Effectiveness of a Domestic Abuse Program for Australian Indigenous Offenders

Abstract

The subject of this study is to assess the effectiveness of a culture-neutral domestic abuse program (DAP) developed for offenders with domestic and family violence histories, when used for Australian Indigenous offenders, serving community-based supervised orders. The study employed a retrospective quasi-experimental research design and intention-to-treat program definition for 953 DAP-treated men and propensity score–matched controls, from diverse cultural, religious, and racial backgrounds, 19% being Indigenous Australians. Outcomes measured were program completion, time to first reconviction, and reconviction rates. Indigenous men completed the program similar to non-Indigenous men (58% vs. 63%; NS), although significantly more dropped out (22% vs. 18%); 63% of Indigenous DAP-treated men, remained reconviction free versus 49% of Indigenous controls. Significant therapeutic benefits required program completion, 73% Indigenous and 74% of non-Indigenous men remained reconviction free. Indigenous DAP participants, relative to controls, took significantly longer to first reconviction. Survival was associated with prior criminal histories, but not with rural or remote domiciles. Reconviction rates were predicted by Level of Service Inventory–Revised actuarial risk scores and by DAP completion, but not by Indigenous status. Program effect size was d = .477, mean reconvictions for Indigenous DAP enrollees being 50% lower than controls. Generic domestic violence interventions utilising evidence-based theoretical principles can be effective for Indigenous offenders, despite an absence of specific Indigenous cultural, or healing content, and delivery by Indigenous facilitators or Elders.

Keywords

domestic violence group intervention aboriginal Indigenous offender rehabilitation program evaluation recidivism reconviction rates risk factors survival

Indigenous Australians (Aboriginal and Torres Strait Islander peoples) are more likely to be victims of violence than any other ethnic group in Australia (Gordon et al., 2002; Wild & Anderson, 2007) with higher rates of contact with the criminal justice system than non-Indigenous Australians (Australian Law Reform Commission [ALRC], 2017). In New South Wales [NSW], Australia, Indigenous people comprise 3.4% of the general population but 29% of prison and community-based populations (Australian Bureau of Statistics, 2018a). NSW Indigenous offenders are estimated to be 1.5 times more likely than non-Indigenous offenders to reoffend after release from prison (NSW Bureau of Crime Statistics and Research [BOCSAR], 2019). Domestic and family violence offenders represented 21% of all those serving supervised community-based orders with Corrective Services NSW (Corben, 2011; CSNSW, 2010). To meet the rehabilitative needs of this significant offender population, a generic domestic abuse program (DAP) was implemented in August 2007, initially for offenders serving community-based orders and subsequently for offenders in custody. Overall outcomes of this program were described in Blatch et al., 2016.

Assessing the effectiveness of the DAP for Indigenous men is the subject of this study.

Factors Associated With Family Violence

Family violence victimisation is 2 to 5 times higher among Indigenous than non-Indigenous Australians and many more times likely to result in hospitalisation (Australia’s National Research Organisation for Women’s Safety [ANROWS], 2018; Steering Committee for the Review of Government Service Provision [SCRGSP], 2016). Patterns of victimisation also differ, with assaults on both male and female Indigenous persons much more likely to be perpetrated by a spouse, partner, or other family member than is the case for non-Indigenous persons (Pointer, 2019). Research on the factors associated with violence and victimisation in Indigenous communities suggests no single factor predicts the likelihood that an individual perpetrates violence, but predictive factors are multidimensional, with many fundamental antecedents shared with a range of other health and social problems (Anderson, 2002). Many authors posit that family violence in Indigenous communities can only be understood in the context of the historical impacts of colonisation and marginalisation (Blair et al., 2014); family separation, through the large-scale removal of Indigenous people into institutions (Gordon et al., 2002; Smallacombe, 2004); and detrimental intergenerational impacts resulting from the stolen generation and changes in the social roles of men and women (Cheers et al., 2006; Cripps, 2007; Davis & Taylor, 2002; Day et al., 2012). These significant historical impacts caused breakdowns of traditional culture and kinship practices, disconnection from cultural roots (ANROWS, 2016), which undermined social norms (Memmott et al., 2001). Other authors contend that ongoing contemporary political and socioeconomic disadvantage issues (Al-Yaman et al., 2006; Anderson, 2002; Gordon et al., 2002) also contribute to community violence, the most significant being basic housing, education, employment, and health (Memmott et al., 2009; Weatherburn, 2014; Widders, 2003; Wundersitz, 2010). Socioeconomic stressors beyond the control of the offenders have been inextricably linked to reoffending (Allan & Dawson, 2004; Weatherburn, 2014).

An assessment of theories of Indigenous violence (Snowball & Weatherburn, 2008) found little support for anomie theories that linked violence to a loss of social and cultural meaning, value and purpose resulting from the impacts of colonisation, and only moderate support for social disorganisation theories linking the impacts of colonisation to a breakdown of Indigenous social structures. In a later work examining ways of reducing Indigenous incarceration, Weatherburn (2014) asserted that the impacts of colonisation do not adequately explain continuing high rates of Indigenous overrepresentation as imprisonment rates did not increase until after the processes of colonisation had largely been completed. He also highlighted what he saw as a lack of evidence that systemic bias and racism were a factor in Indigenous overrepresentation. These conclusions have been criticised by Indigenous academics including Guthrie (2015) and Williams (2016), for failing to take into account the full body of evidence on these factors, leading to selective and inadequate conclusions. In the view of these other academics, colonisation and dispossession cannot be dismissed as factors contributing to Indigenous offending—including violent offending—and contact with the criminal justice system.

Less controversially, Snowball and Weatherburn (2008) found some support for social deprivation theories that see economic and social disadvantage as directly contributing to violence. Their study found the strongest support for lifestyle/routine activity theories, mainly resulting from high-risk alcohol assumption emerging as the strongest single predictor of engagement in violence (Snowball & Weatherburn, 2008). Lifestyle and routine activity factors such as living in a problematic neighbourhood, substance misuse, having a disability, and residing with someone who has been charged with an offence were less strong but still significant predictors of Indigenous violence (Snowball & Weatherburn, 2008).

The misuse of alcohol has been identified in other studies as one of the strongest predictors of an Indigenous person being arrested and charged for any offence (Chan, 2005; Hunter, 2001; Ramamoorthi et al., 2014; Weatherburn et al., 2006, 2008) and has been specifically identified as a significant contributor to violent offending and victimisation (Bryant & Willis, 2008; Macklin & Gilbert, 2011; Memmott, 2010; Weatherburn, 2014; Weatherburn et al., 2006). Other studies have also noted the contribution to Indigenous violence of lifestyle factors such as illicit drug use, gambling, childhood exposure to violence, abuse and pornography, low education levels, physical and mental health issues, and destructive coping behaviours (Al-Yaman et al., 2006; Cripps, 2007; Gentle & Taylor, 2002; SCRGSP, 2016; Weatherburn, 2014; Wundersitz, 2010). Poor parenting has also been noted as a significant risk factor for Indigenous violent perpetration in Indigenous communities (Queensland Crime and Misconduct Commission, 2009; Weatherburn, 2001; Weatherburn & Holmes, 2010). Living in remote regions also predicts violent offending and is a significant risk factor for higher rates of hospitalisation due to violent victimisation (Al-Yaman et al., 2006; Pointer, 2019; SCRGSP, 2016).

What appears clear from these studies is that there is no one cause of Indigenous violence—rather, it occurs as the result of a constellation of factors that interact with one another and that have differential effects across time, place, communities, and individuals. Given these multifactorial associations with interpersonal violence, several researchers have recommended that when developing rehabilitative programs for Indigenous offenders, Indigenous issues of disempowerment, loss of culture and lands, Aboriginal culture and law, marginalisation, family separation, socioeconomic disadvantage, and traumas should specifically be addressed (Cunneen, 2010; Day et al., 2003; Macklin & Gilbert, 2011). Cunneen (2001a) identified three themes as central to successful Indigenous rehabilitation and crime prevention programs: (a) enhancing self-determination, (b) holistic approaches, and (c) promoting empowerment rather than dependency. Others posit that successful Indigenous rehabilitative programs require local-level development in consultation with Indigenous communities (Cripps & Davis, 2013; facilitation by Indigenous staff and/or Elders (Brown & Languedoc, 2004); culturally specific content; an individual “healing” focus addressing spiritual and cultural loss (Boersig, 2007; McKendrick et al., 2017; Office of the Status of Women [OSW], 2001); engagement of the family, reconnection with homelands and traditional culture, and relevance to the life experiences of Indigenous people (Indigenous Justice Forum, Attorney-General’s Department, 2012; Royal Commission into Aboriginal Deaths in Custody, 1991; Trevethan et al., 2005; Willis & Moore, 2008). These contentions held by many Australian academics, Indigenous peoples, and their representatives (Brown & Languedoc, 2004; Cripps & Davis, 2012; OSW, 2001) are yet to be empirically tested.

Willis and Moore (2008) have postulated that program efficacy and integrity can be diluted, or lost, in efforts to make culturally specific adaptations to programs. These researchers also reported that some Indigenous male offenders express preferences for mainstream racially mixed, rather than Indigenous-specific, programs. Additional support for treating Indigenous offenders in mainstream programs with other ethnic and racial groups comes from the similarity of risk factors for police charges and incarceration for both Indigenous and non-Indigenous men. These risk factors include alcohol and illicit substance use, unemployment and financial stressors, a lower age, being a sole parent with dependent children, being domiciled in a crime-prone area, social disengagement, low educational levels, socioeconomic disadvantage factors, and childhood abuse experiences (Hunter, 2001; Jones et al., 2006; Snowball & Weatherburn, 2006; Weatherburn et al., 2008; Wundersitz, 2010).

In the United States, Gondolf and Williams (2001) initially posited that separate culturally focused interventions for African American domestic violence (DV) offenders would yield better therapeutic outcomes than racially mixed programs. After comprehensively evaluating three counselling approaches, Gondolf (2008) reported outcomes for culturally focused counselling for African-Americans, to be no better than conventional interventions; completion rates were similar (55%), as were reassault rates (23%). Comparing urban African American experiences with those of Australian Indigenous men may lack surface validity, but the research questions and investigative pathway parallel current Australian debates.

Other Australian researchers have suggested that the effectiveness of family violence programs for Indigenous offenders could be improved by addressing the cognitive-affective instigators and processes preceding violence and aggression (Day et al., 2018) and by applying best practices used in violence prevention programs for general offender populations (Day et al., 2009). The application of Risk-Needs-Responsivity (RNR) principles to programs for Canadian Aboriginal offenders was found to effectively reduce reoffending by 35% when delivered in the community and by 17% in custody (Andrews & Bonta, 2006; Rugge, 2006). Day et al. (2018) identified no studies that confirmed the effective application of RNR principles to Australian Indigenous peoples or that included a control group. A review of Australian offender treatment programs (Heseltine et al., 2011) found no published evaluations of either treatment programs for violent Indigenous male offenders or family violence programs (Cripps & Davis, 2012). Research has primarily addressed family violence prevention (Macklin & Gilbert, 2011), though some more recent work has looked at innovative and promising models for rehabilitative interventions in remote Indigenous communities (Blagg et al., 2018).

The present study aims to assess the effectiveness for Australian Indigenous offenders of a generic DV intervention, based on theoretical principles of empirically demonstrated clinical efficacy.

Description of the Domestic Abuse Program

The DAP behaviour change group program addresses common aspects of domestic and family violence arising in many cultures. DAP content does not explicitly focus on any unique racial, ethnic, or cultural factors in DV contexts. The program primarily targets men at “moderate” to “high” risk of re-offending measured on the Level of Service Inventory–Revised (LSI-R; Andrews & Bonta, 1996). It aims to modify patriarchal perceptions of women, identify and change domestically abusive and violent behaviours perpetrated against partners, and develop new communication and conflict management skills. Content is based on empirically supported interventions and features known to reduce re-offending. Several theoretical approaches are utilised, in particular RNR principles (Andrews & Bonta, 2006; Rugge, 2006) and cognitive behavioural therapy (CBT; Hatcher et al., 2008; Polaschek, 2011a; Thakker & Gannon, 2010). One module contains Duluth-style feminist-based material (Miller, 2010). Participants are required to take responsibility for their own behaviours. DAP is delivered in a closed-group format by two facilitators, a male and female. Twenty manualised sessions of 2 hours each are divided into five modules: (a) Identifying Abuse contains psychoeducational material to develop participants’ knowledge of abuse and CBT techniques to challenge antisocial and pro-violent cognitions (Thakker & Gannon, 2010); (b) Managing Emotions, Beliefs, and Attitudes challenges participants’ cognitions, feelings, and behaviours associated with the promotion, or maintenance of violent behaviours (Hatcher et al., 2008; Polaschek, 2011a; Thakker & Gannon, 2010); (c) Offence Mapping uses behaviour chain analysis to identify offending antecedents, warning signs, and high-risk situations (Polaschek, 2011a; Thakker & Gannon, 2011); (d) Victim Impact introduces psychoeducational and Duluth-style feminist theory concepts (Miller, 2010) to address victim empathy and family impacts (Thakker & Gannon, 2010); (e) Sexual Respect, Relationship Skills, and Safety Strategies model conflict management, communication skills, and dispute resolution techniques. The fourth session of each module has no prescribed content and is used to match the module’s topics to the specific circumstances shared by the group participants. The program’s content, delivery methods, and resources are described in the CSNSW (2012a) Compendium of Correctional Programs and the CSNSW (2012b) Domestic Abuse Program, Training Manual.

The DAP meets criteria for a “Basic Level Rehabilitation Program” (Polaschek, 2011b) given its length, adherence to manualised content, methods of ensuring program integrity, and levels of facilitator training. By 2011, 204 DAP programs with 2,555 men enrolled had been delivered in 32 of 62 NSW Community Corrections Offices, earning a Crime and Violence Prevention Award from the Australian Institute of Criminology in 2011.

First Evaluation of the Domestic Abuse Program

A prior quasi-experimental study (Blatch et al., 2016) evaluated DAP effectiveness for 1,906 mainstream offenders, who reflected the multiracial ethnic and religious mix of the State’s general offending population, were serving community sentences or parole orders, and were enrolled in community-based programs. Applying the five-category Maryland Scale of Scientific Methods (MSSM; Sherman et al., 1998), this study met Level 4 criteria (good statistical power and quasi-experimental research design, matched groups or a statistical control methodology, low measurement error, and control for risk factors). In comparison, a review of 69 studies comparing outcomes for treated sex offenders and a control group, showed only 9% met MSSM Level 4 criteria (Lösel & Schmucker, 2005). This 2016 DAP recidivism study demonstrated that DAP enrolment (n = 953) significantly improved odds of time to a first general reconviction and to first violent reconviction and reduced overall reconviction rates relative to 953 propensity score (PS)-matched controls (Blatch et al., 2016). Program efficacy for Indigenous participants was not analysed.

Subsequent to these findings, the NSW Department of Justice requested development of a culture-specific DV rehabilitation program suitable for Indigenous men. The applicability of the DAP to the needs of Indigenous men was questioned, as the program was neither facilitated by local Indigenous Elders, nor included any culture-specific content as recommended by Cripps and Davis (2012 & 2013) and Brown and Languedoc (2004).

Rationale for the Present Study

Prior to designing a new culture-specific DAP for Indigenous men, the authors reanalysed the data from the 2016 DAP evaluation (Blatch et al., 2016) to ascertain whether Indigenous participants in that cohort derived any therapeutic benefit. Of particular concern was DAP suitability for Indigenous men in remote and rural locations. These men live more traditional lifestyles, with greater educational, socioeconomic and other disadvantages, including lack of transport to attend weekly or biweekly DAP sessions, at supervising Community Corrections District Offices, compared with urban-dwelling Indigenous men. There was also a concern that some Indigenous people fear accessing non-Indigenous services (Laing, 2013).

Research Aims

The aim of the present study was to explore the degree to which the generic DAP was effective for the Indigenous participants as measured by the following:

The proportions of Indigenous men living in rural and remote locations, successfully completing the DAP, benchmarked against completion rates for the non-Indigenous participants;

The elapsed time before a new reconviction (survival time) for the treatment-assigned Indigenous DAP subgroup, compared with (a) the Indigenous control subgroup and (b) the non-Indigenous DAP treatment subgroup;

Rates of reconviction for the Indigenous and non-Indigenous subgroups, adjusted for time free to reoffend during follow-up;

Factors associated with (a) DAP completion and (b) reconviction(s) during follow-up;

Comparing rural/remote and metropolitan dwelling Indigenous men for (a) DAP completion and (b) reconviction rates;

Comparing treatment effect sizes for (a) the full DAP and control cohorts and (b) the Indigenous DAP and control subgroups.

Method

This study uses the full data set utilised for the 2016 evaluation of the DAP (Blatch et al., 2016). The methodology for that publication is repeated here.

Research Design

The study employed a retrospective quasi-experimental design “intention-to-treat” methodology and a PS-matched control group. An “intention-to-treat” design includes all participants enrolled in the DAP, including those but who failed to commence or failed to complete the program (“dropped out”). Research designs that only use program “completers” introduce strong selection biases; it is well documented that “dropouts,” and those who fail to start or even attend one session, are at higher recidivism risk and more likely to reoffend than those who complete (Bennett et al., 2007; Palmer et al., 2007). Polaschek (2011a) argues for evaluations to include all participants, either agreeing to or “required” to attend treatment (“intention to treat”), including those not completing and/or failing to start. The design of this study accordingly reduces selection biases that favour better outcomes (Feder & Wilson, 2005; Bennett et al., 2007). The control group was selected to meet all the essential suitability and eligibility criteria required for enrolment in a DAP, but who did not attend a DAP, either because the program was not yet provided by their supervising Community Corrections Office or no program was available that could be completed within the span of their community order.

Definitions and Variables

a). Data definitions

i. Follow-up start date (index conviction date for supervised community orders, or supervised parole release date for each individual)

ii. Census date (final date for observation of potential reoffending—June 30, 2010)

iii. Follow-up period (time from the follow-up start date to the census date)

iv. Equivalent follow-up durations (as the maximum follow-up period for the DAP treatment group was shorter [991 days], the follow-up period for the control group was capped at 991 days to create equivalent follow-up durations)

v. DAP completion status (“completed” defined as attending a minimum 16 sessions out of 20; “dropped out” defined as attended one or more sessions but <16; “did not start” defined as enrolled in a program, but failed to attended any sessions; not enrolled [control])

vi. Days free “at risk” (from each offender’s “follow-up period,” all days in custody, on remand, or in police cells were deducted, to give days free “at risk.” Poisson regression techniques were used to calculate rates of reconvictions, with the offset variable of total “days at risk” to adjust for differences in follow-up time)

vii. LSI-R risk measures (total actuarial LSI-R score and nine discrete domain scores: criminal history, education/employment, financial, leisure/recreation, companions, accommodation, alcohol/drug, emotional/personal, attitude/orientation; Level of Service Inventory–Revised, Andrews & Bonta, 1996)

viii. Indigenous status (formally verified)

ix. Age (as at index order/release start date)

x. Rural/remote or metropolitan domicile (“metropolitan” defined as a Community Corrections Office’s location in greater Sydney [the capital of NSW]; all other Offices classified as “rural/remote”)

xi. DAP enrolment status (“completed” defined as attending ≥16 sessions of 20, “dropped out” defined as attending a least one session but <16; “did not start” defined as enrolled but failed to attend any sessions, and not enrolled [controls])

Variables xii, xiii, xiv, xv, xvi, xvii are taken from the NSW BOCSAR database that records of all NSW convictions and custodial sentences:

xii. Most serious index offence (MSO); if concurrent orders were being served for a number of convictions, only the MSO was ranked, based on The National Offence Index [NOI]; Australian Bureau of Statistics, 2018b)

xiii. Conviction/reconviction (if several charges were finalised on the same date, only the most serious offence [MSO] is recorded/counted)

xiv. Number of convictions in 5 years prior to index conviction relating to the supervision order or custodial sentence start date (MSO only)

xv. Number of custodial sentences in 5 years prior to index conviction/parole release date (custodial sentences served concurrently, without a release between them, are counted as a single custodial episode)

xvi. Time to first reconviction (survival) (the number of days between an offender’s follow-up start date and their first reconviction date [(if any])

xvii. Number/rates of reconvictions (reconvictions summed from the index offence date for participants on community orders, or the custody release date for participants released on supervised parole, up to December 31, 2010, 6 months after the census date of June 30, 2010, to allow offences committed before the census date to be prosecuted and finalised. Treated participants’ reconvictions before the DAP commencement date were included to ensure equivalent reconviction start dates for the treatment and control groups

xviii. Reconviction status (no reconviction during follow-up [“reconviction free”] or at least one reconviction).

b). Dependent variables

Four dependent variables were used to measure outcomes, as applied in (Blatch et al., 2016):

Time to first reconviction (survival)

Reconviction status (none, one or more)

Number/rates of reconvictions during follow-up

DAP completion status

c). Independent variables

LSI-R risk measures

Most serious index offence (MSO)

Number of convictions in the 5 years prior

Indigenous status

Age

Type of community order (parole or community order)

Rural/remote or metropolitan domicile

DAP enrolment status

Most serious index offence (MSO)

Number of convictions in the 5 years prior

Number of custodial sentences in 5 years prior

Participants

A study group of DAP enrollees and a control cohort of non-enrollees were developed and subsequently dichotomised into subgroups of Indigenous and non-Indigenous offenders, as described below.

Identifying the potential DAP treatment group

CSNSW’s electronic Offender Information Management System (OIMS) provided listings of offenders eligible for the 2016 study, namely those supervised by a Probation and Parole Officer from a Community Corrections District Office, between October 1, 2007 and June 30, 2010 (the “census date”) with domestic abuse as an identified need in their case plan; program eligibility as assessed by a Probation and Parole Officer using DAP enrolment criteria, absence of florid mental health issues/psychoses, drug and alcohol issues, cognitive or intellectual deficits, that would negatively impact group dynamics and impede meaningful group participation; and sufficient time remaining in their order to complete the program. Additional study criteria included a minimum of one day in the community from the index offence conviction date to the study census date.

Those study selection parameters yielded 1,146 records, from which were removed six duplicate records, 63 participants enrolled in a DAP after the census date, 25 participants with missing data (e.g., LSI-R scores, age, supervision order types, and/or gender), 95 participants who received earlier nonstandardised DV interventions, and 10 with insufficient follow-up time. This yielded a total of 957 male participants.

Identifying the control group

Population-based matching techniques identified potential control participants from the OIMS offender database, with similar risk and demographic characteristics to the DAP treatment group. This potential control cohort was extracted via queries from OIMS if members met eligibility criteria for DAP enrolment, but without actual enrolment; were serving a community-based supervision order in the CSNSW criminal justice system; and had an index offence court date and a completed case management plan which matched the DAP treatment cohort’s period (between July 1, 2007 and June 31, 2010; DV recorded in the offender’s Community Corrections case plan as a treatment need). Exclusion factors included not meeting eligible criteria for DAP interventions, florid mental health issue/psychoses, cognitive or intellectual disorders, or significant drug and alcohol issues that would prevent meaningful group participation.

These selection and exclusion criteria generated a potential control pool of 6,624 offenders. Removed from this pool were 775 offenders of female/unrecorded gender, 64 offenders with missing data, 29 duplicate records, six duplicated in the DAP cohort, and 18 serving home/periodic detention and therefore unable to attend a DAP. This left a pool of 5,734 potential controls to be used for the PS-matching procedure.

Factors Predicting DAP Cohort Membership

To aid the control group matching procedure, variables that predicted membership of the DAP treatment group were identified using binary logistic regression. All available risk factors empirically associated with reoffending described above were modelled (Andrews & Bonta, 2006; Delaney, 2010; Girard & Wormith, 2004; Hanson & Wallace-Capretta, 2000a, 2000b; Hilton et al., 2011; Olver et al., 2014).

Ten significant factors predicted membership of the DAP treatment group: LSI-R risk score, type of supervision order (three factors), MSO type (four factors), Indigenous status, and the number of custodial sentences in the 5 years preceding the index offence (Table 1).

Table 1.

Logistic Regression Model—Factors Predicting Membership of the DAP Treatment Group.

Variable	Coefficient	SE
LSI-R score	.184**	.039
Order type
Community service (baseline)
Parole	−.594**	.138
Supervision	−1.078**	.166
Most serious offence
Other MSO (baseline)
Assault/homicide	.372**	.122
Property damage	.218	.194
Regulatory offences/order breaches	.456**	.149
Indigenous status	.180*	.086
No. of custodial sentences in past 5 years	−.041*	.020
Constant	−1.909*	.217

Note. Treated n = 957, untreated n = 5,763. DAP = domestic abuse program; LSI-R = Level of Service Inventory–Revised.

p < .05. **p < .01.

Matching DAP Cases to a Control

When randomisation techniques cannot be used, a PS matching technique is the preferable option to minimise potential differences and biases (Duwe & Goldman, 2009) as it tends to balance all the observed covariates (Joffe & Rosenbaum, 1999). To select a control case for each DAP participant, a one-to-one PS technique was utilised (Duwe & Goldman, 2009).

Using stepwise logistic regression, the nine risk factors previously identified in Table 1, and Indigenous status, were used to calculate a single PS (between 0 and 1) for the 957 men in the DAP treatment group and all potential (untreated) controls (N = 5,734). A “nearest neighbour,” one-to-one procedure without replacement, was employed to match each DAP-enrolled participant to a control (Duwe & Goldman, 2009; Rosenbaum & Rubin, 1983). Treated participants were sorted randomly and sequentially matched to the case in the control pool with a PS closest to their own. When ties occurred, randomisation was used to select one match. The PS of each matched pair was compared. Where differences exceeded 0.1, the pair was removed. Four pairs were removed, yielding a final matched cohort of 953 in each group (Table 2).

Table 2.

Propensity Score Matching and Covariate Balance Between Treated and Untreated Participants Before and After Matching.

Variable		Matched Sample			Unmatched Sample			Total
Variable		Treated	Untreated	Bias	Treated	Untreated	Bias	Bias reduction (%)
Propensity score	M	0.17	0.17	0.01	0.17	0.16	28.78	99.97
Propensity score	SD	0.04	0.05		0.01	0.01
Age (years)	M	33.39	33.46	−0.75	33.39	33.54	−1.6	53.19
Age (years)	SD	9.13	9.58		83.44	92.87
Total LSI-R score	M	23.74	23.8	−0.12	23.74	23.45	3.37	103.59
Total LSI-R score	SD	8.25	8.27		68.1	80.15
Times in custody (past 5 years)	M	0.66	0.65	0.57	0.66	0.94	14.35	103.99
Times in custody (past 5 years)	SD	1.75	1.75		3.05	4.37
Indigenous Y/N	%	19	19	2.17	19	22	12.07	117.0
Indigenous Y/N	SD	0.15	0.15		0.15	0.17
Index MSO (vs. other offences)
Assault/homicide	%	76	76	−1.09	76	72	11.97	109.11
Assault/homicide	SD	0.18	0.18		0.18	0.2
Property damage	%	4	4	9.94	4	5	–4.65	313.64
Property damage	SD	0.04	0.04		0.04	0.05
Order/regulatory	%	12	13	−2.47	11	12	6.38	138.68
Order/regulatory	SD	0.11	0.11		0.1	0.11
Community order:
Supervision	%	81	80	1.69	0.81	0.76	16.98	90.05
Supervision	SD	0.16	0.16		0.16	0.18
Parole	%	12	13	−3.63	0.12	0.2	–40.2	90.98
Parole	SD	0.11	0.11		0.11	0.16

Note. LSI-R = Level of Service Inventory–Revised; MSO = most serious index offence; M = mean; SD = standard deviation.

Treated n = 953, total untreated n = 5,734, matched untreated n = 953. Categorical variables are given as percentages versus the base case. Bias values exceeding 20 are considered significant.

Results of Matching Procedure

After removal, a residual bias assessment (Rosenbaum & Rubin, 1985) quantified the reduction in the differences between the treated and untreated groups, before and after matching. The total PS difference between the groups was 0.01 (Table 2). This relatively small PS change estimate reflects the degree of homogeneity already attained within the treatment and potential control cohorts prior to PS matching, due to population matching techniques and DAP enrolment eligibility criteria being used as a filter to select potential controls from the OIMS offender database.

Subgroups within the Matched Treatment and Control Cohorts

Researchers such as Rassen et al. (2012) have reported that PS-matched cohorts can be used with acceptable accuracy to examine subgroup effects, if there are large numbers of outcome events in the subgroups and particularly if the subgroups are binary (Loughran et al., 2010).

To meet this study’s aims, the PS-matched treatment and control cohorts of 1,906 men (Blatch et al., 2016) were dichotomised into binary subgroups capturing the Australian Indigenous (n = 359) and non-Indigenous offenders (n = 1,547). Two other subgroups, dichotomised into rural/remote and metropolitan enrollees were created, to explore concerns that the DAP might not be suitable for rural Indigenous offenders. Of 691 rural DAP enrollees, 149 were Indigenous men.

Analyses

The following analyses were undertaken:

Two survival analyses employing Cox regression (stepwise log rank) measured survival to first reconviction, for the Indigenous subgroup of DAP enrollees compared with (a) the Indigenous control subgroup and (b) non-Indigenous DAP subgroup;

A Poisson regression calculated rates of reconviction rates using total days free “at risk” as an offset variable, to adjust for differences in follow-up time and using Indigenous status and a rural/metropolitan domicile as covariates;

Two binomial logistic regressions to identify factors associated with (a) DAP completion and (b) and reconviction status during follow-up;

Chi-square analyses examined the proportions of rural and metropolitan offenders (a) completing a DAP and (b) remaining reconviction free;

Cohen’s d determined effects size for (a) the full DAP and control cohorts and (b) the Indigenous DAP and control subgroups, using standardised mean reconviction differences.

Risk Factors in Regression Models

In the present study, all available risk factors empirically associated with recidivism were entered as covariates into the Cox, Poisson, and binomial logistic regression models. These 20 factors included Indigenous status (yes/no); total LSI-R scores and criminogenic need domains (criminal history, education/employment, financial, leisure/recreation, companions, accommodation, alcohol/drug, emotional/personal, attitude/orientation); age in years; MSO type for the index order/sentence; number of incarcerations and convictions in the 5 years prior to the index sentence, or parole release date; DAP completion status (completed, dropped-out, did not start, and control); rural versus metropolitan location.

Insignificant risk factors were removed from all reported models unless specifically relating to the study’s aims (e.g., Indigenous status) or contributing to the model’s explanation (e.g. age).

Results

Cohort Characteristics

The total cohort

The DAP cohort’s mean age was 33.1 and 33.4 years for the controls. Both cohorts contained equal numbers in four of the five LSI-R risk categories (50%) with the exception of more “high-risk” cases in the DAP cohort (52% vs. 48%), introducing a conservative bias. The majority (73%) served their supervised orders outside metropolitan Sydney (rural).

The Indigenous subgroups

To assess equivalence between the Indigenous treatment and the control subgroup, two-tailed t tests (equal variances not assumed) were performed on all continuous risk variables. No factors approached significance, suggesting the two Indigenous subgroups were of acceptable homogeneity. NOI (Australian Bureau of Statistics, 2018b) scores assessed seriousness of the MSO index offence: mean NOI score for the two Indigenous subgroups was 42.5 (DAP treated) and 43.5 (controls); the LSI-R Criminal History Score was 5.6 for the Indigenous DAP subgroup and 5.7 for the Indigenous control subgroup; mean ages were 30.6 and 31.8 years, respectively.

Similarly, the NOI Criminal History Scores for the rural Indigenous and the rural non-Indigenous DAP subgroup were 43.8 and 42.5, respectively. The LSI-R Alcohol and Substance scores were similar (5.2 and 5.0, respectively), although the LSI-R Criminal History Scores were significantly higher for DAP-treated rural Indigenous men compared with non-Indigenous treated men (5.5 and 4.7, respectively).

Survival to First Reconviction: Indigenous DAP Enrollees and Indigenous Controls (n = 359)

Following a Kaplan–Meier estimator to ensure the proportional hazards assumption was not violated, all 20 recidivism risk factors were fitted into a Cox regression model to measure survival time to first reconviction, for the Indigenous DAP-enrolled subgroup compared with their Indigenous control subgroup. The model showed that the DAP-treated men survived significantly longer than the controls (χ² = 7.983, df = 2, p < .05; survival curve at Figure 1). As no covariates, including age, were significantly associated with survival, the only factor affecting time to first reconviction was DAP enrolment (Table 3). Two insignificant covariates were retained in the model, as they contributed to the model’s explanation (Table 3).

Figure 1.

Survival curves for Indigenous DAP enrollees versus Indigenous control subgroups from order state date.

Table 3.

Survival Time to First Reconviction: Indigenous DAP Enrollees Compared With Indigenous Control Subgroup (n = 359).

Variable	Hazard ratio	95% confidence interval
Variable	Hazard ratio	Lower	Upper
LSI-R criminal history score	1.086	0.980	1.204
No. prior offences in past 5 yrs	1.065	0.947	1.198

Note. DAP = domestic abuse program; LSI-R = Level of Service Inventory–Revised.

Prior to DAP commencement, survival was initially similar for the Indigenous DAP and control subgroups (Figure 1). At approximately 4 months post sentencing, or release for custody, survival for the DAP subgroup became superior and remained so, relative to controls, until the conclusion of follow-up. The 4-month mark would coincide with either DAP program completion or near completion, suggesting early therapeutic gains from program participation for the Indigenous men.

Survival to First Reconviction-Indigenous Compared With Non-Indigenous DAP Enrollees (N = 953)

To determine whether DAP enrolment provided similar therapeutic survival benefits for Indigenous men relative to non-Indigenous men, a Cox regression was performed. All variables were initially entered into the model, but only variables that remained significant at the 5% level were retained in the final model (Table 4; age and area of domicile being nonsignificant and not contributing to the model’s explanation were therefore not retained). The final model showed significant differences in time to first reconviction for the Indigenous and non-Indigenous DAP subgroups (χ² = 17.707, df = 2, p < .001). Only two factors were significantly associated with survival: completing a DAP lengthened time to first reconviction by approximately 16% (hazard ratio = 1.165, confidence interval [CI] = [1.010, 1.342], p < .05), and higher LSI-R criminal history risk scores shortened survival by approximately 11% (hazard ratio = 1.109, CI = [1.052, 1.170], p < .001).

Table 4.

Survival to First Reconviction—Indigenous and Non-Indigenous DAP Enrollees (N = 953).

Variable	Hazard ratio	95% confidence interval
Variable	Hazard ratio	Lower	Upper
LSI-R vriminal history	1.109***	1.052	1.170
DAP completed vs. noncompleted	1.165*	1.010	1.342

Note. DAP = domestic abuse program; LSI-R = Level of Service Inventory–Revised.

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

The two survival curves were similar. Indigenous and non-Indigenous survival was equivocal for the first 3 months, but from this time onwards, coinciding with DAP participation and/or completion, Indigenous men’s survival improved (Figure 2) and remained moderately better than the non-Indigenous men’s survival, until approximately 14.5 months. After this, the two survival curves again merged, by which at this time, 30% of both Indigenous and non-Indigenous men had a new reconviction.

Figure 2.

Survival curves for Indigenous versus non-Indigenous DAP enrollees to first reconviction from order start date.

At approximately 22 months, Indigenous men’s survival deteriorated relative to the non-Indigenous men, suggesting that although their initial treatment benefits were superior, they were not as well sustained in the longer term.

Reconviction Rates for DAP Completers Versus the Combined DAP Nonstart/Drop Out Category (N = 1,906)

To determine whether completing a DAP reduced the number of reconvictions for Indigenous men, a Poisson regression measuring reconviction rates was performed, adjusting for each offender’s time at risk in the community, when free to reoffend. All 20 risk factors were entered as covariates, with DAP completion status entered as a binary variable (completed vs. did not start, withdrew, and controls combined into a second category). In the final parsimonious model (Table 5), only significant variables were retained, except for Indigenous status and age, which although nonsignificant, were salient variables. Completing a DAP reduced reconviction rates by approximately 12% (risk ratio = 1.122, CI = [1.045, 1.206], p < .05). Each LSI-R risk increment increased the reconviction risk ratio by approximately 1.5% (risk ratio = 1.015, CI = [1.006, 1.023], p < .001). Being Indigenous did not increase reconvictions, as rates did not differ from those of the non-Indigenous men. The age variable was also nonsignificant, but was retained in the final model, as it contributed to the explanation of the variance.

Table 5.

Poisson Regression Model for Rate of Reconvictions (MSO)—Adjusted for Risk Exposure Time (N = 1,906).

Variable	B	SE	Risk ratio	95% Wald confidence interval
Variable	B	SE	Risk ratio	Lower	Upper
(Intercept)	−1.236	.2268	0.291**	0.186	0.453
Age	0.002	.0036	1.002	0.995	1.009
LSI-R total score	0.014	.0041	1.015**	1.006	1.023
Indigenous status (yes/no)	0.057	.0822	1.058	0.901	1.243
DAP completion (yes/no)	0.116	.0366	1.122*	1.045	1.206
(Scale)	825.411^a

Note. MSO = most serious index offence; LSI-R = Level of Service Inventory–Revised; DAP = domestic abuse program.

df = 1.

Based on the Pearson chi square.

p < .05. **p < .01.

These findings suggest that the main factors associated with reduced reconviction rates were completing a DAP and having lower LSI-R criminal risk scores (Table 5).

Factors Predicting Program Completion/Noncompletion (N = 953)

To identify whether being Indigenous, and having a rural/remote domicile was associated with failing to complete a DAP, a binomial logistic regression was employed, entering all risk factors. Table 6 shows only four factors significantly predicted DAP completion. Decreasing LSI-R risk increments improved the likelihood of completing a program (exp = 39.5, p < .001), as did each increase in age (exp = 10.2, p < .001). A substantial criminal history, measured by both higher numbers of prior incarcerations and convictions, reduced completion odds (exp 5.6 and exp 4.3, p < .05 respectively). Indigenous status (no/yes) did not significantly contribute, suggesting that these men were not at greater risk of program failure/success than non-Indigenous men, with the severity of their past offending and actuarial measures of criminal risk, being independently more important determinants. Area of domicile was not a significant factor.

Table 6.

Binomial Logistic Regression Predicting DAP Completion Versus Noncompletion (N = 953).

Variables in equation	B	SE	Wald	Exp (B)
Step 0 Constant	0.501	.067	56.212	1.650***
Indigenous status	−0.137	.182	0.565	0.872
Convictions in past 5 years	0.123	.059	4.303	1.131*
Incarcerations in past 5 years	−0.119	.050	5.684	0.888*
LSI-R total score	−0.062	.010	39.559	0.940***
Age	0.025	.008	10.213	1.025***

Note. DAP = domestic abuse program; LSI-R = Level of Service Inventory–Revised.

p < .05. **p < .01. ***p < .001.

Factors Predicting a Reconviction in DAP Enrollees (N = 953)

This binomial logistic regression sought to identify whether any particular factors were likely to be associated with a subsequent reconviction in DAP enrollees, particularly for Indigenous men, to inform the need for (a) specific motivational/precursor programs before DAP enrolment and/or follow-up maintenance programs and (b) new program content to better address any salient risk factors. Only three factors significantly contributed to a reconviction during follow-up. These were (a) higher numbers of convictions in the prior 5 years (exp = 7.95, p <.001), (b) higher LSI-R alcohol/other drug domain scores increased reconviction odds by exp 3.82 (p < .05), (c) and not completing the DAP. The odds of a reconviction were 5.45 times higher for those who either dropped out or failed to start their DAP (p < .005). A new reconviction was not associated with being Indigenous, suggesting these men were at no higher reconviction risk than other enrollees (Table 7), after controlling for the effects of criminal conviction histories and alcohol and drug-related risk. Participants’ ages and area of domicile were not significant determinants.

Table 7.

Binomial Logistic Regression Predicting Any Reconviction in DAP Enrollees (N = 953).

Variables in equation	B	SE	Wald	Exp (B)
Step 0 Constant	−0.884	.071	154.090	0.413**
Indigenous status (no/yes)	−0.327	.179	3.328	0.721
DAP completion vs. noncompletion	0.345	.148	5.449	1.412*
LSI-R AOD score	0.070	.036	3.822	1.072*
Convictions in past 5 years	0.132	.047	7.954	1.141**
Constant	−1.279	.250	26.185	0.278**

Note. DAP = domestic abuse program; LSI-R = Level of Service Inventory–Revised; AOD = alcohol and other drugs.

p < .05. **p <.01. df = 1.

Indigenous Offenders Completing a DAP in Rural Versus Metropolitan Areas (N = 359)

To address concerns that rural Indigenous men may be less able to start their prescribed DAP, and/or were possibly more likely to drop out, chi-square analyses compared the commencement, dropout, and completion rates of Indigenous to non-Indigenous men within both the rural and metropolitan locations.

Indigenous men in rural versus metropolitan areas

Contrary to expectation, the proportions of Indigenous men commencing a DAP in rural areas were similar to those of metropolitan Indigenous men (81% vs. 77%) as were the proportions completing a program (58% vs. 64%). However, more rural men dropped out (23% vs. 13%), suggesting that for some enrollees, weekly or biweekly travel to a town for their DAP sessions, was difficult to sustain (Table 8), possibly due to complex cultural and socioeconomic issues and lack of transport in remote areas.

Table 8.

Reconviction by Indigenous and DAP Completion Status.

DAP completion	Indigenous status		No reconviction		Any reconviction		Chi-square df = 1
Completed n = 593	Indigenous	n = 106	77	73%	29	27%	0.094
Completed n = 593	Non-Indigenous	n = 487	362	74%	125	26%	0.094
Withdrew n = 181	Indigenous	n = 40	21	52%	19	48%	5.633*
Withdrew n = 181	Non-Indigenous	n = 141	102	72%	39	28%	5.633*
Did not start n = 179	Indigenous	n = 36	15	42%	21	58%	8.407**
Did not start n = 179	Non-Indigenous	n = 143	97	68%	46	32%	8.407**
N = 953	Total		674	100%	279	100%

Note. DAP = domestic abuse program.

p < .05. **p < .01, two-tailed.

Indigenous versus non-Indigenous men

Using non-Indigenous rural men as a comparison benchmark, rural Indigenous men commenced the program at similar rates (81% vs. 80% non-Indigenous); proportions completing were reasonably similar (58% vs. 62%), but significantly more rural Indigenous men dropped out (23% vs. 18%; χ² = 1.000, p < .01, df = 1).

In metropolitan Sydney, the Indigenous men completed at a rate comparable with non-Indigenous men and fewer actually dropped out, but more failed to start (23% vs. 15.5%; Appendix Table A1).

Overall, these findings suggest that Indigenous men in rural areas are not always at a greater disadvantage than metropolitan Indigenous men; they were more likely to start a program, but as more dropped out, fewer went on to complete a DAP.

Reconvictions by Indigenous and DAP Completion Status

To ascertain the degree of therapeutic benefit to Indigenous men completing the DAP, chi-square analyses compared the reconviction status for treatment completers and noncompleters, relative to non-Indigenous men (Table 8). The proportions of Indigenous and non-Indigenous men completing a DAP and remaining reconviction free were similar (73% and 74%, respectively).

However, the negative ramifications of not completing DAP were of a greater magnitude for Indigenous men. Of Indigenous men who completed their program, 27% had a reconviction, but for those dropping out, the likelihood of having a reconviction increased to 48% (χ² = 5.633, p < .02) and then to 58% if they failed to attend any DAP sessions (χ² = 8.047, p < .005). These data suggest that the benefits of completing a DAP are of a greater magnitude for Indigenous enrollees.

Reconviction Free Status—Rural and Metropolitan Indigenous and Non-Indigenous DAP Enrollees

A comparison of rural Indigenous to metropolitan Indigenous enrollees, showed reasonably similar proportions remained reconviction free (61% vs. 69%, NS), but significantly fewer had no reconviction, compared with non-Indigenous rural enrollees (61% compared with 73%, χ² = 8.684, df = 1, p < .005). These data suggest that although Indigenous men living in both remote/rural and metropolitan communities are reasonably likely to be reconviction free, compared with non-Indigenous men in both rural and metropolitan areas, they are at significantly higher risk of having a new reconviction.

Effect Sizes

Cohen’s d was used to calculate effect sizes using standardised mean reconviction differences between both the full DAP treatment cohort (which included nonstarters and dropouts), the control cohort and the Indigenous DAP and Indigenous control subgroups. Equivalent follow-up durations for all four groups was achieved by truncating reconvictions at 991 days post the index offence conviction date.

For the Indigenous DAP enrollee subgroup relative to their control subgroup, the effect size was d = .48, which is a “moderate” effect size, but superior to the effect size of d = .26, for the full DAP enrollee cohort, relative to the full control cohort (Coe, 2002; Cohen, 1988; Sullivan & Feinn, 2012; Table 9).

Table 9.

Effect Sizes—Standardised Mean Reconviction Differences Between DAP Treatment and Controls with Equivalent Risk Exposure (N = 1,906).

Total cohort	DAP treatment n = 953	Controls n = 953	Total N = 1,906	Cohen’s d
Mean reconvictions	.3696	0.6466	.5233	d = .259
SD	.7225	1.0184	.9138
Percent without reconviction	71	58	66
Indigenous cohort	DAP Indigenousn = 180	Control Indigenousn = 179	All IndigenousN = 359	Cohen’s d
Mean reconvictions	.4514	0.9143	.6828	d = .477
SD	.73694	1.1473	.9706
Percent without reconviction	62	22	46

Note. DAP = domestic abuse program.

The higher effect size (“moderate” effect’) for the Indigenous subgroups, is due to the greater relative difference in mean reconvictions for the treatment group relative to their Indigenous controls. Treated Indigenous men had a reconviction mean of .4514, which was 50.4% lower than the control’s reconviction mean of .9143 (Table 9). In comparison, there was a relative difference of only 40% for the non-Indigenous men’s treatment and control subgroups (a mean of .3506 compared with .5847). When the full DAP and treatment cohorts were compared, the inclusion of the Indigenous men improved the difference in mean reconvictions between the full treatment and control cohorts, up to 42% (a mean of .3696 compared with .6466; Table 9).

These differences in mean reconvictions for Indigenous compared with non-Indigenous men, highlight the apparent greater therapeutic efficacy of the DAP for Indigenous men, despite 28% more rural/remote Indigenous men dropping out of the program and having significantly higher criminal history risk scores on the LSI-R.

Discussion

This study explored whether a generic DAP provided by CSNSW might be effective for Indigenous offenders, thereby contributing to the discussion on the role of culture-specific programs for Indigenous offenders in correctional rehabilitation. Concerns were also held that although the DAP might be effective for urbanised Indigenous men, it might be unsuitable and not therapeutically effective for Indigenous men living in rural/remote areas in the rest of the state, where distance, or more traditional lifestyles, might affect participation and program outcomes. The study therefore examined program effectiveness for Indigenous offenders supervised in remote or rural areas, compared with Indigenous men in metropolitan Sydney, and also to non-Indigenous participants as benchmarks.

Summary of Main Findings

For the first time, an Australian sample of domestically violent/abusive Indigenous men, serving supervised community-based sentences or orders and enrolled in a generic family violence rehabilitative program, was found to have reduced reconvictions during follow-up, compared with other participants. Despite the lack of culture-specific content in the DAP and/or facilitation by Indigenous staff, Indigenous men could derive significant therapeutic benefits, but only if the full program was completed, as dropouts and nonstarters did no better than controls. The higher dropout rates for rural/remote Indigenous men may be associated with cultural or structural factors, which were not assessed in this retrospective study.

Survival time to first reconviction was positively affected by DAP participation. At approximately 4 months post order commencement and approaching program completion, survival curves improved for both Indigenous and non-Indigenous participants, relative to controls. If the DAP was completed, survival for Indigenous men was similar to that of non-Indigenous DAP-treated men. Indigenous status was not statistically associated with reconviction rates over the follow-up, suggesting Indigenous men’s risk status and follow-up behaviours were similar to that of the non-Indigenous men, who included men from diverse cultural and ethnic backgrounds, and who generally reflect the normative multiracial population in NSW, Australia. The universal therapeutic approach of the DAP, addressing no specific cultural mores, appears successful and endorses Andrews and Bonta’s (2006) conclusions that programs that succeed in changing criminogenic needs are the most effective in reducing recidivism risk.¹

DAP commencement and completion rates—Metropolitan and rural offenders

“Failure-to-start” rates were similar for both rural Indigenous and non-Indigenous men (19% and 20%, respectively), suggesting that motivation and transportation factors were not initially major deterrents to attendance for rural Indigenous men, although higher dropout rates (24% vs. 18% for non-Indigenous), meant fewer rural Indigenous men successfully completed; only 58% of Indigenous rural men completed, compared with 64% dwelling in metropolitan Sydney. Rural men are likely to face different barriers to successful program completion. The Sydney Indigenous cohort completed at almost identical rates to their non-Indigenous counterparts (64%–65%, respectively), suggesting that their challenges and abilities to engage in the program were similar to those of other residents in this capital city. These latter completion rates approximated the 67.5% rate obtained in a 24-session mainstream DV pro-feminist group intervention in the United Kingdom (Bowen & Gilchrist, 2006) and 55% from three programs in Gondolf’s (2008) comparative study of racially mixed and nonracially mixed interventions.

Although being Indigenous was not significantly associated with overall program completion/attrition, discrepancies between rural and metropolitan Indigenous offenders warrant further investigation and development of more intensive retention strategies to assist rural Indigenous men to complete their DV intervention.

Factors associated with DAP completion

Protective factors associated with program completion were the same for Indigenous and non-Indigenous men, suggesting similar risk homogeneities. These protective factors—higher age, lower LSI-R risk scores, fewer incarcerations, and convictions in the previous 5 years—have previously been documented (Jewell & Wormith, 2010). Deleterious risk factors—high scores on the LSI-R AOD (alcohol and other drugs) use and criminal history domains—have been also previously associated with program attrition (Andrews & Bonta, 2006; Jones et al., 2006; Weatherburn et al., 2008).

Relationship of DAP completion to a follow-up reconviction

DAP completion was similarly therapeutic for both Indigenous and non-Indigenous offenders, with 73% of Indigenous and 74% of non-Indigenous participants having no follow-up reconvictions. This finding highlights the importance of practical motivational strategies to maximise program commencement/retention and completion, and the need for intense supervisory support, particularly for rural Indigenous offenders, as the ramifications of not completing DAP are of a greater magnitude for Indigenous men. For Indigenous men being reconviction free during the follow-up period, decreased from 73% if the program was completed, to 52% if some sessions were attended, then to 42% for those who failed to attend any sessions. Ramifications of noncompletion were less insidious for non-Indigenous men, as the likelihood of remaining reconviction free declined more slowly, from 74% for those completing, to 72% for dropouts, then to 68% for those failing to attend any sessions. These comparisons highlight the therapeutic efficacy for Indigenous men if they complete a DAP and the benefits they appear to derive from the new behavioural, cognitive, communication, conflict resolution, and other skills learnt. With treatment effects primarily confined to the DAP “completing” cohort (approximately 24% better than controls), mandatory precursor group “readiness” programs, or one on one motivational interventions, such as described by Anstiss et al. (2011), may enhance positive program commitment and engagement.

Time to first reconviction—Indigenous versus non-Indigenous DAP enrollees

Three discrete time intervals emerged in this survival analysis. In the initial 3 months post index-conviction/parole order, Indigenous enrolled men had similar survival to non-Indigenous enrollees, suggesting that pre-DAP enrolment and attendance, both treatment groups behaved and reoffended similarly. At the 3-month follow-up mark, mostly coinciding with advanced DAP attendance/program completion, survival for Indigenous men improved, even relative to non-Indigenous, suggesting they responded very well to both the intervention itself and possibly also the supervision and support provided by their Community Corrections Officers. But at 14 months, when the majority would have completed both the DAP and also the most common 12-month community-based order, Indigenous reconvictions accelerated, mirroring those of non-Indigenous men, but remaining superior to Indigenous controls, whose reconviction curve decreased linearly.

Several factors may be associated with these discrete time line survival changes. Once an offender’s community-based order is completed, legal reasons to desist from negative behaviours would no longer be paramount. The offender would no longer be subject to mandatory reporting/supervision, random urine/drug testing or court-mandated program, and case plan attendance requirements. Other complex cultural, lifestyle, socioeconomic, structural factors, and social realities may also impact negatively on desistance capabilities. Indigenous (and non-Indigenous) men may benefit from a follow-up DAP and/or a general maintenance program, immediately prior to completing their community supervision or parole order, to reinforce desistance skills and learning (Kurlychek et al., 2012).

Survival time to first reconviction compared with untreated controls

Survival improvements for DAP-treated Indigenous men relative to untreated Indigenous controls were both significant and sustained over the 2-year follow-up period, suggesting that for a proportion, the therapeutic efficacy of DAP enrolment is retained over time. It is not clear which elements of the DAP contribute to this efficacy. Therapeutic benefits may be a complex synergy of newly acquired behaviours, skills and knowledge, attitude or cognitive changes, and a realisation of the relationship of alcohol/other drug use to offending behaviours. As current use or a strong history of drugs and/or alcohol misuse emerged as a singular criminogenic need associated with a new reconviction, it is advisable that men with high LSI-R AOD risk scores are also referred to AOD programs during their sentence, to address this rehabilitative need.

Why does the DAP work for Indigenous men? How to improve?

DAP program content focuses on interpersonal communication and relationship difficulties, precursors and consequences, as well as the way patriarchal beliefs manifest to different degrees in most cultures. The program emphasises that each individual must take responsibility for their offending behaviour(s) and the way their behaviours impact on themselves, their children, families, and the community at large. Anecdotally, DAP facilitators reported that this emphasis on individual responsibility and accountability, together with requirements to conform to group rules and expectancies, appears valued by Indigenous men and increases motivation to participate meaningfully. Many Indigenous participants, at program conclusion feedback, maintained that DAP content was as relevant to them as to the non-Indigenous participants; “they have the same problems as us”. They reported favouring equal treatment and consideration in all program aspects and did not like being treated more leniently (or more harshly). “Different” treatment appeared to be perceived by some as patronising/demeaning, impacting on estimations of self-worth/self-esteem and dignity perceptions. Similarly, results of stakeholder interviews conducted by Willis and Moore (2008) demonstrated that many Indigenous men prefer inclusion in mainstream programs. Cost-effectiveness issues also need consideration. Providing one standardised program and processes, as opposed to developing and delivering culture-specific programs that target small specific populations, is costly, more difficult to provide with regularity and to maintain.

The structure of the DAP, with periodic sessions without prescribed content, allows ample space to discuss the relevance of topics to the day-to-day reality of participants’ lives, including their culture. Competent facilitators (who have completed cultural inclusiveness training) can apply responsivity factors to spontaneously shared experiences and personal difficulties raised by participants, to stimulate group interactions, challenge different cultural mores, and facilitate learning.

Study limitations

A major study limitation was that reconviction data could not be disaggregated into intimate partner/family violence convictions. When this study was undertaken, all assaults were coded in one generic court conviction category. This anomaly was rectified by NSW legislation in 2011. Future studies will be able to analyse intimate partner assaults separately from “other” assaults.

As this study was retrospective, many relevant sociocultural factors could not be included, as available data were limited to variables routinely recorded in CSNSW’s OIMS and the NSW BOCSAR court conviction database (ROD). Prospective research designs could possibly include data on factors contributing to ongoing family violence in Indigenous communities, such as physical and mental health issues, substance abuse, low self-esteem (Cripps, 2007; Robertson, 1999; Weatherburn et al., 2006), socioeconomic and educational disadvantage, unemployment and welfare dependency, as well as “disruption and distress attributable to” . . . “the removal of Indigenous children from their families” (Cripps & Davis, 2012).

While our findings affirm the effectiveness of the DAP program for Indigenous men, regardless of the program facilitators’ cultural background (with the caveat that all facilitators must be appropriately trained and skilled), future studies could empirically test this hypothesis, as some authors (e.g., Day et al., 2003) advocate program facilitation by Indigenous staff and Elders. Although some programs in this study were delivered by facilitators with Indigenous backgrounds, this proportion was unknown, as facilitator ethnicity was not routinely recorded. A few areas in NSW with high Indigenous populations provide DAPs with exclusive Indigenous membership but without any culturally relevant adaptations from the standardised manualised version. Future research could assess whether different outcomes are obtained in programs including Indigenous members only and/or led by Indigenous facilitators.

As Indigenous men had higher program dropout rates than non-Indigenous men, focus groups with Indigenous stakeholders may help identify specific reasons for dropout and help development of innovative strategies to improve attendance. Indigenous men dropping out, or failing to start in rural areas, may represent those with more traditional lifestyles from outlying rural areas, thus exaggerating the retention of less traditional Indigenous men. Nongovernment organisations, or local community agencies, may provide travel assistance, to overcome difficulties with long biweekly journeys from remote areas. Additional, or more intense precursor or motivational readiness programs and individual motivational interventions, delivered before DAP commencement, may also enhance therapeutic commitment and attendance.

Meta-analysis comparisons

No meta-analyses of DV/domestic abuse interventions directly comparable with this quasi-experimental study were identified. Other studies frequently used measures such as victim reports, standardised instruments to measure change in risk outcomes, or the use of program dropouts for comparison or as controls, which introduces strong biases, due to higher recidivism risk (Delaney, 2010; Feder & Wilson, 2005; Palmer et al., 2007). Many programs were of noncomparable length or follow-up. A meta-analysis by Arias et al. (2013) of four DV quasi-experimental studies also used official court recidivism data and similar theoretical approaches. This latter meta-analysis reported a Cohen’s h effect size of .40, whereas our study reports a more favourable effect size of .48 (Cohen’s d) for the two Indigenous groups. However, in Arias et al.’s (2013) meta-analysis, the treatment group only included program completers, which is known to strongly bias favourable outcomes. In our study, the treatment groups also included dropouts and nonstarters, who were 41% less likely to remain reconviction free. Babcock et al.’s (2004) meta-analysis of 22 methodologically sound quasi- or experimental evaluations also included dropouts and reported “small” effect sizes for Duluth interventions (d = .32, k = 7) and CBT programs (d = .12, k = 4). Again, our study’s outcomes for Indigenous men compare favourably.

Others analyses report reconviction rates of 30% by 6 months, regardless of the intervention strategy (Stover et al., 2009). In comparison, in our study, it took approximately 10 months for 30% of Indigenous DAP enrollees to have a reconviction, suggesting stronger therapeutic efficacy.

Conclusion

Despite the inclusion of program dropouts and nonstarters in our treatment groups, introducing a strong negative bias, this study’s findings demonstrate reduced Indigenous reoffending rates and extensions of time to first reconviction for participants who completed a DAP. The results support the use of validated best practice theoretical approaches and frameworks, including RNR and cognitive behavioural principles, in the design of future programs for Indigenous offenders, especially for Indigenous DV.

In this cohort, Indigenous men reported similar relationship and communication difficulties associated with abusive behaviours and violence in interpersonal and family contexts, to non-Indigenous men from diverse cultural groups. Findings suggest positive program engagement and outcomes can be obtained from group rehabilitative programs that integrate Indigenous men with offenders from other races and ethnicities. Including Indigenous men in generic DAP programs may have contributed to lower reconviction rates in this cohort. Exposing them to the notion that domestic abuse and violence is not unique to Indigenous communities, but occurs in all cultures, appeared to empower the men to take responsibility for their offending behaviours.

Our findings also suggest that segregating Indigenous men into culture-specific programs may also increase feelings of difference, isolation, and cultural helplessness.

For the first time, this validation of a “treatment as usual” “one size fits all” (Polaschek, 2011b) intervention provides a benchmark against which other, or alternative, therapeutic approaches for Indigenous peoples can be studied. A question not addressed in this study is whether the outcomes described are the best that can be achieved for Indigenous men. Our results do not support a view that effective Indigenous offender programs can only be successful if developed and delivered/facilitated by Indigenous people. It is possible that therapeutic approaches that include additional Indigenous topics, or parallel culture specific programs, may further improve outcomes. This needs to be established, as many of the above observations contrast with the views of some commentators, who endorse the development and delivery of rehabilitative programs by local Indigenous peoples and elders, with scant requirements for formal facilitator training, competencies, or qualifications. These views are primarily based on consultation, rather than on empirical outcomes.

Indigenous advisors also emphasise that rehabilitative programs should contain a “healing” content (Cunneen, 2001b, 2010; Day, 2008; Day et al., 2003; Macklin & Gilbert, 2011). The DAP contains no “healing” content specific to the needs of any one racial group, but addresses cognition and behavioural change, risk and criminogenic needs, skill, and strategy acquisitions. Parra-Cardona et al. (2013) report that no research has found culturally focused interventions to be effective for ethnic minority groups. However, Gondolf (2008) argues that culture-specific/racially focused programs should not be entirely dismissed in favour of racially mixed approaches with better outcomes. Culture- specific approaches may just need refinement and be open to varied theoretical applications (Babcock et al., 2016).

Findings from this study do not minimise the complex needs, disadvantage issues and cultural loss experienced by many Indigenous peoples, nor suggest that therapeutic benefits cannot be gained by “healing” programs developed and provided by Indigenous peoples or agencies. Additional therapeutic gains may be achieved if such programs are attended in conjunction with programs of proven effectiveness. Careful evaluation is required to determine the comparative efficacy of generic multiracial treatment programs and Indigenous treatment approaches, delivered either separately or in combination. Future research should also ascertain whether improved program completion rates and outcomes are associated with not only the competencies and skills of the facilitators, but also their cultural heritage. We believe there is room for a diversity of approaches and believe the foregoing findings contribute to dialogues on this topic.

Footnotes

Appendix

Table A1.

Reconviction Status of Metropolitan and Rural Indigenous and Non-Indigenous DAP Enrollees (N = 953).

Reconviction status	Domicile
	Rural				Metropolitan
	Indigenousn	%	Non-Indigenousn	%	Indigenous n	%	Non-Indigenous n	%
No reconviction	92	61%	399	73%	21	69%	162	71%
>1 conviction	58	39%	145	27%	9	31%	67	29%
Total	150	100%	544	100%	30	100%	229	100%
Chi square^a	χ² = 8.684				χ² = 0.144
p	<.005^a				NS

Note. DAP = domestic abuse program.

df = 2; two-tailed test.

Acknowledgements

Conviction data were provided by the New South Wales Bureau of Crime Statistics and Research.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: New South Wales Corrective Services funded this study with the assistance of Forensic Psychology Master’s student interns from the University of NSW.

ORCID iD

Chris Blatch

Notes

References

Allan

Dawson

(2004). Assessment of the risk of reoffending by Indigenous male violent and sexual offenders. Australian Institute of Criminology.

Al-Yaman

Van Doeland

Wallis

(2006). Family violence among Aboriginal and Torres Strait Islander peoples. Australian Institute of Health and Welfare. https://www.aihw.gov.au/getmedia/34fea687-d980-4e54-8826-256b6acfdd0f/fvaatsip.pdf.aspx?inline=true

Anderson

(2002). Understanding Indigenous violence. Australian and New Zealand Journal of Public Health, 26(5), 408–409.

Andrews

Bonta

(1996). The Level of Service Inventory-Revised: User manual. Multi-Health Systems.

Andrews

Bonta

(2006). The psychology of criminal conduct (4th ed.). Anderson Publishing.

Anstiss

Polaschek

D. L. L.

Wilson

(2011). A brief motivational interviewing intervention with prisoners: When you lead a horse to water, can it drink for itself? Psychology, Crime & Law, 17(8), 689–701.

Arias

Ramón Arce

Vilariño

(2013). Batterer intervention programmes: A meta-analytic review of effectiveness. Psychosocial Intervention, 22, 153–160.

Australian Bureau of Statistics. (2018a, June 7). Corrective Services, Australia, 2018 (Cat. No. 4512.0). http://www.abs.gov.au/Ausstats/abs@.nsf/0/2B6A956599FF46AACA25813800176319?OpenDocument

Australian Bureau of Statistics. (2018b). The National Offence Index (Cat. No. 1234.0.55.001).

10.

Australian Law Reform Commission. (2017). Pathways to justice: An inquiry into the incarceration rate of Aboriginal and Torres Strait Islander Peoples: Final report.

11.

Australia’s National Research Organisation for Women’s Safety. (2016). Existing knowledge, practice and responses to violence against women in Australian Indigenous communities: State of knowledge paper.

12.

Australia’s National Research Organisation for Women’s Safety (2018). Fast facts: Indigenous family violence. https://d2rn9gno7zhxqg.cloudfront.net/wp-content/uploads/2019/02/19025923/Fast-Facts-Indigenous-family-violence-1.pdf

13.

Babcock

Cannon

Lauve-Moon

Ferrieira

Cantos

Hamel

. . . Solano

(2016). Domestic violence perpetrator programs: A proposal for evidence-based standards in the United States. Partner Abuse, 7(4), 355–460.

14.

Babcock

Green

Robie

(2004). Does batterers’ treatment work? A meta-analytic review of domestic violence treatment. Clinical Psychology Review, 24, 1023–1053.

15.

Bennett

Stoops

Call

Flett

(2007). Program completion and re-arrest in a batterer intervention system. Research on Social Work Practice, 17, 42–55.

16.

Blatch

O’Sullivan

Delaney

van Doorn

Sweller

(2016). Evaluation of an Australian Domestic Abuse Program for offending males. Journal of Violence, Conflict and Peace Research, 8(1), 4–20.

17.

Blagg

Williams

Cummings

Hovane

Torres

Woodley

K. N.

(2018). Innovative models in addressing violence against Indigenous women: Final report. ANROWS Horizon, 01/2018. ANROWS.

18.

Blair

Chapman

Little

Prentice

Tanner

Walker

(2014). The Yarning Up! Project report. Phoenix House.

19.

Boersig

(2007). Cultural healing in criminal justice service delivery: An identification of best practice and innovation to inform further service design: National Indigenous Justice CEO Forum. National Indigenous Law and Justice Strategy. Indigenousjustice.gov.au/db/publications/278593.html

20.

Bowen

Gilchrist

(2006). Predicting dropout of court mandated treatment in a British sample of domestic violence offenders. Psychology, Crime & Law, 12, 573–587.

21.

Brown

Languedoc

(2004). Components of an Aboriginal-based family violence intervention program. Families in Society: The Journal of Contemporary Social Services, 85(4), 477–483.

22.

Bryant

Willis

(2008). Risk factors in Indigenous violent victimisation (Technical and Background Paper No. 30). Australian Institute of Criminology. http://aic.gov.au/publications/current%20series/tbp/21-40/tbp030.aspx

23.

Chan

(2005). Alcohol issues in domestic violence. Australian Domestic and Family Violence Clearinghouse, University of New South Wales.

24.

Cheers

Binell

Coleman

Gentle

Miller

Taylor

Weetra

(2006). Family violence: An Australian Indigenous community tells its story. International Social Work, 49(1), 51–63.

25.

Coe

(2002, September). It’s the effect size, stupid. Paper Presentation at the Annual Conference of the British Educational Research Association, Exeter, UK.

26.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum.

27.

Corben

(2011, December). Trends in the adult indigenous inmate population in NSW 1998-2010; the impact of changes in measurement (Research Bulletin No. 32). NSW Department of Corrective Services.

28.

Corrective Services NSW. (2010). CSNSW Offender Integrated Management System. Offender database maintained by CSNSW Corporate Research Evaluation and Statistics.

29.

Corrective Services NSW. (2012a). Compendium of correctional programs. NSW Department of Police and Justice Offender Programs Unit.

30.

Corrective Services NSW. (2012b). Domestic Abuse Program training manual. NSW Department of Police and Justice Offender Programs Unit.

31.

Cripps

(2007). Indigenous family violence: From emergency measures to committed long term action. Australian Indigenous Law Review, 11, 7–18.

32.

Cripps

Davis

(2012). Communities working to reduce Indigenous family violence: Indigenous Justice Clearinghouse Brief 12. Australian Institute of Criminology.

33.

Cripps

Davis

(2013). Communities working to reduce Indigenous family violence. Australian Domestic and Family Violence Clearinghouse Newsletter, 54, 6–7.

34.

Cunneen

(2001a). The impact of crime prevention on Aboriginal communities. Institute of Criminology.

35.

Cunneen

(2001b, October). Residential alternatives for Indigenous offenders. Paper presentation at the Conference of the Australian Institute of Criminology on Best Practice Interventions in Corrections for Indigenous People. Sydney, Australia.

36.

Cunneen

(2010). Alternative and improved responses to domestic and family violence in Queensland Indigenous communities. Department of Communities, Queensland Government.

37.

Davis

Taylor

(2002). Voices from the margins Part 1: Narrative accounts of the support needs of Indigenous families experiencing violence. Contemporary Nurse, 14(1), 66–75.

38.

Day

Jones

Martin Nakata

McDermott

(2012). Indigenous family violence: An attempt to understand the problems and inform appropriate and effective responses to criminal justice system intervention. Psychiatry, Psychology and Law, 19(1), 104–117.

39.

Day

(2008). Reducing the risk of reoffending in Australian Indigenous offenders: What works for whom? Journal of Offender Rehabilitation, 3(2), 1–15.

40.

Day

Chung

O’Leary

Carson

(2009). Programs for men who perpetrate domestic violence: An examination of the issues underlying the effectiveness of intervention programs. Journal of Family Violence, 24, 203–212.

41.

Day

Howells

Casey

(2003). The rehabilitation of Aboriginal prisoners: An Australian perspective. Journal of Ethnicity in Criminal Justice, 1(1), 115–133.

42.

Day

Vlais

Chung

Green

(2018). Standards of practice in domestic and family violence behaviour change programs in Australia and New Zealand. Australian and New Zealand Journal of Family Therapy, 39, 501–513.

43.

Delaney

(2010). Corrective Services NSW’s Domestic Abuse Program: An evaluation of recidivism outcomes (Unpublished master’s dissertation, University of NSW, Sydney, Australia).

44.

Duwe

Goldman

R. A.

(2009). The impact of prison-based treatment on sex offender recidivism: Evidence from Minnesota. Sex Abuse, 21, 279–307.

45.

Feder

Wilson

(2005). A meta-analytic view of court mandated batterer intervention: Can courts affect abusers’ behaviour? Journal of Experimental Criminology, 1(2), 239–362.

46.

Gentle

Taylor

(2002). A community taking control of family violence. Aboriginal and Islander Health Worker Journal, 26(5), 4–6.

47.

Girard

Wormith

J. S.

(2004). The predictive validity of the Level of Service Inventory–Ontario Revision on general and violent recidivism among various offender groups. Criminal Justice and Behavior, 31, 150–181. https://doi.org/10.1177/0093854803261335

48.

Gondolf

E. W.

(2008). Program completion in specialized batterer counselling for African-American men. Journal of Interpersonal Violence, 23, 94–116.

49.

Gondolf

E. W.

Williams

O. J.

(2001). Culturally focused batterer counseling for African American Men. Trauma, Violence & Abuse, 2(4), 283–295.

50.

Gordon

Hallahan

Henry

(2002). Putting the picture together: Inquiry into response by government agencies to complaints of family violence and child abuse in Aboriginal communities. State Law Publisher.

51.

Guthrie

(2015). Book review: D. Weatherburn. Arresting incarceration: Pathways out of Indigenous imprisonment. The Australian Journal of Anthropology, 26(1), 132–134.

52.

Hanson

Wallace-Capretta

(2000a). A multi-site study of treatment for abusive men (Report No. 2000-05). Department of the Solicitor General.

53.

Hanson

Wallace-Capretta

(2000b). Predicting recidivism among make batterers. A multi-site study of treatment for abusive men (Report No. 2000-06). Department of the Solicitor General.

54.

Hatcher

Palmer

McGuire

Hounsome

Bilby

Hollin

(2008). Aggression replacement training with adult male offenders within community settings: A reconviction analysis. Forensic Psychology and Psychiatry, 19, 517–532.

55.

Heseltine

Day

Sarre

(2011). Prison-based correctional offender rehabilitation programs: The 2009 national picture in Australia. Australian Institute of Criminology. https://aic.gov.au/publications/rpp/rpp112

56.

Hilton

N. Z.

Harris

G. T.

Popham

Lang

(2011). Risk assessment among incarcerated male domestic violence offenders. Criminal Justice and Behavior, 37(8), 815–832.

57.

Hunter

(2001). Factors underlying Indigenous arrest rates. NSW Bureau of Crime Statistics and Research.

58.

Indigenous Justice Forum, Attorney-General’s Department. (2012). Transcript from National Justice Chief Executive Officer’s Indigenous Justice Forum. Australian National University.

59.

Jewell

Wormith

(2010). Variables associated with attrition from treatment programs targeting domestic violence batterers. A meta-analysis. Criminal Justice and Behavior, 37(10), 1086–1113.

60.

Joffe

M. M.

Rosenbaum

P. R.

(1999). Invited commentary: Propensity scores. American Journal of Epidemiology, 150(4), 227–333.

61.

Jones

Hua

Donnelly

McHutchinson

Heggie

(2006). Risk of reoffending among parolees (Crime and Justice Bulletin No. 91). NSW Bureau of Crime Statistics and Research.

62.

Kurlychek

M. C.

Bushway

S. D.

Brame

(2012). Long term desistance and recidivism patterns: Evidence from the Essex County convicted felon study. Criminology, 50(1), 71–103.

63.

Laing

(2013). “It’s like this maze that you have to make your way through”: Women’s experiences of seeking a domestic violence protection order in NSW. Faculty of Education and Social Work, University of NSW.

64.

Lösel

Schmucker

(2005). The effectiveness of treatment for sexual offenders: A comprehensive meta-analysis. Journal of Experimental Criminology, 1, 117–146.

65.

Loughran

T. A.

Mulvey

E. P.

Schubert

C. A.

Chassin

L. A.

Steinberg

Piquero

A. R.

. . . Losoya

(2010). Differential effects of adult court transfer on juvenile offender recidivism. Law and Human Behavior, 34(6), 476–488.

66.

Macklin

Gilbert

(2011). Working with Indigenous offenders to end violence (Indigenous Justice Clearinghouse, Brief No. 11). Australian Institute of Criminology.

67.

McKendrick

Brooks

Hudson

Thorpe

Bennett

(2017). Aboriginal and Torres Strait Islander healing programs: A literature review. Healing Foundation.

68.

Memmott

(2010). On regional and cultural approaches to Australian Indigenous violence. Australian & New Zealand Journal of Criminology, 43(2), 333–355.

69.

Memmott

Passi

Go-Sam

Thomson

Sheppard

(2009). Family violence in the Torres Strait. Queensland Department of Communities.

70.

Memmott

Stacy

Chambers

Keys

(2001). Violence in Indigenous communities: Full report. Crime Prevention Branch, Attorney-General’s Department.

71.

Miller

(2010). Discussing the Duluth curriculum: Creating a process of change for men who batter. Violence Against Women, 16, 1007–1021.

72.

NSW Bureau of Crime Statistics and Research. (2019). Re-offending statistics for NSW. https://www.bocsar.nsw.gov.au/Pages/bocsar_pages/Re-offending.aspx

73.

Office of the Status of Women. (2001). Working together against violence: The first three years of partnership against violence. Australian Government, Department of Prime Minister and Cabinet.

74.

Olver

M. E.

Stockdale

K. C.

Wormith

S. J.

(2014). Thirty years of research on the Level of Service Scales: A meta-analytic examination of predictive accuracy and sources of variability. Psychological Assessment, 26(1), 156–176.

75.

Palmer

McGuire

Hounsome

Hatcher

Bilby

Hollin

C. R.

(2007). Offending behaviour programs in the community: The effects on reconviction of three programs with adult male offenders. Legal and Criminology Psychology, 12(2), 251–264.

76.

Parra-Cardona

J. R.

Escobar-Chew

A. R.

Holtrop

Carpenter

Guzman

Fernanadez

Ramiriez

D. G.

(2013). Latino immigrants’ satisfaction with a culturally informed intervention for men who batter. Violence Against Women, 19, 107–132.

77.

Pointer

S. C.

(2019). Hospitalised injury among Aboriginal and Torres Strait Islander people, 2011-12 to 2015-16. Injury research and statistics series no. 188. Cat. No. INJCAT 198. AIHW.

78.

Polaschek

(2011a). High intensity rehabilitation for violent offenders in New Zealand: Reconviction outcomes for high and medium risk prisoners. Journal of Interpersonal Violence, 26, 664–682.

79.

Polaschek

(2011b). Many sizes fit all: A preliminary framework for conceptualising the development and provision of cognitive behavioural rehabilitation programs for offenders. Aggression and Violent Behavior, 16, 20–35.

80.

Queensland Crime and Misconduct Commission. (2009). Restoring order: Crime prevention, policing and local justice in Queensland’s Indigenous communities.

81.

Ramamoorthi

Jayaraj

Notaras

Thomas

(2014). Alcohol-related violence among the Australian Aboriginal and Torres Strait Islanders of the Northern Territory: Prioritizing an agenda for prevention: Narrative review article. Iranian Journal of Public Health, 43(5), 539–544.

82.

Rassen

Glynn

R. J.

Rothman

K. J.

Setoquchi

Schneeweiss

(2012). Applying propensity scores estimated in a full cohort to adjust for confounding in a sub group analysis. Pharmacoepidemiology and Drug Safety, 21(7), 697–709.

83.

Robertson

(1999). The Aboriginal and Torres Strait Islander Women’s Taskforce on Violence report. Queensland Department of Aboriginal and Torres Strait Islander Policy and Development.

84.

Rosenbaum

P. R.

Rubin

D. B.

(1985). Constructing a control group using multivariate matched sampling methods that incorporate a propensity score. The American Statistician, 39(1), 33–38.

85.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41–55.

86.

Royal Commission into Aboriginal Deaths in Custody. (1991). Royal Commission into Aboriginal Deaths in Custody national report: Overview and recommendations. Australian Government Publishing Service.

87.

Rugge

(2006). Risk assessment of male Aboriginal offenders: A 2006 perspective. Public Safety and Emergency Preparedness. https://www.publicsafety.gc.ca/cnt/rsrcs/pblctns/rskssssmnt-ml/rsk-ssssmnt-ml-eng.pdf

88.

Sherman

L. W.

Gottfredson

D. C.

MacKenzie

D. L.

Eck

J. E.

Reuter

Bushway

S. D.

(1998, July). Preventing crime: What works, what doesn’t, what’s promising. National Institute of Justice, Research in Brief. https://www.ncjrs.gov/pdffiles/171676.PDF

89.

Snowball

Weatherburn

(2006, September). Indigenous over-representation in prisons: The role of offender characteristics. Crime and Justice Bulletin No. 99. NSW Bureau of Crime Statistics and Research.

90.

Snowball

Weatherburn

(2008). Theories of Indigenous violence: A preliminary empirical assessment. Australian & New Zealand Journal of Criminology, 41(2), 216–235.

91.

Steering Committee for the Review of Government Service Provision. (2016). Overcoming Indigenous disadvantage: Key indicators 2009. Productivity Commission. https://www.pc.gov.au/research/ongoing/overcoming-indigenous-disadvantage/2016/report-documents/oid-2016-overcoming-indigenous-disadvantage-key-indicators-2016-report.pdf

92.

Smallacombe

(2004). Speaking positions on Indigenous violence. Hecate, 30(1), 47–55.

93.

Stover

C. S.

Meadows

A. L.

Kaufman

(2009). Interventions for intimate partner violence: Review and implications for evidence-based practice. Professional Psychology: Research and Practice, 40(3), 223–233.

94.

Sullivan

G. M.

Feinn

(2012). Using effect sizes- or why the p value is not enough. Journal of Graduate Medical Education, 4(3), 279–282.

95.

Thakker

Gannon

T. A.

(2010). Rape treatment: An overview of current knowledge. Behaviour Change, 27, 227–250.

96.

Trevethan

Moore

J. P.

Allegri

(2005). The “in search of your warrior” violence prevention program for Aboriginal Offenders: A preliminary evaluation. Correctional Service of Canada. https://www.csc-scc.gc.ca/research/092/r172-eng.pdf

97.

Weatherburn

(2001, February). What causes crime? (Crime and Justice Bulletin No. 54). NSW Bureau of Crime Statistics and Research.

98.

Weatherburn

(2014). Arresting incarceration: Pathways out of Indigenous imprisonment. Aboriginal Studies Press.

99.

Weatherburn

Holmes

(2010). Re-thinking Indigenous over-representation in prison. Australian Journal of Social Issues, 45(4), 559–576.

100.

Weatherburn

Snowball

Hunter

B. H.

(2006). The economic and social factors underpinning Indigenous contact with the justice system: Results from the 2002 NATSISS survey. Crime and Justice Bulletin No. 104. NSW Bureau of Crime Statistics and Research.

101.

Weatherburn

Snowball

Hunter

B. H.

(2008). Predictors of Aboriginal arrest: An exploratory study. Australian & New Zealand Journal of Criminology, 41(2), 216–235.

102.

Widders

(2003). Domestic violence, sexual assault and the health of Aboriginal people. The Child, Youth and Family Work Journal, 6, 40–45.

103.

Wild

Anderson

(2007). Ampe akelyernemane meke mekarle: “Little children are sacred”: Report of the Northern Territory. Northern Territory Government.

104.

Williams

(2016). Book Review: Arresting incarceration: Pathways out of Indigenous imprisonment, Don Weatherburn (Aboriginal Studies Press, Canberra, 2014, ISBN 9781922059550 [paperback], 189 pp.). Australian Journal of Public Administration, 75(1), 119–120.

105.

Willis

Moore

J. P.

(2008). Reintegration of Indigenous prisoners. Research and Public Policy Series No. 90. Australian Institute of Criminology. http://www.aic.gov.au/publications/currentseries/rpp/81-100/rpp90.aspx

106.

Wundersitz

(2010). Indigenous perpetrators of violence: Prevalence and risk factors for offending. Research and Public Policy Series No. 105. Australian Institute of Criminology.