View this issue as a PDF: CrimHighlightsV15N4.pdf
This issue of Criminological Highlights addresses the following questions:
- Are high risk sex offenders high risk forever?
- Why should governments be wary of cost-savings claims for private prisons?
- Why does poor treatment of ordinary citizens by the police lead to more offending?
- What types of restorative justice programs have been shown to be effective in reducing re-offending?
- Is it useful to talk about a single imprisonment rate for federal jurisdictions like Australia?
- What do the different rates of imprisonment in the US states reflect?
- Is it possible to provide effective reintegration services for prisoners serving short sentences?
- For what type of offender is police contact most likely to lead to increased violent offending?
The likelihood of someone convicted of a sex offence reoffending decreases substantially the longer that person remains in the community offence-free.
Sexual offenders are often believed to be the least likely offenders to change, even though the evidence suggests that their rate of recidivism overall does not distinguish them from other groups of offenders (See Criminological Highlights 3(3)#3, 5(1)#4, 6(3)#3, 6(6)#8, 9(2)#5, 13(2)#6 ). The result of this incorrect belief is that various jurisdictions put special controls on those released from prison after serving sentences for sex offences.
However, sex offenders as a group vary considerably in their likelihood of committing another sex offence. If sex offenders are like more general offenders, the greatest risk for new criminal offending occurs immediately after release from prison. The purpose of this study is to see whether the time that a sex offender has been free in the community relates to recidivism risk.
Using an aggregate sample of 7,740 sex offenders from 21 different samples, this study estimates the overall sex offence recidivism rates (charges or convictions) for those 5 and 10 years after release, for both typical offenders and low- and high-risk offenders (categorized by the most widely used sexual offender risk tool that is used in mental health and corrections, the Static-99R).
The overall sexual offence recidivism rate was highest in the first few years after release from prison. For the sample as a whole, 10.1% committed new sex offences within 5 years, 14.2% within 10 years, and only 2.4% more or 16.6% within 16 years.
For the highest risk group, 22% had reoffended with a sex offence within 5 years. Within 10 years, only an additional 7% (28.8% overall) had reoffended, and at 15 years, only an additional 3% (31.8% overall) reoffended. No high risk sex offender who had been offence-free for 16 years reoffended. For low risk sex offenders 2.2% had reoffended within 5 years, 3.1% within 10 years, and 4.7% within 15 years. For the large ‘moderate risk’ group (62% of the overall sample), 6.7% reoffended; within 10 years 10.4% reoffended, and within 15 years 12.6% reoffended.
Conclusion: “The risk of sexual recidivism was highest in the first few years after release, and decreased the longer [these offenders] remained offence free in the community” (p. 2804-5). The decline over time was greatest for those who had been categorized as highest in risk. “If high-risk sexual offenders do not reoffend when given the opportunity to do so, then there is clear evidence that they are not as high risk as they are initially perceived [to be]” (p. 2805). What this implies, of course, is that given that all measures we have (risk prediction measures as well as simple criminal history records) are fallible indicators of risk-relevant propensities to re-offend, individuals who score high either may never actually have been high risk or they may genuinely have changed. The data do show that, as a group, those classified as high risk are, indeed, more likely to reoffend than those with lower classifications. Hence it would make sense, initially, to focus resources (service and monitoring) on them initially. However, the results also suggest that “sexual offenders who remain offence-free could eventually cross a ‘redemption’ threshold in terms of recidivism risk, such that their current risk for sexual crime becomes indistinguishable from the risk presented by non-sexual offenders” (p. 2806).
Reference: Hanson, R. Karl, Andrew J. R. Harris, Leslie Helmus, and David Thornton (2014). High-Risk Sex Offenders May Not Be High Risk Forever. Journal of Interpersonal Violence, 29(15), 2792-2813.
Recidivism rates of those who served their sentences in a private prison in Minnesota were, if anything, higher than recidivism rates of those who served their sentences in state-operated facilities.
Though the number of prisoners in privately operated prisons in the United States is not large (estimated at 8% of all US prisoners), interest in the “effectiveness” of private prisons has been high.
Various concerns have been raised about the privatization of punishment (see Criminological Highlights 3(6)#1, 3(6)#2), as well as the fact that private prisons do not seem to be less expensive for the state when all costs are considered (Criminological Highlights, 2(5)#2). Furthermore, the evidence seems to show that they may foster more prison disturbances (Criminological Highlights 5(2)#1) and do not appear to reduce recidivism rates (Criminological Highlights 7(3)#3, 8(1)#1). Generally speaking, it has been found that cost savings are non-existent or negligible, and there is little evidence that recidivism rates are lower among prisoners who served all or part of their sentences in private facilities.
This paper examines the recidivism rate of 1,766 prisoners who spent time in a Corrections Corporation of America prison in Minnesota, comparing their post-release behaviour to 1,766 comparable prisoners who did not spent time in the private facility. All were released between 2007 and 2009. As is generally the case elsewhere, only a subset of Minnesota prisoners were eligible to serve their sentences in private facilities. Hence, in this study, propensity score matching, based on 20 relevant variables, was used to create pairs of prisoners who had similar characteristics, but where one prisoner spent time in the private prison whereas the other did not. Prisoners who served their time in the private facility spent, on average, a year there.
The follow-up period, after release from prison during which reoffending was measured, averaged about 30 months. Reoffending was assessed using 4 different measures: Rearrest, reconviction, reincarceration, and revocation of parole. For each of these outcomes, three different analyses were carried out: the impact of any time in private prison, the number of days spent in private prison, and the proportion of the sentence spent in the private facility. In addition, analyses were carried out using only those who served at least a year in the private facility as well on those who spent at least 50% of their prison time in the private facility.
In all, then, 20 different analyses were carried out, each of which was in the direction of higher recidivism rates in private facility. In 8 of these 20 analyses, the recidivism rates were significantly higher for those who spent time in the private prison. Reconviction rates were higher for all 5 operationalizations of ‘time in private prison.’
The study did not, unfortunately, provide any clear data on why recidivism rates – most notably reconviction rates – might have been higher for those who were incarcerated in the private prison. There were programming differences between the two facilities (more of a variety of programs in the state-operated prison) and it was more difficult for people in the community to visit prisoners in the private facility.
Conclusion: The daily per-prisoner costs to the state of the private and state-run institutions were very similar. However, given the higher reconviction rates for those who spent time in the private prison, it would appear there were higher costs, ultimately, to the state if the prisoner went to a private prison. Hence, “the evidence from this evaluation and prior studies indicates that private prisons are not a superior alternative to state-run prisons” (p. 391). More cautiously, the results suggest that one cannot assume superiority of privately run prisons.
Reference: Duwe, Grant and Valerie Clark (2013). The Effects of Private Prison Confinement on Offender Recidivism: Evidence from Minnesota. Criminal Justice Review, 38(3), 375-394.
Treatment by the police that is perceived to be unfair reduces citizens’ willingness to be law abiding because being treated badly leads people to feel angry or resentful which, in turn, makes them less likely to follow the law and obey the police.
There is considerable evidence that procedurally fair treatment by the police is important in motivating ordinary citizens to cooperate with them and to follow the law (Criminological Highlights 4(4)#1, 7(1)#4, 11(4)#1, 12(5)2, 15(1)#5, 15(3)#2). This paper, reporting the results of a survey and an experiment, examines the psychological mechanism whereby unfair treatment appears to reduce the view that obeying the law and the police is important.
In this study, a representative sample of Australians (drawn from voting lists) was asked to fill out a survey questionnaire in 2007, and again in 2009. The study focuses on those who had contact with the police in the previous 12 months. Procedural justice was measured by such questions as whether the police were polite, respectful, and fair. Those who reported that they were treated fairly, etc., by the police were less likely to report being angry, resentful, frustrated, etc., after the interaction with the police (controlling for age, gender, level of education, and income). Those who reported having been treated in a procedurally fair manner were also more likely to report willingness to comply with the law and to obey the police. However, when the reported emotional response of the respondent to the encounter was controlled for, the effect of procedural justice disappeared. This pattern of findings suggests that the relationship between being treated in a procedurally just fashion and compliance with the law and the police is mediated by negative affect created by procedurally unjust treatment. Said differently, being treated in a procedurally unjust fashion leads to feelings of frustration and anger which, in turn, reduce the likelihood of future compliance with the police and with the law.
In the second (experimental) study, Australian university students were given descriptions of one of two scenarios in which they were to imagine being stopped by the police for exceeding the speed limit by 5 km/hour. In one scenario (given to half the respondents) the police officer was described as courteous, friendly, and giving an explanation for the stop. For the other half of the respondents, the police officer was described as rude, condescending, and not explaining the purpose of the stop. Once again, being treated in a procedurally unjust fashion led respondents to report more negative affect. In addition, they reported they would, in the future, be less likely to be careful to follow all road rules, and generally would be less likely to follow the law. However, once again, when negative affect was controlled for, the effect of procedural justice disappeared suggesting that being treated in a procedurally unfair manner leads people to be angry, etc., which in turn makes them less likely to follow the law in the future.
Conclusion: “Procedural justice appears to be consistently important for influencing both emotional reactions and compliance [with the law and the police]…. By engaging with the public in a polite, respectful, and empathetic manner, police officers will be able to reduce negative sentiments and emotion directed at them, thereby increasing people’s willingness to comply with them both immediately and in the future” (p. 269). “If the police wish to be able to effectively manage citizen behaviour and promote compliance with the law, the findings… suggest that they ought to treat people with procedural justice” (p. 270).
Reference: Barkworth, Julie M, and Kristina Murphy (2015). Procedural Justice Policing and Citizen Compliance Behaviour: The Importance of Emotion. Psychology, Crime & Law, 21(3), 254-273.
Restorative Justice conferences involving victims and offenders carried out, largely, on those who had been found guilty, led to small reductions in subsequent offending.
Many claims are made about the effectiveness of restorative justice practices in the criminal justice system. Few have been substantiated with adequate data. In an attempt to evaluate whether restorative justice conferences reduce repeat offending, this study carefully examined 519 studies that purported to have relevant findings. Unfortunately, only 10 of these studies were adequately carried out to the extent that inferences about their effectiveness in reducing re-offending could be made.
“The diverse nature of [restorative justice] practices makes it difficult to answer the question of whether ‘restorative justice’ defined so broadly works better than conventional justice … Most of the practices described as restorative justice have never been subjected to controlled field tests” (p. 2). Restorative justice conferences (RJC) have, however, been evaluated. In this paper, studies were examined in which the offence involved an actual victim (in contrast to offences such as impaired driving) and in which cases were assigned on a random (or quasi-random) basis to be resolved with an RJC or by conventional criminal justice approaches. The ten studies were carried out in the US (1 study), the UK (7 studies) and Australia (2 studies). Assessment of the effectiveness of the RJC treatment was assessed on the basis of the treatment to which people were randomly assigned. Some of those assigned to the RJC treatment, in the end, did not receive it. To avoid selection effects, the effect of the assigned (not received) treatment was examined. Between 68% and 92% of those assigned to RJC received this treatment. Four studies involved diversion from prosecution or caution. In 4 studies, the RJC took place post-plea, presentence. Two involved RJC after sentence (while the offender was in prison or on probation). 6 studies involved adult offenders; 4 involved young offenders.
There was an average repeat offending rate decrease (across all studies) of 0.15 standard deviations in the two years following the assignment. Hypothetically, if in a given study people without the RJC were likely to commit 2 offences on average and most (95%) committed between 0 and 4 offenses, this would mean that on average the RJC participants would have on average 1.85 repeat offences. There was some suggestion that RJC showed a slightly larger effect when the crime involved was violent, and involved adults rather than youths. These differences were not, however, statistically significant.
Conclusion: “Restorative justice conferences… appear [to] be likely to reduce the future frequency of detected and prosecutable crimes among the kinds of offenders who are willing to consent to Restorative Justice Conferences when victims are also willing to give consent to the process…. The operational basis of holding such conferences at all depends on consent…” (p. 19). However, “the effects of RJCs on the frequency of repeat offending are especially clear as a supplement to conventional justice, with less certainty about its effects when used as a substitute. Yet RJCs may be seen as most appealing when they can both reduce crime and save money – starting with diversion from expensive court process. The use of restorative processes in this way has grown rapidly in some countries without rigorous testing” (p. 20).
Reference: Sherman, Lawrence W., Heather Strang, Evan Mayo-Wilson, Daniel J. Woods, and Barak Ariel (2015). Are Restorative Justice Conferences Effective in Reducing Repeat Offending? Findings from a Campbell Systematic Review. Journal of Quantitative Criminology, 31(1), 1-24.
Four Australian states show relatively similar patterns of increases in imprisonment rates since the 1970s, yet the local cultures of these 4 states result in different overall rates of imprisonment.
As in the US (see Criminological Highlights 15(4)#6), the Australian states vary substantially in their imprisonment rates. Although the 2014 national average was 186 per thousand adults, Western Australia’s rate was 265 and Victoria’s was 134. As in the US during the period 1970-2010, rates of imprisonment increased in each state (though not exactly at the same time). This paper examines the state-specific drivers of penal policy, and its result – imprisonment rates – in four Australian states.
In New South Wales (NSW) between 1976 and 1986, policies were implemented to reduce imprisonment. Beginning in 1986, however, ‘law and order’ politics prevailed and sentences were explicitly lengthened (e.g., by abolishing remission). But in addition, pretrial release was made more difficult as a result of 23 legislative changes between 1992 and 2008 and of changes made a few years later in response to 3 highly publicized crimes. Judges, it seems, attempted to “resist what they see as legislative and executive incursions into judicial discretion, whether [these changes were] intended to reduce or increase sentences” (p. 353).
Victoria’s increase in imprisonment occurred under left- and right-leaning governments which imposed changes designed to increase imprisonment . Increases occurred a bit later than in NSW. In the 2010 election, crime became an issue, notwithstanding evidence that Victorians were more accepting of alternatives to imprisonment than their politicians appeared to believe. Politicians promoted the belief that imprisonment was effective in controlling crime. Overall, Victoria’s policies appear to be driven more by symbolic and emotional pressures “mostly triggered by high-profile media events” (p. 356).
Imprisonment rates in South Australia (SA) started increasing in 1984 and almost tripled since then as a result of actions of both main political parties. Remission was abolished in 1994, a policy supported apparently by both main parties. Various types of offenders were targeted (e.g., sex offenders, certain drug offenders, and those violating conditions of their parole). SA has a remand rate that is about 50% higher than that of the rest of Australia. Both major parties currently favour ‘tough on crime’ policies.
Western Australia (WA) has historically had high rates of imprisonment. In 1981, a committee established to examine this phenomenon concluded that this was due largely to the number of people imprisoned rather than the length of their prison terms. WA also has Australia’s highest rate of Indigenous over-representation. The 1990s brought minimum sentence legislation and the abolition of remission. In 2009 an explicit attempt was made to reduce the use of parole.
The imprisonment rate for Indigenous people in Australia is 13 times that of non-Indigenous people. For Indigenous women and children it is 24 times the non-indigenous rate. Though Indigenous incarceration does not explain the differences across states, “it is still arguably the key penal problem in the Australian context” (p. 364). One explanation is that laws that are, on their face, neutral with respect to race have different effects on Indigenous people. A complementary explanation is that Indigenous people, because of Australia’s colonial history, have experienced forms of deprivation that are criminogenic and are, as a result, more likely to be involved in crime (especially crimes likely to result in incarceration).
Conclusion: All states broadly show increased imprisonment rates between 1970 and 2014, though the timing of the increasing rates varies. Differences in imprisonment rates across states are stable across time. In all 4 states that were studied in detail, crime became more politicized and no party in any state was consistently willing to distance itself from high imprisonment policies. The views of judges are important in understanding the effects of legislative changes on the sentences imposed. But in addition, local traditions, cultures, and values are important in understanding differences in imprisonment trends.
Reference: Tubex, Hilde, David Brown, Arie Freiberg, Karen Gelb, and Rick Sarre (2015). Penal diversity within Australia. Punishment & Society 17(3), 345-373.
Between 1970 and 2010, the 50 US states were united in one important way: all 50 states (and the federal government) increased their imprisonment rates by at least 100%. But the amount that state imprisonment rates increased is a reflection of value differences across the various states.
In the early 1970s, before imprisonment rates in the US started increasing, state imprisonment rates (excluding jails and federal prisons) across the US states varied from a low of 25 prisoners per 100,000 residents in North Dakota to a rate of 183 in North Carolina. By 2010, the range was from a low of 150 in Maine to 862 in Louisiana. This paper suggests that the increases were linked to, and may reflect, other social policies of exclusion.
In the current debate about how to bring American imprisonment rates down to those in other western countries, the focus seldom is on the 51 independent jurisdictions that may need to change in order to accomplish this goal. Other countries that have reduced their imprisonment rates dramatically (e.g., Finland – See Criminological Highlights 3(5)#1, 13(4)#8) have found that many legislative changes over a long period of time were needed, although in some cases (Alberta, Canada, for example: Criminological Highlights 14(3)#6) reduced imprisonment rates occurred quickly as a result of local administrative changes. For the US, then, 51 separate initiatives would be necessary to bring down imprisonment in each jurisdiction.
In important ways, then, there is no US imprisonment rate. The build-up of imprisonment in the US is interesting in part because, like Australia (see Criminological Highlights 15(4)#5) increases happened in each state but the size of the increase was not uniform. Maine and Minnesota increased their imprisonment rates by ‘only’ 100 and 144 (prisoners per 100,000 residents), respectively. At the other end of the spectrum Mississippi and Louisiana increased their rates by 616 and 749, respectively. Variability, is then, the rule, and more interestingly, the states that increased their imprisonment rates most between the early 1970s and 2006-2010, also started with relatively high imprisonment rates. In other words, the most punitive states in the early 1970s tended to increase their punitiveness the most, suggesting that punitive sentencing policies are linked to other more fundamental values.
Not surprisingly, then, those states in which capital punishment was permitted (in 2012) were more likely to have high imprisonment rates in 1970-74 and in 2006-10, and to have the largest increases in imprisonment compared to abolitionist states. The states that disenfranchise those convicted of felonies (see Criminological Highlights 5(5)#1) also had larger increases in imprisonment and higher imprisonment rates in 2006-10 than those that did not remove this right of citizenship from those who have served their sentences for a felony. Similarly, those states that have low minimum wages (below the federal rate) or that ban certain forms of welfare assistance for ex-felon drug offenders increased their imprisonment rate more and ended up with higher rates than those states with more generous minimum wages or states that did not remove the right to welfare from ex-felon drug offenders. Said differently, harsh sentencing policies are associated with harsh policies in other social domains.
Conclusion: “That imprisonment rates are linked to … other social policies suggests that high imprisonment cannot be addressed in a vacuum that ignores social and cultural factors or values. Intuitively, it seems reasonable to expect that states that practice capital punishment, do not allow ex-felons to vote or receive certain welfare benefits after the expiry of their sentences, and have a low minimum wage… would likely need different types of political pressures to reduce their (high) imprisonment rates than are states on the other end of each of these dimensions. Hence changing imprisonment policy in the US may present a greater challenge than in many non-federal states” (p. 556).
Reference: Doob, Anthony N. and Webster, Cheryl Marie (2014) Creating the Will to Change: The Challenges of Decarceration in the US. Criminology & Public Policy 13(4), 547-559.
Interventions designed to provide reintegration services for those serving short prison sentences can reduce both violations of conditions of release as well as reoffending.
Although rehabilitation programs in prison have received a good deal of attention for decades, a focus on the reintegration of prisoners released from prison is fairly new (see Criminological Highlights 6(1)#1). This paper reports the evaluation of a Minnesota program aimed at providing basic support to prisoners released from prison who had previously violated conditions of supervised release. The prisoners who were the focus of the program, then, were those deemed to be high risk to fail.
The study randomly assigned prisoners released after serving no more than 6 months in prison to one of two conditions. They either received ‘treatment as usual’ (involving setting the conditions of release and giving help in finding appropriate housing) or they received a more concentrated ‘Second Chance’ treatment program involving access to a one-stop community ‘hub’ where a wide range of different services (e.g., housing and employment) were available in one location. But in addition, those who received the ‘Second Chance’ treatment also were eligible for up to 75 days of transitional housing (sometimes receiving cash assistance for housing), up to 16 weeks of subsidized employment as well as referrals to non-subsidized employment programs and training programs for a year. The ‘second chance’ group also had access to weekly ‘life skills” programming and were given bus passes (for a year). For the treatment group, contact with the community service providers began before release so that when release occurred, all of the necessary help was in place and prisoners knew how to access services. For this reason, in order to be eligible for the program, prisoners needed to have at least five months left in their sentence before being assigned to receive the program. Although there were some problems in the assignment of prisoners (e.g., some were assigned to the treatment program but were later found not to meet the criteria), the control and treatment groups appeared to be comparable on most variables.
The ‘second chance’ group clearly received more services than the control group in terms of community based cognitive-behavioural programming, employment assistance, and transportation assistance and participation in hub-based ‘one-stop’ services. On housing the services received seemed comparable in quantity of assistance, though the exact nature of the help probably differed.
The two groups were followed for at least one year. The ‘second chance’ prisoners were less likely than the control group to have their release revoked because of a violation of condition of release. More importantly, perhaps, ‘second chance’ prisoners were less likely (23%) to be reconvicted for a new felony offence than those in the control group (31%). Other analyses suggested that “employment assistance, including subsidized employment, was one of the most influential factors in reducing recidivism risk” (p. 210).
Conclusion: This prisoner re-entry program was designed for a particular group of prisoners who were known to be at risk for violating conditions of release. Its success, therefore, should not be interpreted as meaning that any ‘re-entry program’ will work equally well. In this study, it was demonstrated that “Dedicating some extra planning and resources toward… high risk short-term offenders [who have previously violated conditions of release] can significantly reduce recidivism” (p. 211). “In line with established principles of effective correctional treatment…. the targeting of high-risk prisoners for intensive programming, the use of highly trained staff in a structured program” (p. 212) may account for the apparent success of the program.
Reference: Clark, Valerie A. (2015). Making the Most of Second Chances: An Evaluation of Minnesota’s High-Risk Revocation Reduction Reentry Program. Journal of Experimental Criminology, 11, 193-215.
Contact with the police can increase the likelihood of future violent offending for those already involved in small amounts of violent crime, but not for those who, previously, were not involved in violent crime.
There is substantial evidence that for young people, contact with the youth court is more likely to increase future offending than to reduce it (see Criminological Highlights 14(6)#1). This paper examines the effect of contact with the police on subsequent offending, taking advantage of the fact that many youths who commit offences do not get apprehended for these offences.
In this study, boys in Rochester, NY, were interviewed every six months starting when they were approximately age 13 until they were about age 17. The focus of the study was on self-reported violent crime. Since most violent crime (e.g., fights between youths) does not come to the attention of the police, this was probably the best measure of involvement in violence for these youths. The measure used was the number of different kinds of violence (attacking someone with a weapon, throwing things at someone, robbery, etc.) the youth engaged in. Because the youths were interviewed regularly during this period of time, it was possible to classify different youths into different groups in terms of their involvement in crime. In this case, youths appeared to fall into three relatively distinct groups: those who reported close to no violence in early adolescence (39% of the sample), those involved in relatively little violence (49% of the sample) and those involved in a substantial amount of violence (11% of the sample). Not surprisingly, these three groups varied dramatically in how much police contact they had experienced before age 16.
Within each of the three groups, youths were identified who had and had not been picked up and formally questioned by the police for suspected involvement in crime between age 16 and age 17-18. Youths who had been in contact with the police were then matched with those who had not had police contact on a wide range of measures (using their predicted likelihood of having police contact based on their previous reported behaviour and other measures such as race , neighbourhood characteristics, family structure, peer associations, prior justice system contact, etc.). For the high offending group, separate from the other groups, however, acceptable matching was not possible. Hence it is not possible to look at the impact of police contact on this group. However, since this group would likely have had contact with the police earlier in their lives, it is likely that an additional police contact would not have much additional impact on them.
It appeared that contact with the police had very little, if any, impact on the non-offending group. For the low-offending group, however, there was an effect: police contact appeared to increase subsequent involvement in violence in the 1.5 years following the contact. “When individuals are successfully matched on 40 [variables], there is empirical evidence for a short-run labeling effect of the police contact treatment for the low offending… group” (p. 458-9).
Conclusion: The fact that police contact with youths who have, thus far in their lives, engaged in some, but not very much violence, has the effect of increasing subsequent violence suggests that “the police are [faced with] a most difficult task. [In responding to possible offending by these youths] police intervention may unintentionally make the offending problem worse in the short run” (p. 459). The effect of police contact was not found for the (previously) non-offending group, in part perhaps, because they show stronger attachment to parents and school and have fewer delinquent friends.
Reference: Ward, Jeffrey T., Marvin D. Krohn, and Chris L. Gibson (2014). The Effects of Police Contact on Trajectories of Violence: A Group-Based, Propensity Score Matching Analysis. Journal of Interpersonal Violence, 29(3), 440-475.