some other countries, after middle school begins, youth spend significant amount of their time after school in tutoring, and thus away from parental supervision.

Prior empirical studies (see Moon et al., 2009; Morash & Moon, 2007) suggest that core propositions of GST apply to youth in Korea. For example, Moon et al. (2009) found that various strains (i.e., family conflict, parental/teacher punishment, criminal victimization) are significantly and positively related to delinquent behaviors, and that situational anger had positive effects on various delinquent behaviors. Their additional finding that strained Korean youths less often engaged in delinquency when they had positive relationships with parents or had higher problem-solving ability further supported GST.

Only two previous studies (Morash & Moon, 2007; Yun, Kim, & Morris, 2014) examined GST’s utility for explaining gender differences in delinquency causation in Korea. Morash and Moon (2007) found that boys more often than girls experienced examination-related strain and physical and emotional punishment by teachers. However, they found no gender differences in levels of financial strain, and teachers’ punishment significantly predicted delinquency regardless of gender. Yun et al. (2014) indicate that males and females experience different types of strains, in that male students tend to experience parental physical punishment and peer stress, but female students are more likely to report victimization, health problems, and appearance stress, which have significant effects on deviance through their influence on anger and conditioning factors. These prior studies have limitations. Moon et al. (2009) used gender as a control variable and did not examine the moderation effect of a sum of the conditioning factors on the connection of strain to delinquency or on the connection of negative emotions to delinquency. Morash and Moon (2007) tested only part of the model suggested by GST, failing to examine the mediating/moderating effects of negative emotions and conditioning factors on delinquency. Moreover, Yun et al. (2014) measured limited types of deviant behaviors and only trait-based negative emotions. In a comparison of tests of the GST-derived explanatory models for girls and boys, the present research attempts to address these limitations.

Hypotheses

Based on theory and prior research, we predict that boys are more likely to experience emotional/physical punishment by parents and teachers, criminal victimization, examination-related strain, and financial strain (so-called male strains), but females are more likely to report higher levels of family conflict and gender discrimination (so-called female strains). Second, we expect that male strains are more likely to lead to violent and property-related deviant behaviors, and female strains are more likely to result in status delinquency. Finally, because males are higher in delinquency-promoting conditioning factors, we predict that there will be gender differences in the moderating effects of a combined measure of multiple conditioning factors on the connection of both strains and negative emotions to delinquency.

Method

Sample

Data were collected from middle school students in South Korea in two waves of a longitudinal study. The first and second waves of the longitudinal data were collected from middle school students in South Korea during the summers of 2005 and 2006. To adequately represent middle school students in South Korea, students who attended middle schools were randomly selected in two metropolitan cities (Incheon and Daegu) and one medium-sized city (Cheongju). Incheon and Daegu are metropolitan cities with a population of approximately 2.6 million and 2.5 million, respectively. Cheongji is a medium-sized city with a population of approximately 600,000. A comparison of several social economic status indicators (i.e., poverty, divorce) for these three cities and for other cities/provinces revealed that the three cities are not noticeably different from other population centers in South Korea.

In the summer of 2005, a list of schools in these cities was assembled: There were 90 middle schools in Incheon, 108 middle schools in Daegu, and 30 middle schools in Cheongju. From the list of middle schools in each of the research sites, one school was randomly selected. With the approval of school administrators and teachers, all eighth-grade students in the three selected schools and their parents were informed about the purpose of the research, and youth whose parents provided a signed consent form were asked to voluntarily complete questionnaires. To maximize confidentiality and accuracy of the responses, researchers distributed questionnaires, allowed time for their completion, and collected them after 1 hr. Attendance records for each class indicated low absentee rates at the time of the data collection. Of 900 questionnaires distributed, 817 were collected; however, 30 collected questionnaires with extensive missing data were discarded. Overall, the response rate was 87% (787 out of 900). Approximately 1 year later (summer 2006), students who participated in the first wave of data collection were asked to voluntarily complete the second wave questionnaires in their classrooms. Six hundred sixty-four ninth graders completed the second wave questionnaire, yielding a response rate of 84% (664 out of 787). The attrition rate from the first wave to the second wave data collection was 16%. We conducted a series of t-tests on key Wave I variables considered in the present study to compare students who did and did not continue to participate and found no statistically significant differences for variables used in the analysis. For the current study, 659 valid cases were used for analyses after listwise deletion of missing values.

Measurement

Both independent and dependent variables are presented in Table 1. Most items used to operationalize variables were taken from prior research. To clarify time order, independent variables were measured at Wave I, and dependent variables were measured at Wave II. The scales were coded so that a higher score indicates a higher level of each strain, negative emotion, and delinquency. Recoding was used for some measures of conditioning factors so that for each, a high score indicated the presence of a variable thought to increase delinquency.

Table 1.

Comparisons of Key Variables by Gender.

	Males (n = 255)		Females (n = 404)
Variables	M	SD	M	SD	t value
Family C.	2.32	1.97	2.78	2.03	−2.84*
Parental P	2.12	2.87	1.87	2.39	1.16
Teacher P	2.36	3.49	1.24	2.01	4.68*
Exam S	5.94	2.38	6.05	2.08	−0.58
Financial S	2.63	2.20	2.75	2.19	−0.69
Gender D	1.12	2.25	0.82	1.58	1.82
Victim	0.51	1.84	0.20	0.72	2.61*
Sanger1	1.89	2.26	2.05	2.16	−0.92
Depres1	2.01	2.42	3.08	2.61	−5.24*
Sanger2	1.18	2.03	1.43	2.11	−1.51
Depres2	1.42	2.28	2.21	2.64	−4.05*
Sanger3	1.41	2.49	1.07	2.04	1.83
Depres3	1.07	2.18	1.14	2.17	−0.40
Sanger4	1.92	2.62	1.43	2.15	2.47*
Depres4	2.04	2.84	2.75	2.75	−3.19*
Sanger5	0.84	1.75	0.61	1.42	1.80
Depres5	1.02	1.83	1.38	2.13	−2.29*
Sanger6	0.81	1.91	1.05	2.06	−1.54
Depres6	0.69	1.78	1.24	2.33	−3.39*
Sanger7	0.51	1.74	0.30	1.36	1.60
Depres7	0.35	1.45	0.21	1.02	1.31
N. Parents R	9.85	3.09	8.98	2.92	3.67*
L. Parental C	14.64	3.65	13.35	3.26	4.61*
Legitimacy	9.37	4.09	8.20	3.30	3.85*
Low S.C	53.98	12.31	54.66	9.08	−0.76
T. Conditioning	9.03	1.65	8.39	1.47	5.07*
Violent delinquency at Wave I	0.76	2.47	0.14	0.61	3.91*
Property delinquency at Wave I	0.25	0.85	0.10	0.35	2.69*
Status delinquency at Wave I	0.57	1.90	0.35	0.76	1.81
Violent delinquency at Wave II	0.69	1.62	0.18	1.18	4.37*
Property delinquency at Wave II	1.05	2.63	0.20	1.57	4.67*
Status delinquency at Wave II	0.87	1.97	0.41	1.18	3.33*

Note. Due to the space limitation, we abbreviated key variables’ names in the Tables throughout the article. Family conflict as Family C; Parental punishment as Parental P; Teacher punishment as Teacher P; Examination-related strain as Exam S; Financial strain as Financial S; Gender discrimination as Gender D; Victimization as Victim; Anger to family conflict as Sanger1; Depression to family conflict as Depres1;Anger to parental punishment as Sanger2; Depression to parental punishment as Depres2; Anger to teacher punishment as Sanger3; Depression to teacher punishment as Depres3; Anger to examination-related strain as Sanger4; Depression to examination-related strain as Depres4; Anger to financial strain as Sanger5; Depression to financial strain as Depres5; Anger to gender discrimination as Sanger6; Depression to gender discrimination as Depres61; Anger to victimization as Sanger7; Depression to victimization as Depres7; Negative relationship with parents as N. Parents R; Low parental control as L. Parental C; Legitimacy of violence as Legitimacy; Low self-control as Low S.C; Composite condition factor as T. Conditioning.

p ≤ .05.

Independent variables

Seven strains

The family conflict scale (Cronbach’s α = .75) was created by combining three items, measuring the frequency of verbal arguments, tensions, and conflicts of a study participant with one or more parents during the last 6 months. The response options ranged from strongly disagree (1) to strongly agree (4). The parent emotional and physical punishment scale consisted of four items adopted from Piquero and Sealock (2000). The scale (Cronbach’s α = .72) measures how often a respondent was emotionally and physically punished by parents through acts such as name calling, negative comparison with others, and hitting. The teachers’ emotional and physical punishment scale (Cronbach’s α = .77) was created by using a parallel set of the items used for parental punishment. The response options for the punishment scales ranged from 0 (none) to 4 (10 or more times). The examination-related strain scale (Cronbach’s α = .64) consists of the three items: “I feel a lot of stress about studying,” “I am not satisfied with my grades,” and “My parents stress studying too much.” Strongly disagree (1) to strongly agree (4) ratings were used with these items and for the financial strain items. The financial strain scale (Cronbach’s α = .62) combines responses to the three items: “I am not satisfied with the amount of money I have,” “My family has too little money for clothing or food,” and “My family does not have enough money to support me.” Four items adapted from Landrine and Klonoff (1996) comprise the gender discrimination scale (Cronbach’s α = .76). These four items, originally developed to measure racial discrimination, were slightly rephrased to measure respondents’ experiences of discrimination because of their gender, specifically that they were treated with less courtesy and respect. Finally, the victimization scale (Cronbach’s α = .66) consists of five items, measuring respondents’ and family members’ experience of being victims of theft, robbery, burglary, sexual assault, and physical assault. We recognize the possibility that a person’s own victimization experience is distinct from family members’ victimization (a vicarious strain). However, we assume that the measurement does not greatly reduce any effect of victimization on negative emotions and deviant behaviors, as any family member’s criminal victimization experience could have a negative effect on other family members, causing high levels of stress and negative emotions (see Agnew, 2002 for more information on the effects of vicarious strains on delinquency), especially in a collective society such as South Korea, which is characterized by strong family ties and interdependency among family members.

Situation-based negative emotions: Anger and depression¹

For each of the seven strains for which respondents indicated any presence, study participants were asked to report whether they responded with uncontrollable outbursts of temper, urges to beat or harm someone, and urges to break things. These three indicators of anger were combined to reflect situational anger for each strain. The measure of situation-based depression consists of three items reflecting respondents’ experience of sadness, worthlessness, and depression in response to any of the seven strains that was reported. Responses to the three items were combined to create measures of situational depression for each type of strain reported. Overall, seven situation-based anger and seven situation-based depression indicators were created; each of them was associated with one of the seven strains. The response options, which ranged from 0 for never to 3 for always, reflected how often respondents experienced the reaction in response to a specific type of strain.

Four conditioning factors

As suggested by Agnew (2006), we included four conditioning variables hypothesized to increase the likelihood of strained and emotionally distraught adolescents engaging in delinquent behaviors. Four items—measuring respondents’ perception of parents’ understanding, interest, and openness toward respondents—are combined to create a negative relationship with parents scale (Cronbach’s α = .78). The low parental supervision scale (Cronbach’s α = .76) consists of five items, measuring respondents’ perception of their parents’ knowledge of their whereabouts and deviant behaviors. The scale is coded so that a higher value indicates a low level of parental control. The legitimacy of violence scale (Cronbach’s α = .89) included five items, measuring whether violence can be justifiable in various situations, for example, to defend one’s rights, achieve respect, and avoid appearing weak. The low self-control scale (Cronbach’s α = .81) was created by summing 24 items derived from Grasmick, Tittle, Bursik, and Arneklev (1993), measuring six dimensions of low self-control. For all conditioning variables, four-point Likert-type (from strongly disagree to strongly agree) ratings were used as response options.

In constructing a composite conditioning index to reflect tendencies toward delinquent coping, because Agnew (2006) did not hypothesize different effect sizes of the conditioning factors affecting deviant behaviors, we assumed equal influence of the four conditioning factors. Thus, first the summed score for each conditioning scale was divided by the number of items (e.g., the self-control measure was divided by 24 and the low parental control measure by 5). Second, the resulting values for the four conditioning influence scales were summed to create the composite measure.

Dependent variables—three types of delinquency

At Wave II, participating students were asked how often they engaged in various types of delinquent behaviors over the last year. The violent delinquency scale (Cronbach’s α = .71) consists of six items, measuring the frequency of involvement in behaviors such as “used force or strong-arm methods to get money or things from others” and “hit or threatened to hit fellow students.” The property-related delinquency scale (Cronbach’s α = .86) was created by summing 11 items such as “broken or tried to break into a building or vehicle to steal something” and “sold or tried to sell a stolen good to others.” The status delinquency scale (Cronbach’s α = .68) was created by summing responses to four items: “drinking alcohol,” “smoking,” “running away,” and “skipping classes without a reason.” The response options for each item were 0 (never), 1 (1 to 3), 2 (4 to 6), 3 (7 to 9), and 4 (10 or more times).

Analysis

The analysis first compares the males and females on all independent and dependent variables included in the multivariate models to predict delinquency. Then, to see the effect of strains and the direct and mediating effects of composite emotions, separately for each gender group, we (a) regress each type of delinquency on strains, and (b) then regress each type of delinquency on the strains plus composite indices of anger and depression. Finally, we examine gender differences in the interaction effects of the composite measure of the four conditioning factors on the relationship of strain to delinquency as well as the relationship of negative emotions to delinquency. To avoid multicollinearity problems with multiple strains and negative emotions, we used composite measures of total strains, total situational-based anger, and total situational-based depression. These composite variables were created by adding measures of seven strains, anger, and depression, respectively. Furthermore, before we constructed interaction terms, we grand mean centered the composite measures of strain, anger, depression, and the conditioning index to reduce the potential for multicollinearity (Aiken & West, 1991). In each of the six models tested (2 gender groups × 3 types of delinquency), three interaction terms (Total strain × Total conditioning, Total anger × Total conditioning, and Total depression × Total conditioning) were included as independent variables.

Ai and Norton (2003) described the difficulty of interpreting the results of interaction effects in nonlinear models. Thus, we estimated the interaction effects of the composite conditioning index using a series of ordinary least squares regressions. For each gender group and type of delinquency, three models were estimated: (a) a base model plus the interaction term between composite strain and the composite conditioning factor, (b) a base model plus the interaction term between composite anger and the composite conditioning factor, and (c) a base model plus the interaction term between composite anger and the composite conditioning factor.

Results

Table 1 presents the results of t-tests for gender differences in all variables included in the present research. Consistent with GST, boys are more likely to experience emotional/physical punishment by teachers and criminal victimization, whereas girls are more likely to report higher levels of family conflict. Unexpectedly, boys report more gender discrimination than girls, though the difference is not significant. For parental punishment, financial, and examination-related strain, the t-tests show no significant gender differences.

Regarding gender differences in negative emotions related to each strain, girls are more likely than boys to report significantly higher levels of depression in response to all strains except teacher punishment and criminal victimization, for which there were no gender differences. The only gender difference in anger is that boys score significantly higher than girls on anger in response to examination-related strain.

As expected, there are significant gender differences for most conditioning variables and the composite conditioning index. Boys are more likely than girls to report lower levels of parental control and negative relationships with parents and to have positive attitudes toward the use of violence. However, there is no significant gender difference in self-control. With the exception of status offending at Wave I, compared with girls, boys report significantly more of each type of deviant behavior at Waves I and II.

Table 2 presents the results of tests of the full gender-specific and delinquency type-specific models that include the effects of all seven strains, composite indices of anger and depression, and conditioning factors on delinquent behaviors. Negative binomial regression was used because a large number of youths indicated no involvement in delinquency, and thus ordinary least squares model estimates can be misleading (Long, 1997). To further investigate gender differences in the effects of each strain and negative emotion on delinquency, z-statistic comparisons associated with strains and negative emotions were computed (see Paternoster, Brame, Mazerolle, & Piquero, 1998). The findings indicate that family conflict, parental punishment, and examination strain are significant predictors of violent, property, or status delinquent behaviors for males, while gender discrimination has a significant positive effect on violent delinquency for females, after controlling the effects of negative emotions and conditioning factors. Unexpectedly, financial strain is found to have a negative effect on status delinquency for females (see S1 for female status delinquency). Contrary to GST propositions, in the S2 models, for both gender groups the composite measures of anger and depression have no significant relationship to delinquent behaviors and no mediating effect on the link of strains to delinquent behaviors. The results of z-statistic comparisons indicate that gender differences in the effects of most strains and composite measurements of anger and depression are non-significant except that examination-related strain has a stronger effect on boys’ violent delinquency, but gender discrimination has a statistically stronger effect on girls’ violent delinquency.

Table 2.

Negative Binomial Regression Models Predicting Delinquency—Full Models.

	Male violence		Female violence		Male property		Female property		Male status		Female status
	S1	S2	S1	S2	S1	S2	S1	S2	S1	S2	S1	S2
Family C	.14*(.07)	.15*(.07)	.08(.14)	.06(.15)	.02(.10)	.02(.10)	.02(.14)	.02(.14)	.23*(.09)	.24*(.10)	.11(.07)	.12^†(.07)
Parent P	.04(.04)	.05(.04)	−.07(.13)	−.09(.14)	.12*(.06)	.12*(.06)	−.11(.13)	−.06(.14)	−.01(.05)	−.01(.05)	.07(.05)	.07(.05)
Teacher P	−.00(.04)	.00(.04)	.20(.13)	.26^†(.14)	−.04(.05)	−.04(.05)	.17(.12)	.17(.12)	.03(.05)	.04(.05)	.01(.06)	−.01(.06)
Exam S	.17*(.06)	.20*(.07)	−.15(.12)	−.19(.13)	.10(.08)	.11(.08)	.10(.12)	.13(.12)	.07(.07)	.08(.07)	−.03(.06)	−.03(.06)
Financial S	.07(.05)	.08(.04)	−.05(.11)	−.05(.12)	.08(.07)	.08(.07)	.03(.11)	.03(.11)	.03(.07)	.03(.07)	−.13*(.06)	−.12^†(.06)
Gender D	.03(.05)	.05(.05)	.33*(.13)	.39*(.15)	−.04(.08)	−.02(.08)	−.15(.18)	−.11(.18)	−.06(.07)	−.06(.07)	.08(.07)	.07(.07)
Victim	.09(.07)	.08(.07)	.44(.29)	.41(.29)	.19^†(.11)	.17(.11)	.14(.28)	.10(.29)	.06(.09)	.05(.09)	.06(.14)	.02(.14)
T. Anger		.01(.02)		−.05(.04)		.02(.02)		−.02(.04)		.01(.02)		.03(.02)
T. Depres		−.03^†(.02)		.02(.03)		−.03^†(.02)		−.02(.03)		−.02(.02)		−.01(.01)
N. Parents R	−.00(.05)	−.02(.05)	−.01(.10)	.03(.11)	−.02(.06)	−.03(.06)	.07(.11)	.08(.11)	−.08(.06)	−.09(.06)	−.01(.06)	−.03(.06)
L. Parental C	.08*(.04)	.09*(.04)	−.09(.09)	−.11(.10)	.10*(.05)	.11*(.05)	−.04(.08)	−.05(.08)	.16*(.05)	.16*(.05)	.08^†(.05)	.08^†(.05)
Legitimacy	.03(.04)	.03(.04)	.14*(.07)	.17*(.07)	−.02(.05)	−.01(.05)	.09(.07)	.10(.07)	−.09*(.04)	−.08^†(.05)	−.12(.04)	−.04(.04)
Low S.C	.01(.01)	.02(.01)	.05^†(.03)	.05^†(.03)	.02(.02)	.02(.02)	.04(.03)	.04(.03)	.01(.01)	.01(.01)	.05*(.02)	.04*(.02)
T1 Delinqu	.05(.06)	.06(.06)	.61^†(.32)	.75*(.34)	−.02(.24)	.03(.24)	1.36*(.43)	1.41*(.47)	.27*(.11)	.26*(.11)	.94*(.14)	.93*(.14)
Pseudo R²	.11	.11	.13	.13	.05	.05	.11	.12	.07	.07	.13	.14
Model χ²	58.03	61.87	39.42	40.93	28.56	31.08	37.29	38.13	38.76	39.37	85.79	87.84

Note. Robust standard errors in parentheses. Time 1 delinquency is consistent with the dependent variable in each model.

†

p ≤ .10. *p ≤ .05.

Table 3 presents the tests for moderating effects of the composite conditioning index on the strain-delinquency and on the negative emotions-delinquency relationships. All three interaction terms are significantly related to violent delinquency for males. Strained, angry, or depressed male adolescents are more likely to engage in violent delinquency when they have a higher risk for delinquent coping as indicated by the composite conditioning index. However, none of the three interaction terms involving strain, negative emotions, and a total conditioning index has a significant effect on violent delinquency for females. For property delinquency, no interaction terms are found to be a significant predictor for either males or females. For status offenses (see S2 for males), there is a tendency (p ≤ .10) for angry males to commit status offenses when they report a higher tendency toward criminal coping. Like males, angry females report more status offending when conditioning factors reflect tendencies toward illegal behavior.

Table 3.

Ordinary Least Squares Regression Models Predicting Delinquency Outcomes With Interactions and Comparison by Gender.

	Violent delinquency						Property delinquency						Status delinquency
	Male			Female			Male			Female			Male			Female
	S1	S2	S3	S4	S5	S6	S1	S2	S3	S4	S5	S6	S1	S2	S3	S4	S5	S6
Delinquency I	0.03(.04)	0.06(.04)	0.06(.04)	0.47*(.10)	0.47*(.10)	0.47*(.10)	0.25(.22)	0.22(.22)	0.23(.22)	1.44*(.22)	1.43*(.22)	1.40*(.22)	0.21*(.07)	0.22*(.07)	0.22*(.07)	0.57*(.07)	0.59*(.07)	0.57*(.07)
T. Strain	0.35*(.13)	0.05*(.01)	0.05*(.01)	−0.04(.07)	−0.01(.01)	−0.01(.01)	0.51*(.23)	0.06*(.02)	0.06*(.02)	−0.06(.10)	−0.01(.01)	−0.01(.01)	0.23(.17)	0.03^†(.02)	0.03^†(.02)	0.06(.07)	0.01(.01)	0.01(.01)
T. Anger	0.02(.01)	−0.03(.14)	0.04(.14)	0.02*(.01)	0.19*(.09)	0.03*(.01)	0.04^†(.03)	0.21(.25)	0.27(.24)	0.04*(.01)	0.24*(.12)	0.04*(.01)	0.01(.02)	−0.13(.19)	−0.06(.19)	0.02*(.01)	0.14^†(.08)	0.03*(.01)
T. Depression	−0.02(.01)	−0.02(.01)	−0.02(.01)	−0.02*(.01)	−0.01^†(.01)	−0.13^†(.08)	−0.05^†(.02)	−0.04^†(.02)	−0.05^†(.02)	−0.03*(.01)	−0.03*(.01)	−0.27*(.10)	−0.01(.02)	−0.00(.02)	−0.01(.02)	−0.02*(.01)	−0.02*(.01)	−0.19*(.07)
T. Conditioning	0.28*(.10)	0.26*(.11)	0.35*(.11)	0.03(.06)	0.04(.06)	0.03(.06)	0.15(.19)	0.15(.19)	0.23(.19)	0.04(.08)	0.05(.08)	0.04(.08)	0.16(.14)	0.17(.14)	0.23(.14)	0.06(.06)	0.06(.06)	0.06(.06)
T. Strain × T. Conditioning	0.38*(.09)			−0.07(.06)			0.14(.16)			−0.07(.07)			0.13(.11)			0.08(.05)
T. Anger ×T. Conditioning		0.27*(.09)			0.00(.05)			0.22(.16)			0.06(.07)			0.21^†(.11)			0.12*(.05)
T. Depression × T. Conditioning			0.33*(.11)			−0.08(.05)			0.26(.19)			−0.07(.07)			0.20(.13)			−0.04(.05)
Constant	0.53*(.15)	−0.20(.24)	−0.07(.24)	0.12(.10)	0.37*(.17)	0.07(.16)	0.96*(.27)	0.20(.42)	0.31(.41)	0.15(.13)	0.47*(.22)	−0.02(.21)	0.69*(.20)	0.13(.31)	0.23(.30)	0.20*(.10)	0.17(.16)	−0.13(.16)
R ²	.26	.23	.24	.09	.08	.09	.11	.11	.11	.13	.13	.13	.13	.13	.13	.20	.21	.20

Note. Standard errors in parentheses. Delinquency 1 is consistent with the dependent variable in each model.

†

p ≤ .10. *p ≤ .05.

The z-statistic comparisons show significant gender differences in the effects of interactions with conditioning factors on delinquency. Four interaction terms (For violent delinquency, Strain × Conditioning, Anger × Conditioning, and Depression × Conditioning; and for status delinquency, Depression × Conditioning) had significantly stronger effects on violent and status delinquent behaviors for boys.

Discussion and Conclusion

The present study tested hypotheses deduced from key GST propositions by exploring gender differences in strains, negative emotions, and conditioning factors and by showing whether and how these variables relate to delinquent outcomes. The findings provide partial support for the theory’s hypothesized gender differences in types of strains that males and females experience. As expected, males are more likely to report higher levels of teachers’ physical/emotional punishment and criminal victimization, and females report higher levels of family conflict. However, contrary to the theory’s prediction, no significant gender differences in financial strain, examination-related strain, and gender discrimination are found. As Broidy and Agnew (1997) suggested, all contemporary adolescents may be encouraged to succeed financially and academically in an economically developed and consumerist society, and thus they may experience similar levels of financial and examination strain. The finding of no significant gender difference in gender discrimination is unexpected, especially considering that despite girls’ equal access to education, Korea is still a highly patriarchal society (Patterson, Bae, & Lim, 2013). It is plausible that in a patriarchal society, adolescent males see themselves as being treated with less courtesy and respect than females. This is consistent with our finding of more severe punishment by teachers being directed at male compared with female students. We also speculate that the young age of study participants may be related to this unexpected finding, since as girls mature, they may face more gender discrimination, for example in the workforce. Also, parents’ and teachers’ higher level of monitoring and supervision of girls may lead to their spending most of their time at school or home and thus having limited exposure to various situations in which they experience gender discrimination.

Consistent with GST, some strains (parental punishment, examination-related strain, financial strain, and criminal victimization) are significantly related to violent or property delinquency, but only for males, and two strains (family conflict and gender discrimination) have significant effects on violent or status delinquency only for females. Furthermore, examination-related strain has a stronger effect on violent delinquency for males, while gender discrimination is found to have a stronger effect on status delinquency for females. However, variables predictive of boys’ delinquency were not always the so-called “male” strains. Notably, family conflict, which is assumed to be a “female strain” and was found to be significantly higher for girls, still significantly predicted boys’ violent, property, and status delinquency.

Examination strain appeared to significantly contribute to violence and status delinquency for male adolescents. This is inconsistent with Agnew’s (2006) prediction that strains such as failure to achieve educational success goals and excessive demand associated with conventional pursuits are less likely to be related to deviant behaviors because they are less likely to be perceived as unjust and do not create pressure for delinquent coping. To understand why examination-related strain, which studies in the United States do not consider, is related to Korean boys’ delinquent behaviors, the college entrance examination system and the importance of being admitted into prestigious universities in South Korea needs to be understood. Quite different from the United States, there is extreme emphasis on preparation for college entrance examinations throughout East Asia, but especially in South Korea. Most high school students are required to remain at school until late evening for self-study or to attend private, after-school academic institutions to improve their test scores. The resulting pressures on Korean students to achieve high examinational grades and succeed academically contribute to their high levels of stress and anxiety (Lee & Larson, 2000). A recent survey (Pang Jong-Hwan Research Center, 2010) of 5,437 Korean schoolchildren found that approximately half reported dissatisfaction with their lives. Compared with youth from the 26 member-countries of the Organization of Economic Cooperation and Development, Korean adolescents were the unhappiest. School-generated strain (i.e., examination pressures, negative interaction with teachers) was one of main causes of their dissatisfaction. Examination-related strain, which may be non-criminogenic in the United States, may become a criminogenic strain when a society and culture overemphasizes the goal of academic success.

In the present research, negative emotions generally did not significantly mediate the connection of strains to delinquency. These findings are unexpected, especially considering that situation-based negative emotions in response to each strain were measured. Possibly youth’s experience of these distressing strains have damaging effects and thus result in delinquent behaviors, regardless of experience of negative emotions (see Moon et al., 2009; Piquero & Sealock, 2004). Alternatively, our use of longitudinal data may account for the findings if it fails to capture the effects of negative emotions on delinquency accurately. GST (Agnew, 1992) presents the relationship among strains, negative emotions, and delinquency as contemporaneous in that angry/depressed youth in response to strains engage in deviant coping within a relatively short period of time.

The tests of models that include the interaction effects of strains and emotions with the composite conditioning index provide limited support for GST. Strained, angry, or depressed boys are more likely to engage in violent delinquency when they report a higher risk for delinquent coping. For females, just one interaction term (Anger × Conditioning index) was significantly related to delinquency (i.e., status offending) in the expected positive direction. The interactions between conditioning factors and both strains and negative emotions are not useful in explaining property delinquency for either gender group. For males, the current study provides some support for the moderating effects of a composite conditioning index on the relationships between strain and violent delinquency, and for females it provides minimal support. These findings (also see Lin et al., 2011) raise a question about Agnew’s (2006) argument (p. 110) that “total risk” for criminal coping can better explain strain-delinquency and negative emotions-delinquency relationships. However, we believe that future research would benefit from including more or different conditioning factors (e.g., association with delinquent peers, inadequate conventional coping skills) that may be more likely to condition the effects of strains and negative emotions on delinquency (Agnew, 2006).

Limitations of the research should be considered in interpreting results. First, we measured situational-based anger and depression by combining several items widely used in prior GST research. We recommend the employment of more items measuring these negative emotions, for instance by including the many dimensions contained in the Beck Depression Inventory (Sharp & Lipsky, 2002), to better understand situation-based negative emotions and their effects on deviance. Also, of course it would be useful to have more detailed longitudinal data to sort out the time order of strains, negative emotions, and delinquency. Given the immediacy of emotion in response to a negative event and emotion’s effect on deviant coping behaviors, methods like maintaining a diary or frequent interviews (longitudinal data with short time gaps) would be most useful (Agnew, 1992). Third, GST indicates that other forms of negative emotions such as guilt, fear, shame, frustration, and anxiety are important in explaining gender differences in various types of delinquency. It may be that emotions other than anger and depression are relevant to Korean girls’ delinquency, and for both gender groups, these should be explored in future study. Fourth, several key conditioning variables (i.e., delinquent peer association and criminal skills/resources) and non-delinquent coping strategies were not measured. We recommend that future research use a comprehensive list of conditioning factors and non-delinquent coping strategies to better understand the interaction effects of conditioning factors on the strain-delinquency relationship.

In conclusion, the current study provides some support for GST’s utility in explaining differences in how girls and boys become delinquent (Broidy & Agnew, 1997). Findings also advance the empirical development of Broidy and Agnew’s GST explanation of the influence of gender. Moreover, it provides partial evidence of the GST’s generalizability in explaining gender differences in pathways to delinquency outside of the Western context in which the theory was initially developed and tested empirically. It also demonstrates the importance of including culturally unique and influential strains, in the case of Korea, teachers’ punishment and examination-related strain, as hypothesized influences on delinquency.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biographies

Byongook Moon is an associate professor in the Department of Criminal Justice at the University of Texas, San Antonio. His research focuses on the application of criminology theories to explaining delinquency in South Korea. He also has been involved in school bullying and victimization.

Merry Morash is a professor at the School of Criminal Justice, Michigan State University. In addition to research on girls in the juvenile justice system, her research interests include the effects of probation and parole practices on women offenders, domestic violence internationally and among immigrants to the U.S., crime and delinquency causation in the U.S. and South Korea, and women working in policing. She is editor (with Chesney-Lind) of Feminist Theories of Crime: A Library of Essays in Theoretical Criminology (Ashgate), and author of Women on Probation and Parole (Northeastern U. Press) and Advances in Understanding Gender, Crime and Justice (Sage).

References

Agnew

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

Agnew

(2002). Experienced, vicarious, and anticipated strain: An exploratory study on physical victimization and delinquency. Justice Quarterly, 19, 603-632.

Agnew

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

F. T.

Wright

J. P.

Blevins

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

Agnew

Brezina

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

Norton

E. C.

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

Aiken

L. S.

West

S. G.

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

Broidy

Agnew

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

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J.

Jr. Arneklev

B. J.

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

Hoffmann

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 & Delinquency, 34, 46-78.

10.

Jang

S. J.

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

11.

Jang

S. J.

Johnson

B. R.

(2005). Gender, religiosity, and reactions to strain among African Americans. Sociological Quarterly, 46, 323-357.

12.

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.

13.

Kaufman

J. M.

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

14.

Landrine

Klonoff

E. A.

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

15.

Leadbeater

B. J.

Blatt

S. J.

Quinlan

D. M.

(1995). Gender-linked vulnerabilities to depressive symptoms, stress, and problem behaviors in adolescents. Journal of Research on Adolescence, 5, 1-29.

16.

Lee

Larson

(2000). The Korean “Examination Hell”: Long hours of studying, distress, and depression. Journal of Youth and Adolescence, 29, 249-271.

17.

Lin

Cochran

J. K.

Mieczkowski

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

18.

Long

J. S.

(1997). Regression models for categorical and limited dependent variables. London: Sage.

19.

Mazerolle

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

20.

Mazerolle

Maahs

(2000). General strain and delinquency: An alternative examination of conditioning influences. Justice Quarterly, 17, 753-777.

21.

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.

22.

Moon

Morash

McCluskey

C. P.

Hwang

(2009). A comprehensive test of general strain theory: Key strains, situational-and trait-based negative emotions, conditioning factors, and delinquency. Journal of Research in Crime & Delinquency, 46, 182-212.

23.

Morash

Moon

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

24.

Pang Jong-Hwan Research Center. (2010). 한국 어린이ㆍ청소년 행복지수의 국제비교 (International comparison of Korean children & adolescents). Seoul, Korea. Author.

25.

Paternoster

Brame

Mazerolle

Piquero

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

26.

Patterson

Bae

S. O.

Lim

J. Y.

(2013). Gender equality in Korean firms: Recent evidence from HR practitioners. Asia Pacific Journal of Human Resources, 51, 364-381.

27.

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, 125-157.

28.

Piquero

N. L.

Sealock

M. D.

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

29.

Sharp

Lipsky

(2002). Screening for depression across the lifespan: A review of measures for the use in primary care settings. American Family Physician, 66, 1001-1008.

30.

Yun

Kim

Morris

(2014). Gendered pathways to delinquency: An examination of general strain theory among South Korean youth. Youth Violence and Juvenile Justice, 12, 268-292.

Gender and General Strain Theory

Abstract

Keywords

Introduction

Gender, General Strain Theory, and Empirical Findings

General Strain Theory in the Context of South Korea

Hypotheses

Method

Sample

Measurement

Independent variables

Seven strains

Situation-based negative emotions: Anger and depression 1

Four conditioning factors

Dependent variables—three types of delinquency

Analysis

Results

Discussion and Conclusion

Footnotes

Declaration of Conflicting Interests

Funding

Notes

Author Biographies

References

Situation-based negative emotions: Anger and depression¹