Australian Institute of Criminology

Skip to content

Identifying risk factors for Indigenous violence: Some conceptual issues

The ecological systems approach

It is widely recognised that there is no single cause of violence in Indigenous communities. Instead, such behaviours stem from, or are associated with, a multitude of variables operating at different levels in the environment. Memmott and National Crime Prevention (2001) divide the risk factors for violence into three categories:

  • precipitating causes—the specific event or series of events that trigger a particular incident;
  • situational factors—located at either the community, family or personal level that impact on an individual, such as unemployment and poverty; and
  • underlying factors—that constitute an historical pattern of disruption involving Indigenous systems of law, morals, authority and punishment that triggered the onset of widespread social and psychological problems which are now being passed from generation to generation.

This categorisation is a variant of an ecological systems theory for understanding crime causation. This theory, originally developed by Bronfenbrenner (1979) in his writings on child developmental psychology, has since been applied in a diverse range of fields, including criminology, where Zubrick and Robson (2003), among others, have noted its applicability to understanding Indigenous offending. This theory recognises that the risk factors for offending are located not only within the individual, but also in the wider environment within which that individual resides. This environment can be conceived as a series of concentric circles, radiating out from the individual and his/her immediate family at the centre (the microsystem), to the individual's local neighbourhood, school and work environment (the mesosystem). From there, it radiates to the broader community within which the individual's local networks sit (the exosystem) and finally, to the wider society (the macrosystem) which, through its cultural values, customs and laws, help shape the characteristics of an individual's local community. To these four levels, Bronfenbrenner (1979) added a time dimension—the chronosystem which, at a personal level, encompasses the changes arising from normal maturation processes and other life events, and at a macro-level, reflects the broad historical forces impacting on a society, such as colonisation.

Each of these levels generates different sets of risk (and protective) factors for violent offending. Those situated within or near the individual him/herself (such as mental illness, drug abuse, level of social support and family conflict) are classified as 'proximal' factors, in that they occur in close proximity to the offending behaviour itself. In contrast, broader community characteristics (such as historical events, socioeconomic inequality, poverty and unemployment) are classified as 'distal' factors and, as the term implies, operate as some distance from the offending behaviour. The probability that a specific characteristic will directly 'cause' violent behaviour diminishes as analysis moves from the proximal to the distal factors, from the micro- to the macro-system. Even among the proximal factors, many will not play a direct causative role in violence but may nevertheless indirectly increase the likelihood that such behaviour will occur.

An ecological systems approach is particularly relevant to understanding Indigenous violent offending for a number of reasons:

  • By recognising that there is no single 'cause' of violence, it redirects attention away from a search for such a 'cause' (which may ultimately be a fruitless exercise) to an exploration of the interconnections between the various risk factors and how these interactions increase the probability that violence will occur.
  • The framework's focus on 'risk' rather than 'causation' acknowledges that, while a particular characteristic may appear to be strongly linked to violence, not all individuals who exhibit that characteristic actually engage in such behaviour. For instance, while there seems to be a higher incidence of violence among those who abuse alcohol, not all persons who drink excessively commit acts of violence and not all perpetrators of violence misuse alcohol.
  • By focusing on the interconnectedness of risk factors, it also draws attention to the fact that the relationship between any two variables in this framework is not uni-directional. While community characteristics such as poor living conditions may increase the risk of violence, in turn, violent behaviour potentially reinforces those negative community characteristics.
  • It also recognises that a particular risk factor may operate at different levels within the framework. For example, a high level of substance abuse is both a community characteristic as well as a characteristic of specific individuals and so could be considered as both a distal and a proximal risk factor for violence.
  • It makes it clear that in order to reduce violence, initiatives must be targeted not just at the individual, but at the broader family, community and societal framework within which they are located.
  • It also allows for the fact that, even if a number of the identified factors pose a low risk for violence, their effect on an individual's behaviour may be cumulative and hence, significant (Zubrick & Robson 2003).

For these reasons, this report uses the ecological systems approach to explore the issue of Indigenous violence. It identifies a range of factors operating at each of the different levels in the model and, when examining a particular correlate of violence, it explores the extent to which this characteristic exists at both an individual level (ie within Indigenous offender populations) as well as in the broader Indigenous community. It also adopts the term 'risk' rather than 'causation', although in so doing it does not take the further step advocated by Zubrick and Robson (2003) of differentiating between those factors that act as predictors for the onset of offending (which they designate as 'risk' factors) and those that predict the continuation or persistence of such offending (referred to as 'prognostic' variables).

Theories of Indigenous violence

A number of general criminological theories may help to explain why particular factors, such as unemployment, poor educational standards or cultural disintegration, increase the likelihood that some individuals in some situations will become perpetrators of violence. One exposition of those theories that may have been applied to, or may have relevance for, Indigenous violence comes from Snowball and Weatherburn (2008). These include:

  • cultural theory;
  • anomie theory;
  • social disorganisation theory;
  • social deprivation theory; and
  • lifestyle/routine activity theory.

Of these, Snowball and Weatherburn (2008) found strong support for lifestyle/routine activity theory and moderate support for the other theories. The exception was cultural theory for which they found no support.

Cultural theory

Cultural theorists argue that violence, including family and inter-tribal violence, was an integral and legitimate part of traditional Aboriginal society and constituted a socially acceptable way of achieving specified goals and redressing perceived wrongs done to either the individual or the group. One recent proponent of this theory maintained that contemporary family violence in Indigenous communities has its roots in inherently violent and misogynist traditional law and practices (Nowra 2007). If this argument is correct, then violence should be higher in those communities located within traditional homelands, where Indigenous adherence to traditional law, clan obligations and ceremonies remain strong. However, there seems to be no evidence of this or of the claim that high levels of violence within Indigenous communities could be attributable to Aboriginal law (see for example Anderson & Wild 2007; Gordon, Hallahan & Henry 2002). In fact, the argument that a 'propensity for violence is a feature of Indigenous culture is rejected by most scholars' (Snowball & Weatherburn 2008: 218), a number of whom have roundly condemned such views (see Cripps 2007; Robertson & Cunneen cited in Sneddon 2007). Such views are also resented by Indigenous people themselves on the grounds that it 'reinforces prejudice and ignorance [and] masks the complex nature' of violence (Wild & Anderson 2007). In fact, the term 'bullshit law' has been used to describe those situations where Indigenous perpetrators of sexual abuse have attempted to use traditional law as justification for the assault and rape of women (Payne 1990).

Anomie theory

This theory focuses on the often brutal process of colonisation and dispossession, with its attendant destruction of Indigenous cultural values and roles. This, it argues, created feelings of alienation and anomie among Indigenous people, especially among males who, having been deprived of their status as 'law-makers' and religious leaders, no longer have a sense of purpose or identity (Langton cited in Snowball & Weatherburn 2008). For these individuals, violence may either be a means of trying to regain some level of dominance or authority within their family and community, or an expression of their sense of frustration, anger and alienation.

Social disorganisation theory

This theory is closely linked to anomie theory. It posits that 'colonisation and dispossession produced a breakdown of Indigenous informal social controls' (Snowball & Weatherburn 2008: 219). In the initial years of European expansionism, the loss of key lawmakers and community members through disease and deliberate killings meant that traditional authority structures and knowledge bases within Indigenous communities were quickly undermined. This process continued during the post-settlement phase, fostered by such policies as the deliberate disempowerment of Indigenous adults by 'white' missionary and reserve managers, and the forced removal of children of mixed parentage from their families. This meant that those surviving members who would traditionally have been responsible for inculcating and enforcing adherence to community values and behavioural norms lost the authority and capacity to do so. The social disorganisation theory accounts for the apparently higher levels of violence among children of the stolen generation and underpins the expectation of Memmott and National Crime Prevention (2001) of higher levels of violence within those communities that have a long history of functioning as removal centres or missions. Finally, it may also offer some explanation for any apparent 'normalisation' of violence within some communities, in that the absence of formal and informal social controls may potentially create a vacuum within which violence could become established as one of the new behavioural norms.

Social deprivation theory

According to this theory, the broad range of economic and social disadvantages experienced by Indigenous communities (the origins of which can be traced to the negative impact of colonisation and dispossession) generate feelings of anger, frustration and despair that, in turn, result in violence. If such disadvantage could be redressed, then differences in the level of violence between Indigenous and non-Indigenous people should largely disappear. This argument has gathered considerable support over the decades and underpinned much of the Royal Commission Into Aboriginal Deaths in Custody (RCIADIC) reasoning, where '[t]he single significant contributing factor to incarceration is the disadvantaged and unequal position of Aboriginal people in Australia in every way, whether socially, economically or culturally' (RCIADIC 1991: 15).

Lifestyle or routine activity theory

Contrary to the four theories outlined above, those who advocate this approach do not invoke the effects of colonisation and dispossession to explain the currently high levels of violence in Indigenous communities. Instead, they concentrate on factors embedded within the present lifestyle of Indigenous people and in particular, on what they view as the comparatively recent phenomenon of alcohol abuse. Pearson (2001a, 2001b), for example, argues that the high level of alcohol consumption now present in many communities is not only one of the major causes of Indigenous violence but also contributes to ongoing social and economic disadvantages. He argues that the symptom theory of substance abuse (and by extension, violence) is wrong. Instead, 'addiction is a condition in its own right' (Pearson 2001a: 4). He continues:

We must understand that trauma, dispossession et cetera make our communities susceptible to grog and drug epidemics, [but] they do not automatically cause abusive behaviour. Addiction [to alcohol and drugs] is a condition in its own right...[A]n established addiction is…independent of the historical causes of the first voluntary consumption of the addictive substance (Pearson 2001a: 4).

According to this thesis, violence cannot be reduced by focusing on the so called 'underlying causes' such as transgenerational grief, racism, dispossession and so on, but by tackling what Pearson (2001a: 4) sees as the 'core' of the problem—namely addiction and substance abuse. In his view, '[i]f we get on top of our grog and drug problem we will get on top of the worst of our violence problem' (Pearson 2001b: 1).

In summary, it is likely that all of the criminological theories described above (with the exception of the cultural theory) help to explain Indigenous violence, although the relevance of each may vary from one community to another and from one time to another. It also underlines the multi-faceted approach required to prevent and reduce violence, which research indicates addressing structural factors, as well as ensuring there are a range of specific situational and community crime prevention measures in place in individual community settings (Memmott & National Crime Prevention 2001).

Key data sources and their contribution to understanding Indigenous violence

To measure accurately the level of violence perpetrated by Indigenous individuals and to 'unpack' the complex interactions between the host of risk factors seemingly associated with such violence, two types of information are required:

  • accurate statistics on each person's actual involvement (or non-involvement) in violent offending; and
  • comprehensive details per individual on each of the potential proximate and distal risk factors thought to be associated with such violence.

The availability of such data would make it possible to identify key areas of difference between offenders and non-offenders and in turn, using multlvariate analysis, test the relative contribution of each of these differences to the likelihood that an individual will engage in illegal behaviour.

Such empirically-based information does not, however, exist. Instead, while the study was able to draw on three important data sources—namely official crime and criminal justice data, population and offender-based surveys, and government reports and academic papers—each had some limitations, as described below.

Official crime and criminal justice data

In determining the nature and extent of Indigenous violent offending, apprehensions data containing details on the number and characteristics of persons arrested or summonsed by police is most relevant. Courts and corrections data also provide information on offenders, but they exclude those individuals who, although apprehended by police, do not progress through to those stages of the criminal justice system. Hence, while they are useful in identifying how the criminal justice system responds to offenders following apprehension, they provide a less useful insight than do police statistics into those who allegedly offend in the first place. In this report, police data will therefore be used as the primary source of information, with courts and corrections data cited only occasionally.

In using police statistics, however, their limitations should be acknowledged:

  • Apprehensions data do not encompass all offenders but only those who are formally charged or summonsed. A high proportion of incidents are never reported to police or, if reported, never result in the apprehension of a suspect. This is particularly true in the case of sexual assaults.
  • As noted earlier, apprehension statistics relate only to those forms of violent behaviour that are legislatively defined as criminal, such as homicide, physical and sexual assault. Individuals who commit acts of emotional, psychological or economic abuse are less likely to be charged with committing illegal acts and so often go undetected and unrecorded.
  • Even in the case of those acts of violence defined as 'criminal', it is often not possible to extract data on specific offence types, such as family violence or assaults against children or the elderly because of the way in which such data are entered onto the systems.
  • Nor is it possible in most instances to extract data on the dynamics of, or circumstances surrounding, the incident itself to determine whether, for example, the perpetrator acted alone or as part of a group, or whether the violence was part of a sequence of criminal events.
  • Apprehension data are also of little use in assessing potential risk factors for violence. Apart from basic demographic descriptors such as age and sex, most contain little or no information on the personal characteristics of the offender. Nor do they contain any information on the characteristics of the family or community within which the offender resides.
  • Another important limitation for risk assessment is the fact that official crime statistics tell us nothing about those Indigenous persons who do not offend, which is fundamental to identifying risk factors for violence. Instead, they only allow comparisons between Indigenous offenders and non-Indigenous offenders.
  • Determining the Indigenous status of offenders is also problematic. While most police apprehension data across Australia include some indication of Indigenous status, there is no standard procedure currently in place for obtaining such information.
  • Inconsistency in counting rules and offence definitions from one state to another makes it impossible to profile Indigenous offending at a national level.
  • Very little regional or small area data on Indigenous offenders are publicly released by police agencies. The only exception is some Western Australian statistics on violence apprehension rates for Indigenous and non-Indigenous persons across seven of that state's regions for the year 2001 (Fernandez 2003).

There are several national databases that provide some useful statistics on perpetrators of specific types of violence. One of these is the AIC's NHMP, which is sourced from police records and supplemented by information from investigating officers, media reports etc. Another minor source of information is the reports from coronial inquests but, being predominantly case-based, they provide mainly qualitative details on individual offenders. Wherever possible, information from these alternative administrative databases has been included in this report.

Self-report surveys

Population surveys

Broad-based population surveys provide the most appropriate vehicle for collecting the information required to accurately assess both the prevalence and risk factors for Indigenous violence. Not only do they have the potential to ask respondents about their actual criminal behaviour, but they can also collect wide-ranging details on each person and his/her immediate environment.

Two population surveys conducted in Australia that specifically targeted Aboriginal and Torres Strait Islander communities are the National Aboriginal and Torres Strait Islander Survey (NATSIS) of 1994 and its successor, the NATSISS—both administered by the ABS. On the positive side, these surveys collected respondent-specific data on a broad range of socioeconomic, health, welfare, housing and other characteristics that constitute potential risk factors for offending behaviour. They also included questions about whether or not the person had been a victim of physical or threatened violence in the previous 12 months and, more importantly for this exercise, whether they had ever been formally charged and/or arrested or imprisoned during the previous five years. On the downside, however, these surveys did not ask respondents about their actual offending behaviour. Consequently, while they can provide useful insights into differences in characteristics between arrested and non-arrested individuals, it cannot be assumed that such differences also exist, or exist to the same degree, between those who do or do not offend simply because decisions taken by criminal justice agents are influenced by certain factors extraneous to those associated with the actual offending itself. Nevertheless, there is likely to be some overlap between the two and, in the absence of more accurate data, the NATSIS and NATSISS constitute the primary source of data for assessing risk.

The NATSIS and NATSISS have other limitations:

  • Particularly problematic for this current exercise is the fact that the amount of information collected on violence per se was limited. While the NATSIS did record some data on individuals' charge and arrest histories for assaultive behaviour, the NATSISS only asked about charges/arrests for all offences combined, without any reference to violence-specific incidents.
  • Because they are resource-intensive and logistically difficult to administer, especially in remote Indigenous communities, they are not conducted on a sufficiently regular basis to provide adequate time series data that could identify longitudinal changes in the patterns and risk factors for violence. Hence, unlike official crime statistics, which entail continuous data collection, they merely provide a 'snapshot' at a particular point in time.
  • Their resource intensiveness limits the size of the respondent sample which in turn, restricts the extent to which the data can be spatially disaggregated. While the two Indigenous population surveys provide some reliable statistics at a state level, no sub-state or regional analysis is possible.
  • Both surveys targeted respondents living within households, thereby excluding marginalised individuals, such as the homeless and those in institutions, whose risk of offending may be relatively high.
  • The methodologies used, the populations targeted and the questions asked varied not only from one survey to the other but also from one region to another within the same survey, making comparisons between the surveys difficult. The minimum age for inclusion in the surveys also varied, with the NATSIS targeting respondents aged 13 years and over, while the NATSISS focused on those aged 15 years and over.
  • As with all self-report surveys, they rely on the willingness or ability of the respondent to provide honest and accurate answers, which when it comes to sensitive issues such as offending may not always be provided. The fact that some respondents were apparently questioned in the presence of other family members may also increase the risk of inaccurate answers.

Other Indigenous-specific population surveys have also been conducted, including the National Aboriginal and Torres Strait Islander Health Survey and the Western Australian Aboriginal Child Health Survey (NATSIHS). However, neither of these sought information on the respondent's offending behaviour. Thus, although they provide some useful background information on Indigenous adults and children, including the types of stressors to which they are exposed, they are unable to shed any light on Indigenous perpetrators of violence.

Offender-based surveys

Two AIC survey-based monitoring programs targeted at specific groups of offenders proved relevant for this report:

  • Drug Use Monitoring in Australia (DUMA) involves ongoing quarterly interviews with arrestees in selected police stations/watch houses in Queensland, Western Australia, South Australia, New South Wales and more recently, Victoria and the Northern Territory.
  • Drug Use Careers of Offenders (DUCO), a one-off survey of adult male and female prisoners and juvenile detainees in Australian states and territories.

In terms of their ability to contribute to an understanding of the nature and extent of Indigenous violent offending, DUCO proved to be particularly important for two reasons:

  • It provided some data on the self-reported lifetime and 'regular' offending behaviour of detainees, which potentially includes all offences perpetrated by these individuals, even if that offending never resulted in formal contact with the criminal justice system.
  • It asked respondents about their previous involvement in particular types of violence—namely, physical assaults, sex offences and the act of 'killing someone'.

In contrast, DUMA is of more limited use because it only asks respondents whether they had previously been charged/arrested by police and it restricted its questions to 'all' offending, without reference to violence.

Both surveys provided comprehensive insights into the alcohol and illicit drug use patterns of Indigenous offenders, which had not previously been available. However, their ability to contribute to a broader empirical understanding of risk factors for Indigenous violent offending was hampered not only by the small amount of 'personal' information collected, but also by the fact that they did not record data on Indigenous non-offenders. Instead, like official crime statistics, they only enable comparisons to be drawn between Indigenous offenders and non-Indigenous offenders which, for reasons already outlined, cannot provide an accurate insight into the risk factors for Indigenous violence.

Research-based information

A considerable amount of anecdotal information about Indigenous communities may be derived from academic research reports, most of which date from the late 1980s and early 1990s onwards. The authors of these reports come from a range of disciplines, including health, anthropology, criminology, psychology and from the legal and judicial fields (see for example Atkinson 1994, 1991, 1990a, 1990b; Blagg 2000, 1999; Brady 1990; Collmann 1988; Hunter 1991a, 1991b, 1990). These often entail in-depth observations of specific communities and, as such, provide important qualitative details about the community context for violence. However, many focus on more remote and semi-remote communities, with relatively few concentrating on urban dwellers (exceptions include Gale 1972; Gale & Wundersitz 1982).

In addition to these academic publications, there is now a plethora of highly influential reports funded by, and/or arising from, various Commonwealth and state inquiries into Indigenous violence and associated matters.

At a national level, these include:

  • RCIADIC;
  • the National Inquiry into the Separation of Aboriginal and Torres Strait Islander Children from their Families (HREOC 1997);
  • an overview of Indigenous family violence by Memmott and National Crime Prevention (2001) commissioned by the Australian Government Attorney-General's Department (AGD); and
  • the report by Al-Yaman, Van Deland and Wallis (2006) on family violence among Aboriginal and Torres Strait Islander peoples, undertaken by the Australian Institute of Health and Welfare (AIHW) as part of the work of the National Advisory Group on Aboriginal and Torres Strait Islander Health Information and Data, funded by the Australian Health Ministers Advisory Council.

Reports largely funded by state governments include:

  • Queensland's Aboriginal and Torres Strait Islander Women's Task Force on Violence Report (Aboriginal and Torres Strait Islander Women's Task Force on Violence & Department of Aboriginal and Torres Strait Islander Policy and Development 2000);
  • the Cape York Justice Study (Fitzgerald 2001) which aimed to identify the nature and causes of offending in this region's Indigenous communities and to examine the relationship between alcohol, substance abuse and offending;
  • the Western Australian Government's Inquiry into the Reponses of Government Agencies to Complaints of Family Violence and Child Abuse in Aboriginal Communities (Gordon, Hallahan & Henry 2002);
  • Victoria's Indigenous Family Violence Taskforce report (2003);
  • the New South Wales Aboriginal Child Sexual Assault Taskforce report (Ella-Duncan et al. 2006);
  • the report by the Northern Territory Government (Wild & Anderson 2007), known as the 'Little Children are Sacred' report—which helped initiate the Australian Government's NTER; and
  • the Inquiry into Child Sexual Abuse in the Anangu Pitjantjatjara Yankunytjatjara (APY) Lands in South Australia (Mullighan 2008).

These government reports contain a wealth of information on Indigenous violence, gleaned from existing literature and from evidence provided by individual witnesses, public consultations and site visits. They also include broad-ranging discussions about the 'causes' of Indigenous violence but in the main, the material they present tends to be highly descriptive and/or heavily reliant on anecdotal evidence or case studies that do not allow for generalisations. They contain relatively little (if any) statistical data on the nature or extent of Indigenous violence from a perpetrator perspective (although many do include data on Indigenous victimisations). Nor do they employ empirical methods to test the predictive power of any of the variables cited as potential risk factors for violence.

For a detailed discussion of data limitations, see Appendix A.