Using an Integrated Model of Strain and Routine Activities to Explore the Reciprocal Relationship Between Direct/Various Victimization and Criminal Coping Among Serious Adolescent Offenders

Abstract

Using an integrated model of general strain and lifestyle/routine activities theories, the study aimed to prospectively assess the reciprocal relationship between direct victimization, vicarious victimization, and delinquency/crime over time among serious offenders. A cross-lagged path model was conducted using three waves from the Pathways to Desistance Study. Past victimization consistently predicted future victimization, while past delinquency/crime consistently affected future delinquency/crime, demonstrating stability across these variables. Prior vicarious victimization also indirectly increased subsequent direct victimization and delinquency/crime. However, there were no direct or indirect effects found between direction victimization and later vicarious victimization or delinquency/crime, or between delinquency/crime and later direct or vicarious victimization. Sensitivity analyses revealed the contemporaneous effects of victimization were more consequential on offending than the lagged effects.

Keywords

reciprocal relationship general strain theory lifestyle/routine activities victimization crime

Victimization research reveals that a great number of juveniles experience criminal victimization during their childhood and adolescence (e.g., Becker & Kerig, 2011; Copeland et al., 2007). It is also noted that criminal victimization is diverse, both in the ways it is experienced (i.e., direct and vicarious victimization) and its context (e.g., within a family, school, and community). These adverse events include physical violence and child maltreatment (Dixon et al., 2005), witnessing violence at home or in the community (Graham-Bermann et al., 2012), and peer bullying (Park & Metcalfe, 2020).

Studies show that an individual’s risk of future offending is related to their prior victimization (Agnew et al., 2002) and an individual’s risk of future victimization is related to their prior participation in risky/deviant lifestyles (Cohen & Felson, 1979; Peterson et al., 2004). This intersection between victimization and delinquency, often termed the victim-offender overlap, has been explained by several criminological theories. Victimized individuals often take some form of “corrective action” as a response (Agnew, 1992, p. 60). Agnew’s general strain theory (GST) has been employed to account for the positive effect of past victimization on future offending. According to GST, deviant behavior is a form of coping that is used among strained adolescents with negative emotionality and low constraint, because delinquency can serve to alleviate strain and negative emotions, especially when legitimate ways of coping are lacking (Agnew, 1992; Agnew & White, 1992). The association between victimization and offending can be stronger when direct and vicarious victimization occur simultaneously (Lin et al., 2011), and when victimization repeatedly occurs over time (Ousey et al., 2008).

Alternatively, lifestyle/routine activities theory posits the reverse causal pathway by focusing on the impact of past offending on future victimization and the similarities between victims and offenders (Hindelang et al., 1978). As individuals become involved in risky/deviant lifestyles, potential victims become closer to offenders. The physical proximity between potential victims and offenders can make these potential victims an attractive target and facilitate a criminal incident (Tillyer et al., 2011). The likelihood of being victimized is expected to be much higher for juveniles who are involved in unstructured or unsupervised social activities, such as gang activities (Osgood et al., 1996).

The separate models proposed by each theory suggest a limited explanation regarding the reciprocal relationship between victimization and offending. That is, each model can only describe unidirectional relationships of what research suggests is a bidirectional relationship. Iratzoqui (2018) and Schreck et al. (2006) suggest integrating these two theories to explain the victim-offender overlap and propose a continuous and prospective analysis. Many prior studies are limited in this regard by relying on a model in which victimization and delinquency are measured at only two different time points. These models fail to unravel whether the likelihood of being victimized results from prior involvement in risky/deviant lifestyles that stemmed initially from prior exposure to victimization. In addition, the distinction between direct and vicarious victimization is rarely made in these assessments, and the association between more serious forms of victimization and chronic offending is less understood. A prospective assessment of how direct and vicarious victimization interplay with each other to influence criminal coping among serious offenders, as well as how this criminal coping affects future direct and vicarious victimization experiences, can provide a better understanding of the nexus between victimization and offending over time and among more chronic offenders.

To this end, the current study focuses on the reciprocal relationship between direct criminal victimization, vicarious criminal victimization, and delinquency/crime using data from the Pathways to Desistance study, which is composed of a sample of serious adolescent offenders transitioning into adulthood. Expanding upon existing research, an integrated theory combining GST and lifestyle/routine activities theory is applied to guide the research. A cross-lagged path model is used to assess the pathways from direct victimization, vicarious victimization, and delinquency/crime to subsequent victimization and delinquency/crime across three points in time. The model enables consideration of the impact of past events and conditions on later events and conditions, as well as the distinct effects of direct and vicarious victimization on delinquency/crime.

Integrating General Strain Theory and Lifestyle/Routine Activities Theory

Agnew’s general strain theory (Agnew, 1983, 1984, 1985, 1992) delineated a relationship between strain and delinquency, such that delinquent activities are more likely to be committed by strained individuals who experience negative emotions. According to Agnew’s (1992) assertions, criminal victimization is one form of strain—a presentation of negative stimuli—that can contribute to criminal behavior (Agnew, 2001).¹ The introduction of criminal victimization meets all four characteristics of strain that can lead to criminal coping (Agnew, 2001, 2002; Kort-Butler, 2010). Criminal victimization is perceived as unjust, a fact that elicits negative emotions (Agnew, 2002; Agnew et al., 2002; Hoskin, 2013). Most criminal victimization is caused by the voluntary and intentional behavior of others, not those of victims (Agnew & Brezina, 1997; Baron, 2009; Ousey et al., 2015). Criminal victimization produces a strain that is high in magnitude, since it can be repeated across time and/or in multiple forms. Such experiences can modify personality traits and levels of social support (Agnew, 2002) and transform the perceived costs of criminal versus noncriminal coping (Radliff et al., 2016). Criminal victimization is associated with low social control (Agnew, 2002). Victimizations at the hands of parents (e.g., child maltreatment) and beyond the scope of parental monitoring and supervision (e.g., school bullying) can lack the element of social or parental control and hamper social bonds between children and parents (Augustyn et al., 2019; Moon et al., 2012). Criminal victimization committed by intimate groups creates pressure or incentive to engage in criminal coping (Agnew, 2001; Baron, 2009) by influencing victims’ beliefs about antisocial behavior and fostering favorable attitudes toward aggressive and antisocial behavior (e.g., the intergenerational cycle of maltreatment: Kim, 2009; Thornberry & Henry, 2013; the victim-offender overlap of school bullying: Connell et al., 2016).

GST distinguishes between direct victimization and vicarious victimization (Agnew, 2001). The direct experience of criminal victimization or physical violence is often a focus within the theory (Agnew, 1992). However, vicarious victimization, defined as witnessing or hearing about criminal situations committed against others, is also common and can have consequences on behavior (Kort-Butler, 2010). While these two forms of strain are distinct, Agnew (2002) proposes that both are closely interrelated with delinquency. Individuals who experience direct victimization in the past tend to respond to current indirect exposure to violence by perceiving others’ victimization as their own strain (Agnew, 2002). Similarly, individuals who have previously observed the criminal victimization of people close to them can respond to their own victimization in a much more aggressive manner (Agnew, 2002; Kort-Butler, 2010). Ultimately, criminal coping is expected in the above situations, as individuals want to avoid unpleasant situations based on what they have learned from their own and others’ experiences.

While Agnew (2002) did not explicitly suggest that the effect of direct victimization would be stronger than that of vicarious victimization or vice-versa, prior studies found that direct victimization is more likely to result in future offending than vicarious victimization (Agnew & White, 1992; Lin et al., 2011). Still, some studies recognized vicarious victimization was linked more so to property crimes and substance use, while direct victimization was linked to violent crimes (Lee & Kim, 2018; Pinchevsky et al., 2014). Together, these findings imply that the effects of direct victimization and vicarious victimization may be different and should be explored separately. From a policy standpoint, understanding how the two forms of victimization effect criminal behavior, as well as how the two interplay to influence coping strategies, is useful in tailoring treatments for victims that can prevent criminal responses to these two varying forms of strain.

The lifestyle/routine activities perspective is employed to describe the relationship between offending and victimization, suggesting that the risk of victimization can be increased through involvement in offending behavior in the routine activities of everyday life (Cohen & Felson, 1979). According to this framework, the positive association between criminal offending and subsequent victimization is dependent on the shared circumstances and lifestyles between victims and offenders (Armstrong & Griffin, 2007; Jensen & Brownfield, 1986; Lauritsen et al., 1991). The convergence of time and place in daily activities increases contact between victims and offenders (Cohen & Felson, 1979; Jensen & Brownfield, 1986). An individual whose lifestyle maintains a physical/residential proximity with violent offenders, such as engaging in criminal and delinquent activities, using alcohol and drugs, hanging out at night, and residing in crime-ridden communities, is expected to have a greater risk of being a victim (Cho et al., 2016; Cohen & Felson, 1979; Schreck et al., 2004). Not only do risky behaviors introduce the initial victimization (Cohen & Cantor, 1981; Jensen & Brownfield, 1986; Schreck & Fisher, 2004), but risky activities also amplify the probability of multiple types of victimization and re-victimization (Turanovic et al., 2018).

Although GST and lifestyles/routine activities theory have often been applied in prior research to explain the link between victimization and offending, two distinct models are proposed in each perspective and neither can fully explore potential reciprocal associations. Stated differently, each theory is restricted to explaining one pathway of the association between victimization and delinquency. That is, GST’s framework is not applicable to considering the impact of delinquency on victimization, while the lifestyle/routine activities perspective cannot explain the origin of deviant behaviors as a result of victimization. For the purpose of the current study, the two theories are seen as complementary to each other (Iratzoqui, 2018; Schreck et al., 2006). In this way, a model combining both theories can address the overall and prospective reciprocal relationship between victimization and delinquency across time points.

Empirical Research on the Reciprocal Relationship Between Victimization and Delinquency

Several studies have investigated the reciprocal relationship between victimization and offending, which describes how new victimizations and crimes result from previous victimizations and crimes. The findings suggest that criminal victimization and offending are reciprocally and positively related, net of the effects of a host of demographic variables (e.g., age, sex, SES, family condition) and prior and current delinquency (Berg et al., 2012; Lauritsen & Laub, 2007; Lauritsen et al., 1991; Ousey et al., 2011; Wilcox et al., 2006).

Still, some of these prior studies employed two separate models: one analysis to examine the impact of time 1 victimization on time 2 delinquency, and another to describe the impact of time 1 delinquency on time 2 victimization (e.g., Berg et al., 2012). In this respect, there is a need to further understand how victimization and delinquency are related to each other in a more continuous setting (Ousey et al., 2011). An integrated theory of GST and lifestyle/routine activities theory suggests that prior victimization and delinquency are related to subsequent victimization and delinquency. In this way, victimization and delinquency need to be measured at different points in time to explore whether the risk of victimization at a later point directly results from initial victimization experiences or indirectly derives from risky/deviant lifestyles as a means of coping with the initial victimization. The research exploring this reciprocal effect is less prevalent (see Iratzoqui, 2018; Schreck et al., 2006).

The prospective approach presented by Schreck et al. (2006) and Iratzoqui (2018) may be a more compelling longitudinal analysis to capture the continuous reciprocal relationship. Schreck et al. (2006) explored how low self-control, risky lifestyles, victimization, and delinquent behaviors are related to one another. Using panel data from the Gang Resistance Education Training program, the results of structural equation models (SEM) show that victimization at time 1 is positively and directly related to victimization at time 3, and indirectly affects victimization at time 3 through association with delinquent peers and involvement in delinquent behavior at time 2. Iratzoqui (2018) found that childhood maltreatment leads to current violent victimization through both negative emotions (i.e., depression) and delinquent coping (i.e., running away). As contrasted with other prior studies, these two studies employed measurements collected at three different points in time as a means of considering the link between earlier and later events of victimization/delinquency. For example, Schreck et al. (2006) demonstrated that preexisting pro-social attachments to parents and other pro-social individuals can fluctuate as a result of victimization, which, consequently, can change the odds of subsequent victimization and delinquency (Schreck et al., 2006). Iratzoqui (2018) showed that the lack of effective guardianship and parental attachment can provoke youths’ further involvement in highly risky/deviant behavior, which, in turn, elevates their probability of subsequent victimization.

However, there are several areas for expansion from these two prospective studies. Both studies relied on school-based samples. The juveniles who were not included in the study may represent a higher likelihood of involvement in delinquent behaviors and victimization experiences. The exclusion of these potentially high-risk juveniles may also lead to issues in fully explaining between-individual differences due to low participation in delinquency among the sample (see Iratzoqui, 2018 for discussion). Also, the measure of delinquent coping relies on minor deviant behaviors only and does not capture more serious offending (e.g., violent and property crimes). Therefore, the reciprocal relationship between more serious forms of victimization and offending is less understood.

The scope of victimization and delinquency can also be expanded. Specifically, Schreck et al. (2006) included violent victimization and property victimization, while Iratzoqui (2018) focused on child maltreatment. Neither of the studies considered the distinction between direct and vicarious victimization in their prospective model. From a theoretical standpoint, it is appropriate to consider the interplay between direct and vicarious victimization in analyses of the reciprocal relationship between victimization and offending. Specifically, the probability of direct victimization can be increased by individuals’ exposure to vicarious victimization and vice versa, since they share some common underlying factors (e.g., family conditions, neighborhood factors; Finkelhor et al., 2009; Turner et al., 2011). Then, those with prior direct and vicarious victimization experiences may be more sensitive to and strained by current victimization experiences, whether direct or vicarious, given the increased fear created by earlier victimization (Agnew, 2002). Finally, risky/deviant lifestyles, such as drinking and illegal activities, could affect individuals’ cognitive perception of victimization, as well as emotional fear of being victimized (Choi & Dulisse, 2019; Melde, 2009). Individuals who are involved in criminal victimization and risky/deviant lifestyles are different from those with only one of the two events in terms of moral disengagement (Runions et al., 2019).

Current Study

The current study applies an integrated theoretical model to explore the non-recursive relationship between direct victimization, vicarious victimization, and delinquency across three waves. This study focuses on a high-risk sample from the Pathways to Desistance data, who have a high prevalence for both direct and vicarious victimization experiences and have been involved in past criminal and delinquent behaviors. Four research questions are considered:

Is exposure to direct victimization and vicarious victimization, respectively, at an earlier point in time related to an increase in delinquency/crime at a subsequent point in time?

Is exposure to direct victimization at an earlier point in time related to an increase in delinquency/crime at a subsequent point in time via vicarious victimization? Also, is exposure to vicarious victimization at an earlier point in time related to an increase in delinquency/crime at a subsequent point in time via direct victimization?

Is involvement in delinquency/crime at an early point in time directly and/or indirectly related to an increase in direct victimization and vicarious victimization, respectively, at a subsequent point in time? If so, does the harmful effect of criminal/delinquent behaviors on subsequent victimization vary by the types of victimization?

These research questions extend existing knowledge about the victimization-delinquency relationship in several ways. They reflect a model that theoretically integrates GST with lifestyles/routine activities theory to consider reciprocal effects (e.g., Iratzoqui, 2018). These questions account for the continuous relationship between victimization and delinquency by assessing the relationship across three different time points (e.g., Schreck et al., 2006). Distinct from prior studies, the research questions separate direct and vicarious victimization experiences to better understand how the two interplay with each other to affect delinquent coping (e.g., Cadely et al., 2019). Finally, the study considers the consequences of more serious forms of victimization and crime.

The GST and lifestyle/routine activity integrated theoretical model can contribute to discussions surrounding theoretical expansion in that the combination of these two theories can overcome the limitations of each to explain the reciprocal relationship between direct victimization, vicarious victimization, and offending. That is, the combined model consists of bidirectional assumptions of the causal pathways between the two events (i.e., victimization and delinquency/crime), which often occur in real-world settings. This model can also contribute to existing theoretical understandings regarding the consequences of the inter-relationship between these two forms of victimization on delinquency, thereby contributing to GST’s existing propositions.

The use of a high-risk sample assures that the variables of interest have variability across respondents, with many samples having low prevalence of victimization and offending. In this sense, this study can offer an improved explanation of the complex relationships between victimization and delinquency. Based on existing theory and prior research, individuals who previously experienced direct (vicarious) victimization are expected to have a higher likelihood of experiencing later vicarious (direct) victimization, which, in turn, should increase the likelihood of subsequent offending. Also, individuals who previously participated in delinquent/criminal activities are anticipated to have a higher likelihood of experiencing direct and vicarious victimization, respectively.

Data and Sample

The data for the current study were taken from the Pathways to Desistance study, made available through the Interuniversity Consortium for Political and Social Science Research (ICPSR). The Pathways to Desistance study is a multi-site panel study, with data collected in two locales—Maricopa County (Phoenix), Arizona, and Philadelphia County, Pennsylvania. Respondents within the study were classified as serious adolescent offenders, such that they were convicted of a serious offense between the ages of 14 to 17. Among those enrolled, a baseline interview was conducted between November 2000 and January 2003. Follow-up interviews were then conducted with the respondents in the following 6, 12, 18, 24, 30, 36, 48, 60, 72, and 84 months. The original purpose of the study was to recognize desistance patterns among adolescent offenders as they transition into adulthood and to explore the impact of social context and developmental factors, as well as sanctions and interventions, on antisocial behaviors (see Mulvey et al., 2004 for more detail on the study).

The current study uses waves 1 through 3, with each of these waves having a 6-month recall period. Several control variables are also taken from the baseline interview (explained further below). At the first three waves, most of the respondents are juveniles, aged 14 to 17 (the following number of respondents are in these age ranges at each wave: 962 in Wave 1, 776 in Wave 2, and 574 in Wave 3) and are thus at the prime age for victimization (Agnew, 2006; Kim et al., 2005) and delinquent behaviors (Agnew, 2013; Defoe et al., 2013). Among the 1,354 eligible participants at the baseline interview, 1,265 respondents (93.43%) agreed to participate and complete the survey at the first follow-up interview (Wave 1), with an average retention rate of about 90% or above over the next two waves. The final sample used in the analyses included 1,156 respondents.²

Key Variables of Interest

Because of the reciprocal nature of this study, the variables of interest are treated interchangeably as exogenous and endogenous variables and are therefore not classified as one or the other in this section. Overall Delinquency/Crime is a time-varying frequency score of 22 different illegal behaviors, including violent offenses, property offenses, and substance-related offenses (see Appendix A for details). Respondents are asked to report the number of times they were involved in these illegal behaviors over the recall period at Waves 1 to 3. As a count, this variable reflects the number of criminal acts in which the person engaged at each wave. As shown in Table 1, the variable ranges from 0 offenses to 3,250 offenses, with higher values indicating a greater frequency in offending by the respondent.

Table 1.

Descriptive Statistics (n = 1,156).

	Wave 1		Wave 2		Wave 3		Min	Max
	Mean	SD	Mean	SD	Mean	SD	Min	Max
Dependent variables
Delinquency/crime (frequency)	35.366	199.904	34.920	161.644	47.153	218.017	0	3,250
Direct victimization (count)	0.279	0.695	0.202	0.589	0.200	0.589	0	6
Vicarious victimization (count)	1.174	1.474	1.067	1.434	0.962	1.401	0	7
Time-varying controls
Age	16.527	1.153	17.042	1.155	17.513	1.138	14	21
Time on streets	0.527	0.438	0.568	0.433	0.667	0.420	0	1
Emotional intensity	2.806	0.673	2.812	0.682	2.842	0.681	1	4
Risk factor index	4.041	2.102	3.982	2.130	4.152	2.147	0	13
Time-invariant controls
Male (%)	85.97	—	—	—	—	—	0	1
Race/ethnicity		—	—	—	—	—
White (%)	21.11	—	—	—	—	—	0	1
Black (%)	40.66	—	—	—	—	—	0	1
Hispanic (%)	33.65	—	—	—	—	—	0	1
Other (%)	4.58	—	—	—	—	—	0	1
Family structure (1 = both biological parents, %)	15.31	—	—	—	—	—	0	1
Family socioeconomic status (SES)	4.305	0.944	—	—	—	—	1	6
Parental warmth	3.053	0.690	—	—	—	—	1	4
Parental monitoring	2.762	0.704	—	—	—	—	1	4
Neighborhood conditions	2.338	0.748	—	—	—	—	1	4

Note. SD = standard deviation; Min = minimum; Max = maximum.

Also, participants are asked to report on their experience with direct victimization and vicarious victimization, separately, over the past recall period. Direct Victimization is measured by asking participants about six related behaviors, including (1) being chased, (2) being beaten up, mugged, or seriously threatened by another person, (3) being raped, a victim of attempted rape, or sexually attacked, (4) being attacked with a weapon, (5) being shot at, and (6) being shot. Vicarious Victimization is measured with seven related behaviors, including (1) observing someone else being chased and thought they could be seriously hurt, (2) observing someone else being beaten up, mugged, or seriously threatened, (3) observing someone else being raped, a victim of attempted rape, or sexually attacked, (4) observing someone else being attacked with a weapon, (5) observing someone else being shot at, (6) observing someone else being shot, and (7) observing someone else being killed as a result of violence. The respondents answered “yes” (1) or “no” (0) to each question. The resulting variables reflect a variety score of the number of direct and vicarious victimization experiences at each wave. In this respect, Direct Victimization ranges from 0 to 6 exposures and Vicarious Victimization ranges from 0 to 7 exposures (see Table 1), with higher values indicating a greater number of either direct or vicarious exposures to criminal victimization by the respondent.

Time-Varying Control Variables

Several time-varying control variables are included that are theoretically related to GST and the lifestyles/routine activities perspectives (see Table 1). Age is a time-variant factor, which captures the age in years of the respondents at each wave. Time on Street reflects the proportion of time spent on the streets during each recall period. Higher scores indicate more time spent on the streets. Emotional Intensity is another time-variant factor that captures the adolescents’ ability to regulate emotions. This variable was originally created by the Pathways team and represents the mean of nine items that are adapted from Walden et al.’s (1992) Children’s Emotional Intensity Child Reports. Higher scores indicate a greater ability to regulate emotions.

Following the recommendation of Agnew (2006, 2013) and prior studies (e.g., Craig et al., 2017; Thaxton & Agnew, 2018), a Risk Factor Index is also created. This index combines 13 factors into a single additive index: peer delinquency, moral disengagement, perception of chances for success, substance abuse, future orientation, personal rewards of crime, religious attendance, low self-control, employment status, school status, gang involvement, family criminality, and relationship status (see Appendix A for details about these measures). Five of these variables are already dummy coded: employment status (1 = unemployed), school status (1 = not enrolled in school), gang involvement (1 = gang activity), family criminality (1 = having delinquent/criminal family members), and relationship status (1 = unmarried or no romantic relationship). Aside for these five dummy coded variables, all other variables are recoded on the upper quartile (the most extreme 25%) of these scales as 1 to represent a higher risk in each of the factors, and those observations on the other quartiles are coded as 0 (see Park & Metcalfe, 2020). For the purpose of the index, perceptions of chances for success, religious attendance, future orientation, and low self-control are reverse coded initially. Each of these dummy variables are added together to form the risk index, with higher values representing greater risk factors for delinquency/crime. Appendix B provides the correlation matrix and descriptive statistics of the items in the index.

Time-Invariant Control Variables

The current study also considers several control variables that are found to be related to victimization and/or delinquency based on the theory and research guiding this study but either do not vary or are not adequately captured at each wave. All of these variables are taken from the baseline interview (see Table 1), including sex (1 = Male), four racial/ethnic categories (White, Black, Hispanic, and Other), Family Socioeconomic Status (SES), Family Structure, Parental Warmth, Parental Monitoring, and Neighborhood Conditions (see Appendix A for details). Variance inflation factors (VIF) and the correlation matrices suggest no issues of multicollinearity (available upon request).

Analysis

Path analysis is used to examine the set of research questions that aims to explore the longitudinal and prospective scope of the non-recursive relationship between direct victimization, vicarious victimization, and delinquency captured at three different waves. The multiple pathways between direct victimization, vicarious victimization, and delinquency are explored by estimating direct and indirect effects simultaneously within a single model. Specifically, a cross-lagged model is used to analyze the impact of the Time 1 measures (i.e., direct victimization, vicarious victimization, delinquency/crime) on their Time 2 and 3 counterparts, while simultaneously estimating the effect of the other covariates (see Bui et al., 2000; Matsueda & Anderson, 1998). By exploring lagged specifications, issues related to causal ordering are also considered. Figure 1 presents the model specified without inclusion of the control variables.

Based on the goals of the study, structural equation modeling (SEM) with full information maximum likelihood (FIML) estimation is utilized.³ The SEM approach is useful in understanding relational data in multivariate systems. The directional parameters between the key variables of interest across waves (e.g., Direct Victimization at Wave 1 to Direct Victimization, Vicarious Victimization, and Delinquency/Crime at waves 2 and 3) enable tests related to whether and what extent an early event or condition has a stable and/or cross-lagged influence on the variables of interest at a later point in time. The model includes covariances between all exogenous variables within and across waves. Specifically, the covariances are measured between Direct Victimization, Vicarious Victimization, and Delinquency/Crime at Wave 1, between the key variables at Wave 1 and the control variables, and between the control variables across waves. Also, the error terms for Direct Victimization, Vicarious Victimization, and Delinquency/Crime at waves 2 and 3 are allowed to covary. As mentioned above, the model consists of four time-variant control variables, including Age, Time on Street, Emotional Intensity, and the Risk Factor Index that are taken from waves 2 and 3, respectively. These control variables are linked to the key variables of interest at the same wave, such that the lagged effects are not explored for these specific variables.⁴

As recommended by Markus (1979), unstandardized estimates are presented “because correlations and standardized regression coefficient values are affected by changes in variances across populations” (p. 49) and the lagged effects of each of the endogenous variables (i.e., direct victimization, vicarious victimization, and delinquency/crime) are included in the models. Due to the assumption of a normal distribution in SEM models, the frequency score of Delinquency/Crime and the counts of Direct Victimization and Vicarious Victimization are log-transformed to more closely meet these assumptions.

Results

As noted, a path analysis is used to simultaneously model the direct and indirect effects between direct victimization, vicarious victimization, and delinquent/criminal behavior across three waves. Thus, the results presented below are based on a single model. Fit statistics for this model indicate the model fits the data well (Comparative Fit Indices—CFI = .985, Tucker-Lewis Indices—TLI = .921, Root Mean Square Error of Approximation—RMSEA = .037, Akaike’s Information Criterion—AIC = 52,921.813, Bayesian Information Criterion—BIC = 54,437.629).⁵ For CFI and TLI, a value of .90 or above indicates a good fit (Bentler & Bonett, 1980). For RMSEA, a value that is close to 0 is optimal for purposes of good fit, and a value of .06 or below indicates a good fit (Brown & Cudeck, 1993).

Tables 2 and 3 report direct effects, specific indirect effects, total indirect effects, and total effects for Direct Victimization. Direct Victimization at Wave 1 has direct effects on Direct Victimization at Wave 2 (b = .127, p < .001) and Direct Victimization at Wave 3 (b = .074, p < .01). Also, Direct Victimization at Wave 1 is indirectly linked to Direct Victimization at Wave 3 (b = .028, p < .001), but this total indirect effect appears to operate specifically through Direct Victimization at Wave 2 (b = .028, p < .001). Individuals who are exposed to direct victimization at an earlier point in time tend to experience direct victimization at a later point in time, indicating a stability in direct victimization. Unexpectedly, Direct Victimization at Wave 1 does not directly or indirectly influence Vicarious Victimization or Delinquency/Crime at the later waves.

Table 2.

Direct Effects of Path Model Examining the Longitudinal Impact of Direct Victimization on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	b	SE
Direct victimization at W1 → Direct victimization at W2	.127***	0.027
Direct victimization at W1 → Vicarious victimization at W2	−.013	0.050
Direct victimization at W1 → Delinquency/crime at W2	.203	0.129
Direct victimization at W1 → Direct victimization at W3	.074**	0.027
Direct victimization at W1 → Vicarious victimization at W3	.049	0.048
Direct victimization at W1 → Delinquency/crime at W3	.150	0.139
Direct victimization at W2 → Direct victimization at W3	.219***	0.032
Direct victimization at W2 → Vicarious victimization at W3	.007	0.056
Direct victimization at W2 → Delinquency/crime at W3	.024	0.162

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient; SE = standard error.

p < .05. **p < .01. ***p < .001 (two-tailed).

Table 3.

Indirect and Total Effects of Path Model Examining the Longitudinal Impact of Direct Victimization on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	Indirect b	Total b
Direct victimization at W1 → Direct victimization at W2 → Direct victimization at W3	.028***
Direct victimization at W1 → Vicarious victimization at W2 → Direct victimization at W3	−.009^a
Direct victimization at W1 → Delinquency/crime at W2 → Direct victimization at W3	−.002^b
Direct victimization at W1 → Direct victimization at W3	.028***	.102***
Direct victimization at W1 → Direct victimization at W2 → Vicarious victimization at W3	.009
Direct victimization at W1 → Vicarious victimization at W2 → Vicarious victimization at W3	−.003
Direct victimization at W1 → Delinquency/crime at W2 → Vicarious victimization at W3	.003
Direct victimization at W1 → Vicarious victimization at W3	.008	.058
Direct victimization at W1 → Direct victimization at W2 → Delinquency/crime at W3	.003
Direct victimization at W1 → Vicarious victimization at W2 → Delinquency/crime at W3	−.002
Direct victimization at W1 → Delinquency/crime at W2 → Delinquency/crime at W3	.047
Direct victimization at W1 → Delinquency/crime at W3	.048	.197

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient.

The coefficient was multiplied by 100 to obtain a non-zero value.

The coefficient was multiplied by 10 to obtain a non-zero value.

p < .05. **p < .01. ***p < .001 (two-tailed).

Tables 4 and 5 consider the direct and indirect pathways stemming from Vicarious Victimization. Vicarious Victimization at Wave 1 is both directly and indirectly linked to Direct Victimization, Vicarious Victimization, and Delinquency/Crime at Wave 3. Vicarious Victimization at Wave 1 has positive and direct effects on Direct Victimization at Wave 2 (b = .052, p < .01), Vicarious Victimization at Wave 2 (b = .308, p < .001), Delinquency/Crime at Wave 2 (b = .355, p < .001), and Vicarious Victimization at Wave 3 (b = .141, p < .001). When looking at total indirect effects, Vicarious Victimization at Wave 1 is indirectly associated with Vicarious Victimization at Wave 3 (b = .088, p < .001), as well as Direct Victimization at Wave 3 (b = .013, p < .05) and Delinquency/Crime at Wave 3 (b = .140, p < .001). The latter two are noteworthy given that the direct effects between Vicarious Victimization and these variables were nonsignificant.

Table 4.

Direct Effects of Path Model Examining the Longitudinal Impact of Vicarious Victimization on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	b	SE
Vicarious victimization at W1 → Direct victimization at W2	.052**	0.017
Vicarious victimization at W1 → Vicarious victimization at W2	.308***	0.032
Vicarious victimization at W1 → Delinquency/crime at W2	.355***	0.082
Vicarious victimization at W1 → Direct victimization at W3	.026	0.018
Vicarious victimization at W1 → Vicarious victimization at W3	.141***	0.032
Vicarious victimization at W1 → Delinquency/crime at W3	−.081	0.091
Vicarious victimization at W2 → Direct victimization at W3	.007	0.018
Vicarious victimization at W2 → Vicarious victimization at W3	.257***	0.031
Vicarious victimization at W2 → Delinquency/crime at W3	.183*	0.089

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient; SE = standard error.

p < .05. **p < .01. ***p < .001 (two-tailed).

Table 5.

Indirect and Total Effects of Path Model Examining the Longitudinal Impact of Vicarious Victimization on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	Indirect b	Total b
Vicarious victimization at W1 → Direct victimization at W2 → Direct victimization at W3	.011**
Vicarious victimization at W1 → Vicarious victimization at W2 → Direct victimization at W3	.002
Vicarious victimization at W1 → Delinquency/crime at W2 → Direct victimization at W3	−.004^a
Vicarious victimization at W1 → Direct victimization at W3	.013*	.039*
Vicarious victimization at W1 → Direct victimization at W2 → Vicarious victimization at W3	.004
Vicarious victimization at W1 → Vicarious victimization at W2 → Vicarious victimization at W3	.079***
Vicarious victimization at W1 → Delinquency/crime at W2 → Vicarious victimization at W3	.005
Vicarious victimization at W1 → Vicarious victimization at W3	.088***	.229***
Vicarious victimization at W1 → Direct victimization at W2 → Delinquency/crime at W3	.001
Vicarious victimization at W1 → Vicarious victimization at W2 → Delinquency/crime at W3	.056*
Vicarious victimization at W1 → Delinquency/crime at W2 → Delinquency/crime at W3	.083***
Vicarious victimization at W1 → Delinquency/crime at W3	.140***	.059

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient.

The coefficient was multiplied by 10 to obtain a non-zero value.

p < .05. **p < .01. ***p < .001 (two-tailed).

When looking more specifically at these indirect effects, early exposure to vicarious victimization at Wave 1 increases the likelihood of experiencing direct victimization at Wave 3 mainly by increasing direct victimization in Wave 2 (b = .011, p < .01). Criminal coping in Wave 3 is enhanced by exposure to vicarious victimization in Wave 1 mainly through increasing delinquency/crime at Wave 2 (b = .083, p < .001) and vicarious victimization at Wave 2 (b = .056, p < .05). Also, early exposure to vicarious victimization at Wave 1 increases the likelihood of experiencing vicarious victimization at Wave 3 mainly by increasing vicarious victimization in Wave 2 (b = .079, p < .001).

Direct and indirect relationships between Delinquency/Crime at Wave 1 and subsequent victimization and deviant coping are shown in Tables 6 and 7. In terms of the direct effects, Delinquency/Crime at Wave 1 has a positive impact on Delinquency/Crime at Wave 2 (b = .250, p < .001) and Delinquency/Crime at Wave 3 (b = .209, p < .001). However, the direct effects of Delinquency/Crime at Wave 1 on Direct Victimization and Vicarious Victimization at later waves are statistically nonsignificant. Turning to the total indirect effects, Delinquency/Crime at Wave 1 is significantly associated with an increase in Delinquency/Crime at Wave 3 (b = .061, p < .001), specifically by increasing Delinquency/Crime at Wave 2 (b = .058, p < .001), demonstrating a stability in offending across waves.

Table 6.

Direct Effects of Path Model Examining the Longitudinal Impact of Delinquency/Crime on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	b	SE
Delinquency/crime at W1 → Direct victimization at W2	.010	0.006
Delinquency/crime at W1 → Vicarious victimization at W2	.013	0.012
Delinquency/crime at W1 → Delinquency/crime at W2	.250***	0.030
Delinquency/crime at W1 → Direct victimization at W3	−.002	0.006
Delinquency/crime at W1 → Vicarious victimization at W3	−.008	0.011
Delinquency/crime at W1 → Delinquency/crime at W3	.209***	0.033
Delinquency/crime at W2 → Direct victimization at W3	−.001	0.007
Delinquency/crime at W2 → Vicarious victimization at W3	.015	0.012
Delinquency/crime at W2 → Delinquency/crime at W3	.233***	0.034

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient; SE = standard error.

p < .05. **p < .01. ***p < .001 (two-tailed).

Table 7.

Indirect and Total Effects of Path Model Examining the Longitudinal Impact of Delinquency/Crime on Direct Victimization, Vicarious Victimization, and Delinquency/Crime Using FIML (n = 1,156).

Path	Indirect b	Total b
Delinquency/crime at W1 → Direct victimization at W2 → Direct victimization at W3	.002
Delinquency/crime at W1 → Vicarious victimization at W2 → Direct victimization at W3	.009^a
Delinquency/crime at W1 → Delinquency/crime at W2 → Direct victimization at W3	−.003^b
Delinquency/crime at W1 → Direct victimization at W3	.002	.004^b
Delinquency/crime at W1 → Direct victimization at W2 → Vicarious victimization at W3	.001
Delinquency/crime at W1 → Vicarious victimization at W2 → Vicarious victimization at W3	.003
Delinquency/crime at W1 → Delinquency/crime at W2 → Vicarious victimization at W3	.004
Delinquency/crime at W1 → Vicarious victimization at W3	.008^†	.001^b
Delinquency/crime at W1 → Direct victimization at W2 → Delinquency/crime at W3	.002^b
Delinquency/crime at W1 → Vicarious victimization at W2 → Delinquency/crime at W3	.002
Delinquency/crime at W1 → Delinquency/crime at W2 → Delinquency/crime at W3	.058***
Delinquency/crime at W1 → Delinquency/crime at W3	.061***	.269***

Note. The model includes the time-variant and time-variant control variables noted. b = unstandardized coefficient.

The coefficient was multiplied by 100 to obtain a non-zero value.

The coefficient was multiplied by 10 to obtain a non-zero value.

^†p < .10. *

p < .05. **p < .01. ***p < .001 (two-tailed).

Supporting prior research on persistence in victimization, the results of the path model show that prior direct victimization experiences increase the subsequent chance of exposure to direct victimization (total effect: b = .102, p < .001). Also, prior vicarious victimization experiences increase the subsequent chance of exposure to direct victimization (total effect: b = .039, p < .05) and vicarious victimization (total effect: b = .229, p < .001). In a similar vein, stability in deviant/criminal behavior across waves is found (total effect: b = .269, p < .001). However, prior direct victimization experiences do not significantly influence, either directly or indirectly, subsequent vicarious victimization and criminal coping. Similarly, prior deviant/criminal behaviors do not have significant impacts on subsequent chances of being a victim or committing a crime, either directly or indirectly.

It should also be noted that some of the control variables have consistent relationships with the endogenous variables across waves. To be specific, the Risk Factor Index at waves 2 and 3 is positively and significantly related to Direct Victimization, Vicarious Victimization, and Delinquency/Crime at waves 2 and 3. Time on Street at waves 2 and 3 is also positively and significantly associated with Direct Victimization and Delinquency/Crime at both waves, but not with Vicarious Victimization. Offending is significantly higher for males at waves 2 and 3 compared to females, while the risk of direct/vicarious victimization does not vary by sex. There are no variations in offending and victimization by race/ethnicity. Individuals with lower SES have an increased risk of Direct Victimization (at Wave 3), Vicarious Victimization (at Wave 3), and Delinquency/Crime (at Wave 2), but the associations are not consistent across waves. Expectedly, poorer neighborhood conditions are positively and significantly related to Vicarious Victimization (at Wave 2 and Wave 3) and Delinquency/Crime (at Wave 3), whereas a high level of parental monitoring is negatively associated with victimization (at Wave 2) and Delinquency/Crime (at Wave 2). An unanticipated finding is that the risk of Vicarious Victimization (at Wave 2) can increase as the level of parental warmth increases.

Sensitivity Analyses

Several sensitivity analyses are conducted to further consider the assumptions of the traditional SEM model, limitations of the cross-lagged model, and potential for greater contemporaneous effects. The main model is re-analyzed with bootstrapped standard errors using 50 replications to ensure the assumption of normality of the key variables of interest, even after the log-transformed versions of direct/vicarious victimization and delinquency/crime are used.⁶ The results regarding the key variables of interest are substantively similar to those reported. The magnitude and significance of the relationships among Direct Victimization, Vicarious Victimization, and Delinquency/Crime are comparable to those reported.

Also, the main model is re-analyzed using a generalized structural equation model (GSEM). In doing so, the frequency scores of Delinquency/Crime at waves 1 to 3 are used, which is truncated to the 95th percentile to remove the outliers (the following numbers are the upper limit of the variable at each wave: 135 in Wave 1, 153 in Wave 2, and 201 in Wave 3). The count scores of Direct Victimization (range between 0 and 6) and Vicarious Victimization (range between 0 and 7) are employed. The GSEM using a negative binomial regression estimator (n = 1,083; AIC = 18,286.74; BIC = 18,760.55) shows similar results to the main model reported.

In addition to these two specifications, an alternative version of the cross-lagged model is analyzed, which can minimize issues related to unobserved confounders and autocorrelation by having all covariance paths constrained between the disturbance terms of the endogenous variables to zero (Brunton-Smith, 2011; Sturgis et al., 2004). The effects of the endogenous variables (i.e., Direct Victimization, Vicarious Victimization, and Delinquency/Crime) are constrained on themselves to represent the average effect overtime, such that the relationship of each of these endogenous variables on itself is fixed to the average effect across the waves. This version of the model shows similar results as those found in the main model (CFI = .976, TLI = .900, RMSEA = .042, AIC = 52,938.016, BIC = 54,423.516). In contrast with the main model, though, Delinquency/Crime at Wave 1 has a direct and marginal impact on Vicarious Victimization at Wave 2 (b = .019, p < .10), and a total indirect effect on Vicarious Victimization at Wave 3 (b = .008, p < .05), specifically through its impact on Vicarious Victimization at Wave 2 (b = .004, p < .10). This finding supports theoretical expectations that prior involvement in delinquency/crime increases the subsequent risk of vicarious victimization.

Lastly, a series of negative binomial regression analyses at each wave are conducted to consider the contemporaneous relationship between victimization and offending, given that a preliminary analysis indicated that the within-wave control variables are more consequential than the lagged control variables. Also, some prior studies suggest that the concurrent impact is more crucial than the lagged effects in terms of criminal coping (e.g., Barnes et al., 2014; Brezina, 1999). These models include the frequency score of Delinquency/Crime at waves 1 to 3, which is truncated to the 95th percentile. The count scores of Direct Victimization (range between 0 and 6) and Vicarious Victimization (range between 0 and 7) are also employed. The results reveal significant concurrent/immediate associations between the variables across waves. That is, a positive and significant association is found between direct victimization, vicarious victimization, and delinquency/crime at all waves assessed. Also, the risk factor index is positively and significantly related to direct/vicarious victimization and offending across waves.

Discussion and Conclusion

The current study applied an integrated theoretical model built on general strain and lifestyle/routine activities theories to examine the continuous and reciprocal relationship between criminal victimization and offending across time among serious offenders. Thus, exposure to direct and vicarious victimization at an early point in time was posited to directly and indirectly influence offending behaviors at a later point in time, and vice versa. Also, an association between direct victimization and vicarious victimization was expected.

Three key conclusions emerged from the study’s findings. First, prior vicarious victimization was a significant predictor of future offending, supporting GST’s argument and several prior studies (e.g., Lin et al., 2011; Menard et al., 2015). However, prior exposure to direct victimization did not have a significant lagged effect on subsequent offending, conflicting with prior studies that suggest between-individual differences in offending based on prior direct victimization experiences (e.g., Ousey et al., 2015; Watts & McNulty, 2013). These results are consistent with a recent study that compared the harmful effect of victimization across contexts, revealing that vicarious victimization was significantly related to an increase in various forms of dating aggression among adolescents, while direct victimization was either insignificantly or negatively associated with dating aggression (Cadely et al., 2019). It should be recognized that more respondents in the study experienced vicarious victimization than direct victimization. This difference may be related to the types of victimization included in the Pathways data. The questions were designed to capture serious forms of victimization, such as homicide, rape, and physical attacks with weapons. Due to the severity of these victimization experiences, adolescents likely observed and heard about others’ victimization experiences more often than experiencing them directly. Also, severe forms of vicarious victimization could yield a more harmful effect when witnessed vicariously by juveniles, specifically. As Agnew (1997) noted, juveniles are a unique social group, which are susceptible to external factors, including criminal behaviors committed toward others.

Second, the current study did not find evidence to support the cross-lagged impacts of delinquency/crime at an early point in time on direct/vicarious victimization at a subsequent point in time. Inconsistent with the lifestyle/routine activity perspective and some prior findings in the literature (e.g., Choi et al., 2016; Turanovic et al., 2018), neither direct victimization nor vicarious victimization at Wave 3 was influenced by involvement in deviant/criminal activities at Wave 2 or Wave 1. Third, the expected mediating role of direct victimization and vicarious victimization was not found for the association between direct and vicarious victimization experiences at an earlier point in time and delinquency/crime at a subsequent point in time. However, the likelihood of direct victimization was significantly increased by previous exposure to vicarious victimization. This finding partially supported general strain theory’s propositions (Agnew, 1992, 2002), which predicted a positive and significant effect of one form of criminal victimization on the other form of criminal victimization. Still, this interrelationship between the two forms of victimization did not translate into offending at later waves, as was theoretically expected.

Importantly, though, sensitivity analyses designed to further explore the limited lagged effects found that the lack of an effect of delinquency/crime on victimization, as well as the lack of a mediating impact of direct and vicarious victimization, could be due to the timeframe between waves. As noted in the supplementary analysis, deviant/criminal coping was influenced more by victimization experiences that occurred at the same wave than those that occurred in the previous waves. Even though the two interviews were 6 months apart, the immediate effects of criminal victimization were more consequential than the lagged effects. In accordance with Agnew’s (1992) recency arguments, the harmful impacts of criminal victimization may be more immediate and short-term, with less of an effect after a certain period of time (Park & Metcalfe, 2020). When considering the reciprocal nature of victimization and offending, further research is needed that explores the duration of time in which victimization experiences become less consequential to criminal behavior and vice versa.

As noted, the findings of the current study contributed to the theoretical literature on the victim-offender overlap by integrating propositions from two complementary theoretical arguments to consider the continuous and prospective association between victimization and delinquency. The integration enabled the study to overcome the limitations of each theoretical perspective and provided greater explanatory power to support the reciprocal relationship proposed beyond the propositions of each theory on its own (Bernard & Ritti, 1990; Elliott et al., 1979). Focusing first on GST, the study further considered the interrelationship between direct and vicarious victimization. We found minimal support for their interrelationship over consecutive waves but did uncover strong associations between the two forms of victimization within waves. An analysis of a concurrent model of these relationships could broaden the understanding of the link between direct victimization, vicarious victimization, and delinquency/crime.

Regarding lifestyle/routine activities theory, we analyzed a model that could incorporate the effects of delinquency/crime on later victimization. However, the findings suggested that this perspective may benefit from broadening its propositions to further consider the temporal aspects of events and types of criminal victimization. Due to the lack of recognition of explicit time elements of criminal victimization (e.g., elements like recency and duration) in this theory, most prior research has not considered whether deviant/risky lifestyles have a lagged versus contemporaneous effect on criminal victimization. The nonsignificant lagged effect in the present study conflicted with Iratzoqui (2018) and Schreck et al. (2006), possibly due to the characteristics of the high-risk sample and/or the serious forms of criminal victimization that were asked. Given these mixed findings, further studies are needed to examine the time elements of crime and criminal victimization, as well as the consequences of crime and victimization among a high-risk sample versus a general sample.

From a policy standpoint, the results of the current study have implications for delinquency and victimization prevention programs for those with serious antisocial tendencies that could help recognize risk factors related to these events over time. Specifically, the stability found in direct and vicarious victimization over the waves was particularly noteworthy, as well as the stability in deviant/criminal behaviors. These findings suggest that juveniles’ future behaviors and conditions are significantly influenced by their past behaviors, characteristics (e.g., low self-control; Schreck et al., 2006), or environmental factors (e.g., living on the street; Baron, 2004). The knowledge of their past life can be used to transform their circumstances and minimize the risk factors that are conducive to adverse conditions in the future. The findings of a significant harmful impact of vicarious victimization on offending could also inform parental training programs and parental skills programs. These programs can serve to educate parents in ways to improve communication skills and help children deal with their exposure to violence within and outside the home (Agnew, 1999; Anderson, 1990; Piquero et al., 2009). This finding is also relevant for school authorities. Programs in school that are designed to detect and reduce school violence by promoting prosocial behaviors, teaching intervention skills, and enhancing unfavorable perspectives toward delinquency can be helpful (Bradshaw, 2015; McCarty et al., 2016).

Despite these contributions, there are several avenues to consider in future research. First, the sample consists of serious adolescent offenders. While a noteworthy sample to consider, this type of sample restricts the range of the dependent variable, which can attenuate the effect of the independent variables. Therefore, the results presented may not be generalizable to the general juvenile population. Some researchers may also argue that general strain theory and the lifestyle/routine activity theory, as general theories of crime, are more applicable to explaining crime among general samples than high-risk samples. In an alternative viewpoint, as general theories, these theories should apply to all populations and samples. A key contribution of this study was its attempt to explore the propositions of the two theories to explain serious adolescent offending for those at greater risk for both criminal victimization and offending.

Second, GST proposes that negative emotions mediate the relationship between strain and criminal coping. Unfortunately, the Pathways to Desistance data did not provide adequate and multiple indicators for the various forms of negative emotions. Also, given the complexity of the model, focus was placed on the mediating roles of victimization and delinquency. Many prior studies suggested that anger was a significant mediator in the link between various forms of strain and different types of delinquency (e.g., Agnew et al., 2002; Patchin & Hinduja, 2011). Also, Iratzoqui (2018) found that depression, fear, and hopelessness were directly and indirectly associated with the impact of child maltreatment on risk behaviors and subsequent violent victimization. Future research should focus greater attention on the mediating role of negative emotions within the reciprocal relationship between victimization and offending.

Third, gender differences might be explored further in terms of victimization. We did not find gender differences, but the dataset included mostly male adolescents at the baseline interview (86.40%) and this imbalance in the gender ratio was maintained throughout the waves. According to Broidy and Agnew (1997), gender differences can derive from the different types and magnitudes of strain, negative affective states, and coping mechanisms (i.e., personal characteristics and external supports). Despite these gender differences found in prior research (e.g., Hay, 2003; Ostrowsky & Messner, 2005), relatively few studies have examined gender differences in relation to vicarious victimization and subsequent delinquency, and those that have explored this topic report inconsistent findings (see Hoffmann & Su, 1997; Kaufman, 2009; Lee & Kim, 2018). Future research can broaden the understanding of the reciprocal association between victimization and offending by focusing on differences between males and females.

Finally, GST is applicable for describing the variations in delinquent and criminal coping among different racial and ethnic groups, although this was not a focus of the current study. Four distinct racial/ethnic groups were considered as control variables, and a significantly lower likelihood of offending for Blacks than Whites was found in some instances, with no racial/ethnic differences in terms of victimization. According to studies testing the Racialized General Strain Theory (RGST) (Agnew, 1999, 2006; Kaufman et al., 2008), minorities are at greater risk of criminal coping (e.g., Jennings et al., 2009; Spohn & Wood, 2014) through aggression, and (Agnew, 1999; Jang & Johnson, 2003) depression (Peck, 2013). They also often experience reduced levels of social support (Agnew, 2006) and/or detrimental environments (Agnew, 1999; Anderson, 1990). Although there are mixed findings (e.g., Isom-Scott & Grosholz, 2019), an intersectional perspective could be used to further explore the impact of one’s coexistent identities (e.g., age, race, sexual orientation, and SES), as well as the interrelationship between these characteristics, victimization, and offending (Isom-Scott, 2018; see also Crenshaw, 1989, 1991; Potter, 2008, 2015).

In conclusion, the current study set out to extend GST and lifestyle/routine activities theory, as well as prior studies related to the victim-offender overlap, by integrating two theoretical perspectives into a model that explores the continuous and prospective reciprocal relationship between victimization and delinquency among serious offenders. Although there are mixed findings based on the two theoretical perspectives, the results reveal important patterns of victimization and offending over time, as well as draw attention to the consideration of lagged versus contemporaneous effects of victimization and offending. Future research should continue to explore the consequences of this timing between victimization and delinquency.

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.

ORCID iD

Yeoju Park

Notes

Author Biographies

Yeoju Park, Ph.D., is an Assistant professor in the Department of Police Administration at the Busan University of Foreign Studies (Busan, South Korea). Her research focuses on theoretical development, victimization, bullying, exposure to violence, and developmental and life-course criminology.

Christi Metcalfe, Ph.D., is an Associate professor in the Department of Criminology and Criminal Justice at the University of South Carolina (Columbia, SC USA). Her research focuses on criminal courts, developmental and life-course criminology, attitudes about crime and the criminal justice system, and quantitative methods.

References

Agnew

(1984). Goals achievement and delinquency. Sociology and Social Research, 68, 435–447.

Agnew

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

Agnew

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

Agnew

(1997). Stability and change over the life course: A strain theory explanation. In Thornberry

T. P.

(Ed.), Developmental theories of crime and delinquency (pp. 101–132). Transaction Publishers.

Agnew

(1999). A general strain theory of community differences in crime rates. Journal of Research in Crime and Delinquency, 36, 123–155.

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

(2002). Experienced, vicarious, and anticipated strain: An exploratory study on physical victimization and delinquency. Justice Quarterly, 19(4), 603–632.

Agnew

(2006). Pressured into crime: An overview of general strain theory. Roxbury Publishing Company.

Agnew

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

10.

Agnew

Brezina

(1997). Relational problems with peers, gender, and delinquency. Youth & Society, 29(1), 84–111.

11.

Agnew

Brezina

Wright

J. P.

Cullen

F. T.

(2002). Strain, personality traits, and delinquency: Extending general strain theory. Criminology, 40, 43–72.

12.

Agnew

White

H. R.

(1992). An empirical test of general strain theory. Criminology, 30, 475–500.

13.

Agnew

R. S.

(1983). Social class and success goals: An examination of relative and absolute aspirations. Sociological Quarterly, 24, 435–452.

14.

Allison

P. D.

(2001). Missing data. SAGE Publications.

15.

Anderson

(1990). Streetwise: Race, class, and change in an urban neighborhood. University of Chicago Press.

16.

Armstrong

G. S.

Griffin

M. L.

(2007). The effect of local life circumstances on victimization of drug-involved women. Justice Quarterly, 24, 80–105.

17.

Augustyn

M. B.

Thornberry

T. P.

Henry

K. L.

(2019). The reproduction of child maltreatment: An examination of adolescent problem behavior, substance use, and precocious transitions in the link between victimization and perpetration. Development and Psychopathology, 31, 53–71.

18.

Barnes

J. C.

Golden

Mancini

Boutwell

B. B.

Beaver

K. M.

Diamond

(2014). Marriage and involvement in crime: A consideration of reciprocal effects in a nationally representative sample. Justice Quarterly, 31(2), 229–256.

19.

Baron

S. W.

(2004). General strain, street youth and crime: A test of Agnew’s revised theory. Criminology, 42, 457–484.

20.

Baron

S. W.

(2009). Street youths’ violent responses to violent personal, vicarious, and anticipated strain. Criminal Justice Journal, 37, 442–451.

21.

Becker

S. P.

Kerig

P. K.

(2011). Posttraumatic stress symptoms are associated with the frequency and severity of delinquency among detained boys. Journal of Clinical Child & Adolescent Psychology, 40(5), 765–771.

22.

Bentler

P. M.

Bonett

D. G.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588–606.

23.

Berg

M. T.

Stewart

E. A.

Schreck

C. J.

Simons

R. L.

(2012). The victim-offender overlap in context: Examining the role of neighborhood street culture. Criminology, 50(2), 359–390.

24.

Bernard

T. J.

Ritti

R. R.

(1990). The role of theory in scientific research. In Kempf

K. L.

(Ed.), Measurement issues in criminology (pp. 1–20). Springer.

25.

Bradshaw

C. P.

(2015). Translating research to practice in bullying prevention. American Psychologist, 70(4), 322–332.

26.

Brezina

(1999). Teenage violence toward parents as an adaptation to family strain: Evidence from a national survey of male adolescents. Youth & Society, 30, 416–444.

27.

Broidy

Agnew

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

28.

Brown

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structure equation models (pp. 136–162). SAGE.

29.

Brunton-Smith

(2011). Untangling the relationship between Fear of crime and perceptions of disorder: Evidence from a longitudinal study of young people in England and Wales. The British Journal of Criminology, 51(6), 885–899.

30.

Bui

K. V. T.

Ellickson

P. L.

Bell

R. M.

(2000). Cross-lagged relationships among adolescent problem drug use, delinquent behavior, and emotional distress. Journal of Drug Issues, 30(2), 283–303.

31.

Cadely

H. S.

Mrug

Windle

(2019). Comparisons of types of exposure to violence within and across contexts in predicting the perpetration of dating aggression. Journal of Youth and Adolescence, 48, 2377–2390.

32.

Cho

Wooldredge

Sun Park

(2016). Lifestyles/routine activities and bullying among South Korean youths. Victims & Offenders, 11(2), 285–314.

33.

Choi

Dulisse

(2019). Behind closed doors: The role of risky lifestyles and victimization experiences on fear of future victimization among South Korean inmates. Journal of Interpersonal Violence. Advance online publication. https://doi.org/10.1177/0886260519888186

34.

Choi

K. S.

Cronin

Correia

(2016). The assessment of capable guardianship measures against bullying victimization in the school environment. Police Practice and Research, 17(2), 149–159.

35.

Cohen

L. E.

Cantor

(1981). Residential burglary in the United States: Lifestyle and demographic factors associated with the probability of victimization. Journal of Research in Crime and Delinquency, 18, 113–127.

36.

Cohen

L. E.

Felson

(1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44, 588–608.

37.

Connell

N. M.

Morris

R. G.

Piquero

A. R.

(2016). Predicting bullying: Exploring the contributions of childhood negative life experiences in predicting adolescent bullying behavior. International Journal of Offender Therapy and Comparative Criminology, 60(9), 1082–1096.

38.

Copeland

W. E.

Keeler

Angold

Costello

E. J.

(2007). Traumatic events and posttraumatic stress in childhood. Archives of General Psychiatry, 64(5), 577–584.

39.

Craig

J. M.

Cardwell

S. M.

Piquero

A. R.

(2017). The effects of criminal propensity and strain on later offending. Crime and Delinquency, 63(13), 1655–1681.

40.

Crenshaw

(1989). Demarginalizing the intersection of race and sex: A Black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics. The University of Chicago Legal Forum, 1(8), 139–167.

41.

Crenshaw

(1991). Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review, 43(6), 1241–1299.

42.

Defoe

I. N.

Farrington

D. P.

Loeber

(2013). Disentangling the relationship between delinquency and hyperactivity, low achievement, depression, and low socioeconomic status: Analysis of repeated longitudinal data. Criminal Justice Journal, 41, 100–107.

43.

Dixon

Howie

Starling

(2005). Trauma exposure, posttraumatic stress, and psychiatric comorbidity in female juvenile offenders. Journal of the American Academy of Child and Adolescent Psychiatry, 44(8), 798–806.

44.

Elliott

D. S.

Ageton

S. S.

Canter

R. J.

(1979). An integrated theoretical perspective on delinquent behavior. Journal of Research in Crime and Delinquency, 16(1), 3–27.

45.

Finkelhor

Ormrod

R. K.

Turner

H. A.

(2009). Lifetime assessment of poly-victimization in a national sample of children and youth. Child Abuse & Neglect, 33, 403–411.

46.

Gould

Pitblado

(Eds.). (n.d.). Guidelines for bootstrap samples: Resource document. STATA. https://www.stata.com/support/faqs/statistics/bootstrapped-samples-guidelines//

47.

Graham-Bermann

S. A.

Castor

L. E.

Miller

L. E.

Howell

K. H.

(2012). The impact of intimate partner violence and additional traumatic events on trauma symptoms and PTSD in preschool-aged children. Journal of Traumatic Stress, 25(4), 393–400.

48.

Hay

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

49.

Hindelang

Gottfredson

Garofalo

(1978). Victims of personal crime: An empirical foundation for a theory of personal victimization. Ballinger Publishing.

50.

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 and Delinquency, 34(1), 46–78.

51.

Hoskin

(2013). Experiencing prejudice and violence among Latinos: A general strain theory approach. Western Criminology, 14(1), 25–38.

52.

Hoyle

R. H.

Panter

A. T.

(1995). Writing about structural equation models. In Hoyle

R. H.

(Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 159–176). SAGE Publications.

53.

Iratzoqui

(2018). Strain and opportunity: A theory of repeat victimization. Journal of Interpersonal Violence, 33(8), 1366–1387.

54.

Isom-Scott

D. A.

(2018). Disentangling the impact of female victimization across racial and ethnic lines. Victims & Offenders, 13(5), 711–737.

55.

Isom-Scott

D. A.

Grosholz

J. M.

(2019). Unpacking the racial disparity in crime from a racialized general strain theory perspective. Deviant Behavior, 40(12), 1445–1463.

56.

Jang

S. J.

Johnson

B. R.

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

57.

Jennings

W. G.

Piquero

N. L.

Gover

A. R.

Pérez

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. Criminal Justice Journal, 37, 404–417.

58.

Jensen

G. F.

Brownfield

(1986). Gender, lifestyles, and victimization: Beyond routine activity. Violence and Victims, 1, 85–99.

59.

Kaufman

J. M.

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

60.

Kaufman

J. M.

Rebellon

C. J.

Thaxton

Agnew

(2008). A general strain theory of racial differences in criminal offending. Australian and New Zealand Journal of Criminology, 41(3), 421–437.

61.

Kim

(2009). Type-specific intergenerational transmission of neglectful and physically abusive parenting behaviors among young parents. Children and Youth Services Review, 31(7), 761–767.

62.

Kim

Y. S.

Koh

Y. J.

Leventhal

(2005). School bullying and suicidal risk in Korean Middle school students. Pediatrics, 115(2), 357–363.

63.

Kort-Butler

L. A.

(2010). Experienced and vicarious victimization: Do social support and self-esteem prevent delinquent responses? Criminal Justice Journal, 38(4), 496–505.

64.

Lauritsen

J. L.

Laub

J. H.

(2007). Understanding the link between victimization and offending: New reflections on an old idea. In Hough

Maxfield

(Eds.), Surveying crime in the 21st century (pp. 55–75). Criminal Justice Press/Willow Tree Press.

65.

Lauritsen

J. L.

Sampson

R. J.

Laub

J. H.

(1991). The link between offending and victimization among adolescents. Criminology, 29, 265–292.

66.

Lee

Kim

(2018). Examining the gendered effect of experienced and vicarious victimization: A general strain theory perspective. American Journal of Criminal Justice, 43, 181–196.

67.

Lin

W. H.

Cochran

J. K.

Mieczkowski

(2011). Direct and vicarious violent victimization and juvenile delinquency: An application of general strain theory. Sociological Inquiry, 81(2), 195–222.

68.

Markus

G. B.

(1979). Analyzing panel data (quantitative applications in the social sciences). SAGE Publications.

69.

Matsueda

R. L.

Anderson

(1998). The dynamics of delinquent peers and delinquent behavior. Criminology, 36, 269–308.

70.

McCarty

Teie

McCutchen

Geller

E. S.

(2016). Actively caring to prevent bullying in an elementary school: Prompting and rewarding prosocial behavior. Journal of Prevention & Intervention in the Community, 44(3), 164–176.

71.

Melde

(2009). Lifestyle, rational choice, and adolescent fear: A test of a risk-assessment framework. Criminology, 47, 781–812.

72.

Menard

Covey

H. C.

Franzese

R. J.

(2015). Adolescent exposure to violence and adult illicit drug use. Child Abuse & Neglect, 42, 30–39.

73.

Moon

Morash

McCluskey

J. D.

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

74.

Mulvey

E. P.

Steinberg

Fagan

Cauffman

Piquero

A. R.

Chassin

Knight

G. P.

Brame

Schubert

C. A.

Hecker

Losoya

S. H.

(2004). Theory and research on desistance from antisocial activity among serious adolescent offenders. Youth Violence and Juvenile Justice, 2, 213–236.

75.

Osgood

D. W.

Wilson

J. K.

O’Malley

P. M.

Bachman

J. G.

Johnston

L. D.

(1996). Routine activities and individual deviant behavior. American Sociological Review, 61(4), 635–655.

76.

Ostrowsky

M. K.

Messner

S. F.

(2005). Explaining crime for a young adult population: An application of general strain theory. Criminal Justice Journal, 33, 463–476.

77.

Ousey

G. C.

Wilcox

Brummel

(2008). Déjà vu all over again: Investigating temporal continuity of adolescent victimization. Journal of Quantitative Criminology, 24(3), 307–335.

78.

Ousey

G. C.

Wilcox

Fisher

B. S.

(2011). Something old, something new: Revisiting competing hypotheses of the victimization-offending relationship among adolescents. Journal of Quantitative Criminology, 27, 53–84.

79.

Ousey

G. C.

Wilcox

Schreck

C. J.

(2015). Violent victimization, confluence of risks and the nature of criminal behavior: Testing main and interactive effects from Agnew’s extension of general strain theory. Criminal Justice Journal, 43, 164–173.

80.

Park

Metcalfe

(2020). Bullying victimization as a strain: Examining changes in bullying victimization and delinquency among Korean students from a developmental general strain theory perspective. Journal of Research in Crime and Delinquency, 57(1), 31–65.

81.

Patchin

J. W.

Hinduja

(2011). Traditional and nontraditional bullying among youth: A test of general strain theory. Youth & Society, 43(2), 727–751.

82.

Peck

J. H.

(2013). Examining race and ethnicity in the context of general strain theory, depression, and delinquency. Deviant Behavior, 34(9), 706–726.

83.

Peterson

Taylor

T. J.

Esbensen

F. A.

(2004). Gang membership and violent victimization. Justice Quarterly, 21, 793–815.

84.

Pinchevsky

G. M.

Fagan

A. A.

Wright

E. M.

(2014). Victimization experiences and adolescent substance use: Does the type and degree of victimization matter? Journal of Interpersonal Violence, 29(2), 299–319.

85.

Piquero

A. R.

Farrington

D. P.

Welsh

B. C.

Tremblay

Jennings

W. G.

(2009). Effects of early family/parent training programs on antisocial behavior and delinquency. Journal of Experimental Criminology, 5(2), 83–120.

86.

Potter

(2008). Battle cries: Black women and intimate partner abuse. New York University Press.

87.

Potter

(2015). Intersectionality and criminology: Disrupting and revolutionizing studies of crime. Routledge.

88.

Radliff

K. M.

Wang

Swearer

S. M.

(2016). Bullying and peer victimization: An examination of cognitive and psychosocial constructs. Journal of Interpersonal Violence, 31(11), 1983–2005.

89.

Runions

K. C.

Shaw

Bussey

Thornberg

Salmivalli

Cross

D. S.

(2019). Moral disengagement of pure bullies and bully/victims: Shared and distinct mechanisms. Journal of Youth and Adolescence, 48, 1835–1848.

90.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177.

91.

Schreck

C. J.

Fisher

B. S.

(2004). Specifying the influence of family and peers on violent victimization: Extending routine activities and lifestyles theories. Journal of Interpersonal Violence, 19, 1021–1041.

92.

Schreck

C. J.

Fisher

B. S.

Miller

J. M.

(2004). The social context of violent victimization: A study of the delinquent peer effect. Justice Quarterly, 21(1), 23–47.

93.

Schreck

C. J.

Stewart

E. A.

Fisher

B. S.

(2006). Self-control, victimization, and their influence on risky lifestyles: A longitudinal analysis using panel data. Journal of Quantitative Criminology, 22, 319–340.

94.

Spohn

R. E.

Wood

S. D.

(2014). Spare the rod, endanger the child? Strain, race/ethnicity, and serious delinquency. Journal of Ethnicity in Criminal Justice, 12, 159–193.

95.

Sturgis

Smith

Berrington

(2004). Using repeated measures data to analyse reciprocal effects: The case of economic perceptions and economic values [Conference session]. Conference Papers From the Department of Sociology, University of Surrey (Vol. 4).

96.

Thaxton

Agnew

(2018). When criminal coping is likely: An examination of conditioning effects in general strain theory. Journal of Quantitative Criminology, 34(4), 887–920.

97.

Thornberry

T. P.

Henry

K. L.

(2013). Intergenerational continuity in maltreatment. Journal of Abnormal Child Psychology, 41(4), 555–569.

98.

Tillyer

M. S.

Fisher

B. S.

Wilcox

(2011). The effects of school crime prevention on students’ violent victimization, risk perception, and fear of crime: A multilevel opportunity perspective. Justice Quarterly, 28, 249–277.

99.

Turanovic

J. J.

Pratt

T. C.

Piquero

A. R.

(2018). Structural constraints, risky lifestyles, and repeat victimization. Journal of Quantitative Criminology, 34(1), 251–274.

100.

Turner

H. A.

Finkelhor

Hamby

S. L.

Shattuck

Ormrod

R. K.

(2011). Specifying type and location of peer victimization in a national sample of children and youth. Journal of Youth and Adolescence, 40, 1052–1067.

101.

Walden

Lemerise

Gentile

(1992). Emotional competence and peer acceptance among preschool children. Paper presented at the Conference on Human Development, Atlanta, GA.

102.

Watts

S. J.

McNulty

T. L.

(2013). Childhood abuse and criminal behavior: Testing a general strain theory model. Journal of Interpersonal Violence, 28(15), 3023–3040.

103.

Wilcox

May

D. C.

Roberts

S. D.

(2006). Student weapon possession and the “fear and victimization hypothesis”: Unraveling the temporal order. Justice Quarterly, 23(4), 502–529.

104.

Young

Johnson

D. R.

(2015). Handling missing values in longitudinal panel data with multiple imputation. Journal of Marriage and Family, 77(1), 277–294.