Australian Institute of Criminology

Skip to content

The prevalence and nature of Indigenous violent offending

Qualitative evidence indicates that 'Indigenous violence is widespread and disproportionately high compared to non-Indigenous violence in Australian society' (Memmott & National Crime Prevention 2001: 2). The relatively limited amount of empirical data currently available supports this conclusion. Using statistics derived predominantly from police apprehensions, this section explores:

  • the extent to which Indigenous persons are charged with offences of violence and how this compares with other non-violent offending;
  • the different types of violence involved;
  • the racial identity of the victims of Indigenous violence and their relationship to the offender; and
  • levels of recidivism.

Wherever possible, comparisons are drawn with non-Indigenous offenders to identify both differences and similarities in patterns of violence.

In the absence of national data on this topic, the discussion will draw heavily on the very small number of states (notably Western Australia and South Australia) that publish annual statistics on Indigenous apprehensions. However, no direct comparisons between these two jurisdictions can be drawn because of differences in criminal justice processes, data extraction methods, offence classificatory systems etc. For example, the WA data described below pertain to the most serious charge per apprehension report, while much of the SA data include all charges laid, irrespective of the number of apprehension reports or discrete individuals involved.

In using police apprehension data, the term 'violence' will be applied to those offences which are listed by police as either 'inter-personal' or 'against the person'. In both Western Australia and South Australia, these include homicide and related offences, assault, sexual offences and kidnapping/abduction. In Western Australia, dangerous operation of a vehicle is also included in this category, although in South Australia this offence is defined as a driving offence.

Indigenous apprehensions for violent offending

In both Western Australia and South Australia, Indigenous people are substantially more likely to be apprehended by police for violent offences than non-Indigenous people.

In Western Australia during 2005, there were 4,911 police apprehensions involving Indigenous persons where the most serious charge was an offence of violence (Loh et al. 2007). This represents an apprehension rate of 111.1 per 1,000 Indigenous population aged 10 years and over. These apprehensions involved 3,796 discrete individuals (Loh et al. 2007: Table 2.1). The fact that this latter figure is lower than the total apprehensions indicates that at least some persons were apprehended more than once during the 12 month period. Overall, for every 1,000 Indigenous population aged 10 years and over, 85.9 individuals were formally proceeded against by police at least once during 2005. Indigenous rates were also substantially higher than those of the non-Indigenous population. As Figure 1 shows, the Indigenous apprehension rate in 2005 was 22 times as high as the non-Indigenous rate and was 19 times as high for discrete persons apprehended.

SA data for 2006 reflect a similar pattern (OCSAR 2007). Although lower than in Western Australia, the rate of Indigenous apprehensions for violent offences in this state was almost 18 times as high as the non-Indigenous rate and was 15.3 times as high for discrete persons apprehended (see Figure 1).

Comparison between Indigenous violent and non-violent apprehension rates

When compared with other types of criminal behaviour, such as property offending or illicit drug use, violence features prominently in the charge profiles of Indigenous persons apprehended by police.

As shown in Figure 1, Western Australian Indigenous apprehension rates for violent offending in 2005 were higher than for any other offence type. At 111.1 per 1,000 Indigenous population aged 10 years and over, it was 1.1 times the apprehension rates for 'against justice' offences and 1.3 and 1.5 times as high as good order and property offence rates respectively. It was also a substantial nine times the apprehension rate for illicit drug offences (derived from Loh et al. 2007: Table 2.4).

Figure 1: Apprehension reports and discrete persons apprehended for a violent offence by Indigenous status, Western Australia 2005 and South Australia 2006 (rates per 1,000 relevant population 10 years and over)

Apprehension reports and discrete persons apprehended for a violent offence by Indigenous status, Western Australia 2005 and South Australia 2006

Note: Rates are based on ABS unadjusted population figures for Western Australia and South Australia extracted from the 2006 census (ABS cat. no. 2068.0). The Indigenous population figures are lower than the estimated figures released by the ABS (see cat. no. 4705.0) which include adjustments for those respondents who did not record their Indigenous status. However, actual census data rather than estimated data were used because they provide age specific and gender specific breakdown which is not available for the estimated data

In Western Australia, only the most serious charge per apprehension report or per individual is counted. In South Australia, each apprehension report which includes an offence against the person is counted, even if it does not represent the most serious charge in that report. Similarly, each person charged with at least one against the person offence in 2006 is counted, even if they are also charged with more serious offences

In both states, Indigenous status is primarily determined by police according to the physical appearance of the individual

Source: derived from Loh et al. 2007: Tables 2.1 and 2.4; OCSAR 2007: Table 6.27

When discrete individuals, rather than apprehensions, are considered, the dominance of violent offences becomes even more pronounced. Figure 2 shows that in 2005, 85.9 Indigenous persons per 1,000 population aged 10 years and over in Western Australia had an offence of violence recorded as their most serious charge over that 12 month period, which was 2.4 times as high as that of property offences which, as the second ranked category, recorded a rate of just 36.3 per 1,000 population (Loh et al. 2007).

Figure 2: Indigenous apprehension reports and Indigenous persons apprehended by type of offence, Western Australia 2005 (rates per 1,000 Indigenous population 10 years and over)

Indigenous apprehension reports and Indigenous persons apprehended by type of offence, Western Australia 2005

Note: See description of counting rules under Figure 1

Source: Derived from Loh et al. 2007: Tables 2.1 and 2.4

Table 1: Indigenous apprehension reports and Indigenous persons apprehended by type of offence, South Australia 2006 (rates per 1,000 Indigenous population 10 years and over)
Charge type Apprehensions Discrete individuals apprehended
Violent offences 75.0 58.4
Robbery and extortion 3.2 3.1
Offences against property 104.4 67.1
Driving offences 89.1 65.1
Drug offences 4.2 4.0
Offences against good order 199.1 97.3

a: Rates are based on ABS unadjusted population figures for South Australia extracted from the 2006 census (ABS cat. no. 2068.0). The Indigenous population figures are lower than the estimated figures released by the ABS (2006b cat. no. 4705.0) which include adjustments for respondents who did not record their Indigenous status. However, actual census data rather than estimated data were used here because they provide age and gender specific breakdowns not available for the estimated data. See description of counting rules under Figure 1

Source: Derived from OCSAR 2007: Table 6.27

Patterns were somewhat different in South Australia. Table 1 shows that, while the rate of violent offences in 2006 was still comparatively high, it was lower than those recorded for good order, property and driving offences. This finding applied irrespective of whether analysis focused on all apprehension reports or the number of discrete persons apprehended at least once during the 12 month period.

While these differences between South Australia and Western Australia may indicate actual variations in Indigenous offending patterns, they may also be due to differences in the counting rules used to extract the data. Also relevant is the fact that, as noted earlier, Western Australia includes dangerous driving as a violent offence, whereas South Australia does not.

Indigenous apprehension rates for different types of violence

Using apprehension data from Western Australia and South Australia, together with some comparatively early police statistics from New South Wales and national homicide data from the NHMP, this section examines:

  • the type of violent offences charged against Indigenous offenders; and
  • the extent to which these differ from those recorded by non-Indigenous persons. For simplicity, only data relating to apprehension rates are presented, while rates per discrete individual apprehended during the 12 month period are not included.

The data again indicate broad similarities between these jurisdictions. In particular:

  • Indigenous apprehensions are far more likely to involve a charge of assault than any other violent offence; and
  • Indigenous apprehension rates are substantially higher than non-Indigenous rates within each violent offence category.
Table 2: Apprehension reports by type of violent offence and Indigenous status, Western Australia 2005 (rates per 1,000 relevant population 10 years and over)
Charge type Indigenous rate per 1,000 Non-Indigenous rate per 1,000
Homicide and related offences 0.5 0.4
Acts intended to cause injury (ie assault) 78.5 2.9
Sexual assault and related offences 3.4 0.3
Dangerous or negligent acts endangering persons (ie dangerous operation of a vehicle) 28.5 1.8
Abduction and related offences 0.2 <0.1
Total 111.1 5.1

Note: there were 70 apprehensions involving a violent offence where information on Indigenous status was not recorded

See description of counting rules under Figure 1

Source: Derived from Loh et al. 2007: Table 2.4

Western Australia

In Western Australia, the Indigenous apprehension rate for assaults (78.5 per 1,000 Indigenous population aged 10 years and over) was almost three times that of the next highest violence category, dangerous operation of a vehicle (28.3 per 1,000; see Table 2). In contrast, sexual assault and homicide rates were low.

Table 2 also shows that within each charge subcategory, Indigenous apprehension rates were higher than non-Indigenous rates. Assaults recorded the greatest difference, where the apprehension rate for Indigenous persons was 27 times as high as that of non-Indigenous persons. Large variations were also observed for sexual offences, where the Indigenous rate (3.4) was 11 times that of the non-Indigenous rate (0.3). The smallest difference was recorded for homicides, where the Indigenous rate (0.5 per 1,000 population) was only 1.2 times higher than the non-Indigenous apprehension rate (0.4). However, the total number of homicide apprehensions was small for both groups (Loh et al. 2007).

South Australia

In South Australia, as in Western Australia, common assault dominated the charge profile of Indigenous offenders, with an apprehension rate of 62.4 per 1,000 Indigenous population aged 10 years and over. This was over nine times as high as the rate recorded for assault occasioning actual or grievous bodily harm (6.7 per 1,000 population). A comparatively low apprehension rate was recorded for sexual offences (3.8 per 1,000 Indigenous population aged 10 years and over) which, at least in part, may be due to the lower reporting and lower detection rates associated with this type of offence, particularly within Indigenous communities.

Across all of the violent charge categories listed, Indigenous apprehension rates were higher than non-Indigenous rates. These inter-group differences were particularly pronounced for both common assault and assault occasioning actual bodily harm, with Indigenous apprehension rates approximately 20 times as high as the non-Indigenous rates for each. The Indigenous rate for sexual offences (3.8) was almost five times the non-Indigenous rate (0.8).

Table 3: Apprehension reports by type of violent offence and Indigenous status, South Australia 2006 (rates per 1,000 relevant population 10 years and over)
Charge type Indigenous rate per 1,000 Non-Indigenous rate per 1,000
Homicide related offences 0.4 0.1
Assault 82.4 4.3
Assault occasioning actual/grievous bodily harm 6.7 0.3
Common assault 75.7 4.0
Kidnapping/abduction 0.7 0.1
Ill-treatment of children 0 0
Stalking 0 <0.1
Other non-sexual offences against the person 5.8 0.4
Sexual offences 3.8 0.8
Rape 1.5 0.2
Indecent assault 1.1 0.2
Unlawful sexual intercourse 0.4 0.2
Incest 0 <0.1
Other sexual offences 0.8 0.2
Total 93.1 5.6

Note: In this table, all charges included in apprehension reports are counted

Source: Derived from OCSAR 2007: Tables 6.15 and 6.16

New South Wales

Although New South Wales does not publish regular statistics on Indigenous violent offenders, one-off data detailing the number of arrests per selected offence category during the 2000 calendar year show that, as in South Australia and Western Australia:

  • the Indigenous arrest rate exceeded the non-Indigenous rate in every offence category listed;
  • within the Indigenous group, arrest rates were highest for domestic violence assault (1,993.2 per 100,000 population), followed by the property offence of break/enter (1,895.9 per 100,000 population). These were also the two most prominent charge types laid against non-Indigenous offenders (280.1 and 191.7 per 100,000 population respectively); and
  • although arrest rates for murder were comparatively low for both groups, the Indigenous rate (9.7 per 100,000 population) was almost six times as high as the non-Indigenous rate (1.7 per 100,000 population; Weatherburn, Fitzgerald & Hua 2003).

The NSW data also contain separate breakdowns for the sexual assault of children. Among both Indigenous and non-Indigenous offenders, these rates were lower than those recorded for all sexual assaults. However, the Indigenous rate (65.5 per 100,000 population) still exceeded the non-Indigenous rate (19.2 per 100,000 population) by a factor of 3.4.

Table 4: Arrest rates by Indigenous status, New South Wales 2000 (major charge per arrest)
Offence type Indigenous rate per 100,000 population Non-Indigenous rate per 100,000 population Degree of difference
Violent offences
Murder 9.7 1.7 +5.7
Sexual assault 133.8 35.5 +3.8
Sexual assault against children 65.5 19.2 +3.4
Assault—domestic violence related 1,993.2 280.1 +7.1
Assault—grievous bodily harm 228.3 20.6 +11.1
Property offences
Robbery 402.1 65.7 +6.1
Break/enter 1,895.9 191.7 +9.9
Motor vehicle theft 689.7 92.0 +7.5

Note: These data cannot be compared with those from South Australia and Western Australia because the NSW data relate only to arrests and exclude persons apprehended by other means (such as a summons)

Source: Weatherburn, Fitzgerald & Hua 2003: 67

National homicide data

Arguably the only source of national data on Indigenous apprehensions for violence is that provided by the AIC's NHMP. It indicates that a disproportionately high percentage of homicides in Australia are committed by Indigenous offenders. In 2005–06, Indigenous persons constituted 22 percent of those 314 individuals charged with homicide (where racial identity was recorded), although they constituted less than three percent of the Australian population (Davies & Mouzos 2007).

Indigenous overrepresentation was particularly pronounced in the Northern Territory, where this group accounted for all 27 persons apprehended for this offence in 2005–06. In Western Australia, 41 percent of persons apprehended for homicide were Indigenous (12 of the 29), which was 9.2 times higher than expected, given that this group represented only three percent of that state's total population. They also accounted for 13 of the 62 homicide offenders in Queensland (21%), 12 of the 111 in New South Wales (11%) and two of the 23 in South Australia (7%). Of the 53 persons apprehended for this offence in Victoria in 2005–06, none were Indigenous. The same applied in Tasmania, where only two offenders were apprehended that year for homicide (Davies & Mouzos 2007).

Even these disproportionately high figures are likely to underestimate the actual level of Indigenous homicides as they do not include victims (predominantly females) whose deaths are ostensibly attributed to other factors (such as renal failure) but which are, in effect, the culmination of long-term spousal abuse (Memmott & National Crime Prevention 2001).

Self-reported levels of Indigenous violent offending

Indigenous population surveys: 1994 NATSIS and 2002 NATSISS

According to the 1994 NATSIS, over 20 percent of Indigenous respondents aged 13 years and over living in urban and rural/remote Australia reported that they had been arrested by police at least once in the preceding five years. Of these, 17 percent indicated that their most recent arrest had been for assault. This was lower than the proportion arrested for disorderly conduct/public drinking (32%) and drink driving (23%). However, it was higher than the percentage arrested for theft/burglary (15%; Mukherjee et al. 1998). At the time of the 2002 NATSISS, the proportion of Indigenous respondents arrested in the preceding five years had declined to 16 percent (while the age range of persons surveyed by NATSIS and NATSISS was slightly different, namely 13 years and over compared with 15 years and over (ABS 2004), this is unlikely to have affected the results). However, no specific breakdowns of the types of offences involved were collected.

Targeted population surveys: The NSW School Survey

In contrast to the national Indigenous population surveys, a 1999 survey of 3,600 secondary school students in New South Wales sought information on the prevalence and frequency of each individual's actual offending behaviour, irrespective of whether it came to official notice (Weatherburn, Fitzgerald & Hua 2003). As summarised in Table 5, a higher proportion of Indigenous students indicated they had committed at least one offence in the previous 12 months than non-Indigenous students. This difference applied across all six offence categories examined, including assault. Among Indigenous students, one in five (19%) indicated that they had committed at least one assault in the preceding 12 months, while among those Indigenous students classified as 'active offenders', just over half (51%) admitted committing more than five assaults during that period. Both figures were more than double those recorded by non-Indigenous students. Among this latter group, only eight percent admitted to committing at least one assault in the preceding 12 months while 22 percent of those students classified as 'active offenders' reportedly committed five or more assaults during the same period.

Offender-based surveys

As noted earlier, a particularly useful source of information on self-reported violence among Indigenous offenders is the DUCO survey. However, because the respondents were all incarcerated at the time of interview, the results apply only to a subset of relatively serious offenders and should not be generalised to all offenders.

Table 5: Self-reported offending among Indigenous and non-Indigenous school students in New South Wales, 1999
Offence % who committed at least 1 offence in past 12 months Of those who were active offenders, % who committed more than 5 offences in past 12 months
Indigenous % Non-Indigenous % Students per category (n) Indigenous % Non-Indigenous % Students per category (n)
Assault 19 8 324 51 22 107
Vehicle theft 10 2 100 50 20 39
Break/enter 9 4 139 43 18 50
Receiving 24 17 615 40 19 171
Shoplifting 16 8 291 45 28 135
Malicious damage 36 32 1,132 33 19 309

Source: Weatherburn, Fitzgerald & Hua 2003: Table 1: 68

During one phase of this project, information on self-reported offending was collected from 2,135 adult males imprisoned in Western Australia, Queensland, Tasmania and the Northern Territory in mid 2001. Approximately one-quarter of these individuals were Indigenous. Based on their self-reported lifetime offending behaviour, one-third (34%) of Indigenous respondents were classified as regular violent offenders while one-third (32%) were also listed as regular multiple offenders (Makkai & Payne 2003).

When asked about the particular types of violence committed at any stage during their lifetime (see Figure 3) over seven in 10 Indigenous respondents (72%) admitted to physically assaulting another person, while one in three (29%) reportedly did so on a regular basis. The proportion who admitted to sexual offending, either 'ever' or 'regularly', was substantially lower (at 13% and 3% respectively; Putt, Payne & Milner 2005).

The percentage of Indigenous prisoners who reported that they had committed at least one physical assault 'ever' or 'regularly' was significantly higher than that reported by non-Indigenous prisoners. In contrast, no significant inter-group differences were observed for sex offences or for the act of 'killed someone'.

That physical assault featured prominently in the offending profile of these Indigenous prisoners is further indicated in Table 6, which compares levels of self-reported behaviour across a range of violence and non-violent offence types. Assault was the highest ranking offence 'ever' committed and was the third highest offence committed on a regular basis, behind buying illegal drugs and break/enter.

A second component of DUCO conducted in 2003 surveyed 470 adult female prisoners in six Australian jurisdictions (Johnson 2004); of these, 27 percent were Indigenous. As was the case with males, almost three-quarters of the Indigenous female respondents (73%) admitted to physically assaulting another person at some stage in their lives, while of these, 16 percent did so on a regular basis. These figures were much higher than those recorded by non-Indigenous females, 40 percent of whom admitted to an assault 'ever' while of these, only five percent admitted to regular involvement in this type of offending. As a result, the escalation rate (ie the percentage of those who, having committed the initial offence, went on to become regular offenders) was higher among Indigenous than non-Indigenous women (22% compared with 13%).

As Table 7 shows, in terms of the lifetime offending patterns of Indigenous women, assault was the highest ranked offence, followed by drug offences (61%) and break/enter (40%). However, among regular Indigenous offenders, assault was ranked well down the list, behind drug offences, stealing and break/enter. As a result, the escalation rate for assault was low compared with the other categories listed. In other words, while a higher proportion of Indigenous female prisoners had committed this type of violence at some stage in their lives compared with property or drug offending, far fewer went on to become regular violent offenders. A relatively similar pattern was observed among non-Indigenous female prisoners.

Figure 3: Self-reported lifetime and regular offending among adult male prisoners by Indigenous status and type of violence, DUCO 2001 (%)

Self-reported lifetime and regular offending among adult male prisoners by Indigenous status and type of violence

Source: Putt, Payne & Milner 2005: 3

Table 6: Self-reported lifetime and regular offending among Indigenous adult male prisoners by type of offence: DUCO 2001 (%)
Offence type Indigenous prisoners who reported ever committing this offence Indigenous prisoners who reported regular involvement in this offence
Physical assault 72 29
Break/enter 61 32
Bought illegal drugs 56 46
Motor vehicle theft 52 23
Stealing without break in 48 24
Traded stolen goods 34 21
Vandalism 33 9
Sold illegal drugs 30 18
Robbery without weapon 24 9
Armed robbery 19 8
Sex offence 13 3
Fraud 12 4
Killed someone 8 0

Source: Putt, Payne & Milner 2005: 3

Table 7: Self-reported lifetime and regular offending among adult women prisoners by Indigenous status and offence type, DUCO 2003 (%)
Offence type Indigenous Non-Indigenous
Ever Regular Escalation Ever Regular Escalation
Physical assault 73 16 22 40 5 13
Break and enter 40 24 60 32 15 47
Stealing without breaking in 44 30 68 55 33 60
Traded stolen goods 35 22 63 44 29 66
Vandalised property 20 16 2 13
Fraud, forgery 23 12 52 48 22 46
Robbery without weapon 14 5 36 11 2 18
Armed robbery 12 14 2 14
Drug offences 61 52 85 72 66 92

Source: Johnson 2004: 95

While the self-reported offending profiles of Indigenous male and female adult prisoners were relatively similar, the juvenile component of DUCO (involving mainly male respondents) produced some different results (Prichard & Payne 2005). This survey of 371 young people held in custody in 2004 across Australia showed that Indigenous youths (who comprised 59% of the total sample) were less likely to engage in physical assault than in drug offences, break and enter, stealing and trading stolen goods either during their lifetime or on a regular basis (see Figure 4).

Figure 4: Self-reported lifetime and regular offending among Indigenous juvenile detainees by offence type, DUCO 2004 (%)

Self-reported lifetime and regular offending among Indigenous juvenile detainees by offence type

Source: Prichard & Payne 2005: 84

Also, contrary to the adult findings, significantly fewer Indigenous than non-Indigenous juveniles had committed an assault ever (65% compared with 84% respectively), while slightly (but not significantly) fewer of these were regularly involved in this type of behaviour (28% of Indigenous and 31% of non-Indigenous youths). Overall, the escalation rate for physical assault was relatively similar (at 43% for young Indigenous detainees and 37% for non-Indigenous detainees).

Extent of Indigenous violent offending: A summary

While anecdotal information suggests that violent behaviour is widespread within many Indigenous communities, most of the available statistics measure either the individual's level of contact with the criminal justice system or the self-reported offending behaviour of small, potentially unrepresentative, groups of Indigenous offenders (notably prisoners) currently being dealt with by the criminal justice system. Despite these limitations, police apprehension data from Western Australia and South Australia indicate that, in recent years:

  • Indigenous people were substantially more likely to be apprehended by police for an offence of violence than non-Indigenous people. In both states, the Indigenous apprehension rate for violent offending was between 15 and 20 times the non-Indigenous rate.
  • In Western Australia, Indigenous persons were more likely to be apprehended for a violent offence than for any other type of offence, such as property, drug or good order matters. These findings did not apply in South Australia and this may be due to the different counting rules used to extract the data.
  • By far, the most common violent offence charged against Indigenous persons was common or minor assault. In contrast, rates of apprehension for sexual assault were very low in both states, although this may be more reflective of low reporting and detection levels rather than low involvement in this type of behaviour.
  • Irrespective of the type of violence involved, Indigenous apprehension rates consistently exceeded non-Indigenous rates.

Indigenous population surveys and self-report studies of students and prisoners identified similar patterns.

  • According to the 1994 NATSIS, one in five Indigenous persons had been arrested by police at least once in the preceding five years, and of these, 17 percent had been charged with assault at the time of their most recent arrest.
  • A NSW survey of school students found that one in five Indigenous students admitted to assaulting another person in the preceding 12 months, while among those classified as 'active offenders', one in two reported committing more than five assaults in the same period. These figures were double those reported by non-Indigenous students.
  • Among adult male and female prisoners surveyed as part of DUCO, approximately three-quarters admitted to assaulting another person at least once in their lifetime, while 29 percent of Indigenous males and 16 percent of Indigenous females apparently did so on a regular basis. These figures were substantially higher than those recorded by non-Indigenous prisoners. In contrast, although levels of assaultive behaviour were still high among Indigenous juvenile detainees, they were lower than those recorded by non-Indigenous detainees.

Recidivism among violent Indigenous offenders

In the absence of any population-based surveys designed to obtain details from Indigenous respondents on their actual levels of offending and re-offending, most Indigenous recidivism studies undertaken in Australia have had to rely on official criminal justice data. These have defined recidivism either as re-apprehension, re-conviction or re-imprisonment and as such, have measured re-contact with the system itself, rather than actual re-offending behaviour. Although the overwhelming majority of these have focused on all Indigenous offenders rather than violent offenders per se, they have consistently identified much higher levels of re-contact among Indigenous than non-Indigenous offenders, irrespective of age or gender (for an overview of some of the findings on Indigenous re-contact in general see SCRGSP 2007: s 9.2).

Similar findings have emerged from the handful of analyses that have attempted to assess levels of re-contact by Indigenous violent offenders (see below). However, in presenting these results, it should be noted that such studies are beset by classification problems because of the fact that most recidivists do not specialise in only one type of offending. The usual strategy adopted is to classify offenders according to the offence for which they were most recently charged, convicted or imprisoned. But this is artificial because it defines as 'non-violent' those individuals who, although currently being dealt with for another type of offence, such as a property or drug matter, may have had prior episodes of violence. A more accurate approach would be to classify perpetrators according to the number of violent offences for which they were charged, convicted or imprisoned over a specified time period (eg 1 or 5 years). However, such data are not easily extracted from official criminal justice databases. This definitional limitation should be borne in mind when interpreting the results outlined below.

Prisoners' prior contact with the custodial system

The annual census of adults in Australian prisons includes statistics on the proportion who had previously served a term of incarceration, according to the type of offence for which the individual was currently imprisoned (ABS 2006a).

As Figure 5 indicates, of the 396 Indigenous persons incarcerated across Australia on 30 June 2006 for a homicide or related offence, 65 percent had a record of prior incarceration, as did:

  • Seventy-five percent of those currently imprisoned for acts intended to cause injury;
  • Sixty-six percent of sexual offenders;
  • Eighty-seven percent of those incarcerated for dangerous/negligent acts; and
  • Sixty percent for those involved in abduction or a related offence.
Figure 5: Sentenced prisoners who had a prior imprisonment by most serious current offence and Indigenous status, Australia, 2006 (%)

Sentenced prisoners who had a prior imprisonment by most serious current offence and Indigenous status, Australia, 2006

Note: for comparative purposes, only a selection of non-violent offences have been included here
Source: ABS 2006a: Table 8

These figures were generally consistent with those recorded by Indigenous offenders currently imprisoned for a non-violent offence. Among this latter group, the percentage who had previously been incarcerated ranged from 69 percent of those currently imprisoned for deception or property damage, up to 85 percent of those currently in jail for a road traffic/motor vehicle regulatory offence.

Figure 5 also shows that a much higher proportion of Indigenous prisoners currently serving time for a violent offence had a prior record compared with their non-Indigenous counterparts. Indigenous levels ranged from 1.3 times greater for abduction up to 2.1 times greater for sexual offences. Similar variations were also observed between Indigenous and non-Indigenous non-violent offenders.

Predicting re-contact with the prison system

Rather than focusing on an individual's prior contact with the criminal justice system, an alternative approach is to use survival analysis to determine the probability that an individual will re-offend in the future. One of the few (and by now, somewhat dated) studies undertaken in Australia examined the likelihood of re-incarceration among first-time Indigenous and non-Indigenous prisoners released from WA jails between July 1975 and June 1984 (Broadhurst et al. 1988). The study paid particular attention to differences in the probability of re-contact among male prisoners, depending on the type of offence for which individuals were imprisoned at the time of release. It also examined the time taken between initial release and re-imprisonment (ie the time to 'fail').

While levels of re-incarceration among Indigenous male releasees were very high across all offence types, Indigenous sex offenders and 'against person' offenders had slightly lower probabilities of re-incarceration (75% per category) than did motor vehicle theft, property and good order offenders, all of whom had probabilities exceeding 80 percent (see Figure 6).

Figure 6: Probability of re-incarceration among male prison releasees by Indigenous status, Western Australia 1975–84 (%)

Probability of re-incarceration among male prison releasees by Indigenous status, Western Australia 1975-84

Note: The very small numbers of Indigenous persons with a most serious offence of homicide (n=20) or drug offences (n=10) prevented reliable calculation of recidivism probabilities for these two offence types

Source: Broadhurst et al. 1988: 94

While non-Indigenous releasees exhibited greater variability in re-incarceration probabilities from one offence type to another, the patterns were similar to those of Indigenous releasees. Non-Indigenous males imprisoned for a sex offence or an offence against the person had a lower probability of being re-incarcerated than those previously imprisoned for a motor vehicle, property, good order or traffic offence.

However, across all offences types, Indigenous male releasees had a higher probability of re-incarceration than non-Indigenous male releasees. The greatest discrepancy between the two groups was recorded for sex offences (where the Indigenous probability of re-incarceration was 2.5 times greater than the non-Indigenous probability) and against person offences (1.9 times greater). In contrast, the Indigenous probability of re-incarceration for traffic offences was only 1.3 times higher than the non-Indigenous figure.

In terms of the median time taken by Indigenous males to 'fail' (ie to be re-incarcerated) there were some slight variations between offence types. For those most recently imprisoned for a sex offence or an 'against person' offence, it took only 12.2 months and 10.4 months respectively for them to be re-incarcerated (see Figure 7). These were slightly higher than the results for property offenders (who had a median 'time to fail' of 8.5 months) but lower than for traffic offenders (14.2 months). A similar pattern was evident among non-Indigenous male releasees. Those who were serving time for a sex offence or an 'against person' offence at the point of release took slightly longer to be re-incarcerated than property offenders, but recorded a shorter time to 'fail' than traffic offenders and, most noticeably, drug offenders. Interpreting the findings is difficult, however, because the time taken to be re-incarcerated is a product of both the time taken to re-offend and the time taken by the criminal justice system to reconvict and re-imprison that individual. The more serious the offence, the longer it usually requires to complete the judicial process, particularly if the defendant enters a 'not guilty' plea.

Figure 7: Median time to fail among male prison releasees by Indigenous status, Western Australia, 1975–84 (months)

Median time to fail among male prison releasees by Indigenous status, Western Australia 1975-84

Note: The very small numbers of Indigenous persons with a most serious offence of homicide (n=20) or drug offences (n=10) prevented reliable calculation of recidivism probabilities

Source: Broadhurst et al. 1988

A follow-up study of prisoners released from Western Australian jails between 1 July 1975 and 30 June 1987 focused on re-contact levels among 560 male sexual offenders. Rather than classifying individuals according to the offence for which they were imprisoned at the time of release, the study defined a sex offender as any prisoner who had been incarcerated for a sexual offence at some stage during their recorded criminal careers. It also considered the type of offence for which these individuals were re-incarcerated. Results indicated that the chance of re-incarceration was significantly higher for Indigenous than non-Indigenous sex offenders (80% compared with 35% respectively). For both groups, the probability of re-incarceration was higher among younger individuals and those with a prior record, although the same was also true of non-sex offenders. There was also a relatively high probability that sex offenders would be re-incarcerated for a violent offence, although the degree of 'specialisation' in sex offences per se was low. The study therefore concluded that 'aggression rather than perversion is the more salient characteristic of sex offenders' (Broadhurst & Maller 1991).

The finding that Indigenous violent offenders are more likely to be re-incarcerated and within a shorter timeframe than their non-Indigenous counterparts has since been replicated by other studies. For instance, an analysis of nearly 9,000 violent male offenders released from prison in all Australian jurisdictions over a two year period (2001 and 2002) found that proportionately, more Indigenous than non-Indigenous prisoners (55% and 31% respectively) were re-incarcerated within two years of release and, on average, they returned to prison more quickly, with almost one-quarter (24%) re-incarcerated within six months of initial release, compared with only 12 percent of non-Indigenous releasees. Indigenous violent offenders were also far more likely than their non-Indigenous counterparts to be re-imprisoned for a violent crime. This was particularly true of assault, with 44 percent of Indigenous prisoners re-admitted for this offence, compared with only 20 percent of non-Indigenous releasees. In contrast, the latter were more likely to have committed robbery, break and enter or theft offences (Willis & Moore 2008).

A study of Northern Territory adult prisoners released in 2001–02 also found that Indigenous offenders were three times more likely to return to prison within two years than non-Indigenous offenders (45% compared with 15% respectively). Although no Indigenous breakdowns were provided, the highest rate of return to prison was recorded by those who, at the time of release, had been serving time for assault. This group also recorded the highest rate of return for committing the same type of offence (31%; Northern Territory Office of Crime Prevention 2005).

Indigenous status as an independent predictor of the likelihood of re-offending

While Indigenous offenders seem to have higher recidivism levels (as defined by re-contact with the criminal justice system) than non-Indigenous offenders, a range of other variables (such as age, gender and prior criminal record) are also associated with higher rates of recidivism. Hence, Indigenous status itself may not be the key issue, but instead, Indigenous offenders are more likely than their non-Indigenous counterparts to possess characteristics that constitute risk factors for violence. In other words, 'Aboriginality may be a factor that catches a number of stigmatising characteristics (such as truancy, unemployment, substance abuse) and in a sense operates as a shorthand predictive model' for both re-offending and re-contact with the system (Broadhurst 1997: 417).

Some studies have attempted to explore this issue by testing whether Indigenous offenders continue to record higher re-contact levels when the effects of some other factors are being controlled. While not focused on violent offenders per se, their findings are worth noting here.

One study involved 3,352 sentenced prisoners released from Victorian prisons in 2002–03, five percent of whom were Indigenous (Hollard, Pointon & Ross 2007). Like the earlier WA study described above, it found that:

  • Indigenous prisoners returned to jail at significantly higher rates than non-Indigenous prisoners, with 50 percent being re-incarcerated within two years of release compared with 34 percent of non-Indigenous prisoners.
  • The time taken to return to prison was shorter for Indigenous prisoners, with a 60 percent higher rate of return in the first six months than would have been expected if recidivism levels had remained constant over the two year follow-up period.

To identify those factors that were potentially predictive of these re-imprisonment trends, six variables were tested:

  • age at time of release;
  • gender;
  • Indigenous status;
  • whether the prisoner was serving a sentence for property offences at the time of initial release;
  • time served; and
  • number of prior terms of imprisonment.

Results indicated that the strongest predictor of a return to prison was the number of prior imprisonment terms experienced by the individual, followed by their age at the time of release and whether or not a property offence was involved. In combination, these three variables correctly predicted 73 percent of all 'return to prison' cases. Indigenous status was not found to be a significant predictor when the effects of these other variables were controlled for. The higher recidivism rates for Indigenous prisoners could therefore be explained by the fact that this group was generally younger than their non-Indigenous counterparts (with an average age of 28.8 years compared with 32.1 years for non-Indigenous prisoners) and had a greater number of prior imprisonments (at an average of 3.4 per person compared with 2.1 per person for non-Indigenous prisoners).

Other research though has produced different results. One study measured the risk of re-offending (defined as reappearance in court for an offence allegedly committed after release) among a group of NSW prisoners granted parole in 2001–02 (Jones et al. 2006). It found that Indigenous offenders were 1.4 times more likely than non-Indigenous releasees to reappear in court, even when factors such as prior custodial episodes, prior drug convictions, age, type of parole, time spent in custody and the offence for which they were imprisoned prior to being granted parole had been partialled out. Indigenous status proved to be the third strongest predictor of the time taken to re-offend. Given the possibility that these inter-group differences could be due to differential rates of detection rather than to differential offending, the study re-analysed the data excluding those offences most susceptible to police discretion. It found that the remaining Indigenous offenders still had a higher risk of re-offending than their non-Indigenous counterparts.

Another study, which focused on youths who first appeared in the NSW Children's Court in 1995, examined re-offending over an eight year period, with re-offending again defined as a reappearance in court (Chen et al. 2005). Results indicated that among the 693 Indigenous youths in the sample, the average number of reappearances was 8.3 compared with 2.8 for the 4,783 non-Indigenous youths. Indigenous status proved to be a significant predictor of the number of reappearances, even when the effects of age at first court appearance, gender and principal offence at first appearance were controlled for. The study also found that being Indigenous increased the likelihood of appearing in an adult court once the individual turned 18 years of age, with nine in 10 Indigenous males who had appeared at least once in the Children's Court being almost certain to appear in an adult court within eight years of their first juvenile appearance. Non-Indigenous males had a lower (6 in 10) chance of a subsequent adult court appearance. The likelihood of an adult court appearance by female Indigenous offenders was also much higher than that of non-Indigenous females with eight in 10 likely to appear in an adult court compared with less than three in 10 non-Indigenous females.

It should be noted, however, that these studies were unable to control for more than a handful of variables known to be associated with re-offending. If a broader range of information on each individual had been available for testing, the role of Indigenous status in predicting re-offending may have been significantly weakened.

Victims of Indigenous violent offending

Many government inquiries into violence in Indigenous communities do not 'explore or state the race or cultural identity of the victims and perpetrators' but instead, imply that 'not only are all members of Indigenous communities Indigenous people, but that both parties to the assault are Indigenous' (Keel 2004: 6). Allied with this is an assumption that most intra-community violence is, in fact, family violence. However, very little data are available that shed any light on these issues. What little can be gleaned from the literature is summarised below.

Indigenous status of victims and offenders

Data from the NHMP indicates that, at a national level, well over nine in 10 homicides are intra-racial in nature—that is, both the victim and offender were from the same racial group. As Table 8 shows, excluding those incidents where relevant details were not recorded, over eight in 10 (n=203 or 83%) of the 245 homicides brought to police attention in 2004–05 involved a non-Indigenous person as both perpetrator and victim, while in 14 percent of cases (n=35), both were Indigenous. Only seven homicides (3%) were classified as interracial and of these, the majority (5 of the 7) involved an Indigenous offender and a non-Indigenous victim (SCRGSP 2007).

Earlier data from Western Australia indicate a similar profile for sexual assault incidents. Of those for which relevant data were available, 95 percent were intra-racial, with 10 percent involving an Indigenous person as both victim and offender. Very few were interracial (36 of 678, or 5%) and of these, the majority (n=26) involved an Indigenous perpetrator and non-Indigenous victim. A different pattern was evident, however, for assaults. As Table 8 shows, a much lower proportion (78%) were intra-racial while conversely, 22 percent were interracial. And of the 852 inter-racial assaults, over nine in 10 (n=794) comprised an Indigenous perpetrator and a non-Indigenous victim. This finding is somewhat contrary to the common perception that the overwhelming majority of Indigenous violence occurs within the group.

Interesting findings also emerge when these data are analysed from the perspective of the offender. While the overwhelming majority (88%) of the 40 homicides perpetrated by an Indigenous offender involved an Indigenous victim, this was not the case for sexual assaults and, more particularly, for assaults. Of the 94 sexual assaults perpetrated by an Indigenous offender, over one-quarter (28%) involved a non-Indigenous person as did almost one-half (47%) of the 1,680 assaults committed by Indigenous offenders. Although the WA statistics, in particular, are now somewhat dated and are hampered by the high percentage of incidents where the Indigenous status of both victim and perpetrator is not recorded (see footnote to Table 8), these findings point to a higher level of interracial offending by Indigenous perpetrators than has generally been acknowledged.

Table 8: Racial identity of victims and offenders (%)
Homicide Australia 2004–05a Assault WA 1993b Sexual assault WA 1993b
Indigenous offender, Indigenous victim 14 23 10
Non-Indigenous offender, non-Indigenous victim 83 55 85
Total intra-racial 97 78 95
Indigenous offender, non-Indigenous victim 2 20 4
Non-Indigenous offender, Indigenous victim 1 2 1
Total interracial 3 22 5
Total 100 100 100
(n=245) (n=3,905) (n=678)

Note: There were 13 homicides where the racial identity of either the victim or offender was unknown. The number of unknowns was higher for the WA data, accounting for 4,202 of the 8,107 assaults (52%) and 891 of the 1,569 sexual assaults (57%). However, the majority of these incidents (3,661 and 820 respectively) involved a non-Indigenous victim and an unknown offender. If, as the above Table suggests, non-Indigenous victims of sexual assaults are usually targeted by non-Indigenous offenders, then it could be argued that the racial profile for sexual assaults would not differ much from that outlined in the Table even if data were available for these missing cases. The situation for assaults is more ambiguous

a: SCRGSP, attachment Table 3A.10.6

b: Broadhurst, Ferrante & Harding 1995: 29

Results are less surprising when a victim perspective is adopted. In line with general expectations, the majority of Indigenous victims are attacked by other Indigenous persons. More specifically, of the 37 homicides recorded in Australia in 2004–05 that involved an Indigenous victim, 35 (95%) were committed by an Indigenous perpetrator, as were 94 percent of the 944 assaults and 87 percent of the 78 sexual assaults involving an Indigenous victim recorded in Western Australia in 1993. But while these findings are in line with expectations, the role played by non-Indigenous offenders should not be ignored, particularly in relation to sexual assaults where, according to the above data, they account for 13 percent of all sexual assaults against Indigenous victims. Anecdotal evidence supports this finding. For example, the Inquiry into child sexual assault in the Northern Territory found evidence of what it described as non-Aboriginal 'paedophiles' who infiltrated Indigenous communities to sexually abuse children. Such individuals often held positions of influence and trust within the community (Wild & Anderson 2007: 61) which meant that at times, the families of the young female victims did not try to prevent the abuse because of a fear of losing entitlements and other benefits which the perpetrator had helped them obtain (Coorey 2001: 7). The Inquiry also cited examples of more organised offending where non-Aboriginal men ran 'an elaborate scheme that involved taking young Aboriginal girls from the remote community to town and trading sex with the girls for drugs' (Wild & Anderson 2007: 64). It also identified a 'rampant informal sex trade' between Indigenous girls (some as young as 12 years old) and non-Indigenous workers from the local mining company, with the girls receiving alcohol, cash and other goods in return. Yet, because of the girls' age, this constituted unlawful sexual intercourse. Another example of transactional sex included the exploitation by taxi drivers of young Indigenous girls who either exchanged sex for free taxi rides, or were procured by the taxi drivers for non-Indigenous clients (Wild & Anderson 2007: 61–64).

While not empirically tested, the extent to which Indigenous people are the victims of non-Indigenous violence may be higher in urban areas. For example, a community survey conducted in Adelaide found that, of all Indigenous sexual assaults identified, 42 percent were perpetrated by non-Indigenous males while 41 percent were committed by Indigenous males (Carter 1987). The remaining 17 percent were pack rapes involving both Indigenous and non-Indigenous offenders acting in concert. The study also found that where the perpetrator was Indigenous, he was usually known to the victim, whereas in those instances where the perpetrator was non-Indigenous, he was more likely to be a stranger or only known to the victim by sight.

While the results from the Adelaide survey are now somewhat dated and cannot be generalised to other types of violence or to other Indigenous communities, they do suggest that the extent to which Indigenous violence involves non-Indigenous persons as either the victims or perpetrators may warrant closer investigation. At the very least, it seems important to acknowledge that some acts of violence within Indigenous communities are committed by non-Indigenous perpetrators, thereby challenging what seems to be a widespread tendency to often lay the blame for such behaviour at the feet of Indigenous people.

Relationship of offender to victim

Numerous reports contain details on the Indigenous victim's relationship to the offender. As summarised by Bryant and Willis (2008), these show that in the overwhelming majority of cases, the perpetrators of such violence are identified by the victim as family members. However, very little information is available on the converse—that is, on the Indigenous offender's relationship to the victim. In the absence of such data, it seems reasonable to assume that, if victims report that their attackers are predominantly family members, then the converse will also hold true; namely that Indigenous perpetrators of violence will predominantly target Indigenous victims.

This assumption gains some support from the NHMP data. During 2004–05, in over three-quarters (n=27 or 77%) of the 35 'Indigenous' homicides, the victim was related to the offender, either as an intimate partner or as a family member. In a further 20 percent of cases, the victim was a friend or acquaintance. Only one (3%) of the 35 'Indigenous' homicides involved a stranger (SCRGSP 2007). The results for non-Indigenous homicides were quite different. A much smaller proportion of such incidents were directed against an intimate partner or family member (40%) while a higher proportion of victims were either a friend or acquaintance of the offender (30%) or a stranger (16%). In a further 11 percent of non-Indigenous homicides, the victim's relationship to the offender was recorded as 'other'. Of the seven interracial homicides recorded in 2004–05, none involved a family member, while the majority involved a stranger (SCRGSP 2007).

That a high proportion of 'Indigenous' homicides are perpetrated against a family member is not surprising, given that all are intra-racial. What is not clear, however, is whether these patterns would hold true for other forms of violence where, at least according to the WA assault data cited previously, a much higher proportion of the victims of Indigenous violence (47%) may be non-Indigenous.

Recidivism and victim/offender characteristics: A summary

Despite the paucity of empirical data, some tentative conclusions can be drawn from the preceding discussion. In particular:

  • Indigenous violent offenders are substantially more likely than their non-Indigenous counterparts to be re-incarcerated and to be re-incarcerated for a violent offence, particularly assault.
  • The average time between initial release and re-imprisonment is also shorter for this group compared with non-Indigenous violent offenders.
  • The extent to which Indigenous status is predictive of re-offending once the effect of some other factors, such as age, gender and prior criminal record, have been taken into account is unclear. While some analyses have indicated that Indigenous status remains an independent predictor of re-offending, others have shown that this relationship disappears when other variables are factored in. This latter finding seems the most logical given that, as a group, Indigenous people are substantially disadvantaged across a range of indicators, many of which constitute risk factors for violence. If the effects of all relevant risk factors could be partialled out rather than just some of them, it seems unlikely that Indigenous status per se would continue to be predictive of violence.
  • In line with qualitative information, data from the NHMP and from Western Australia indicate that the majority of officially-reported homicides, sexual assaults and assaults are intra-racial in nature. However, this was less true for assaults than for the other two offences, with over one in five classified as interracial.
  • The common perception that most acts of violence involving an Indigenous victim take place at the hands of an Indigenous offender is also supported by the limited data available.
  • Whether the converse holds true—that is, whether Indigenous offenders predominantly target Indigenous victims—is less clear. This seems to be the case for the very serious offence of homicide. However, WA data indicate that for the less serious violent offence of assault, almost one-half of the victims were non-Indigenous. The same applied to about one-quarter of sexual offences. While this suggests a potentially higher level of involvement by Indigenous offenders in interracial violence than generally acknowledged, two points should be stressed. First, the amount of data available on this issue is extremely limited and somewhat tenuous because of the large number of incidents where relevant information on the racial identity of the victim and offender were missing. Second, because the data relate only to those incidents brought to police attention, they are influenced by the victim's willingness to report a matter, which may vary depending on the racial identity of the parties involved.
  • Over three-quarters of recorded homicides committed by Indigenous persons are directed against either partners or family members. This was markedly higher than that observed for non-Indigenous homicides, of which only about two in five involved a family member. This result is not unexpected, given that most Indigenous homicides were intra-racial and given the extended kin networks that characterise many Indigenous communities. However, whether this pattern holds true for less serious forms of Indigenous violence (particularly for assaults), where a higher proportion of victims may be non-Indigenous, cannot be ascertained.

These data raise an inevitable question—why are Indigenous recorded and self-reported crime rates so much higher than non-Indigenous rates? One reason may be the differences in age profiles. Other contributing factors may include higher levels of unemployment, more harmful use of alcohol, poorer health and living standards etc. The link between these factors and Indigenous violent offending will be explored in the third section of this report.