Evaluation of a Collaborative Operation to Improve Child Sexual Abuse Reporting in Western Australian Indigenous Communities

Abstract

This study presents an evaluation of Operation RESET, a community engagement intervention designed to help remote Indigenous communities and human service agencies to uncover, respond to, and prevent child sexual abuse. The primary aim of this evaluation was to determine whether the intervention was associated with increased reporting. Data were obtained for six Western Australian regions between 2007 and 2012. Number of reports and arrests significantly increased in the intervention areas during the intervention compared with the pre-intervention time period but not in the control areas. Arrest rates significantly increased during the intervention and increased further following the intervention. There were no changes in arrest rates in regions that did not participate in the operation. This evidence suggests that the reforms led to a marked improvement in some key outcomes for Indigenous victims of child sexual abuse and supports the adoption of this collaborative approach by other jurisdictions.

Keywords

child sexual abuse aboriginal Indigenous remote community engagement initiative intervention

Incidents of substantiated harm, or risk of harm, to Australian Aboriginal and Torres Strait Islander children are estimated at between 5 and 8 times that of non-Indigenous children, despite lower rates of reporting and convictions (Australian Institute of Health and Welfare, 2011; Bromfield & Holzer, 2008; Stanley, Tomison, & Pocock, 2003). The prevalence of child sexual abuse in Indigenous¹ communities has been the subject of numerous government inquiries (Aboriginal Child Sexual Assault Taskforce, 2006; Berlyn, Bromfield, & Lamont, 2011; Mullighan, 2008; NSW Ombudsman, 2012) and media attention, especially during the Northern Territory Emergency Response (Northern Territory Emergency Response Review Board, 2008). Nonetheless, quantitative evaluation outcomes have been lacking in child protection systems across the globe, constraining the ability of researchers to provide conclusive answers about the impact of certain procedures and constraining service providers’ ability to make evidence-based decisions about policy and procedure (Snell, 2003). Absence of good evaluation data related to the management of child abuse cases has made it difficult for service providers to evaluate outcomes and has provided governments little return on the investments they have made in financing reforms (NSW Ombudsman, 2012).

One of the chief issues with collecting information and data in the area of child sexual abuse is that of under-reporting. Calma (2006) referred to this as the code of silence that exists in a many Indigenous communities. Although under-reporting of this crime type is a universal problem, there are a range of factors that further exacerbate the problem for Indigenous people. Some of the barriers to reporting these crimes include an inadequate understanding by community members of the nature of the crime and the laws that relate to it, as well as systemic barriers, such as shame and fear of reprisal, that make it difficult for Indigenous people to engage with authorities (Aboriginal Child Sexual Assault Taskforce, 2006; Lievore, 2003; Stanley et al., 2003). Community engagement and the combined response of the relevant governmental departments working together is required to address the complex and multifaceted nature of this problem (Thompson, Greville, & Param, 2008) and tackle the “unresolved grief associated with multiple layers of trauma that has spanned many generations” (Stanley et al., 2003, p. 7).

The current study presents an evaluation of Operation RESET, a community engagement initiative aimed at facilitating remote Indigenous communities and human service agencies to uncover, respond to, and prevent incidents of child sexual abuse. This trial strategy was a joint initiative of the Western Australian Police Sex Crime Division and the Department for Child Protection and Family Support’s ChildFIRST service, and operated in 13 remote towns and communities in the Mid West Gascoyne and two in the Pilbara regions of Western Australia (WA). The proactive, collective impact approach between government, non-government organizations, and communities aimed to empower families and communities to improve the overall safety and well-being of children, as well as acknowledge the underlying causes and contexts of child sexual abuse. This approach supports the model advocated by Calma (2006) for a proactive system aimed at building and supporting healthy communities, rather than a passive reaction supporting a dysfunctional system.

Operation RESET delivered a different type of service to what had previously been offered throughout Australia and had a number of distinctive features. These included the establishment of genuine consultation between the operation team, local stakeholders and community members, proactive service delivery through capacity building, and a strong emphasis on self-determination. The model was created to provide a tangible process through which the different stakeholders could come together to identify strengths and problems, design collective solutions, and craft a unified plan for the region or community. The intentions underlying the intervention were that Indigenous communities themselves would effectively prevent the sexual abuse of their children (with the support of service providers) and that complaints made to the police and child protection service providers would be investigated thoroughly and expeditiously.

Children in Indigenous families are generally more likely to experience the multiple types of disadvantage that are associated with alcohol and substance abuse, family violence, poverty, mental health issues, and adults with histories of neglect or abuse (Calma, 2006; Wild & Anderson, 2007) than non-Indigenous children. Areas selected for Operation RESET had reported crimes symptomatic of these types of disadvantage. Consultation with stakeholder groups indicated, anecdotally, that the number of child sexual abuse reports was not representative of the day-to-day experience of Indigenous people living and working in these communities (Mace & Powell, 2012). Following the selection of the intervention areas, a core team for the project was assembled, which included a senior investigation officer, detective sergeant, four specialist child abuse detective investigators from the Sex Crime Division’s Child Abuse Squad, and Specialist Forensic Child Interviewers attached to the Department for Child Protection’s ChildFIRST service and the Police Child Assessment Interview Team. Once the operation team was established, extensive consultation with community members and a wide range of local service providers was conducted to share information about child sexual abuse and then collectively devise solutions for application.

The action phase of the intervention involved implementing the community engagement strategy. The RESET team was engaged in proactive, pre-planned visits during which the team took part in activities designed to build rapport and capacity as well as conducting investigations. Activities included raising awareness of child abuse by sharing information, providing training to school staff and families on protective behaviors and abuse indicators, increasing support to victims after arrests, connecting victims and families to local support services, building rapport between service providers and children in the classroom, and attending sporting activities. Reactive visits to investigate specific incident reports were also undertaken and were assigned according to priority and available resources. Documentation was carried out throughout this phase to provide a framework for planning and informing the strategic and equitable distribution of services. Furthermore, the operation had an exit strategy built into the project to enable successful initiatives to be adopted by local agencies and maintained in the long term. Costs incurred by Operation RESET were funded separately from normal business, and resources were not diverted from other areas of WA to fund the operation. For a more thorough description of the intervention, please refer to Mace and Powell (2012).

The current study is the second evaluation of the effectiveness of Operation RESET in improving child sexual abuse outcomes. The previous evaluation used a qualitative analysis method to examine stakeholders’ perceptions of the strategic model (Mace, Powell, & Benson, 2015). In-depth interviews were collected from community members, victims’ families, and professionals who provided services across the region, including those involved in the delivery of Operation RESET. The analysis attempted to determine whether there were improved relationships between Indigenous community members and government agencies, and improved service delivery in relation to the detection, response, and prevention of child sexual abuse. Because prior research in this field is limited and this initiative signaled a significantly new approach, the detailed and non-directive nature of a qualitative study was deemed to be appropriate for this first level of analysis. This initial review determined that there were four distinctive features that could be attributed to the success of the reform. These included (a) proactive outreach (i.e., the model brought specialized services to the people), (b) capacity building (i.e., the model improved skills, competencies, knowledge, and abilities of professionals and communities), (c) brokering both holistic and integrated service provision, and (d) genuine engagement based on trust. The review concluded that Operation RESET had been a major step forward in improving relationships between Indigenous community members and human service agencies, and had considerably assisted in maximizing justice outcomes and victims’ well-being.

Increasing reporting rates was one of the primary aims of the intervention, and therefore the overall aim of the current study was to test whether the new model of intervention was associated with actual increases in reports and arrests in the areas where Operation RESET was conducted. The first specific aim was to determine whether reports of child sexual abuse had increased in intervention areas compared with regions without the intervention. Second, the study aimed to investigate any corresponding increases in arrests in intervention areas compared with non-intervention areas. Finally, for the intervention to be deemed successful in the longer term, the arrest rate (rate of arrest per incident report) would need to increase during Operation RESET and be maintained or increased further following the intervention. Increases in arrest rates would not be apparent in the regions that did not participate in the operation.

Method

Sample

Analyses in this study were divided into two sections. The first section used data collected during the operation in the Mid West Gascoyne, and the second section used data collected from the WA Police Information Management System.

Operation Data

The sample for the first section of the evaluation consisted of all cases of child sexual abuse that were reported in the Mid West Gascoyne region and referred to the Operation RESET team and included information on reports, investigations conducted, victim interviews, and arrests. Information that tracked the team’s day-to-day activities was also collected. This included data on which community was attended by team members, the services they interacted with, schools visited, participation in sports days and sausage barbecues, and the type of training that was delivered and to whom. Activities were defined as “proactive” when the team entered and engaged with the community and “reactive” when the team responded to an incident report. Proactive days were characterized by the team conducting meetings with local organizations, training, participating in social events, and managing investigations. Table 1 contains a breakdown of proactive and reactive presence by community. The table also outlines the significant amount of proactive activity conducted as part of this operation, with more than 88% of the team’s days spent in proactive engagement. Only four communities received reactive visits, indicating that the intervention activities were mostly conducted during proactive visits. In non-intervention regions, all interactions would be described as reactive visits.

Table 1:

Operation RESET Team Proactive and Reactive Days by Community

Community	Proactive	Reactive	Total
Burringurrah	8	0	8
Carnarvon	27	3	30
Cue	12	0	12
Gascoyne Junction	5	0	5
Geraldton	38	13	51
Karalundi	5	0	5
Meekatharra	21	6	27
Mount Magnet	23	2	25
Mullewa	14	0	14
Pia Wadjarri	5	0	5
Sandstone	2	0	2
Yalgoo	12	0	12
Yulga Jinna	9	0	9
Total	181	24	205

During Operation RESET, there were 45 training sessions presented by the RESET team; these covered mandatory reporting, criminal law, Australian National Child Offender Register, protective behaviors, signs and indicators of child abuse, and how to deal with district intelligence (now incorporated in the Crime Intelligence Coordination Unit). In addition, 3 days of Child Assessment Interview Training (Forensic) was delivered in Carnarvon along with Specialist Child Interview Training, specifically for staff in the Mid West Gascoyne region. A total of 330 meetings were conducted by the Operation RESET team with local organizations and communities during this 18-month time period, and are presented by the organizations in Table 2. Local police, the Education Department, Indigenous communities, and Department for Child Protection and Family Support make up the majority of the meetings (63%). Information from qualitative interviews (Mace et al., 2015) indicated that the meetings were well attended.

Table 2:

Meetings Conducted by Operation RESET Team by Organization

Organization	Meetings	%	Cumulative %
Police	74	22.4	22.4
Education department	55	16.7	39.1
Indigenous community	41	12.4	51.5
Child Protection and Family Support (CPFS)	38	11.5	63.0
Health department	25	7.6	70.6
Shire	18	5.5	76.1
Justice	16	4.8	80.9
Community health agencies	14	4.2	85.1
Victim support services	14	4.2	89.3
Private school	10	3.0	92.3
Counseling services	9	2.7	95.0
Corrective services	8	2.4	97.4
Local training	6	1.8	99.2
CPFS and police combined session	1	0.3	99.5
Safety alliance	1	0.3	99.8
Total	330	100.0	100.0

Data From WA Police Information Management System

The number of reports of arrests made in six WA regions was obtained from the WA Police Incident Management System between 2007 and 2012. Two of these regions, the Mid West Gascoyne and the Pilbara, received interventions in selected areas. The two main towns in Mid West Gascoyne (Geraldton and Carnarvon) also referred cases to the RESET team and have therefore been included as being within the intervention area. Overall, around 90% of the Indigenous population of the Mid West Gascoyne was within the targeted area. The intervention in the Pilbara was on a smaller scale; the two targeted towns accounted for around 11% of the total Indigenous population across the region.

The remaining four regions where data were collated (Goldfields, Great Southern, Wheatbelt, and Southwest) were selected to act as controls and will hence be referred to as non-intervention areas. The Kimberley was excluded from the sample as it received a different type of intervention early in this time period. The Goldfields region had received a small RESET intervention in a remote town with a small population at a similar time period to the Mid West Gascoyne, however, the data retrieved from the Incident Management System for the intervention towns were too small to analyze, and the region was therefore retained as a non-intervention region. To test for any bias that may have arisen from including the Goldfields data in the current study, analyses were repeated without these data; no differences in the results were found. There were no observed differences between intervention and non-intervention areas that could be considered to be confounding factors.

Procedure

The evaluation design was approved by the Deakin University Human Research Ethics Committee, steering groups and committees from within the intervention communities, and the research committees associated with the WA Department for Child Protection and Family Support and WA Police.

Operation data were provided by the Operation RESET team from the Mid West Gascoyne area and recorded onto Microsoft Excel spreadsheets. Prosecutorial outcomes from the arrests and charges made during the intervention in the Mid West Gascoyne were collected on a case-by-case basis by the RESET team once these cases had proceeded to trial.

Data from WA Police Information Management System were extracted by a police analyst who performed a search for child sexual abuse reports by setting parameters for location, date, and type of offense, and collating the relevant case identification numbers. Each case was then checked by a research assistant who extracted information on region, town, incident and reporting dates, and the outcome of the investigation (such as arrest, insufficient evidence, withdrawn, or no offender detected).

In the Mid West Gascoyne, the RESET team was active in communities from April 2009 to September 2010. Therefore, three equal time periods, before, during, and after the RESET intervention period, were used for comparison: Time Period 1 (TP1) refers to the 18 months prior to Operation RESET (October 2007 to March 2009), Time Period 2 (TP2) refers to the 18 months during RESET (April 2009 to September 2010), and Time Period 3 (TP3) refers to the 18 months following RESET (October 2010 to March 2012). In the Pilbara, the intervention was in place for 20 months, and again three equal time periods were used: TP1 refers to the 20 months prior to the intervention (December 2007 to July 2009), TP2 refers to the 20 months during the intervention (August 2009 to March 2011), and TP3 refers to the 20 months following the intervention (April 2011 to November 2012).

Data from the four non-intervention regions were segmented into the same time periods as the Mid West Gascoyne for comparison to ascertain whether results obtained for the intervention area were due to the effect of the operation in these areas, or if they were the effect of other trends that would be reflected in the data from different regions. Report and arrest data from the Mid West Gascoyne and the Pilbara were segmented between intervention and non-intervention areas. Data from all other regions were analyzed by region. Population estimates for the Indigenous population and total population were obtained from the Australian Bureau of Statistics (2013a, 2013b).

Data Analysis

First, the team’s activities from the operation data were analyzed and documented. Following this, a time study investigating the relationship between the team’s presence in a community and the date of individual incident reports was conducted.

For the second section of the analysis, the number of reports, number of arrests, and arrest rates (rate of arrest per incident report) for each region (or area) by time period were compiled. Generalized estimating equations (GEEs) were used to investigate interaction effects for within-subject (time) and between-subject (intervention) comparisons (Liang & Zeger, 1986) using the SPSS version 22. GEEs are an effective tool for analyzing repeated categorical response data to make inferences about population or subpopulation averages (Heagerty & Zeger, 2000). With this technique, the mean response can be modeled and parameter estimates will be consistent even if the covariance structure has not been correctly specified. The GEE technique is suitable for the analysis in the current study as it can allow for different types of unmeasured dependence between outcomes and has been found to be a stable analysis method for these data (Ballinger, 2004). Number of reports, arrests, and arrest rates per region were analyzed in three separate models for intervention and non-intervention areas. Post hoc paired comparisons were used to analyze differences between time periods. Significance levels were set at p < .05 for model tests and p < .01 for paired comparisons, due to the number of tests being conducted.

Results

Section 1: Data Collected During Operation RESET

The relationship between the RESET team’s presence and reports of child sexual abuse was investigated. The data collected during Operation RESET included 135 incident reports of child sexual abuse, with 124 field investigations conducted, leading to 24 arrests. To explore the effect of the presence of the RESET team in the communities, the proximity of the team’s presence to the incident reports was investigated. More than 72% of incidence reports were made during or within 3 working days of Operation RESET’s presence (proactive and reactive) in the communities, as shown in Figure 1.

Figure 1:

Number of Reports of Child Sexual Abuse (CSA) by Number of Working Days Since RESET Team Presence in Communities

Further information was obtained regarding the prosecutorial outcomes of those arrested and charged during the course of Operation RESET and revealed an overall successful prosecution rate of 54% (13 out of the 24 cases led to convictions). Of the 11 unsuccessful cases, 9 were dismissed by the Director of Public Prosecutions and 2 were acquitted. Further analysis revealed that for the first 9 months of the operation, of the 12 cases that were charged, 7 (58%) proceeded to court. Of those, 5 were convicted, giving a successful prosecution rate of 71%. For the second 9 months of the operation, 8 out of 12 cases proceeded to court (66.7%), and all of these cases were successfully prosecuted (100%). These results indicate that there was an improvement in conviction rates as the model used by the RESET team was implemented.

Section 2: Data Collected Through the WA Police Incident Management System

Data for reports, arrests, and arrest rates for the six regions, with the Mid West Gascoyne and Pilbara regions split by intervention/non-intervention, are presented in Table 3. As can be seen by the population estimates for the Indigenous population, the areas of the Mid West Gascoyne that were not covered by the intervention were small in comparison with those covered by the intervention, making up less than 10% of the total Indigenous population for the region.

Table 3:

Number of Reports, Arrests, and Arrest Rates; Indigenous Population Estimates; and Percentage of Indigenous Population per Total Population by Time Period

Time Period	Reports	Arrests^a	% A of IR^b	Indigenous Population Estimates^c	Indigenous per Total Population
a. MWG intervention^d				5,392	11.1
TP1–pre	49	18	36.7
TP2–during	74	41	55.4
TP3–post	63	38	60.3
Total	186	97	52.2
b. MWG non-intervention^d				584	5.4
TP1–pre	13	7	53.8
TP2–during	24	12	50.0
TP3–post	8	4	50.0
Total	45	23	51.1
c. Pilbara intervention				801	32.5
TP1–pre	14	6	42.8
TP2–during	18	8	44.4
TP3–post	12	6	58.3
Total	44	21	47.7
d. Pilbara non-intervention				6,410	11.2
TP1–pre	53	26	49.1
TP2–during	45	20	44.4
TP3–post	27	8	29.6
Total
e. Goldfields				4,723	11.3
TP1–pre	67	38	56.7
TP2–during	66	29	43.9
TP3–post	53	22	41.5
Total	186	89	47.8
f. Great Southern				1,650	3.0
TP1–pre	68	42	61.8
TP2–during	69	42	60.9
TP3–post	54	26	48.1
Total	191	110	57.6
g. Wheatbelt				3,747	4.9
TP1–pre	55	23	41.8
TP2–during	56	23	41.1
TP3–post	52	28	53.8
Total	163	74	45.4
h. Southwest				3,714	2.2
TP1–pre	76	29	38.2
TP2–during	86	36	41.9
TP3–post	83	38	45.8
Total	245	103	42.0

Note. TP1-pre = the time period prior to the intervention; TP2-during = the time period of the intervention; TP3-post = the time period post the intervention.

Arrests includes “Offender Processed.”

% A of IR = percentage of arrests from incident reports.

Australian Bureau of Statistics (2013a).

Mid West Gascoyne Regions

The number of reports increased between TP1 and TP2 (pre-intervention and during intervention) for the Mid West Gascoyne intervention and non-intervention areas (see Table 3a and b), which could be attributable to the presence of the Operation RESET team in this region. Given that the size of the Indigenous population in the non-intervention Mid West Gascoyne areas is only approximately 10% of the total region, and that there could be expected spillover from the intervention into these nearby communities, it is not surprising that there are also increases in reports in the non-intervention areas of the Mid West Gascoyne. The number of reports also increased in the Pilbara intervention region (see Table 3c), but not in the Pilbara non-intervention areas (see Table 3d). To test the significance of changes between time periods by intervention status, three GEE models were produced for reports, arrests, and arrest rates. Means with standard errors for number of reports per time period by intervention status for the models are shown in Table 4.

Table 4:

Means and Standard Errors of Generalized Estimating Equations Analyses by Intervention and Time Period for Reports, Arrests, Arrest Rates

	Reports	Arrests	Arrest Rates
	M (SE)	M (SE)	M (SE)
Non-intervention
TP1-pre	55.33 (8.37)	27.50 (4.61)	50.23 (3.36)
TP2-during	57.67 (7.99)	27.00 (4.08)	47.04 (2.78)
TP3-post	46.17 (9.61)	21.00 (4.78)	44.81 (3.17)
Intervention
TP1-pre	21.00 (6.55)	8.00 (1.63)	41.74 (4.16)
TP2-during	30.67 (8.76)	16.33 (4.58)	53.31 (4.56)
TP3-post	25.00 (11.03)	14.67 (5.89)	65.29 (6.77)

Note. TP1-pre = the time period prior to the intervention; TP2-during = the time period of the intervention; TP3-post = the time period post the intervention.

Reports

The GEE analysis of the number of reports indicated that the intervention by time interaction effect was not significant, Wald χ²(2, N = 24) = 4.61, p = .10. There were significant differences for both the main effects—Intervention: Wald χ²(1, N = 24) = 5.09, p = .02; Time: Wald χ²(2, N = 24) = 60.85, p < .001. Paired comparisons indicated that mean number of reports increased from TP1 to TP2 in the intervention areas (M difference = 9.67, SE = 2.60, p < .001), but not for the non-intervention areas (M difference = 2.33, SE = 2.66, p = .38). There was a non-significant decrease in the intervention areas (M difference = 5.67, SE = 2.59, p = .03) and a significant decrease in reports for the non-intervention areas between TP2 and TP3 (M difference = 11.20, SE = 2.85, p < .001).

Arrests

The GEE analysis of number of arrests indicated that the intervention by time interaction effect was not significant, Wald χ²(2, N = 24) = 5.41, p = .07. There were significant differences for both the main effects—Intervention: Wald χ²(1, N = 24) = 4.54, p = .03; Time: Wald χ²(2, N = 24) = 15.94, p < .001. Paired comparisons indicated that there was a significant increase in number of arrests from TP1 to TP2 in the intervention areas (M difference = 8.33, SE = 3.07, p =.007), but not for the non-intervention areas (M difference = 0.50, SE = 2.29, p = .83). There was no significant change in number of arrests for the intervention areas between TP2 and TP3 (M difference = 1.67, SE = 1.66, p = .31), and a non-significant decrease in arrests for the non-intervention areas between TP2 and TP3 (M difference = 6.00, SE = 3.01, p = .05).

Arrest Rates

The GEE analysis of arrest rates indicated that the intervention by time interaction effect was significant, Wald χ²(2, N = 24) = 22.32, p < .001. Paired comparisons indicated that differences between intervention and non-intervention areas increased over time. There was a significant increase in the arrest rate from TP1 to TP2 (M difference = 11.56, SE = 4.30, p = .007) as well as between TP2 and TP3 in the intervention areas (M difference = 11.98, SE = 3.47, p < .001), with no corresponding increases in the non-intervention areas (TP1–TP2: M difference = 3.19, SE = 2.07, p = .12; TP2–TP3: M difference = 2.24, SE = 3.86, p = .56).

In summary, reports and arrests increased between TP1 and TP2 in the intervention areas, but not in the non-intervention areas. Arrest rates increased in the intervention areas from TP1 to TP2 as well as from TP2 to TP3. There was no improvement in arrest rates between either time period for the non-intervention areas.

Discussion

This study contains both the robust study design and reliable quantitative data required to evaluate changes in reporting rates of child sexual abuse during an intervention designed to improve outcomes for children, their families, and communities. Qualitative information gained from the first evaluation of the intervention (Mace et al., 2015) indicated that there was a positive response from service professionals and community members in the communities where the intervention was implemented. The current study provides quantitative evidence that the intervention achieved its principal aim of increasing reporting rates.

The number of reports significantly increased in the intervention areas during the intervention compared with prior to the intervention, but did not increase in the non-intervention areas. Arrests likewise rose significantly in intervention areas but not in areas without the intervention between these time periods. A decrease in reports was evident in all intervention and non-intervention areas between TP2 and TP3; however, the decreases seen in the intervention areas were not significant. One possible explanation of the overall decline in reports after the intervention is that professionals may have had a better understanding of mandatory reporting criteria over this time. On January 1, 2009, mandatory reporting legislation came into effect in WA, making it a legal requirement for all doctors, nurses, midwives, teachers, and police officers to report all reasonable beliefs of child sexual assault. This caused a rise in reporting over all regions of WA as the legislation became known (Western Australian Department for Child Protection, 2008). During the implementation of these changes, misunderstandings regarding the types of behavior that should be reported resulted in many reports being lodged that were outside the bounds of the legislation. During 2010, an education program was promoted across WA to improve understanding of child sexual abuse and behavior that required mandatory reporting. This may have resulted in a decline in reports throughout the year. No other major changes that could have affected reporting were observed, anecdotally or by police, during this time.

The decline in reporting in the intervention areas between TP2 and TP3 could be due to the better understanding of mandatory reporting as outlined above. Alternatively, it could be due to offenders having been removed from intervention areas by police, which then reduced the number of incidents occurring. A further possible explanation could be that after withdrawal of the intervention team, Indigenous persons were less inclined to report due to the support from and relationships with the RESET team no longer being available. The large arrest rate increase from TP2 to TP3 in the intervention areas, however, appears to indicate that the processes put in place by the team for the period post the intervention were successful. Arrest rates increased between TP1 and TP2 as well as between TP2 and TP3 in the intervention areas, whereas arrest rates remained stable in the non-intervention areas.

Analysis of the data collected on the team’s activities during Operation RESET indicated that there was a strong association between the group’s activities in the intervention areas and subsequent reporting behavior. Victims were most likely to report when the team was active in the intervention area. Importantly, successful conviction rates rose over the course of the operation’s presence in these communities, suggesting that the processes put in place to support victims going through the court system were increasingly successful as the team and support services built rapport with community members over time.

One of the limitations of this study is the inherent difficulty in working with secondary data sources such as police databases, where data are collected primarily for operational rather than research purposes. To ensure the highest quality data, a data reconciliation process between the incident reports supplied by the Operation RESET team and comparative data from the Incident Management System for the RESET intervention areas was carried out. This revealed some discrepancies between the data sets, with a number of cases that appeared in the RESET team data set not included in the Incident Management System. Further analysis revealed that many of these cases in the Incident Management System had been entered as “child abuse” but not “child sexual abuse” and were therefore not within the search parameters. These cases were predominant where no offender was found and the case did not proceed. In cases where this practice had happened and the case had progressed, the anomaly would have been amended as further data were added. It is reasonable to assume that this practice would be systematically equivalent across the entire Incident Management System data set, and therefore we can consider that report and arrest outcomes across regions within this data set would be comparable.

Prosecution and conviction rates between intervention and non-intervention areas were not contained in the Incident Management System, and it was not possible to obtain these at the time of publication. A comparative assessment of successful prosecutions between intervention and non-intervention sites using data for all regions in WA would be an important next step for further research.

Overall, improvements in the number of reports and arrests as well as arrest rates in the intervention areas are consistent with broader reform in the area of child sexual abuse in Indigenous communities, and support a rollout of the model used by Operation RESET. These findings, in conjunction with the prior qualitative study (Mace et al., 2015), suggest that these reforms have contributed to a major cultural shift and level of engagement between victims and service providers. The central concept underlying the development of the reforms is a more accessible, efficient, coordinated, and user-friendly model of service delivery, where local solutions are developed for local problems. The findings of the current evaluation are entirely consistent with this concept.

As with any new system, there are opportunities for improvement. Further investment is required to isolate which elements of the intervention are the most important to implement to make the most efficient use of time and resources. Data need to be collated from numerous sources for ongoing evaluation, and extensive record keeping is required. As more information becomes available through the rollout of future intervention programs, researchers can begin to determine the most significant components of the intervention and the most cost-effective means of implementing change.

The evidence from this evaluation, based on police data with a large and representative sample of communities, suggests that the reforms have corresponded with a marked improvement in some key outcomes for Indigenous victims of child sexual abuse. We recommend that all Australian jurisdictions consider moving to the more proactive model that has been used in the Operation RESET intervention, and that these findings may be applicable to other Indigenous communities internationally.

Footnotes

Acknowledgements

The authors would like to thank detective senior sergeant Bradley Burnett, senior investigating officer for Operation RESET, Bronwen Manger for supplementary data analysis and editorial assistance, and Lucy Busija for statistical advice.

Notes

Cate Bailey is a PhD candidate in psychology at Deakin University (Melbourne campus), and a research fellow with the Centre for Investigative Interviewing. She has particular expertise in statistical analysis, including analysis of large, longitudinal data sets and tracking of child sexual assault cases.

Glenn Mace is a district director of the Department for Child Protection and Family Support, Western Australia, and co-developer of Operation RESET.

Martine Powell is a professor of psychology at Deakin University (Melbourne campus) and founding director of the Centre for Investigative Interviewing. She is a leading expert in the evaluation of reforms related to child sexual offenses.

Mairi Benson is a researcher and industry trainer with the Centre for Investigative Interviewing, with particular expertise in interviewing of child complainants of sexual abuse.

References

Aboriginal Child Sexual Assault Taskforce. (2006). Breaking the silence: Creating the future. Addressing child sexual assault in Aboriginal communities in NSW. Sydney, Australia: NSW Attorney General’s Department.

Australian Bureau of Statistics. (2013a). Estimated resident Aboriginal and Torres Strait Islander and Non-Indigenous population, SLA and LGA—30 June 2011. Retrieved from http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3238.0.55.001June%202011?OpenDocument

Australian Bureau of Statistics. (2013b). QuickStats: A simple at-a-glance summary of Census statistics for your selected area. Retrieved from http://www.abs.gov.au/websitedbs/censushome.nsf/home/quickstats?opendocument&navpos=220

Australian Institute of Health and Welfare. (2011). Child protection Australia 2009-10 (Child Welfare Series No. 51. Cat. No. CWS 39). Canberra: Author.

Ballinger

G. A.

(2004). Using generalized estimating equations for longitudinal data analysis. Organizational Research Methods, 7, 127-150.

Berlyn

Bromfield

Lamont

(2011). Child protection and Aboriginal and Torres Strait Islander children. Melbourne: Australian Institute of Family Studies.

Bromfield

Holzer

P. J.

(2008). A national approach for child protection: Project report. Melbourne: Australian Institute of Family Studies.

Calma

(2006). Ending family violence and abuse in Aboriginal and Torres Strait Islander communities—Key issues: An overview paper of research and findings by the Human Rights and Equal Opportunity Commission, 2001–2006. Canberra, Australia: Human Rights and Equal Opportunity Commission.

Heagerty

P. J.

Zeger

S. L.

(2000). Marginalized multilevel models and likelihood inference (with comments and a rejoinder by the authors). Statistical Science, 15, 1-26.

10.

Liang

K.-Y.

Zeger

S. L.

(1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13-22.

11.

Lievore

(2003). Non-reporting and hidden recording of sexual assault: An international literature review. Canberra, Australia: Commonwealth Office of the Status of Women.

12.

Mace

Powell

M. B.

(2012). New child protection strategy for Aboriginal communities in Western Australia. Current Issues in Criminal Justice, 24, 273-283.

13.

Mace

Powell

M. B.

Benson

(2015). Evaluation of Operation RESET: An initiative for addressing child sexual abuse in Aboriginal communities. Australian & New Zealand Journal of Criminology, 48, 82-103.

14.

Mullighan

(2008). Children on Anangu Pitjantjatjara Yankunytjatjara (APY) Lands: Commission of inquiry: A report into sexual abuse. Adelaide: South Australian Government.

15.

NSW Ombudsman. (2012). Responding to child sexual assault in Aboriginal communities. Sydney, Australia: Author.

16.

Northern Territory Emergency Response Review Board. (2008). Northern Territory Emergency Response: Report of the NTER Review Board. Canberra: Commonwealth of Australia.

17.

Snell

(2003). Child advocacy centers: One stop on the road to performance-based child protection. Los Angeles, CA: Reason Foundation.

18.

Stanley

Tomison

A. M.

Pocock

(2003). Child abuse and neglect in Indigenous Australian communities. Canberra: Australian Institute of Family Studies.

19.

Thompson

S. C.

Greville

H. S.

Param

(2008). Beyond policy and planning to practice: Getting sexual health on the agenda in Aboriginal communities in Western Australia. Australia & New Zealand Health Policy, 5, 3-11.

20.

Western Australian Department for Child Protection. (2008). Mandatory reporting of child sexual abuse in Western Australia: A guide for mandatory reporters. Retrieved from http://mandatoryreporting.dcp.wa.gov.au/Resource%20Documents/Information%20booklet%20for%20mandatory%20reporters.pdf

21.

Wild

Anderson

(2007). Ampe akelyernemane meke mekarle: “Little children are sacred”: Report of the Northern Territory Board of Inquiry into the protection of Aboriginal children from sexual abuse 2007. Darwin, Australia: Office of Indigenous Policy, Department of the Chief Minister.