Trajectories of Exposure to Psychological Domestic Abuse and the Relevance of Differential Development for Predicting Violent Offending in Adulthood

Abstract

This study sought to examine patterns of change and continuity in psychological domestic abuse as a strain-related process that may be predictive of violent offending in adulthood. This study also examined the mediating role of negative affect in this relationship. The Pathways to Desistance data were utilized in analyses. This sample comprised 1,354 juvenile offenders. Group-based trajectory modeling was used to identify latent patterns of development in exposure to psychological domestic abuse. Negative binomial regression models were used to examine the direct effect of trajectory group assignment on violent offending in adulthood and the mediating role of negative affect. Results indicated that a four-group model best fit the data. Assignment to the Declining group was associated with decreased violent offending seriousness in adulthood, relative to assignment to the Accelerating group. Negative affect partially mediated this relationship, accounting for around 35% of magnitude of the direct effect observed.

Keywords

psychological abuse juvenile offenders group-based trajectory modeling intimate partner violence emotional violence

Introduction

Intimate partner violence or domestic violence/abuse is defined as violent behavior perpetrated by an individual against another individual with whom they are engaged in a sexual or romantic relationship of some kind (Breiding et al., 2015). This type of abuse often presents particular issues, as gender and power dynamics involved often lead to patterns of this form of abuse and diminished capacity to exit such relationships (Brosi & Rolling, 2010; Conroy, 2014; Jewkes, 2002). While physical and sexual violence are often the focus of researchers studying relationship violence like this, psychological abuse perpetrated by domestic partners can also be extremely damaging as well. This type of abuse can be characterized by things like belittling one’s partner and/or destroying or getting rid of their partner’s property, among others. Being exposed to these types of abuse can be extremely stressful for victims. Consistent with past research in criminology, exposure to chronic and acute stress like this may have implications for behavioral outcomes also (Hay & Evans, 2006; Lin et al., 2011; Warner & Fowler, 2003). While there is a growing literature on the relationship between psychological abuse and behavior, there remain significant gaps in the literature. There has been little work that has focused on the potential strain effects that psychological domestic abuse may have on the violent offending of victims during adulthood. Further, while some research has taken a developmental perspective in attempting to provide greater understanding of patterns of change and continuity in exposure to psychological domestic abuse, there has yet to be any study which does so among a sample of juvenile offenders. This omission may be particularly problematic, as juvenile offenders present a population that is at particular risk for engagement in violent offending. The present study sought to fill these gaps in the literature by identifying developmental patterns of exposure to psychological domestic abuse and examining the potential that chronic or acute exposure predict increased risk for violent offending in adulthood.

Psychological Domestic Abuse and General Strain Theory

Exposure to psychological domestic abuse has been indicated to be related to the development of numerous mental health problems among victims (Coker et al., 2002; Jones et al., 2001; Romito et al., 2005). It is unsurprising that having a partner who consistently insults and belittles you may increase your risk for developing mental health issues, like depression and anxiety. While it is well established that exposure to physical violence is a risk factor for violent offending (Baskin-Sommers & Baskin, 2016; Fix et al., 2018; Nofziger & Kurtz, 2005), somewhat less attention has been paid to psychological abuse as a potential risk factor predicting violent offending risk. This specific form of abuse within intimate partner relationships may be of great importance. Unlike other forms of relationship abuse, past research has found that males and females report similar risk for exposure to psychological abuse in the context of these relationships (Hines & Saudino, 2003; Romito & Grassi, 2007). This makes psychological domestic abuse a unique form of relationship violence, and it may be necessary to understand the behavioral outcomes associated with such exposure. The limited research on this topic indicates that exposure to such abuse may indeed function to increase risk for violent offending, but that this relationship may be conditioned by gender. Pesta et al. (2019) found that individuals who reported greater masculinity were more likely to react to psychological abuse, perpetrated by an intimate partner with violence, directed at other individuals outside of the relationship. These findings were consistent with Agnew’s (1992) general strain theory. This framework posits that exposure to noxious stimuli, like psychological domestic abuse, results in the arousal of states of negative affect like depression and anger. It is then posited that individuals will choose some type of adaptation to this strain meant to mitigate these undesired mental states. Violent offending has been examined by past research as a behavioral response to strain, as individuals may seek to alleviate anger through engagement in violent behavior (Connolly & Beaver, 2015; Wojciechowski, 2019). As noted earlier, past research on exposure to psychological abuse like this indicates that it may result in the arousal of negative affective states in a manner consistent with general strain theory, but there remains a dearth of research examining this phenomenon from the perspective of this framework. One population that may be particularly relevant for these types of processes, yet remains understudied, is juvenile offenders. Juvenile offenders are at elevated risk for strain exposure (Baglivio et al., 2014). Further, these individuals are at increased risk for precocious sexual debut, meaning that they may be at risk for exposure to psychological abuse by intimate partners at earlier ages by virtue of early entry into such relationships (Farrington, 1991; Ogilvie et al., 1999). This identification of potential for early exposure indicates the importance of taking a developmental perspective on the relationship between psychological domestic abuse also[The meaning of the sentence seems to be unclear. Please consider rephrasing for clarity]. Understanding differences in developmental patterns of exposure may aid in understanding how individuals come to desist from offending across the life-course.

Developmental Implications of Psychological Domestic Abuse and Desistance

Despite the potential importance of understanding the developmental nature of the timing and seriousness of psychological domestic violence exposure, there remains a dearth of research focused on the topic. Little research has taken a developmental perspective for understanding both exposure to and perpetration of intimate partner violence by identifying heterogeneity in patterns of change and continuity of these outcomes across the life-course. Individuals who are in relationships characterized by psychological abuse often have difficulty exiting such relationships for a myriad of reasons (Brosi & Rolling, 2010; Khoury & Wehbi, 2016). Because of this, identifying the relevance of chronic patterns of exposure and its importance for understanding behavioral outcomes should be a paramount concern. Of the two studies identified as seeking to elucidate developmental patterns of intimate partner violence exposure, one of these focused on physical violence (Richards & Gillespie, 2019). This leaves only one study that has examined the development of this construct, using methodologies, which would allow for identification of latent patterns of change and continuity across time. Orpinas et al. (2012) studied differential developments of this phenomenon among a general population sample of high school students. This research identified a three-group model of development, delineating groups of students into low, increasing, and high victimization groups. While the high and low groups maintained a great deal of stability throughout the study period, the increasing group demonstrated large gains in victimization exposure from grades 6 to 12. Despite this, all groups maintained their rank stability throughout the study period. To some extent, this study indicated that those individuals exposed to high levels of psychological domestic abuse also were at greater risk for violence toward peers, though this relationship was potentially confounded by a combining of high perpetration with high victimization as being predictive of this behavioral outcome.

While the highlighted study provided unique insight into the development of this outcome, there remained further gaps in these findings. First, this cross-section of the life-course did not include early adulthood, the period of the life-course that is characterized by the emergence of high salience of serious romantic relationships for offending risk (Capaldi et al., 2008; Sampson & Laub, 1990). This may omit relational ties established during this period of the life-course, which may be strong and potentially expose individuals to chronic and long-term patterns of psychological abuse. Second, this study examined violent offending during adolescence when offending risk is generally at its peak (Kim & Bushway, 2018; Liu, 2015). While examination of this outcome during adolescence is relevant, it is also important to examine continuity of this outcome into adulthood. It is during this period of the life-course that offending risk declines for most individuals. However, some individuals continue to offend after the transition to adulthood and may end up as chronic offenders. This would be consistent with Moffitt’s (1993) life-course persistent offenders, as these individuals demonstrate high offending risk throughout their lives and progress to more serious offending as they age. If individuals face chronic exposure to psychological domestic abuse into adulthood, this may act as a continued strain that may arouse negative affect and require continued adaptation. Violent offending may be one adaptation chosen to this chronic strain in adulthood. Despite these identified mechanisms, there has yet to be any study which examines the potential that exposure to chronic patterns of psychological domestic abuse may result in increased risk for violent offending in adulthood. The present study sought to address these gaps in the literature by testing the following hypotheses:

Hypothesis 1: A pattern of chronic exposure to psychological domestic abuse will be identified.

Hypothesis 2: Exposure to chronic psychological domestic abuse will result in increased violent offending seriousness in adulthood.

Methods

This study utilized data from wave 1 and waves 6 through 11 of the Pathways to Desistance data. These data comprised the responses of 1,354 juvenile offenders followed across seven years, directly after adjudication for a serious offense. Serious offenses, which qualified participants for inclusion, consisted of all felony offenses and serious misdemeanor charges of weapons-related charges and sexual assault. Participants were recruited from study sites in Maricopa County, Arizona, and Philadelphia, Pennsylvania. Attrition from the study reached its peak during the final wave of data collection, with nearly 85% of the original sample being retained. Of all qualified juvenile offenders who were approached during the recruitment process, 20% declined the opportunity to take part in the study. Given that this sample was collected purposively, a cap of 15% of the total sample was applied to male drug offenders in order to ensure sample heterogeneity.

The final sample used in analyses was slightly smaller than the full Pathways to Desistance sample because of more stringent inclusion criteria necessitated by the purpose of this study. Females were excluded from the final sample, leaving only male juvenile offenders as the population of interest. This was done because research has indicated that experiences and development of domestic violence differ substantially by gender (Richards & Branch, 2012; Tillyer & Wright, 2014). This would indicate the importance of modeling males and females separately. However, there were only 184 females included in the original sample. This sample size was insufficient to conduct the necessary analyses, so the final sample used in this study consisted only of males. Further restriction of the final sample was based on the availability of data. Only participants who provided data during at least two of the five waves of outcome data used in trajectory analysis were included in the final sample, because this is the minimum amount of data necessary to study change across time. Furthermore, this approach is consistent with past research, examining domestic violence trajectories using this sample (Richards & Gillespie, 2019).

Data used in the present study were collected via participant self-report. Participants were interviewed by members of the Pathways to Desistance research team. Participants were provided a laptop by the research team, which they used to input responses to verbal prompts administered by the research team. It was believed that maximizing the confidentiality allowed in reporting by using this method that participants would be more honest in reporting. Interviews were conducted in locations that were convenient for the participant, such as criminal justice facilities, participants’ homes, and libraries.

Measures

Violent offending

The main dependent variable of the present study was violent offending seriousness at the age of 23 years. This was measured as a variety score of the number of types of aggressive offenses that participants reported engagement in during the observation period prior to the age of 23 years divided by the total number of violent offense types that participants provided data for. Variety scores like these are useful for measuring overall offending seriousness, as they do not suffer from issues related to recall bias that are often observed with frequency measures of offending (Brame et al., 2004; Steinberg et al., 2015).¹

Psychological domestic abuse victimization

The main independent variable of this study was degree of psychological domestic abuse victimization experienced by participants at waves 4 through 11. This construct was measured using the Domestic Violence Inventory. This scale measured the degree to which participants reported experiencing a series of different forms of psychological abuse by an intimate partner during the previous observation period. The measures used in analyses comprise a count of the total number of types of psychological abuse that participants reported experiencing during a given observation period.²

Negative affect

The role of negative affect in mediating the effects of psychological abuse victimization was also examined in this study. The Brief Symptom Inventory was utilized to assess the degree to which a series of symptoms had bothered participants in the previous week. Individual ordinal items were utilized to assess the presence of hostility symptoms. A mean score was then computed from these items to provide each participant with a single negative affect score at the age of 23 years. Hostility was chosen as the negative affective state of interest because of its proximity to anger, and the relevance that anger has been identified to have for motivating offending (Agnew, 2001).

Control variables

Several important control variables were included in analyses in order to mitigate bias in the estimation of psychological abuse victimization during adulthood. Racial identification was the first of these control variables, as research indicates that violent offending risk may be stratified by race (Hawkins et al., 2000). Further, it is necessary to include a race variable in order to better account for diversity in experiences of exposure to psychological domestic abuse. A four-category nominal variable was used to measure race at baseline, delineating participants into White, Black, Hispanic, and Other Race categories. Four dummy variables were computed from this original variable, with each delineating all participants assigned to a particular race category from all other participants (e.g., 1 = Black; 0 = all other participants). The dummy variable corresponding to White participants was excluded from analyses in order to provide an omitted reference group to understand coefficient effects in comparison to[The meaning of the sentence seems to be unclear. Please consider rephrasing for clarity.

Socioeconomic status (SES) was included as a control variable also. This is because past research indicates that risk for violent offending may differ by social class (Piotrowska et al., 2015). SES was measured as a computed scale score comprising participants’ parents’ occupational prestige combined with parents’ educational attainment. If both parents could provide data for this construct, a mean of the two scores was computed so that each participant had a single SES score at baseline.

Exposure to violence during the study period to measure participants’ exposures prior to the age of 23 years was also included as a control variable. This is because exposure to violence has been identified as a risk factor predicting violent offending (Franzese et al., 2017). Two forms of exposure to violence were included as control variables in analyses: direct victimization and witnessed violence. Both of these variables were measured as count variables, capturing the number of different types of each exposure to violence reported by participants during the observation period prior to the age of 23 years.

Social support has been found to be a protective factor related to the development of mental health issues (World Health Organization, 2014), necessitating its inclusion as a control covariate in analyses. Social support at the age of 23 years was measured as a count of the number of unique caring adults who participants reported provided them with social support in two or more domains. This eschewed more superficial support relationships, characterized by the provision of support in only one domain.

Another control variable included in analyses was self-control, as research has identified low self-control as a risk factor predicting violent offending (Larson et al., 2015). The Weinberger Adjustment Inventory was used to measure self-control at the age of 23 years. A series of hypothetical vignettes asked participants to rate the degree that eight different behaviors were true of their own behavior using an ordinal scale. Seven of the eight items were reverse coded so that lower scores corresponded to lower self-control. A mean score was computed so that each participant had a single self-control score at the age of 23 years.

Deviant peer association was also included as a control variable, as research has identified this variable as an important predictor of violent offending (Haggerty et al., 2013). Deviant peer association at the age of 23 years was measured using seven individual ordinal items, asking participants about the general number of peers who attempted to influence them to engage in a series of antisocial behaviors. A mean score was then computed from these individual items so that each participant had a single deviant peer association score at the age of 23 years.

Another important variable included in analyses as a control on bias in estimation was the number of days in the observation period leading up to measurement at the age of 23years. Given that the dependent variable of interest was a variety score of violent offending, it is logical that having a longer observation period would naturally result in increased risk that more types of violent offenses may be engaged in than someone with a shorter observation period. This variable was measured as a count of the number of days in each participant’s observation period leading up to measurements at the age of 23 years.

The final control variable included in analyses was violent offending seriousness at baseline. This variable was included because there may be continuity of violent behavior stemming from earlier in the study period, which may influence violent offending risk later on. This variable was measured in the same manner as the dependent variable in this study, only that it was captured at baseline measurements.

Analytic Strategy

Analyses for the present study proceeded in two distinct phases. The first phase of analyses utilized group-based trajectory modeling (GBTM) to identify heterogeneity in developmental patterns of exposure to psychological domestic abuse. The GBTM method entails an iterative process of fixing varying numbers and orders of polynomial functions to a set of longitudinal outcome data. Based on the distribution of the outcome data, a choice of probability distribution guiding analyses is made between censored normal, zero-inflated Poisson, and logit. Nested model fit is ascertained using Bayesian Information Criteria (BIC) statistics. As the number of groups added the model changes, and they are each modified by higher/lower function order, model fit will also be altered. Participants are assigned a probability of assignment to each group in a model based on their own individual response trajectory. They are then actually assigned membership to the group to which they have the highest probability of being assigned. The model that demonstrates the best fit to the data based on BIC is chosen as the best fitting model, so long as it meets a number of other selection criteria, as outlined by Nagin (2005). The first additional criterion that a model should meet if it is to be chosen as the best fitting model is that average posterior probabilities of assignment for each trajectory group should exceed 0.7. This metric relates to the average probability of assignment of participants actually assigned to a group. If this exceeds 0.7, it is indicative of individuals assigned to that group having a high probability of being assigned to that group and to none of the others in the model. Another metric used to determine fitness of the model is average odds of correct classification. This indicates the improvement in the odds of correctly selecting the group to which an individual participant was actually assigned, based on their probability of assignment, relative to random selection. These statistics should exceed 5 for each group, representing a 500% improvement in odds. All trajectory groups in a given model should also have 95% confidence intervals, relatively tightly bound around each of them. A final consideration should be given to whether or not adding more groups to the model adds nuance to the heterogeneity. Groups that are added to a model should be distinct from one another or eschewed in favor of a more parsimonious solution. Full information maximum likelihood estimation was utilized to manage missing data in analyses. This results in imputation of missing data based on which trajectory group participants are assigned to, as there is assumed to be no within-group variance for all trajectory groups in a model.

The second phase of analyses utilized a series of negative binomial regression to understand the influence of psychological domestic abuse development on offending seriousness in adulthood. A series of dummy variables were computed from the trajectory groups in analyses, with one variable corresponding to each group in the model. One dummy variable was then omitted from analyses in order to provide a reference category to understand coefficient values in comparison to. While Poisson regression is often used to analyze dependent variables with a notable right-skew, as is the case with the offending seriousness variable, distributional issues led to this not being the optimum model. In order to account for a high degree of dispersion causing a highly different mean and variance for the dependent variable, the negative binomial model was chosen to account for these issues. The first model examined the direct effect of the psychological domestic abuse control variables on violent offending seriousness in adulthood, net of control covariates. The second model included negative affect in the model to examine the potential mediating role of this construct. Coefficient values represent the impact that one unit increases in an independent variable have on the log mean increase in the dependent variable.

Results

The GBTM analytic process resulted in the identification of a four-group model as providing optimum fit to the data. This model provided better fit based on BIC than did the two- and three-group models. While a five-group model provided better fit to the data than a four-group model, based on BIC, there were issues with posterior probabilities not exceeding 0.7 for multiple groups. This problem remained even when varying the order of polynomial functions in the model in numerous ways. For this reason, the four-group model was ultimately chosen as the best fitting model. Due to the distribution of the dependent variable, the zero-inflated Poisson model was chosen as the probability distribution used to model the identified trajectories.

Figure 1 describes the psychological domestic abuse trajectory model identified as best fitting the longitudinal response data. The first group in the model was characterized by a stable course of experiencing no psychological domestic abuse throughout the entire study period. This group was characterized by a linear polynomial function, and 26.9% of the sample was assigned membership to this group. This group is described as the “No Abuse” group. The second group in the model was characterized by relatively low levels of experiencing psychological domestic abuse across the entire study period. This group was described as the “Low Abuse” group, 34.5% of the sample was assigned membership to this group, and it was characterized by a quadratic polynomial function. The third group in the model was characterized by steady declines in exposure to psychological domestic abuse across the entire study period, reaching near-zero levels by 23 years of age. Because of this developmental track, this group is described as the “Declining” trajectory group. This group was characterized by a quadratic polynomial function, and 18.8% of the sample was assigned membership to this group. The final group in the model was characterized by steady increases in exposure to psychological domestic violence across the entire study period. This group is described as the “Accelerating” trajectory group. This group was characterized by a quadratic polynomial function, and 19.9% of the sample was assigned membership to this group.

Figure 1.

Psychological domestic abuse exposure trajectory groups.

Table 1 provides descriptive statistics pertaining to variables included in regression analyses. Table 2 provides results from model 1 of the negative binomial regression modeling. Table 3 provides model 2 results. The Accelerating group was omitted from both models as the reference group. Model 1 results indicate that participants assigned to the Decelerating group reported significantly lower violent offending seriousness in adulthood than the Accelerating group (coefficient = −1.344). Experiencing greater variety of direct victimization, witnessing a greater variety of violence, and having greater deviant peer association predicted greater violent offending seriousness in adulthood. Model 2 results indicated that the inclusion of negative affect in the model resulted in nearly 35% mediation of the magnitude of the Declining trajectory group effect observed in model 1. However, the difference in violent offending seriousness between the two groups remained significant in this model also (coefficient = −1.230). Having greater deviant peer association was associated with increased violent offending seriousness during adulthood in this model.³

Table 1.

Descriptive Statistics.

	Mean/Proportion	Standard Deviation	Minimum	Maximum
Violent offending seriousness at the age of 23	4.428	8.484	0	55
Negative affect at the age of 23	0.425	0.591	0	3.2
Race
Black	0.419	NA	0	1
White	0.191	NA	0	1
Hispanic	0.347	NA	0	1
Other Race	0.043	NA	0	1
Socioeconomic status	51.543	12.087	11	77
Direct victimization at the age of 23	0.189	0.593	0	4
Witnessed violence at the age of 23	1.031	1.445	0	7
Social support at the age of 23	0.670	0.582	0	2
Self-control at the age of 23	3.343	0.953	1	5
Deviant peer association at the age of 23	0.708	0.739	0	4
Recall days in observation period	361.610	30.920	272	427
Violent offending seriousness at baseline	17.078	14.634	0	73

Table 2.

Negative Binomial Regression Covariate Effects of Violent Offending Seriousness at Age 23: Model 1.

	Coefficient	p-Value	95% Confidence Interval
Psychological domestic abuse trajectory group assignment (omitted = accelerating)
No abuse	0.116	0.788	−0.728	0.960
Low	0.205	0.246	−0.141	0.551
Declining	−1.344	p < 0.001	−2.078	−0.611
Race (omitted = White)
Black	−0.374	0.316	−1.104	0.357
Hispanic	−0.346	0.369	−1.101	0.409
Other race	−1.205	0.082	−2.565	0.154
Socioeconomic status	−0.015	0.200	−0.037	0.008
Direct victimization at the age of 23	0.699	0.007	0.195	1.202
Witnessed violence at the age of 23	0.297	0.003	0.099	0.494
Social support at the age of 23	0.177	0.405	−0.240	0.595
Self-control at the age of 23	−0.257	0.082	−0.546	0.032
Deviant peer Association at the age of 23	0.911	p < 0.001	0.513	1.310
Recall days in observation period	<0.001	0.993	−0.008	0.008
Violent offending seriousness at baseline	1.542	0.087	−0.226	3.309
Constant	1.073	0.550	−2.443	4.588

Table 3.

Negative Binomial Regression Covariate Effects of Violent Offending Seriousness at Age 23: Model 2.

	Coefficient	p-Value	95% Confidence Interval
Psychological domestic abuse trajectory group assignment (omitted = accelerating)
No abuse	0.085	0.869	−0.925	1.095
Low	0.191	0.302	−0.172	0.555
Declining	−1.230	0.001	−1.981	−0.480
Negative affect at the age of 23	0.483	0.085	−0.067	1.032
Race (omitted = White)
Black	−0.434	0.300	−1.255	0.387
Hispanic	−0.385	0.377	−1.238	0.469
Other Race	−0.500	0.528	−2.056	1.055
Socioeconomic status	−0.016	0.211	−0.041	0.009
Direct victimization at the age of 23	0.462	0.059	−0.018	0.941
Witnessed violence at the age of 23	0.140	0.184	−0.067	0.347
Social support at the age of 23	0.012	0.962	−0.468	0.491
Self-control at the age of 23	−0.176	0.309	−0.514	0.163
Deviant peer association at the age of 23	0.928	p < 0.001	0.475	1.381
Recall days in observation period	−0.002	0.636	−0.012	0.007
Violent offending seriousness at baseline	1.237	0.224	−0.755	3.230
Constant	1.912	0.353	−2.119	5.942

Discussion

Results from the present study provide unique insight into the relevance of psychological domestic abuse as a form of strain predicting offending risk. The Declining group demonstrated significantly lower violent offending seriousness in adulthood than the Accelerating group. This effect was partially mediated by differences in negative affect at the age of 23 years. Both of these findings are consistent with the postulations of general strain theory. There are numerous relevant practical and theoretical implications of these results of interest to criminal justice professionals and criminologists.

These results provide some support for the theory that psychological domestic abuse may act as strain that may influence violent offending seriousness in adulthood via the mechanism of increased negative affect. This is clear in the differences observed with the Declining and Accelerating groups. However, these findings actually indicate some mixed findings about the nature of this relationship. Though the highlighted results are consistent with general strain theory processes, there were no significant differences in violent offending seriousness in adulthood between the Accelerating group and the No Abuse group. This is surprising considering that, generally, the No Abuse group reported no exposure to psychological domestic violence strain throughout the study period and had about the same level of exposure at the age of 23 years as the Declining group. Because of this, it would be expected that the No Abuse group would report significantly lower violent offending seriousness in adulthood. The fact that only remission from this form of strain exposure was associated with the behavioral outcome of interest may indicate that additional nuance is necessary to understand the experience of strain. Laub and Sampson’s (1993) study on turning points may provide a potential mechanism explaining these mixed findings. Turning points, like getting married or obtaining stable employment, are events in the life-course, which allow individuals to transition from a deviant lifestyle to a more normative one. It may be that actually leaving abusive situations like the ones that would be characterized by elevated levels of exposure to psychological domestic abuse functions as a turning point that facilitates desistance. Individuals who were never exposed to this form of strain may have been exposed to other risk factors influencing their violent offending and may have not had such a turning point to facilitate their desistance. Post hoc analyses appear to support this proposition. Analysis of variance was used to examine trajectory group differences in violent offending seriousness at the age of 19 years, the beginning age of trajectories in the trajectory model. These results indicate that participants assigned to the Declining group reported the greatest violent offending seriousness at this age, providing evidence that psychological domestic abuse functioned in a manner similar to strain at this age. While these bivariate analyses do not provide enough evidence to make a causal argument, they do provide reason to investigate this relationship further. It may be that psychological domestic abuse certainly functions in a manner consistent with strain, but exiting relationships characterized by this type of abuse may act as a turning point in the life-course. If this is what is driving the criminal behavior of individuals, then removal of that strain may, indeed, present the solution to the underlying issue. Future research should investigate this relationship in greater depth. This can be done both using qualitative research to understand the experiences of individuals during and after exposure to psychological domestic abuse and through further analyses of quantitative data to understand the longitudinal relationships between these constructs, following the turning point of exiting abusive relationships.

There is evidence here to support the assertion that exit from psychologically abusive relationships can aid in the desistance process. In terms of programming measures that may be informed by these results, this process may begin by simply raising awareness about what constitutes psychologically abusive behavior in intimate partner relationships. Despite the deleterious effects that such abuse may have, research indicates that domestic violence can sometimes suffer from a definitional issue, as the perception of the behavior in certain circumstances is not uncommon (Canavire-Bacarreza & Rios Avila, 2010; Yount & Li, 2009). By raising awareness about the presence of such abuse, the damage such abuse may cause, and the options that survivors may have, we may observe change in relationship exit behaviors, and this may subsequently influence desistance. Considering the climate of the digital era, the use of technology may be particularly helpful for raising awareness. Research has indicated that adolescents are amenable to smartphone apps being used to raise awareness about these types of dating violence issues (Willis & Bianchi, 2019). The use of such technology may prove useful in attempting to raise awareness so that these types of issues may be identified and dealt with so that they do not continue to act as strain, resulting in increased risk for violent offending.

While this study provided some unique insights into the role that exposure to psychological domestic abuse has for predicting violent offending in adulthood, there remain a number of important limitations. The first limitation pertains to the method used to elucidate developmental patterns of exposure to psychological domestic abuse. While the GBTM method provides a useful tool for providing tractability to aggregate longitudinal response data in a useful manner, the identified trajectory groups should not be viewed as concrete entities that all participants assigned membership actually follow in lockstep. Rather, they should be viewed as relatively rough approximations of developmental patterns that not only are a useful heuristic but also have some degree of error. For these reasons, all interpretations of this and other GBTM models should be done with great caution. Another limitation of the present study relates to the fact that the sample used in analyses consisted solely of males. Because of the limited number of females in the Pathways to Desistance sample, separate GBTM analyses by gender could not be estimated. This left only the males in the sample to estimate these trajectories. Knowing that research indicates that males and females develop in their domestic violence outcomes differently (Richards & Branch, 2012; Tillyer & Wright, 2014), this limits the generalizability of the results of this study. However, while these limitations of using a male-only sample are important, it is also relevant to note that recent research has indicated that victims of psychological domestic abuse that are more masculine are at greater risk for responding to this strain with other-direct violence (Pesta et al., 2019). For this reason, a male sample may hold some special relevance for examining this relationship. While conflation of the gender concept of masculinity and biological male sex is somewhat problematic, it is relatively safe to assume that males in this sample were likely to be more masculine on average, given well-established gender norms of the United States. Additionally, while there are reasons why examining this phenomenon among a sample of juvenile offenders is quite relevant, these results cannot be extrapolated to other populations because of the indicated nature of this sample. In all, these results are quite useful in many ways, but they remain niche in their applicability. Future research should seek to test the robustness of these results, as they relate to the relationship between differential exposure to psychological domestic abuse across time and violent offending risks in adulthood.

While the limitations related to the methods and generalizability of this study temper these findings, the insights identified in analyses remain relevant. These results indicate that declining exposure to psychological domestic violence results in decreased violent offending seriousness in adulthood compared to males who accelerate in their exposure during this same period and even compared to males who experienced no exposure to this type of abuse during the same period. This may indicate that exiting psychologically abusive relationships may act as a turning point for male juvenile offenders, resulting in deceleration and desistance from violent offending. Reasons for differences between accelerating and declining exposure seem to be consistent with strain processes, as individuals exposed to strain to greater extents at certain periods of the life-course mitigate negative affect aroused by strain through violence. The violent offending of other male juvenile offenders who were never exposed to this type of abuse may have been driven by other factors, and identification of the etiology of these individuals’ violent offending bears further examination. The violent offending seriousness of those male juvenile offenders who consistently reported no exposure to psychological domestic abuse did not significantly differ in adulthood from those who reported accelerating exposure. This indicates that some factor other than the strain of psychological abuse is driving the violent offending of these individuals. Future research should seek to identify potential risk factors that may be driving the violent offending of male juvenile offenders who are not exposed to any psychological domestic abuse.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Thomas Wojciechowski

Notes

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30(1), 47–88.

Agnew

(2001). Building on the foundation of general strain theory: Specifying the types of strain most likely to lead to crime and delinquency. Journal of Research in Crime and Delinquency, 38(4), 319–361.

Baglivio

M. T.

Epps

Swartz

Huq

M. S.

Sheer

Hardt

N. S.

(2014). The prevalence of adverse childhood experiences (ACE) in the lives of juvenile offenders. Journal of Juvenile Justice, 3(2), 1.

Baskin-Sommers

A. R.

Baskin

(2016). Psychopathic traits mediate the relationship between exposure to violence and violent juvenile offending. Journal of Psychopathology and Behavioral Assessment, 38(3), 341–349.

Brame

Fagan

Piquero

A. R.

Schubert

C. A.

Steinberg

(2004). Criminal careers of serious delinquents in two cities. Youth Violence and Juvenile Justice, 2(3), 256–272.

Breiding

Basile

K. C.

Smith

S. G.

Black

M. C.

Mahendra

R. R.

(2015). Intimate partner violence surveillance: Uniform definitions and recommended data elements (Version 2.0). CDC.

Brosi

M. W.

Rolling

E. S.

(2010). A narrative journey for intimate partner violence: From victim to survivor. The American Journal of Family Therapy, 38(3), 237–250.

Canavire-Bacarreza

G. J.

Rios Avila

(2010). Domestic violence and labor market outcomes: Evidence from a mixed-race developing country. International Journal of Social Economics, 4(1), 1–36

Capaldi

D. M.

Kim

H. K.

Owen

L. D.

(2008). Romantic partners’influence on men’s likelihood of arrest in early adulthood. Criminology, 46(2), 267–299.

10.

Coker

A. L.

Davis

K. E.

Arias

Desai

Sanderson

Brandt

H. M.

Smith

P. H.

(2002). Physical and mental health effects of intimate partner violence for men and women. American Journal of Preventive Medicine, 23(4), 260–268.

11.

Connolly

E. J.

Beaver

K. M.

(2015). Assessing the salience of gene–environment interplay in the development of anger, family conflict, and physical violence: A biosocial test of General Strain Theory. Journal of Criminal Justice, 43(6), 487–497.

12.

Conroy

A. A.

(2014). Gender, power, and intimate partner violence: A study on couples from rural Malawi. Journal of Interpersonal Violence, 29(5), 866–888.

13.

Farrington

D. P.

(1991). Childhood aggression and adult violence: Early precursors and later life outcomes. The Development and Treatment of Childhood Aggression, 5, 29.

14.

Fix

R. L.

Alexander

A. A.

Burkhart

B. R.

(2018). From family violence exposure to violent offending: Examining effects of race and mental health in a moderated mediation model among confined male juveniles. International Journal of Offender Therapy and Comparative Criminology, 62(9), 2567–2585.

15.

Franzese

R. J.

Menard

Weiss

A. J.

Covey

H. C.

(2017). Adolescent exposure to violence and adult violent victimization and offending. Criminal Justice Review, 42(1), 42–57.

16.

Haggerty

K. P.

Skinner

M. L.

McGlynn

Catalano

R. F.

Crutchfield

R. D.

(2013). Parent and peer predictors of violent behavior of Black and White teens. Violence and Victims, 28(1), 145.

17.

Hawkins

D. F.

Laub

J. H.

Lauritsen

J. L.

Cothern

(2000). Race, ethnicity, and serious and violent juvenile offending. US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

18.

Hay

Evans

M. M.

(2006). Violent victimization and involvement in delinquency: Examining predictions from general strain theory. Journal of Criminal Justice, 34(3), 261–274.

19.

Hines

D. A.

Saudino

K. J.

(2003). Gender differences in psychological, physical, and sexual aggression among college students using the Revised Conflict Tactics Scales. Violence and Victims, 18(2), 197.

20.

Jewkes

(2002). Intimate partner violence: Causes and prevention. The Lancet, 359(9315), 1423–1429.

21.

Jones

Hughes

Unterstaller

(2001). Post-traumatic stress disorder (PTSD) in victims of domestic violence: A review of the research. Trauma, Violence, & Abuse, 2(2), 99–119.

22.

Khoury

Wehbi

(2016). Leaving a violent domestic relationship: Experiences of women in Lebanon. International Social Work, 59(1), 73–85.

23.

Kim

Bushway

S. D.

(2018). Using longitudinal self-report data to study the age–Crime relationship. Journal of Quantitative Criminology, 34(2), 367–396.

24.

Larson

Vaughn

M. G.

Salas-Wright

C. P.

Delisi

(2015). Narcissism, low self-control, and violence among a nationally representative sample. Criminal Justice and Behavior, 42(6), 644–661.

25.

Laub

J. H.

Sampson

R. J.

(1993). Turning points in the life course: Why change matters to the study of crime. Criminology, 31(3), 301–325.

26.

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.

27.

Liu

(2015). Is the shape of the age-crime curve invariant by sex? Evidence from a national sample with flexible non-parametric modeling. Journal of Quantitative Criminology, 31(1), 93–123.

28.

Moffitt

T. E.

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

29.

Nagin

(2005). Group-based modeling of development. Harvard University Press.

30.

Nofziger

Kurtz

(2005). Violent lives: A lifestyle model linking exposure to violence to juvenile violent offending. Journal of Research in Crime and Delinquency, 42(1), 3–26.

31.

Ogilvie

E. L.

Veit

Crofts

Thompson

S. C.

(1999). Hepatitis infection among adolescents resident in Melbourne Juvenile Justice Centre: Risk factors and challenges. Journal of Adolescent Health, 25(1), 46–51.

32.

Orpinas

Nahapetyan

Song

McNicholas

Reeves

P. M.

(2012). Psychological dating violence perpetration and victimization: Trajectories from middle to high school. Aggressive Behavior, 38(6), 510–520.

33.

Pesta

Peralta

R. L.

Novisky

M. A.

(2019). Utilizing a general strain framework to examine behavioral responses to psychological intimate partner violence: Are responses gendered? Journal of Interpersonal Violence, 34(15), 3171–3198.

34.

Piotrowska

P. J.

Stride

C. B.

Croft

S. E.

Rowe

(2015). Socioeconomic status and antisocial behaviour among children and adolescents: A systematic review and meta-analysis. Clinical Psychology Review, 35, 47–55. https://dx.doi.org/10.1016/j.cpr.2014.11.003

35.

Richards

T. N.

Branch

K. A.

(2012). The relationship between social support and adolescent dating violence: A comparison across genders. Journal of Interpersonal Violence, 27, 1540–1561.

36.

Richards

T. N.

Gillespie

L. K.

(2019). Intimate partner violence perpetration, victimization, and overlap among serious juvenile offenders: Trajectories of emerging adulthood. Journal of Interpersonal Violence. https://dx.doi.org/0886260519881000

37.

Romito

Grassi

(2007). Does violence affect one gender more than the other? The mental health impact of violence among male and female university students. Social Science & Medicine, 65(6), 1222–1234.

38.

Romito

Turan

J. M.

De Marchi

(2005). The impact of current and past interpersonal violence on women’s mental health. Social Science & Medicine, 60(8), 1717–1727.

39.

Sampson

R. J.

Laub

J. H.

(1990). Crime and deviance over the life course: The salience of adult social bonds. American Sociological Review, 55, 609–627.

40.

Steinberg

L. D.

Cauffman

Monahan

(2015). Psychosocial maturity and desistance from crime in a sample of serious juvenile offenders. US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

41.

Tillyer

M. S.

Wright

E. M.

(2014). Intimate partner violence and the victim offender-overlap. Journal of Research in Crime and Delinquency, 51, 29–55.

42.

Warner

B. D.

Fowler

S. K.

(2003). Strain and violence: Testing a general strain theory model of community violence. Journal of Criminal Justice, 31(6), 511–521.

43.

Willis

S. M.

Bianchi

A. L.

(2019). Increasing knowledge and awareness about dating violence among college students through a smartphone application. https://sigma.nursingrepository.org/handle/10755/16943

44.

Wojciechowski

T. W.

(2019). Victimization recency, development of anger, and violent offending in early adulthood: A developmental test of general strain theory. Deviant Behavior, 40(7), 866–881.

45.

World Health Organization. (2014). Social determinants of mental health. World Health Organization.

46.

Yount

K. M.

(2009). Women’s “justification” of domestic violence in Egypt. Journal of Marriage and Family, 71(5), 1125–1140.