Trends in violent crime


The public's perception is that violence is increasing, but trends in violent crime reported to police since the early 1990s reveal a mixed story. Homicide has decreased by nine percent since 1990 and armed robbery by one-third since 2001, but recorded assaults and sexual assaults have both increased steadily in the past 10 years by over 40 percent and 20 percent respectively. The rate of aggravated assault appears to have contributed to the marked rise in recorded assault, and for both assault and sexual assault the rate of increase was greater for children aged under 15 years, with increases almost double that of the older age group. Neither population changes among young adult males nor rates of offending seem to explain the trends in recorded violent crime, and indicators of change in reporting to police provide only a partial explanation. Based on self-reported victimisation and reporting to police, it would seem increased reporting of assault is somewhat responsible for the rise in recorded assault rates against adult victims. However, victimisation survey data suggest there has been little change in rates of sexual assault, although reporting to police by women seems to have increased. Victimisation survey data also do not illuminate the most significant recorded increase in violent victimisation, against children, as they are collected less frequently and only apply to those aged at least over 15 years. The paper speculates that the rise could be due to better public understanding of child protection issues and increased reporting due to public awareness of what constitutes physical and sexual assault - especially within the family - but this requires further investigation to examine how many recorded violent crimes against children relate to current and/or past events and of the relationship to the offender.

Violent crime, with the intention of causing (or threatening) physical harm or death to the victim, attracts more attention and debate than other forms of crime. Sustained media attention combined with high-profile incidents - such as the shootings at Port Arthur (Tasmania), and Monash University and Flinders Lane in Melbourne; gang rapes in Sydney; and organised crime-related murders - have prompted a view among the Australian public that violent crime is increasing in Australia. Over two-thirds of Australians (70%) interviewed in the 2003 Australian Survey of Social Attitudes believed crime had increased since 2001; 39 percent of respondents thought it had 'increased a lot' (Indermaur & Roberts 2005). However, commentators on violent crime are more cautious in their interpretation of the trends. Earlier research that examined changes in violent crime detected sustained increases in recorded rates of assault, sexual assault (rape) and robbery that began or intensified in the 1990s (Carcach 2005; Indermaur 1996, 2000; Ross & Polk 2005). While rates of recorded assault and sexual assault continued to rise into the early 2000s, rates of robbery began to decline. In contrast, the homicide rate has remained relatively stable since it peaked in the 1970s. Any year-to-year fluctuations observed in homicide rates is believed to be more a function of the small number of homicides that occur in Australia each year than any real changes in incidence (Mouzos 2000).

Homicide is often used as a gauge of the level of violence occurring in society, so the different patterns observed for other types of criminal violence have necessitated their closer scrutiny to help explain the disparity. Some of the increase in recorded violent crime has been attributed to increased reporting to police. However, Ross and Polk (2005) argued that changes in rates of homicide, assault and robbery more likely reflect real changes in the occurrence of these crimes than in a greater propensity to report. Lack of correlation between patterns in recorded crime data with that drawn from victimisation surveys suggests that such an assumption cannot be made so readily (Carcach 2005).

Given the divergent views it seems that explaining trends in violent crime is no easy undertaking. This paper aims to contribute to the discussion by ascertaining if trends in rates of homicide, assault, sexual assault and robbery identified in earlier papers have continued from the early 1990s into the following decade. Furthermore, it will examine whether discrepancies between recorded crime and victimisation survey data still exist, and explore possible reasons for discrepancies.

Measures of violent crime

Violent crime is generally defined as including the offence categories of homicide, assault, sexual assault and robbery (both armed and unarmed).

As the most serious form of violence, and the least open to interpretation, homicide is considered the most reliable indicator of violent crime. Data on homicide are considered the most accurately and consistently collected of all violent crime data (Davies & Mouzos 2007; Indermaur 2000).

Data on other categories of violent crime are arguably less reliable for a number of reasons. First, some crimes are not reported to police, and violent crimes are particularly affected by a general reluctance to report. Estimates from Australian Bureau of Statistics victimisation surveys (ABS 2006a, 2006b) have suggested that as many as two-thirds of such crimes are not reported. Of the alleged crimes that are reported, some may end up not being recorded by the police (Carcach & Makkai 2002). Second, changes in police policy and attitudes, and improvements in recording practices and databases used to store incident data, suggest that recent recorded violent crime data are more rigorous than earlier data. Another variable is the effect of different jurisdictional definitions of categories of violent crime, particularly with regard to what offences constitute crimes such as assault and sexual assault.

Victimisation surveys are another source of trend data used to interpret patterns in crime. However, self-reported victimisation data have their own inherent problems. These are largely related to issues of recall and, especially for sensitive topics such as sexual assault, an unwillingness to relate an experience of violence. With the use of appropriate methodological approaches, victimisation surveys are considered to better estimate the extent of violent crime occurring in the population, as they capture not only those people who have reported an incident of crime, but also a proportion of those who did not.

Comparing results from different data sources

Caution should be used when interpreting the results from various data sources. Actual rates calculated from recorded crime and victimisation survey data will not necessarily be the same because of the different ways the base data are derived. Furthermore, rates estimated from different victimisation survey data may not tally, as methodological approaches can elicit differential patterns of response. For example, it is feasible that people interviewed in the ABS Women's Safety Survey (WSS)/Personal Safety Survey (PSS) series would be more inclined to report their experience of violent crime than those participating in the ABS Crime and Safety Survey (CSS). This is because respondents to the WSS/PSS have been recruited for a study focusing specifically on experience of physical and sexual violence, whereas the CSS is a supplement to a monthly Labour Force Survey, and covers both household and personal crime. Respondents to the latter survey may be less forthcoming about their experience of sensitive crimes such as sexual assault.

Another consideration is the age range included in the survey. Recorded crime data cover all ages, whereas the CSS comprises people aged 15 years and over, and the WSS/PSS and the International Crime Victims Survey (ICVS: Australian component) includes people aged 18 years and over. Comparing trends between recorded crime data and victimisation survey data could be misleading, particularly if a change in the incidence of violent crime is more apparent within an age group not included in either data source.


Homicide has followed a cyclical pattern since the start of the 20th century, characterised by a trough coinciding with World War II (1939-45), followed by a steady increase to peak at a rate of around 2.0 per 100,000 in the 1970s and 1980s. As homicide offenders are almost always male, historical and social changes affecting the male population in Australia have been used to explain the more marked changes in homicide trends (James & Carcach 1997: 2).

Homicide rates since the early 1990s have fluctuated slightly from year to year (influenced by the relatively small number of victims), but the overall trend has been downward (Figure 1). Since 1989-90, there has been a significant downward trend, with the number of homicide victims declining by nine percent from 330 to 301 (Kendall's tau = -0.43; p<0.05).


Figure 1: Homicide victims, 1989-90 to 2005-06 [see attached PDF for graph]


Assault is the most common form of violent crime; rates of recorded assault have been increasing steadily over the past 10 or more years. Between 1995 and 2006, the rate of recorded assault rose significantly from 562.8 to 829.4 per 100,000 people (Kendall's tau = 0.88, p<0.05), an="" increase="" of="" 47="" percent="">Figure 2). Other research suggests this increase is not a recent phenomenon, but started to rise in the 1970s (Chappell 1995).


Figure 2: Rates of recorded assault, 1995-2006 [see attached PDF for graph]


The rate of recorded assault has increased for males and particularly for females. The rate of increase was also greater for people aged 0-14 years (37% between 1996 and 2003) than for those aged 15 years and over (27%) (Figure 2).

It is the rate of aggravated rather than non-aggravated assault that appears to have contributed to the marked rise in assault rates. For the years where data are available (1999 to 2006), rates of aggravated assault increased by 46 percent (41% for males and 61% for females). In contrast, non-aggravated assault rates rose by only seven percent (3% for males and 19% for females).

Results from victimisation surveys produce a less conclusive picture. Overall rates of assault did not increase significantly over the four survey years of the CSS, nor was a consistent pattern observed from the ICVS (4.4% in 1992; 6.4% in 2000 and 4.7% in 2004). When considering gender, the CSS suggests a an increase in the experience of assault among men and women but neither were statistically significant.

Around one-third of people who experience an assault report the incident to police (ABS 2006a). Reporting rates for assault have increased - 31 percent in 2005, up from 28 percent in 1998 (CSS). Reporting rates increased for males (26% to 29%) and females (30% to 34%), and were higher for females in two of the three reference years (1998 and 2005). The PSS did not detect gender-based differences in reporting rates - 33 percent for both males and females.

Sexual assault

The prevalence of sexual assault is also reported to be increasing. Since 1995, the rate of recorded sexual assault increased by 22 percent, from 72.5 per 100,000 people in 1995 to 88.4 in 2006 (Kendall's tau = 0.67, p<0.05)>Figure 3).


Figure 3: Recorded rates of sexual assault, 1993-2006 [see attached PDF for graph]

A more equivocal pattern emerges from victimisation surveys. The victimisation rates from the CSS for sexual assault (all persons) in 2002 and 2005 were comparable, at 0.2 and 0.3 percent respectively. Rates for females remained steady between 1998 and 2002 at 0.4 percent. A small but not significant decline was observed in the WSS/PSS series, from 1.9 percent of women in 1996 to 1.6 percent in 2005. Less than one percent of men (0.6%) in 2005 reported a recent sexual assault.

An increase in recorded sexual assault among younger Australians is thought to have contributed to the rise in sexual assault. In the 10-year period between 1995 and 2005, the incidence of recorded sexual assault for children aged 0-14 years accounted for around 40 percent of all recorded sexual assaults. Since the early 1990s, the increase in rates of sexual assault for this age group was the same as for people aged 15+ years (23% compared with 24%). However, scrutiny of the period where much of the increase occurred (between 1999 and 2003) shows the increase among the 0 to 14-year-old age group was more than double that of people aged 15+ years (37% compared with 17%). Rates of sexual assault increased for both males and females aged 0-14 years but more so among females (27% increase for females aged 0-14 years between 1996 and 2003, compared with 19% for males of the same age).

If we contrast recorded sexual assault rates for females aged 15+ years with data from the CSS, the initial discrepancy between the two data sources starts to fade. While an increase in rates of recorded sexual assault for the 15+ age group still exists for the period covered by the victimisation surveys, it is considerably less pronounced than when using total population rates.

Most victims of sexual assault are female and few report the assault to police. According to Farrington, Langan and Tonry (2004), victimisation survey data from the late 1990s suggest the propensity for females to report sexual assault is increasing. Certainly, data from the 1996 WSS and 2005 PSS confirmed this observation - 19 percent of women in 2005 said they had reported the most recent incident of sexual assault compared with 15 percent in 1996. An increase was also observed in the percentage of women who reported any experience of sexual assault that occurred since the age of 15 (16% in 2005 up from 9% in 1996). In contrast, the CSS finds the rate to have decreased, from 33 percent in 1998 to 20 percent in 2002. However, without data broken down by sex from the 2005 CSS it is not possible to determine whether this decrease has been sustained.


Robbery is classified as a violent crime, as the loss of property is usually accompanied by use or threat of violence against the victim. For example, in 2005 over one-third of people who had experienced a robbery in the previous 12 months had suffered physical injury as a result of the robbery (ABS 2006a). However, most robberies are committed without weapons. The percentage of armed robberies has fluctuated over the past 10 years but generally accounted for 40 percent of all robberies in any given year.

Recorded rates of armed robbery began to increase in the early 1990s, peaking twice in 1998 and again in 2001 (Figure 4). The second peak was followed by a marked decline to a rate of 36.5 per 100,000 in 2006. Unarmed robbery followed a similar pattern, although characterised by just one peak before dropping to 47.4 per 100,000 in 2006.


Figure 4: Rates of robbery, 1995-2006 [see attached PDF for graph]

Data from the CSS generally support the pattern observed in recorded rates of robbery. Robbery victimisation rates in 1998 and 2002 were similar, at 0.5 percent and 0.6 percent respectively, before dropping in 2005 to 0.4 percent. ICVS robbery rates declined but not significantly (p>0.05). In 1992, 1.3 percent of people reported being a victim of robbery (1.2% in 2000 and 0.8% in 2004).

People tend to report crimes like robbery more readily than they do assault or sexual assault.

According to the CSS, around half of all robbery victims in 1998 and 2002 reported the robbery to police, but by 2005 only 40 percent did so. Future data will help determine whether this decrease in reporting is specific to the 2005 survey or reflective of a general trend since 2002.

Interpreting the trends

Recent commentators examining trends in violent crime have emphasised the difficulty in providing a concise explanation (Carcach 2005; Indermaur 1996, 2000; Ross & Polk 2005), particularly for:

  • the different pattern in rates of homicide compared with those of other violent crimes
  • the differential correlation between trends in violent crime derived from recorded crime data, victimisation survey data and reporting data.

Some observers have found a strong or reasonable correlation between trends (Ross & Polk 2005), whereas others contend there is little or no association (Carcach 2005). What is agreed is that increases detected in rates of violent crime are not necessarily an indicator of increasing violence in Australia.

Australians' experience of robbery, as reported in victimisation surveys, generally support the trend observed in recorded robbery data. Until the early 2000s, the propensity for victims to report robberies remained relatively static, suggesting that the increase in rates of robbery reflected a real rise in victimisation.

Robberies are often committed to help finance drug purchases (Chilvers & Weatherburn 2004) and consequently, shifting patterns in drug availability and cost influence the prevalence of robberies. Changes in heroin supply are thought to have contributed to the rapid rise and then decline in robbery rates in New South Wales between 2000 and 2002 (Chilvers & Weatherburn 2004), but the absence of relevant data from other jurisdictions does not help observers assess if similar scenarios occurred elsewhere. Future data will ascertain whether the nationwide decrease in rates of recorded robbery represents a genuine decline in robbery victimisation or is a consequence of decreasing reporting rates.

Recorded rates of both assault and sexual assault have followed a sustained upward trend since the early 1990s. A simultaneous increase in the reporting of assault suggests this is somewhat responsible for the rise in assault rates. The relationship between rates of recorded sexual assault against those estimated from victimisation surveys is less conclusive, as victimisation surveys produced inconsistent patterns in reporting behaviour. An increased awareness of what constitutes physical and sexual assault (particularly for assaults occurring within the family), a diminishing of associated taboos, a tendency for delayed reporting, and improved police and judicial responses to reports of assault all represent factors likely to have influenced willingness to report (Borzycki 2007; Cook, David & Grant 2001; Lievore 2003; Taylor & Mouzos 2006).

The large increase in recorded assault and sexual assault among males and females aged 0-14 years could be correlated with better public understanding of child protection issues. A coinciding increase in child protection notifications and substantiations has been attributed in part to increased public awareness and hence reporting of child abuse cases (AIHW 2006). Family violence, in particular domestic violence, has also received greater attention recently from political, justice and media spheres. This is likely to have influenced some reporting behaviour among women. Most assaults against women are perpetrated by a partner or family member; almost half (46%) of women physically assaulted since the age of 15 were assaulted by a current or ex-partner and 37 percent by a family member (ABS 2006b).

Some association was found between recorded rates of assault and rates drawn from victimisation surveys, but it was mostly inconsistent for sexual assault data, with victimisation survey data contradicting the increase observed using recorded sexual assault data. Examination of rates of recorded sexual assault data for females aged 15+ years revealed a considerably less graded trend line, suggesting that much of the increase occurred in the 0 to 14-year-old age group.

Age composition effects on crime have been used to explain in part changes in crime rates, specifically relating to a decrease in the proportion of young males (the primary offending age group) with a decrease in crime in the United States during the mid-1980s to late 1990s (Steffensmeier & Harer 1987; Steffensmeier & Harer 1999). However, similar associations have not been observed in other developed countries (Aebi 2004; Gartner & Parker 1990) or for specific crimes such as homicide (Gartner 1990), and this also appears to be the case for Australia. Neither population changes among males aged 15-19 years nor corresponding changes in their rates of offending (per 100,000 population; based on police statistics from Victoria, Queensland and South Australia) - which have both declined over the observation period - exhibit a significant relationship with trends in homicide, robbery or sexual assault (p>0.05). The relationship with assault is ambiguous; while significant, it is small and in the opposite direction to that expected (that is, a decline in the rate of offending among males aged 15-19 was associated with an increase in assault rates) (regression coefficient = -0.03, p<0.05, r2="0.44).">

Another factor for consideration is the change in weapon use. Since 1989–90, the proportion of homicides committed with a firearm has declined, while the proportion committed with a knife or a blunt instrument has increased. However, only the change in firearm-based homicides is significant (Kendall’s tau = -0.62, p<0.05). The trend in firearm- based homicide correlates positively with overall homicide rates (Kendall’s tau = 0.51, p<0.05), which suggests that differential firearm use is associated with, but not necessarily impacting on, patterns of homicide occurring in Australia over the past 20 years.


Crime data can only represent what people are willing to report, either in a formal sense (to police) or informally (as captured in crime victimisation surveys). With increased community awareness and understanding of violent crime, changes in the way the justice system manages violent offences, and a greater commitment on the part of victims to report experience of violence, the capacity for different sources of crime data to reflect real and consistent changes in specific crimes is strengthened. Nonetheless, some inconsistency still exists between the two primary sources of violent crime data. This, combined with the absence of recent national data on crimes such as assault and sexual assault, has made it difficult to determine whether there has been an increase in some forms of violent crime over the past 10-15 years or if experience of these violent crimes is just more likely to be reported now than it was in the past. Greater consistency between data sources in their definition of specific crimes is one suggested approach to smoothing out these differences (Carcach 2005: 323).

If homicide is the yardstick by which the level of violence in society is measured, then the belief that violence is increasing in Australia cannot be substantiated. The significant increase in recorded assault and sexual assault potentially contradicts this view, but without supporting evidence from other sources of information, such an interpretation can only remain provisional. Further research will need to examine how much of the increase in recorded violent crime against children relates to current or past events. Further refinement of crime data sources to improve their comparability, and longer-term assessment of crime patterns, may also offset future problems in interpreting and understanding trends.


URLs were correct at 11 May 2008

  • Aebi MF 2004. Crime trends in Western Europe from 1990 to 2000. European journal on criminal policy and research 10(2-3): 163-186
  • Australian Bureau of Statistics (ABS) 1994-96. National crime statistics. ABS cat. no. 4510.0. Canberra: ABS
  • Australian Bureau of Statistics (ABS) 2006a. Crime and safety, Australia, April 2005. ABS cat. no. 4509.0. Canberra: ABS
  • Australian Bureau of Statistics (ABS) 2006b. Personal safety survey, Australia. ABS cat. no. 4906.0. Canberra: ABS
  • Australian Bureau of Statistics (ABS) 1997-2007. Recorded crime: victims, Australia. ABS cat. no. 4510.0. Canberra: ABS
  • Australian Institute of Health and Welfare (AIHW) 2006. Child protection Australia 2005-06. AIHW cat. no. CWS 40. Canberra: AIHW
  • Borzycki M 2007. Pilot study on sexual assault and related offences in the ACT: stage 3. Research and public policy series no. 79. Canberra: Australian Institute of Criminology
  • Carcach C 2005. Crime and punishment in Australia, 1980-2000, in Tonry M & Farrington DP (eds), Crime and punishment in Western countries, 1980-1999. Chicago: The University of Chicago Press: 295-330
  • Carcach C & Makkai T 2002. Review of Victoria Police crime statistics. Research and public policy series no. 45. Canberra: Australian Institute of Criminology
  • Chappell D 1995. How violent is Australian society? in Chappell D & Egger SJ (eds), Australian violence: contemporary perspectives II. Canberra: Australian Institute of Criminology
  • Chilvers M & Weatherburn D 2004. The impact of heroin dependence on long-term robbery trends. Contemporary issues in crime and justice no. 79. Sydney: Bureau of Crime Statistics and Research
  • Cook B, David F & Grant A 2001. Sexual violence in Australia. Research and public policy series no. 36. Canberra: Australian Institute of Criminology
  • Davies M & Mouzos J 2007. Homicide in Australia: 2004-05 National Homicide Monitoring Program (NHMP) annual reportResearch and public policy series no. 72. Canberra: Australian Institute of Criminology
  • Farrington DP, Langan PA & Tonry M (eds) 2004. Cross-national studies in crime and justice. NCJ 200988. Washington, DC: Bureau of Justice Statistics
  • Gartner R 1990. The victims of homicide: a temporal and cross-national comparison. American sociological review 53: 92-106
  • Gartner R & Parker RN 1990. Cross-national evidence on homicide and the age structure of the population. Social forces 69(2): 351-371
  • Indermaur D 1996. Violent crime in Australia: interpreting the trends. Trends & issues in crime and criminal justice no. 61
  • Indermaur D 2000. Violent crime in Australia: patterns and politics. The Australian and New Zealand journal of criminology 33(3): 287-299
  • Indermaur D & Roberts L 2005. Perceptions of crime and justice, in Wilson S, Meagher G, Gibson R, Denemark D & Western MC (eds), Australian social attitudes: the first report. Sydney: UNSW Press: 141-160
  • James M & Carcach C 1997. Homicide in Australia 1989-96. Research and public policy series no. 13. Canberra: Australian Institute of Criminology
  • Lievore D 2003. Non-reporting and hidden recording of sexual assault: an international literature review. Canberra: Australian Institute of Criminology
  • Mouzos J 2000. Homicidal encounters: a study of homicide in Australia 1989-1999. Research and public policy series no. 28. Canberra: Australian Institute of Criminology
  • Ross S & Polk K 2005. Violent crime in Australia, in Chappell D & Wilson P (eds), Issues in Australian crime and criminal justice. Sydney: LexisNexis Butterworths: 36-51
  • Steffensmeier DJ & Harer MD 1987. Is the crime rate really falling? An 'aging' U.S. population and its impact on the nation's crime rate, 1980-84. Journal of research in crime and delinquency 24(1): 23-48
  • Steffensmeier DJ & Harer MD 1999. Making sense of recent US crime trends, 1980 to 1996/1998: age composition effects and other explanations. Journal of research in crime and delinquency 36(3): 235-274
  • Taylor N & Mouzos J 2006. Two steps forward, one step back: community attitudes to violence against women survey. Melbourne: VicHealth

About the authors

Dr Samantha Bricknell is a research analyst in the Crime Monitoring Program at the Australian Institute of Criminology