Perpetrator Type Moderates the Relationship Between Severity of Intimate Partner Violence and Recidivism

Abstract

The present study investigated the predictive utility of violence severity ratings on recidivism based on behavior-based subtypes (family only [FO] violent and generally violent [GV]) of intimate partner violence (IPV) perpetrators. Participants consisted of 328 men between the ages of 17 to 72 sentenced to probation in Lake County, Illinois between 2006 and 2008. The relationship between ratings of violence severity for the arresting event, based on victims’ and perpetrators’ accounts to responding police officers, and domestic violence recidivism for a 3-year postprobation completion/termination period was examined. Utilizing victims’ accounts, the Kaplan–Meier log rank test revealed a significant main effect for violence severity. In addition, perpetrator type moderated the relationship between violence severity and postprobation recidivism, such that a positive association was found for GV men but not for FO violent men. Results corroborate the predictive utility of assessing violence severity at the arresting event, particularly within GV men of IPV.

Keywords

intimate partner violence domestic violence severity of violence family only violent generally violent recidivism

Intimate partner violence (IPV) represents a major public health concern in the United States. It is strongly associated with physical health problems, depression, psychological distress, injury, substance abuse, and marital dissolution (Coker et al., 2002). Despite reports that the prevalence of IPV decreased substantially between 1994 and 2002 (Catalano, 2012; Truman & Morgan, 2014), IPV occurrences remain relatively high at 6.9 million annual cases of intimate partner-related stalking, physical assaults, and rapes against women and 5.6 million annual cases of these kinds of violence against men (Black et al., 2011). Furthermore, health-related costs of IPV exceed US$5.8 billion each year (National Center for Injury Prevention and Control, 2003). Thus, both the societal impact and financial costs of IPV are immense (Sartin, Hansen, & Huss, 2006).

In recent years, researchers and policy makers have begun to emphasize the heterogeneity of IPV. Two issues that have been identified as critical for understanding the nature and impact of IPV are type of perpetrator and severity of violence. As reviewed below, multiple studies have reported that differences between subtypes of perpetrators and distinctions in severity of violence are predictors of future IPV. However, no known studies have examined the predictive utility of both perpetrator type and severity of violence in the same sample. Consequently, there is no evidence to indicate whether or not they interact in predicting IPV recidivism.

Two theoretical perspectives presented in the current paper suggest that it is worthwhile to examine the potential interaction of perpetrator type and severity of violence on future IPV. One perspective is based on the contention that attachment-related distress is a mechanism underlying the severity of violence among some family only (FO) perpetrators, whereas the alternative perspective is rooted in social control theory of recidivism. The present study addresses a gap that exists in the IPV literature, specifically the absence of empirical studies critically assessing these theoretical perspectives within the context of both perpetrator type and severity of violence. Significant findings could provide preliminary evidence to inform policy and treatment recommendations for high-risk IPV perpetrators in the criminal justice system with the ultimate goal of promoting victim safety.

In addition, prior studies that have utilized self-report as an index of violence severity for an arresting incident have relied on either victim or perpetrator accounts. To date, no known studies have examined both victim and perpetrator accounts of severity of violence at the time of the arresting event and assessed their predictive utility. We contend that consideration of both victims’ and perpetrators’ accounts represents a more comprehensive approach to this field of study.

Although traditional perpetrator interventions suggest that male perpetrators represent a homogeneous group who exercise coercive control over victims (Pence & Paymar, 1993), there is now substantial evidence that IPV perpetrators are heterogeneous (D. G. Dutton & Corvo, 2006). A growing body of evidence indicates that both distal (e.g., family of origin violence, association with delinquent peers, head injury) and proximal (e.g., attitudes supportive of violence, impulsivity) variables differentiate IPV perpetrators. Moreover, several researchers have proposed that there are distinct subtypes of IPV perpetrators (Gondolf, 1988; Hamberger & Hastings, 1985; Holtzworth-Munroe & Stuart, 1994). These perpetrator subtypes have been shown to differ with regard to criminal histories and likelihood of recidivism (Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2003).

In addition, severity of violence has been identified as an important factor in understanding IPV (Woodin & O’Leary, 2006). Although IPV of any degree is serious, a distinction has often been made between “minor” violence (p. 308) or less severe violence, which refers to relatively more frequently occurring forms of physical aggression such as slapping and pushing, versus “severe” violence (p. 308), less frequently occurring and more physically harmful forms such as hitting with objects and the use of lethal weapons (Straus, Hamby, Boney-McCoy, & Sugarman, 1996). Severity of IPV has been associated with poorer mental health and quality-of-life outcomes in victims (M. A. Dutton, Kaltman, Goodman, Weifurt, & Vankos, 2005; Hegarty et al., 2012). Furthermore, several longitudinal studies have found that initial severity of violence is a valid predictor of future IPV (Aldorando, 1996; Feld & Straus, 1989; Quigley & Leonard, 1996).

Violence severity has also been linked to a contemporary theory of different types of IPV. Kelly and Johnson (2008) proposed that there are four qualitatively distinct kinds of violence in couples. Coercive controlling violence, described as a pattern of coercion, intimidation, and control in conjunction with physical violence (Kelly & Johnson, 2008), has been proposed as the most common type of violence occurring among criminal justice populations (Frieze & Browne, 1989; Johnson, 2006). Situational couple violence, which results from conflict situations that escalate to physical violence (Leone, Johnson, Cohan, & Lloyd, 2004), also accounts for a large percentage of IPV incidents (Johnson, 2006). As compared with situational couple violence, coercive controlling violence has been associated with more severe acts of IPV (Johnson, 2006) and more injuries to victims (Johnson & Leone, 2005). Thus, the context of the violent episodes themselves seems to affect the subsequent severity of partner violence.

Interaction Between Severity of Violence and Perpetrator Type

There is reason to believe that perpetrator type may moderate the impact of severity of violence on IPV recidivism. Two types of interactions can be conceptualized. One perspective suggests that the relationship between violence severity and IPV recidivism should be stronger for FO violent men as compared with generally violent (GV) men. There are two different arguments consistent with this perspective. The first rationale is based on the proposition that attachment disturbances and attachment-related distress are mechanisms underlying the severe violence of some FO perpetrators. D. G. Dutton (1998) has argued that early traumatic experiences can lead to insecure attachment, emotional volatility, personality pathology, and high levels of distress. In turn, studies have reported trends indicating that borderline-dysphoric/impulsive perpetrators engage in greater frequency and severity of aggression against intimate partners than other violent perpetrators within the FO category as defined by past history of violence (Boyle, O’Leary, Rosenbaum, & Hassett-Walker, 2008; see also Huss & Ralston, 2008; Saunders, 1992). In line with these findings, a 3-year longitudinal study found similar differences in frequency and severity of IPV between the borderline-dysphoric versus other FO perpetrators (Holtzworth-Munroe et al., 2003). A second rationale for this prediction is based on empirical findings indicating that antisocial tendencies and general aggressiveness are significant predictors of continued IPV (Holtzworth-Munroe et al., 2003). The well-established links between general antisocial behavior and recidivism suggest that GV men should be at increased risk to engage in repeated IPV, irrespective of other individual differences and contextual factors, including initial violence severity. From this perspective, the likelihood of repeated IPV for GV men may be less dependent on various factors (including the severity of earlier violence) than the likelihood of repeated violence for FO men.

Toby’s (1957) stake in conformity hypothesis, which is based on social control theory, provides an alternative view. This perspective suggests that the threat of legal sanctions only deters individuals who have ties to conventional society from reoffending (Sherman, Smith, Schmidt, & Rogan, 1992), such as those who are married, employed, and have attained higher levels of education. Accordingly, individuals with such ties have more to lose as a result of reoffending, and as such, achievement and societal bonds function as protective factors with respect to criminal and violent behavior (Hamberger & Hastings, 1993; Sherman et al., 1992). Consistent with this perspective, domestic violence perpetrators who lack a stake in conformity have been shown to be more likely to have a second offense than their counterparts who have a stake in conformity (Sherman et al., 1992). Cantos, Goldstein, Brenner, O’Leary, and Verborg (in press) also demonstrated that GV men have fewer protective factors with respect to level of education, employment status, and marital status. Accordingly, because the inhibiting impact of consequences provides less of an effective deterrent for GV men, variables related to the violence itself may have a greater impact on the IPV recidivism of GV men. In particular, GV men who have displayed higher levels or severity of violence in the past may be those most likely to display future violence. Thus, attention to violence-related variables may prove to have utility in distinguishing the highest risk GV offenders. As such, an entirely different perspective suggests that the relationship between severity of violence and IPV recidivism may be moderated by type of perpetrator. The stake in conformity hypothesis predicts that severity of violence may be more predictive of recidivism for GV men than for FO violent men.

Assessing Severity of Violence

Although the longitudinal studies summarized above have consistently demonstrated the predictive utility of severity of violence, the results of studies using ratings of a single offense are somewhat inconsistent. Whereas some studies have found a significant relationship between severity of violence of the arresting event and future risk of IPV (Hilton, Harris, & Rice, 2007; Kingsnorth, 2006; Miller & Krull, 1997), other studies have found this relationship to be statistically nonsignificant with trivial effect sizes. (Harrell & Smith, 1996; Miller & Krull, 1997; Murphy, Musser, & Maton, 1998; Paternoster, Bachman, Brame, & Sherman, 1997). These inconsistent findings may reflect differences in the operationalization of severity of violence across studies, with some studies based on perpetrators’ accounts and some based on victims’ accounts, and still others using indirect markers of severity such as victim injury or the use of a weapon. To date, no known studies have examined both victim and perpetrator accounts of severity of violence at the time of the arresting event and assessed their predictive utility. We contend that consideration of both perspectives represents a more comprehensive approach with the potential to elucidate discrepant findings reported above. Therefore, the current study employs a methodological strength by considering both victims’ and perpetrators’ accounts of violence severity at the index offense.

The Current Study

The current study utilizes the typology of FO violent and GV perpetrators (Cantos et al., in press), which is based on taxonomies developed by Holtzworth-Munroe and Stuart (1994) and Boyle et al. (2008). The typology used in the present study expanded on the work of Boyle et al. by classifying male perpetrators based on a combination of self-reported violence and official records of violence over the life span. The present study addresses the following research questions:

Research Question 1: Do victims’ and perpetrators’ accounts of severity of violence for an arresting event, at the time of the arrest, predict risk of domestic violence recidivism?

Research Question 2: Does the relationship between severity of violence and domestic violence recidivism differ based on type of offender?

As noted above, prior findings are not entirely consistent; however, based on the several studies reporting that severity of violence is predictive of recidivism, it was hypothesized that victims’ accounts of severity of violence would be positively related to indices of IPV recidivism. Given prior findings that perpetrators underreport IPV relative to victims (Heckert & Gondolf, 2000; Jouriles & O’Leary, 1985; O’Leary & Murphy, 1992), we anticipated that perpetrators’ accounts would fail to predict IPV recidivism. It was further predicted that the relationship between severity of violence and recidivism would be moderated by type of offender. Specifically, it was hypothesized that the positive association between severity of violence and recidivism would be stronger for FO violent men than for GV men. However, we also examined the possibility that severity of violence would have greater predictive utility for GV men than for FO violent men.

Method

Participants

The sample consisted of men (N = 328) between the ages of 17 and 72 years (M = 33.63, SD = 10.53) who were charged with acts of domestic violence (e.g., domestic battery, aggravated domestic battery, violation of order of protection, stalking) against a former or current intimate partner. All individuals included in the current study were sentenced to probation between 2006 and 2008 in Lake County, Illinois. Probation periods ranged from 1 to 2,849 days (M = 482, SD = 254). The ethnic composition of participants included 43% (n = 141) Caucasian, 34.8% (n = 114) African American, 21% (n = 69) Latino, and 1.2% (n = 4) representing other ethnic backgrounds. To be eligible for this study, individuals had to have met the following criteria: (a) been charged with an act of domestic violence for the arresting event, (b) have a police report for the arresting event on file in the probation department, and (c) the police report must have contained evidence of physical aggression committed against a former or current intimate partner. Complete descriptive statistics for the sample are provided in Table 1.

Table 1:

Individual Characteristics as a Percentage of Violence Classification

Characteristics	Minor Violence (n = 78)	Severe Violence (n = 250)
Education
Post high school	23.4	20.2
Employment
Employed	56.4	54.0
Relationship status
Married	34.6	27.2
Ethnicity
Caucasian	52.6	40.0
African American	28.2	36.8
Latino	17.9	22.0
Other	1.3	1.2
Type of offender
FO violent	66.7	55.2
GV	33.3	44.8

Note. The n values for characteristics within minor violence are smaller for education (n = 77) and employment n = 76). The n values for characteristics within severe violence are smaller for education (n = 238), employment (n = 238), and ethnicity (n = 246). FO = family only; GV = generally violent.

Materials

Data were obtained from each probationer’s electronic file on record in the probation department, which included information pertaining to an individual’s criminal history up to and including the arresting offense, as well as demographic data collected at probation intake. Each electronic file included the following: scanned copy of the probation intake form, a standardized document that included demographic data (e.g., age, education, employment); Law Enforcement Agencies Data System’s (LEADS) criminal history data form, which included official arrest records that probation officers transferred from official LEADS reports to a standardized document; and a scanned copy of the official police report from the arresting event. A small proportion of demographic information (e.g., employment status, level of education) was missing from electronic files and was determined to be missing at random. Corresponding figures are listed in Table 1.

All individuals in the study were assigned a participant identification number. The research analyst employed by the circuit court was provided a list of study participants. He provided postprobation recidivism data for each participant included in the current study in the form of a Microsoft Excel document. The document included the participant identification number and dates of individual probation terms for the arresting event. It also included the following data for the 3-year postprobation period: dates of domestic violence rearrests, dates of nondomestic violence arrests, and dates and number of days spent in custody. A standardized codebook for all variables of interest was developed by a research team member and was utilized throughout data collection by the research team. A Microsoft Excel spreadsheet that was organized by participant identification number was utilized for data entry, which was merged to an SPSS file following completion of data collection.

Procedures

Two graduate-level psychology students were trained in the operationalization of the two perpetrator types, and they coded groups of 30 cases until sufficient interrater reliability was established (κ > .80). Following the attainment of acceptable levels of reliability, all cases were coded. For every 100 cases in the sample, 20 cases were coded by both raters blind to each other to ensure that there was no interobserver drift present. Subsequent to each reliability check, the two coders met to discuss discrepancies and to collaboratively decide upon the code to be entered into the data file.

A behavioral measure of violence severity (see the appendix) was developed by the first author. Following design of the measure, two graduate-level psychology students were trained in the operationalization of the severity levels. Training included psychoeducation on the measure as well as a discussion of case examples. To establish interrater reliability of the measure, all three coders applied the criteria blind to each other for 20 hypothetical cases based on victims’ and perpetrators’ accounts. Sufficient interrater reliability was established with the following results for victims’ accounts among three independent pairs: κ = .93, κ = .93, and κ = 1.0. Coding of perpetrator accounts yielded the following: κ = .93, κ = .77, and κ = .81. Following respective reliability checks, all three coders met to discuss coding discrepancies in a collaborative format. Coders were then permitted to code cases independently. For every 75 police reports in the sample, 20 successive cases were coded by all three raters blind to each other to assess for interobserver drift. Following each of the three reliability checks, coders met to discuss discrepancies and to come up with an agreed upon categorization to be entered into the data file.

Measures

Type of Perpetrator

Fifty-eight percent (n = 190) of the men were classified as FO violent, and 42% (n = 138) were classified as GV. Classifications were made based on individuals’ prior history of violence, documented by LEADS national database and probation records, which included self-report of violence perpetrated against others. In addition to the current convicted offense, which required an official charge of a domestic violence offense, individuals whose criminal histories documented an arrest for at least one aggressive act that was not domestic in nature were classified as GV. That is, all GV men had been charged with at least one prior violent offense against an individual other than a current/former partner or family member. Such acts included simply battery, aggravated assault, armed robbery, and disorderly conduct. Resisting arrest did not qualify as an aggressive act as defined above. In addition to criminal histories, information used to classify difficult cases included conduct problems in youth, including evidence of fights in school or gang activity. Those men who had a history of conduct disorder and aggression problems in childhood were classified as GV. Men were classified as FO violent if records indicated no history of violent behavior except for domestic violence offenses. For every 100 cases coded in the sample, interobserver drift was assessed with the following results: κ = .79, κ = .90, κ = .61, κ = 1.0 (Cantos et al., in press).

Probation Prescreen Intake

The Probation Prescreen Intake is a standardized form completed by probation officers as part of each probationer’s intake. It includes demographic data including age, ethnic status, marital status, level of education, and employment status. LEADS criminal history data for offenses prior to the probation start date for each probationer are also documented. This form acted as a data source for demographic information and criminal history records in the current study.

Severity of Violence

Police reports were coded for severity of perpetrator violence for the index event of IPV. Severity of violence was coded based on victims’ and perpetrators’ accounts of the incident given to the reporting officer at the arresting event. Violence severity was coded as a dichotomous variable, minor (0) or severe (1). Denial of physical violence was also coded (see the appendix for full criteria). Acts of minor and severe violence were drawn from items on the physical assault scale of the Revised Conflict Tactics Scales (CTS2; Straus et al., 1996). The CTS2 is the most commonly used instrument for measuring IPV (Straus & Douglas, 2004) and is psychometrically sound (Straus et al., 1996). In the current study, if multiple acts of violence encompassing more than one severity level were reported, the most severe physical act as part of the index event was coded. Interrater reliability was assessed for three independent pairs. Results for victims’ accounts were as follows: first set—κ = .72, κ = .89, and κ = .72; second set—κ = 1.0, κ = .82, and κ = .82; and third set—κ = .82, κ = .82, and κ = .83. Results for perpetrators’ accounts were as follows: first set—κ = .71, κ = .86, and κ = .73; second set—κ = .94, κ = .88, and κ = .88; and third set—κ = .83, κ = .88, and κ = .82.

Recidivism

Recidivism was coded as a dichotomous variable and was measured by official reports of domestic violence arrest via LEADS, which provides official criminal histories at the national level. Any domestic violence–related arrest (domestic battery, aggravated domestic battery, violation of order of protection, stalking, phone harassment) against a former or current intimate partner following probation monitoring was classified as domestic violence recidivism. Official reports of rearrest were reviewed and collected by a master’s-level research analyst employed by the Circuit Court of Lake County, Illinois.

Follow-Up

To assess for recidivism, all participants were followed for a period of 36 months (1,095 days) beginning the day they were no longer being monitored by probation, which followed either the date their probation terms were successfully completed or the date they were unsuccessfully terminated from probation. Despite a number of men having recidivated between their probation start date and termination date, there was substantial variability in the legal sanctions, clinical interventions, and various other system-level variables during probation sentences. By including time during probation, it is difficult to meaningfully compare the relationship of severity of violence on recidivism without potential for undue influence of extraneous variables. Thus, the current study utilized the last date an individual was being monitored by probation (either completion or unsuccessful termination) as the outset of follow-up for recidivism.

Of the 328 men for whom recidivism data were coded, 23% (n = 76) were incarcerated during the postprobation follow-up. Time incarcerated ranged from 1 to 1,303 days (M = 84.15, SD = 199.50). Given the substantial portion of men who were incarcerated, it was necessary to account for instances in which an offender was not at risk to reoffend following probation. Failure to do so might inherently bias the results by artificially increasing the length of some perpetrators’ at-risk period relative to that of others (Minnesota Department of Corrections, 2011). Therefore, the time perpetrators spent incarcerated as a result of conviction for non-IPV offenses (e.g., theft, drug conviction) was subtracted from their total at-risk period as long as it (a) preceded a domestic violence rearrest, or (b) occurred prior to 1,095 days postprobation termination (end of follow-up period) for those who were not rearrested. This approach allowed for the time to reoffense to be compared equally across men in the sample. The time variable measured the amount of days from the individual’s probation term end date until the date of arrest for a domestic violence charge, or 1,095 days following probation termination date (for those who did not reoffend).

We consider two examples to highlight the current approach. Participant A was convicted of theft and was incarcerated beginning 10 days postprobation term end date for a period of 200 days. He was arrested for domestic battery 14 days following release from prison. Time until recidivism was considered to be 24 days, representing the time spent in the community in which he had the potential to reoffend for a domestic violence offense. Participant B was convicted for possession of a controlled substance and was incarcerated beginning 90 days postprobation term end date for the duration of the 3-year follow-up period. Participant B did not experience the event of interest (domestic violence reoffense), and his follow-up period was calculated as 90 days.

Results

Preliminary Analyses

Chi-square tests of association were conducted to explore potential demographic differences between types of offenders. Chi-square analyses require that all observations are independent and that the sampling distribution of deviations of actual and expected frequency counts is normal in form (McHugh, 2013). Accordingly, given that no more than 20% of expected frequencies should be less than five, four individuals were excluded from the analysis comparing ethnic differences among perpetrator types. Consistent with findings by Cantos et al. (in press), a significant association was found between ethnicity and type of offender, χ²(2, N = 324) = 24.02, p < .001, Φ= .27. GV men included 32.6% (n = 45) Caucasian, 50% (n = 69) African American, 16.7% (n = 23) Latino, and 0.7% (n = 1) represented other ethnic backgrounds. The ethnic composition of FO violent men included 50.5% (n = 96) Caucasian, 23.7% (n = 45) African American, 24.2% (n = 46) Latino, and 1.6% (n = 3) represented other ethnic backgrounds. GV men were significantly less likely than FO violent men to have obtained some postsecondary education, χ²(1, N = 315) = 20.27, p < .001, Φ = .25; to be employed, χ²(1, N = 314) = 17.76, p < .001, Φ= .24; and to be married, χ²(1, N = 324) = 8.63, p = .003, Φ = .16. GV men were significantly younger in age than FO violent men, t(228) = 3.17, p = .002, d = 0.36. Demographic differences between perpetrator types were consistent with findings from a larger sample of male perpetrators of IPV (Cantos et al., in press) from which the current sample was derived.

Perpetrator and victim accounts of violence severity were assessed. According to victim statements, 23.8% (n = 78) of men in the sample engaged in minor violence and 76.2% (n = 250) engaged in severe violence. In contrast, according to perpetrator statements, 56.8% (n = 96) of men engaged in no violence (i.e., they denied committing any physical violence), 30.2% (n = 51) engaged in minor violence, and 13% (n = 22) engaged in severe violence.

We also examined severity of violence for different types of perpetrators. Regarding victim reports, approximately 81.2% (n = 112) of GV men utilized severe violence, whereas 72.6% (n = 138) of FO violent men used severe violence. The relationship between severity of violence and type of offender was found to be statistically nonsignificant, χ²(1, N = 328) = 3.21, p = .07, Φ = .10. Regarding perpetrator reports, 48% (n = 110) of the sample did not have a perpetrator account of violence documented in the police report. For these cases, the perpetrator either fled the scene prior to being interviewed or refused to provide a statement. With respect to documented perpetrator accounts, approximately 13.8% (n = 9) of GV men utilized severe violence, whereas 12.5% (n = 13) of FO violent men used severe violence. In addition, 55.4% (n = 36) of GV men denied engaging in violence as compared with 57.7% (n = 60) of FO violent men. Utilizing perpetrator reports, the relationship between severity of violence and type of offender was found to be statistically nonsignificant, χ²(2, N = 169) = 0.11, p = .95. Because of the vast number of perpetrators who denied engaging in violence, as well as evidence that perpetrators underreport violence severity (relative to victims), subsequent analyses emphasize victim-rated severity of violence.¹

Chi-square analyses were conducted to explore potential demographic differences between levels of violence severity. The relationship between severity of violence and the following demographic variables were found to be statistically nonsignificant: ethnic representation, χ²(2, N = 324) = 3.91, p = .11; educational achievement, χ²(1, N = 315) = 0.36, p = .55; employment status, χ²(1, N = 314) = 0.03, p = .86; and marital status, χ²(1, N = 324) = 1.39, p = .24. Severely violent men were significantly younger as compared with men who engaged in minor violence, t(326) = 2.51, p = .01, d = 0.32.

Primary Analyses

Analyses Addressing Relationships Between Severity of Violence and Recidivism

Overall, 25.9% (n = 85) of the 328 men were arrested for domestic violence in the postprobation follow-up period. Among these men, time until rearrest ranged from 9 to 1,095 days (M = 384, SD = 327). Among individuals who engaged in minor violence, 18% (n = 14) reoffended, compared with 28% (n = 71) of men who engaged in severe violence. A bivariate logistic regression was conducted to examine the potential relationship between severity of violence and risk of domestic violence recidivism. This analysis requires the dependent variable to be binary, correct model fit, independence of observations, and a linear relationship between independent variables and the log odds (Tabachnick & Fidell, 2013). In the current analysis, the independent variable was violence severity, and the dependent variable was odds of domestic violence recidivism measured dichotomously (yes or no). Utilizing bivariate logistic regression, the relationship between severity of violence and domestic violence recidivism was statistically nonsignificant, χ²(1, N = 328) = 3.32, p = .07, odds ratio (OR) = 1.81.

A Kaplan–Meier survival analysis was conducted to compare the survival curves of men who engaged in minor violence and those who engaged in severe violence. Survival analysis is useful as a statistical method when cases differ in their follow-up time (Hilton et al., 2007). The Kaplan–Meier log rank test assumes that at any time participants who are censored have the same survival prospects as those who continue to be followed, survival probabilities are the same for participants recruited at different time points, and the event happens at the time specified (Goel, Khanna, & Kishore, 2010). In the current analysis, the independent variable was severity of violence, and the dependent variable was differences in survival times. The log rank test revealed that the survival functions of men who used minor and severe violence were significantly different, χ²(1, N = 328) = 3.94, p = .047 (see Figure 1). Furthermore, among men who recidivated, those who used minor violence survived (did not reoffend) an average of 437 days (SD = 325), while severely violent men survived an average of 373 days (SD = 328), a difference of just over 2 months. In sum, these results indicate that men who used severe violence had greater risk of recidivism compared with men who used minor violence, and that among recidivists, severity of violence was negatively associated with postprobation survival time.

Figure 1:

Domestic Violence Recidivism Survival Curves for Men Who Used Minor Violence (n = 83) and Severe Violence (n = 269)

Analyses Examining Whether Type of Perpetrator Moderates Relationships Between Severity of Violence and IPV Recidivism

For all multivariate analyses, ORs or hazard ratios (HRs) were reported as measures of effect size. Tolerance and Variance Inflation Factor (VIF) were calculated, and multicollinearity was determined to not be a problem (Tabachnick & Fidell, 2013). A multiple logistic regression was conducted to examine the relative contributions of violence severity and type of offender in the prediction of IPV recidivism. Multiple logistic regression assumes that the observations are independent and that the natural log of the OR and measurement variable have a linear relationship (Tabachnick & Fidell, 2013). Applied to the current analysis, the logistic regression model assumed an equivalent follow-up time period of 3 years for all men, regardless of time spent incarcerated. Independent variables included perpetrator type and severity of violence, and odds of IPV recidivism as a single dichotomous outcome (yes or no) represented the dependent variable. The first model, including the main effects of type of offender and severity of violence, was significant, χ²(3, N = 328) = 22.22, p < .001. Consistent with findings by Cantos, Brenner, Goldstein, Lee, and Fowler (2012), type of offender uniquely predicted the odds of postprobation recidivism, χ²(1, N = 328) = 13.53, p < .001, OR = 2.60. Severity of violence did not reach statistical significance when type of offender was included, χ²(1, N = 328) = 2.28, p = .13.

The second model included the interaction between type of perpetrator and severity of violence in predicting the odds of IPV. The model was significant, χ²(3, N = 328) = 22.22, p < .001. The interaction term of Severity of Violence × Offender Type was significant, χ²(1, N = 328) = 4.47, p = .04, OR = 4.50. Thus, the impact of severity of violence on risk of recidivism was approximately 4.5 times greater for GV men as compared with FO violent men. Within FO violent men, severity of violence was not predictive of risk of recidivism.

A Cox regression survival analysis was also performed to estimate the hazard (or risk) of IPV recidivism based on severity of violence and type of offender entered simultaneously. The multivariate method in the Cox regression allowed us to examine the effect of multiple covariates (severity of violence and type of offender) on survival and to account for censored cases (did not experience event of interest). The Cox regression model assumes that the hazard function for any two levels of a covariate is proportional over the follow-up period (Ata & Sözer, 2007). Independent variables in the Cox regression analysis included perpetrator type and severity of violence, and hazard of IPV recidivism represented the dependent variable. The first model, addressing main effects, was significant, χ²(2, N = 328) = 20.27, p < .001. Type of offender predicted recidivism after controlling for severity of violence, χ²(1, N = 328) = 15.26, p < .001, HR = 2.38. Severity of violence was statistically nonsignificant after controlling for type of offender, χ²(1, 328) = 2.99, p = .08, HR = 1.66 (see Table 2).

Table 2:

Predicting the Hazard of IPV Recidivism Based on Severity of Violence and Perpetrator Type

	β	HR	SE	Wald χ²
Step 1
Severity of violence (Severe = 1)	.51	1.66	.29	2.99
Type of perpetrator (GV = 1)	.87	2.38	.22	15.26**
Step 2
Severity of Violence × Type of Perpetrator	1.27	3.57	.64	3.92*

Note. IPV = intimate partner violence; HR = hazard ratios; GV = generally violent. Step 1: −2 log likelihood = 935.67. Step 2: −2 log likelihood = 931.35.

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

The second model included the interaction of severity of violence and offender type in predicting the hazard of IPV recidivism, which was the dependent variable. The model was significant, χ² (3, N = 328) = 26.95, p < .001. The interaction term was significant, χ²(1, N = 328) = 3.92, p = .048, HR = 3.57. Thus, the impact of severity of violence on risk of recidivism was approximately 3.6 times greater for GV men as compared with FO violent men (see Figure 2). Within FO violent men, severity of violence was not predictive of recidivism.

Figure 2:

Domestic Violence Recidivism Survival Curves for Severity of Violence (Minor and Severe) Based on Perpetrator Type (FO Violent and GV)

In sum, based on victims’ accounts, severity of violence was predictive of domestic violence recidivism utilizing survival analyses. The relationship between severity of violence and domestic violence recidivism was moderated by type of offender. Moderate to large effects were demonstrated. A positive association was found between severity of violence and recidivism for GV men but not for FO violent men.

Discussion

The present study examined the relationship between severity of IPV for an index event and later IPV among male probationers, with type of offender examined as a moderator of this relationship. Consistent with our first hypothesis, survival analyses based on victims’ accounts indicated that severity of violence predicted risk of IPV recidivism measured across a 3-year follow-up. In addition, among perpetrators who recidivated in the 3-year postprobation period, those who had engaged in severe violence recidivated over 2 months faster than those who had engaged in minor violence. The relationship between violence severity and IPV recidivism was statistically nonsignificant in the logistic regression. However, given that the Cox regression model is preferred over the logistic regression when there is censoring as well as information on survival time (Christensen, 1987; Kleinbaum, 1996), we place greater weight on the results of the Cox model. This finding is consistent with the results of several prior findings assessing the relationship between severity of violence and IPV recidivism (Aldorando, 1996; Feld & Straus, 1989; Hilton et al., 2007; Kingsnorth, 2006; Miller & Krull, 1997; Quigley & Leonard, 1996). Utilizing perpetrators’ accounts of severity of violence, no significant relationships emerged.

Analyses also revealed that type of offender moderated the severity of violence–IPV recidivism association. Contrary to our hypothesis, severity of violence was predictive of recidivism among GV men (n = 138) but not among FO violent men (n = 190). These results may help elucidate the mixed prior findings in the literature regarding the relationship between severity of violence and IPV recidivism. That is, the mixed findings may reflect the fact that prior studies differed with respect to the proportions of FO violent and GV men in the samples. Our findings provide preliminary evidence that this relationship differs based on type of perpetrator. Moreover, effect sizes for the interaction of Severity of Violence × Type of Perpetrator appeared to be moderate to large. The Severity of Violence × Type of Perpetrator interaction is consistent with the stake in conformity hypothesis as it pertains to subtypes of male perpetrators. In the present study, GV men were more likely to be unemployed, be single, and have less formal education than FO violent men. In a recent meta-analytic review, Capaldi, Knoble, Shortt, and Kim (2012) highlighted multiple studies demonstrating a significant relationship between these risk factors and IPV recidivism. Consistent with these findings, GV men may have less to lose as a result of ongoing violence and criminality. Accordingly, given the absence of protective factors of achievement and societal ties, prior violence severity may be more indicative of a propensity to engage in ongoing IPV within GV men.

Future Research

It is also possible that individual differences in psychopathic traits contributed to the observed interaction. There is considerable overlap in the personality and behavioral traits identified in GV men and the core features of psychopathy (Boyle et al., 2008; Cleckley, 1976; Moffitt, Krueger, Caspi, & Fagan, 2000; Swogger, Walsh, & Kosson, 2007). In addition, it has been reported that there is an elevated prevalence of psychopathy among IPV perpetrators (15%-30%; Huss & Langhinrichsen-Rohling, 2000). Moreover, the antisocial deviance component of psychopathy is associated with severe violence (Howard, Khalifa, & Duggan, 2014). Given that this component of psychopathy has also been linked with low socioeconomic status (SES) and lower education levels (Farrington, Ullrich, & Salekin, 2010), this perspective also appears consistent with the stake in conformity hypothesis. Additional research examining potential factors associated with violence severity, including psychopathic traits, might further elucidate who, among GV men, is at increased risk of IPV recidivism.

The discrepancy between perpetrators’ and victims’ accounts of violence severity in the current study is consistent with prior findings that male perpetrators are likely to minimize or underreport their level of violence (Heckert & Gondolf, 2000; Jouriles & O’Leary, 1985; O’Leary & Murphy, 1992) relative to victims, and suggests the value of further research addressing the validity of available measures of severity of violence, particularly within samples of male perpetrators who are court-mandated to attend treatment. Future research should expand on the findings in the current study by investigating mechanisms underlying minor and severe violence. In addition, further attention to the distinction between types of violence (e.g., coercive controlling, situational couple), and their relationship to severity of violence and subsequent recidivism among probationers, may further inform this area of study.

Policy and Treatment Implications

Although it is sometimes suggested that official records of IPV should be supplemented by victim accounts whenever possible, we were unable to obtain partner reports of recidivism for this study due to legal restrictions. Kernsmith and Kernsmith (2009) found that 89% of state and local guidelines for perpetrators intervention programs indicate that programs should have no contact with partners of participants except when contact is required by a “duty to warn.” Thus, early in treatment, obtaining victims’ accounts of IPV to assist with treatment planning may prove difficult. Examination of police records for the index arrest to assess violence severity may provide a viable alternative for probation departments and treatment providers to utilize, in conjunction with formal risk assessment tools, in assessing risk of recidivism. Results corroborate the predictive utility of assessing violence severity at the outset of one’s probation term by simply assessing for violence at the arresting event, particularly within GV men.

Results of the current study also have policy implications for IPV risk assessments. In the current study, GV men whose victims reported experiencing severe violence recidivated at a rate of 42%, nearly 3 times the rate of GV men whose victims reported minor violence. As such, severely violent GV men may represent the highest risk offenders among IPV perpetrators. A recent amendment to the Code of Criminal Procedure of 1963 in the state of Illinois indicates that, as a condition of release or as part of monetary bail, the Court may order the alleged offender to undergo a risk assessment evaluation using an evidence-based instrument conducted by an approved partner abuse intervention program provider or department under the court (e.g., probation department). The utility of these risk assessments in differentiating high-risk offenders has implications for both state and national public policy, which may in turn affect victim safety. Results of the current study provide preliminary evidence that would support judges’ decisions to mandate risk assessments for perpetrators who have a history of engaging in both aggression against nonintimates as well as severe IPV. Moreover, results provide preliminary evidence suggesting the value of including indices of severity of violence and type of perpetrator in these risk assessments.

This study also has implications for treatment of male perpetrators of IPV court-mandated to attend treatment. Treatment providers would benefit from utilizing this information in developing and implementing treatments to meet the idiosyncratic needs of male perpetrators. As Paul (1967) indicated, one of the fundamental questions to ask is, “What treatment, by whom, is most effective for this individual with that specific problem, and under what set of circumstances?” (p. 111). Consistent with the stake in conformity hypothesis, it seems likely that male perpetrators who have engaged in both general aggression and severe IPV have distinct treatment considerations, one of which may be the need to address demographic risk factors, such as unemployment and low educational attainment, prior to more targeted interventions related to aggression itself (Cantos & O’Leary, 2014).

Based on the proposition that social control is dependent on the internalization of norms (Aronson, 1988), a multifaceted approach is likely required to decrease domestic violence recidivism. At the criminal justice level, promoting informal controls through reintegration programs, such as job placement services, may help to address economic disparities among GV men. This, in turn, may assist in deterring the highest risk GV men (those who have engaged in severe IPV) from reoffending. At the community level, stimulating informal social control mechanisms may have a deterrent effect. As posited by DeLeon-Granados, Wells, and Long (2005), part of this approach would require an effort to integrate perpetrators back into the segments of conventional society that convey antidomestic violence sentiments. Thus, from a social control perspective, a preventive approach to domestic violence would likely require involvement at both the criminal justice and community level.

Limitations

This study has several noteworthy limitations. First, because severity of violence was measured at the time of the police report, we cannot assume that the arresting event captured the highest degree of violence perpetration within intimate partnerships, and we cannot estimate the stability of these ratings of severity of violence over time. Although it is important to examine the replicability of current findings in studies using other ways of assessing severity of violence, severity of abuse is a robust predictor of law enforcement involvement in response to IPV reports (Bachman & Coker, 1995; Kaufman Kantor & Straus, 1990). Thus, it seems likely that a large proportion of IPV incidents coded in this study were representative of severe incidents in the course of probationers’ respective relationships. Second, there was likely substantial variability in the length and course of partner relationships for participants in the study. Because we did not assess relationship status during follow-up periods, it is possible that some men may have had greater opportunity than other men to reoffend against an intimate partner. However, legal and ethical considerations prohibited contact with these men to inquire as to their relationship statuses once they were no longer on probation and represent a common limitation in studies of criminal justice populations.

Third, the study utilized coded reports of reported violence. Self-reported behavior depends on a high degree of honesty, and neither victim reports nor perpetrator reports are objective. In fact, IPV reporting concordance has been a significant limitation in the assessment of IPV (Armstrong, Wernke, Medina, & Schafer, 2002). Thus, future studies of IPV should incorporate additional indices of violence severity, including the relationship between severity and injury and its relationship to both type of offender and recidivism. Fourth, the high rates of severe violence (76%) in our sample suggest that this sample may have been skewed toward more severe levels of violence. Results related to recidivism may not be generalizable to perpetrators of IPV with potentially less extensive criminal backgrounds. However, several studies utilizing qualitatively different samples have found that severity of violence is predictive of IPV (Aldorando, 1996; Feld & Straus, 1989; Hilton et al., 2007; Kingsnorth, 2006; Miller & Krull, 1997; Quigley & Leonard, 1996). Future studies utilizing publicly recruited samples should also examine whether violence severity interacts with perpetrator type in predicting recidivism.

Fifth, we relied on official arrest records to assess for recidivism in this study. Official arrest records have been found to cite significantly fewer offenses when compared with partner reports (Gondolf, 1997; Rosenfeld, 1992). The use of partner reports for recidivism would likely have provided greater power to detect meaningful differences. Even so, one strength of the current approach was that we were able to obtain data on official arrests for IPV at the national level, which may prove quite difficult by relying on partner reports. Finally, this study, like many studies of IPV, addressed only physical aggression against family members. Studies that also address sexual aggression and emotional abuse may yield different kinds of findings, and studies are also needed to address the predictive utility of the severity of these types of IPV on recidivism.

Conclusion

In summary, current results provide initial evidence that combining information about the severity and generality of violence among IPV perpetrators may improve prediction of recidivism. Criminal risk assessment is one area that should continue to receive adequate attention in the literature, particularly given the substantial health-related costs of IPV. The field would benefit from additional investigations of severity of violence and related areas to further elucidate who, among IPV perpetrators, is at highest risk to reoffend. Only future research can indicate whether violence severity interacts with other factors to improve our ability to predict recidivism and to reduce the likelihood of IPV.

Footnotes

Appendix

Acknowledgements

We would like to express our appreciation to the staff at the Lake County Division of Adult Probation Services for providing us the opportunity to conduct this research. We would like to thank the anonymous reviewers and editor for their helpful comments and suggestions. Finally, we would like to thank Kelsy White and Melanie Koch for their assistance with data collection.

Notes

Daniel A. Goldstein, MA, MS, is a PhD candidate in clinical psychology at Rosalind Franklin University of Medicine and Science. His research interests include investigating the sociocultural, typological, and contextual factors influencing treatment adherence and recidivism among perpetrators of intimate partner violence.

Arthur L. Cantos, PhD, is program director of the PhD program in clinical psychology and director of the Institutional Scientific Research Center at Universidad Carlos Albizu in San Juan, Puerto Rico. He received his PhD degree from Stony Brook University, Stony Brook, New York. His research interests include treatment of perpetrators of intimate partner violence in particular and family violence in general.

Lauren H. Brenner, MS, is a PhD candidate in clinical psychology at Rosalind Franklin University of Medicine and Science. Her research interests include examining the relationship of cognitive factors such as attributions and flexibility on treatment compliance and recidivism in male perpetrators of intimate partner violence.

Robert J. Verborg, MS, is a senior research analyst with the Nineteenth Judicial Circuit Court of Lake County, Illinois. His research interests include evidence-based practices in community correctional setting, and employee engagement, training, and mentoring.

David S. Kosson, PhD, is a professor of psychology at Rosalind Franklin University of Medicine and Science. He studies psychopathy, other personality disorders, substance abuse, and violence with a primary focus on cognitive, affective, physiological, and developmental mechanisms underlying psychopathy.

References

Aldorando

(1996). Cessation and persistence of wife assault: A longitudinal analysis. American Journal of Orthopsychiatry, 66, 141-151. doi:10.1037/h0080164

Armstrong

Wernke

Medina

Schafer

(2002). Do partners agree about the occurrence of intimate partner violence? A review of the current literature. Trauma, Violence, & Abuse, 3, 181-193. doi:10.1177/15248380020033002

Aronson

(1988). The social animal. New York, NY: W.H. Freeman.

Ata

Sözer

M. T.

(2007). Cox regression models with nonproportional hazards applied to lung cancer survival data. Hacettepe Journal of Mathematics and Statistics, 36(2), 157-167.

Bachman

Coker

A. L.

(1995). Police involvement in domestic violence: The interactive effects of victim injury, offender’s history of violence, and race. Violence and Victims, 10(2), 91-106.

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

. . . Stevens

M. R.

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Boyle

D. J.

O’Leary

K. D.

Rosenbaum

Hassett-Walker

(2008). Differentiating between generally and partner-only violent men: Lifetime antisocial behavior, family of origin violence, and impulsivity. Journal of Family Violence, 23, 47-55. doi:10.1007/s10896-007-9133-8

Cantos

A. L.

Brenner

L. H.

Goldstein

D. A.

Lee

H. S.

Fowler

D. R.

(2012, July). Outcome of mandated treatment for generally violent and family only violent perpetrators of intimate partner violence. In Cantos

A. L.

(Chair), Revisiting one size fits all approaches to the treatment of intimate partner violence: New directions. Symposium conducted at the International Family Violence and Child Victimization Research Conference, Portsmouth, NH.

Cantos

A. L.

Goldstein

D. A.

Brenner

O’Leary

K. D.

Verborg

(in press). Correlates and program completion of family only and generally violent perpetrators of intimate partner violence. Behavioral Psychology/Psicologia Conductual, 23(3), 549-569.

10.

Cantos

A. L.

O’Leary

K. D.

(2014). One size does not fit all in treatment of intimate partner violence. Partner Abuse, 5(2), 204-236. doi:10.1891/1946-6560.5.2.204

11.

Capaldi

D. M.

Knoble

N. B.

Shortt

J. W.

Kim

H. K.

(2012). A systematic review of risk factors for intimate partner violence. Partner Abuse, 3, 231-280. doi:10.1891/1946-6560.3.2.231

12.

Catalano

(2012). Intimate partner violence, 1993–2010 (Special Report NCJ 239203). Retrieved from http://www.bjs.gov/content/pub/pdf/ipv9310.pdf

13.

Christensen

(1987). Multivariate survival analysis using Cox regression model. Hepatology, 7, 1346-1358. doi:10.1002/hep.1840070628

14.

Cleckley

(1976). The mask of sanity. St. Louis, MO: Mosby.

15.

Coker

A. L.

Davis

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, 451-457. doi:10.1016/S0749-3797(02)00514-7

16.

DeLeon-Granados

Wells

Long

(2005). Twenty years beyond the Minneapolis Domestic Violence Experiment: A missing link in the preventive effect of arrest on recidivism. Western Criminology Review, 6, 43-58.

17.

Dutton

D. G.

(1998). The abusive personality: Violence and control in intimate relationships. New York, NY: Guilford Press.

18.

Dutton

D. G.

Corvo

(2006). Transforming a flawed policy: A call to revive psychology and science in domestic violence research and practice. Aggression and Violent Behavior, 11, 457-483. doi:10.1016/j.avb.2006.01.007

19.

Dutton

M. A.

Kaltman

Goodman

L. A.

Weifurt

Vankos

(2005). Patterns of intimate partner violence: Correlates and outcomes. Violence and Victims, 20, 483-497.

20.

Farrington

D. P.

Ullrich

Salekin

R. T.

(2010). Environmental influences on child and adolescent psychopathy. In Salekin

R. T.

Lynam

D. R.

(Eds.), Handbook of child and adolescent psychopathy (pp. 251-283). New York, NY: Guilford Press.

21.

Feld

S. L.

Straus

M. A.

(1989). Escalation and desistance of wife assault in marriage. Criminology, 27, 141-161. doi:10.1111/j.1745-9125.1989.tb00866.x

22.

Frieze

I. H.

Browne

(1989). Violence in marriage. In Ohlin

Tonry

(Eds.), Family violence (pp. 163-218). Chicago, IL: University of Chicago Press.

23.

Goel

M. K.

Khanna

Kishore

(2010). Understanding survival analysis: Kaplan-Meier estimate. International Journal of Ayurveda Research, 1, 274-278. doi:10.4103/0974-7788.76794

24.

Gondolf

E. W.

(1988). Who are those guys? Toward a behavioral typology of batterers. Violence and Victims, 3, 187-203.

25.

Gondolf

E. W.

(1997). Patterns of reassault in batterer programs. Violence and Victims, 12, 373-387.

26.

Hamberger

L. K.

Hastings

J. E.

(1985, March). Personality correlates of men who abuse their partners: Some preliminary data. Paper presented at the meeting of the Society for Personality Assessment, Berkeley, CA.

27.

Hamberger

L. K.

Hastings

J. E.

(1993). Court-mandated treatment of men who assault their partner. In Hilton

(Ed.), Legal responses to wife assault: Current trends and evaluation (pp. 188-229). Newbury Park, CA: SAGE.

28.

Harrell

Smith

B. A.

(1996). Effects of restraining orders on domestic violence victims. In Buzawa

E. S.

Buzawa

C. G.

(Eds.), Do arrests and restraining orders work? (pp. 214-242). Thousand Oaks, CA: SAGE.

29.

Heckert

Gondolf

(2000). The effect of perceptions of sanctions on batterer program outcome. Journal of Research on Crime & Delinquency, 37, 369-391. doi:10.1177/0022427800037004002

30.

Hegarty

K. L.

O’Doherty

L. J.

Chondros

Valpied

Taft

A. J.

Astbury

. . . Gunn

J. M.

(2012). Effect of type and severity of intimate partner violence on women’s health and service use: Findings from a primary care trial of women afraid of their partners. Journal of Interpersonal Violence, 2, 273-294. doi:10.1177/0886260512454722

31.

Hilton

N. Z.

Harris

G. T.

Rice

M. E.

(2007). The effect of arrest on wife assault recidivism: Controlling for pre-arrest risk. Criminal Justice and Behavior, 34, 1334-1344. doi:10.1177/0093854807300757

32.

Holtzworth-Munroe

Meehan

J. C.

Herron

Rehman

Stuart

(2003). Do subtypes of maritally violent men continue to differ over time? Journal of Consulting and Clinical Psychology, 71, 728-740. doi:10.1037/0022-006X.71.4.728

33.

Holtzworth-Munroe

Stuart

G. L.

(1994). Typologies of male batterers: Three subtypes and the differences among them. Psychological Bulletin, 116, 476-497. doi:10.1037/0033-2909.116.3.476

34.

Howard

R. C.

Khalifa

Duggan

(2014). Antisocial personality disorder comorbid with borderline pathology and psychopathy is associated with severe violence in a forensic sample. Journal of Forensic Psychiatry & Psychology, 25, 658-672. doi:10.1080/14789949.2014.943797

35.

Huss

M. T.

Langhinrichsen-Rohling

(2000). Identification of the psychopathic batterer: The clinical, legal, and policy implications. Aggression and Violent Behavior, 5, 403-422. doi:10.1016/S1359-1789(98)00038-X

36.

Huss

M. T.

Ralston

(2008). Do batterer subtypes actually matter? Treatment completion, treatment response, and recidivism across a batterer typology. Criminal Justice and Behavior, 35, 710-724. doi:10.1177/0093854808316218

37.

Illinois Code of Criminal Procedure of 1963, 725 ILCS 5/110-5 (2014).

38.

Johnson

M. P.

(2006). Conflict and control: Gender symmetry and asymmetry in domestic violence. Violence Against Women, 12, 1003-1018. doi:10.1177/1077801206293328

39.

Johnson

M. P.

Leone

J. M.

(2005). The differential effects of intimate terrorism and situational couple violence: Findings from the National Violence Against Women Survey. Journal of Family Issues, 26, 322-349. doi:10.1177/0192513X04270345

40.

Jouriles

E. N.

O’Leary

K. D.

(1985). Interspousal reliability of reports of marital violence. Journal of Consulting and Clinical Psychology, 53, 419-421. doi:10.1037/0022-006X.53.3.419

41.

Kaufman Kantor

Straus

M. A

. (1990). Response of victims and the police to assaults on wives. In Straus

M. A.

Gelles

R. J.

(Eds.), Physical violence in American families: Risk factors and adaptations to violence in 8,145 families (pp. 473-487). New Brunswick, NJ: Transaction Publishers.

42.

Kelly

J. B.

Johnson

M. P.

(2008). Differentiation among types of intimate partner violence: Research update and implications for interventions. Family Court Review, 46, 476-499. doi:10.1111/j.1744-1617.2008.00215.x

43.

Kernsmith

(2009). Treating female perpetrators: State standards for batterer intervention services. Social Work, 54, 341-349. doi:10.1093/sw/54.4.341

44.

Kingsnorth

(2006). Intimate partner violence: Predictors of recidivism in a sample of arrestees. Violence Against Women, 12, 917-935. doi:10.1177/1077801206293081

45.

Kleinbaum

D. G.

(1996). Survival analysis: A self-learning text. New York, NY: Springer.

46.

Leone

J. M.

Johnson

M. P.

Cohan

C. L.

Lloyd

(2004). Consequences of domestic violence for low-income, ethnic minority women: A control-based typology of male partner violence. Journal of Marriage and Family, 66, 472-491. doi:10.1111/j.1741-3737.2004.00032.x

47.

McHugh

M. L.

(2013). The chi-square test of independence. Biochemia Medica, 23(2), 143-149. doi:10.11613/BM.2013.018

48.

Miller

J. L.

Krull

A. C.

(1997). Controlling domestic violence: Victim resources and police intervention. In Kaufman Kantor

Jasinski

J. L.

(Eds.), Out of the darkness: Contemporary perspectives on family violence (pp. 235-254). Thousand Oaks, CA: SAGE.

49.

Minnesota Department of Corrections. (2011). The effects of prison visitation on offender recidivism. St. Paul: Minnesota Department of Corrections.

50.

Moffitt

T. E.

Krueger

R. F.

Caspi

Fagan

(2000). Partner abuse and general crime: How are they the same? How are they different? Criminology, 38, 199-232. doi:10.1111/j.1745-9125.2000.tb00888.x

51.

Murphy

C. M.

Musser

P. H.

Maton

K. I.

(1998). Coordinated community intervention for domestic abusers: Intervention system involvement and criminal recidivism. Journal of Family Violence, 13, 263-284. doi:10.1023/A:1022841022524

52.

National Center for Injury Prevention and Control. (2003). Costs of intimate partner violence against women in the United States. Atlanta, GA: Centers for Disease Control and Prevention.

53.

O’Leary

K. D.

Murphy

(1992). Clinical issues in the assessment of spouse abuse. In Ammerman

R. T.

Hersen

(Eds.), Assessment of family violence: A clinical and legal source book (pp. 26-46). New York, NY: John Wiley.

54.

Paternoster

Bachman

Brame

Sherman

L. W.

(1997). Do fair procedures matter? The effect of procedural justice on spouse assault. Law & Society Review, 31, 163-204. Retrieved from http://www.lawandsociety.org/review.html

55.

Paul

G. L.

(1967). Strategy of outcome research in psychotherapy. Journal of Consulting Psychology, 31, 109-118. doi:10.1037/h0024436

56.

Pence

Paymar

(1993). Education groups for men who batter: The Duluth model. New York, NY: Springer.

57.

Quigley

B. M.

Leonard

K. E.

(1996). Desistance of husband aggression in the early years of marriage. Violence and Victims, 11, 355-370.

58.

Rosenfeld

B. D.

(1992). Court ordered treatment of spouse abuse. Clinical Psychology Review, 12, 205-226. doi:10.1016/0272-7358(92)90115-O

59.

Sartin

R. M.

Hansen

D. J.

Huss

M. T.

(2006). Domestic violence treatment response and recidivism: A review and implications for the study of family violence. Aggression & Violent Behavior, 1, 425-440. doi:10.1016/j.avb.2005.12.002

60.

Saunders

D. G.

(1992). A typology of man who batter: Three types derived from cluster analysis. American Journal of Orthopsychiatry, 62, 264-275. doi:10.1037/h0079333

61.

Sherman

L. W.

Smith

D. A.

Schmidt

J. D.

Rogan

D. P.

(1992). Crime, punishment, and stake in conformity: Legal and informal control of domestic violence. American Sociological Review, 57, 680-690. doi:10.2307/2095921

62.

Straus

M. A.

Douglas

E. M.

(2004). A short form of the Revised Conflict Tactics Scales, and typologies for severity and mutuality. Violence and Victims, 19, 507-520. Available from http://www.springerpub.com/

63.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

D. B.

(1996). The Revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283-316. doi:10.1177/019251396017003001

64.

Swogger

M. C.

Walsh

Kosson

D. S.

(2007). Domestic violence and psychopathic traits: Distinguishing the antisocial batterer from other antisocial offenders. Aggressive Behavior, 33, 253-260. doi:10.1002/ab.20185

65.

Tabachnick

B. G.

Fidell

L. S.

(2013). Using multivariate statistics (6th ed.). Boston, MA: Allyn & Bacon.

66.

Toby

(1957). Social disorganization and stake in conformity: Complementary factors in the predatory behavior of hoodlums. Journal of Criminal Law, Criminology, and Political Science, 48, 12-17. doi:10.2307/1140161

67.

Truman

J. L.

Morgan

R. E.

(2014). Nonfatal domestic violence, 2003-2012 (Special Report NCJ 244697). Washington, DC: U.S. Department of Justice. Retrieved from http://www.bjs.gov/content/pub/pdf/ndv0312.pdf

68.

Woodin

E. M.

O’Leary

K. D.

(2006). Partner aggression severity as a risk marker for male and female violence recidivism. Journal of Marital and Family Therapy, 32, 283-296. doi:10.1111/j.1752-0606.2006.tb01607.x