Home → Publications → Reports → Research and public policy series → 64 → 2. Rates of victimisation (in: Crime victimisation in Australia : key results of the 2004 International Crime Victimisation Survey)
Crime victimisation in Australia : key results of the 2004 International Crime Victimisation Survey
Holly Johnson
ISBN 0 642 53881 6 ; ISSN 1326-6004
Canberra: Australian Institute of Criminology: 2005
(Research and public policy series, no. 64)
Download
- PDF print version : chapter 2 (PDF 528kB)
- PDF print version : full report (PDF 1.15MB)
2. Rates of victimisation
Victimisation surveys provide police, policy-makers and researchers with an important source of information for addressing a range of policy issues. This chapter will explore the level of victimisation in Australia as reported to the 2004 ICVS, how victimisation rates have changed since the previous cycle of the ICVS in 2000, correlates of personal and household crime, levels of repeat victimisation, and a detailed look at assaults and threats.
What is the level of victimisation?
The ICVS asked respondents about their experiences of select types of crimes in the preceding five years (back to 1999), and from this, five-year rates and one-year rates for 2003 were calculated. Three personal and six household crimes were included in a standardised questionnaire (see text box Measuring crime victimisation):
- Personal crime
- Assaults and threats
- Robbery (theft of personal property with violence or the threat of violence)
- Personal theft (theft of personal property without violence)
- Household crime
- Burglary
- Attempted burglary
- Motor vehicle theft
- Theft from motor vehicles
- Motorcycle theft
- Bicycle theft
Measuring crime victimisation
Victimisation surveys use highly structured questionnaires. To ensure that estimates of crime victimisation are accurate, it is important that question wording is easily understood, accurately represents the concept it is intended to measure (has a high degree of validity), and is understood by all respondents in the same way (has a high level of reliability). The questionnaire developed for the ICVS was carefully tested and has been used since the first cycle in 1989 (see Appendix for detailed discussion about reliability of the estimates). The following questions were used to measure experiences of crime victimisation (in the order they appear in the questionnaire):
HOUSEHOLD CRIME
Car theft
Over the past five years have you or other members of your household had any of their cars/vans/trucks stolen?
Theft from cars
Apart from this, over the past five years have you or members of your household had something stolen from your car, for example a car stereo, or something that was left in your car? This includes theft of a part of the car, such as a car mirror or wheel. (Excludes thefts from car when car was stolen, and vandalism.)
Motorcycle theft
Over the past five years have you or other members of your household had any of their mopeds / scooters / motorcycles stolen?
Bicycle theft
Over the past five years have you or other members of your household had any of their bicycles stolen?
Burglary
Over the past five years, did anyone actually get into your home without permission, and steal or try to steal something? (Excludes thefts from garages, sheds or lock-ups.)
Attempted burglary
Apart from this, over the past five years, do you have any evidence that someone tried to get into your home unsuccessfully (for example, damage to locks, doors or windows or scratches around the lock)?
PERSONAL CRIME
Robbery
Over the past five years has anyone stolen something from you by using force or threatening you, or did anybody TRY to steal something from you by using force or threatening force?
Theft of personal property
Apart from theft involving force, there are many other types of theft of personal property, such as pick-pocketing or theft of a purse, wallet, clothing, jewellery or sports equipment. Over the past five years, have you personally been the victim of any of these thefts?
Assaults and threats
Apart from the incidents just covered, over the past five years, have you been personally attacked or threatened by someone in a way that really frightened you? Just to explain what we're including, this could have been at home or elsewhere, such as at your workplace, in the street, on public transport, in a pub, at school, or on the beach, etc. And it could have been by someone you know, a close friend, a family member or your partner.
Personal crime
Victimisation rates are calculated per 100 persons and per 100 households. These are prevalence rates which count the percentage of people or households victimised once or more. Overall, half (52%) of the sample reported at least one experience of crime over the five-year period and 17 per cent were victimised during 2003. Twenty-nine per cent had been victims of personal crimes over the five-year period and nine per cent in the previous year. Rates of personal crime were highest for assaults/threats and personal theft and lowest for robbery.
Figure 1: One-year and five-year rates of personal crime
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
Household crime
Rates were higher for household crimes than for personal crimes: 39 per cent of households reported at least one of the household crimes measured by this survey during the five-year period, and 11 per cent experienced a household crime in the 12 months prior to the survey. Theft from motor vehicles occurred most frequently, followed by burglary and attempts. Theft of motor vehicles, motorcycles and bicycles were reported by less than 10 per cent of the sample during the five-year period and less than two per cent in the past year (Figure 2). These rates are based on all households. Limiting the analysis to just those households who were owners of motor vehicles, motorcycles and bicycles, five-year rates are as follows:
- motor vehicle theft: seven per cent of motor vehicle owners;
- theft from motor vehicles: 21 per cent of motor vehicle owners;
- motorcycle theft: five per cent of motorcycle owners; and
- bicycle theft: 10 per cent of bicycle owners.
Five-year rates are not simply five times higher than one-year rates, possibly due to memory loss of incidents that are relatively minor. In addition, there may have been changes to victimisation risk at the personal or household level during the five-year period.
Figure 2: One-year and five-year rates of household crime
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
Have victimisation rates changed over time?
Direct comparisons with the 2000 ICVS should be made with caution because of the methodological differences between the two surveys. For example, the approach taken in 2000 involved the 'White Pages Plus One' method of sample selection which involves selecting residential telephone numbers at random from the telephone directory and altering the last digit. Both the number selected from the directory and the altered number were then used. The 2004 survey used the Random Digit Dialling method which involves retaining the 6 digit prefix of known telephone numbers and randomly generating the last 4 digits. Both these methods are designed to increase the chance of selecting unlisted or not yet listed numbers. In addition, the maximum number of telephone calls made to make contact with a household was 6 in the 2000 survey compared to 15 in 2004. This extended call regime in 2004 was designed to enhance the representation of young people, single person households, and employed people. These differences in approach may have affected comparability in rates of victimisation between the two survey cycles, producing higher rates in the 2004 survey.
As shown in Table 2, the five-year rate of overall victimisation showed a small but statistically significant decline over the two time points, from 55 per cent of persons in 2000 to 52 per cent in 2004. However, the only crime to decline significantly was personal theft not involving burglary or violence. This is a relatively minor but high volume crime. One-year rates are more indicative of current crime conditions and the overall percentage in each sample who reported at least one victimisation in the 12-month period prior to the survey declined from 24 per cent to 17 per cent, a statistically significant drop. Crimes showing significant declines were personal theft, burglary and theft from motor vehicles.
These patterns in victimisation mirror trends in police-recorded crime in Australia. Although the crime categories are more detailed in the ICVS compared with police statistics and therefore are not strictly comparable, police statistics also show recent declines in rates of property theft and burglary (AIC 2004). Similar to the ICVS data, within crimes recorded by police, the most frequently occurring violent crime was assault while the most common property crime was property theft (which would include personal theft, theft from vehicles and bicycle theft in the ICVS) followed by burglary.
Do levels of victimisation vary?
It is a common finding in crime victimisation research that risk of victimisation is not evenly distributed in the population; people and households with certain socio-demographic characteristics report higher rates of victimisation as compared with others. One-year rates of victimisation are used in this analysis as some personal characteristics, such as age, income and others, change over time and may not reflect the person's situation five years previously.
Personal crime
As shown in Table 3, all adults in the population do not have an equal chance of experiencing assaults or threats, robbery or personal theft. In particular, individuals with the following characteristics had significantly higher rates of personal crime:
- young people have higher rates of all types of personal crime and rates decline with age;
- marital status is a significant predictor of victimisation: single people and those living in a de facto relationship report higher rates than those who are married or widowed;
- rates of personal crime are lowest for residents of lowest income households;
- rates of personal victimisation are inversely related to residential stability as measured by time living at current postcode;
- how time is spent, both during the day and in the evening affects risk of victimisation: students, the unemployed, and those who regularly go out in the evening for recreational purposes, report higher rates of personal victimisation;
- those who speak a language other than English at home report lower rates of assault or threat and higher rates of robbery; and
- Indigenous people report higher rates of personal victimisation, although differences are not statistically significant at the .05 level due to low counts of Indigenous people interviewed for this survey.
These results are consistent with the research literature which shows that age, marital status, main activity and night time activity are correlated with higher rates of personal victimisation (ABS 2003; Mihorean et al 2001). According to one theory, routine activity theory, individuals whose lifestyle brings them into close proximity to potential offenders, and situations where guardianship over personal safety is lowered, will have higher rates of personal victimisation (Miethe et al 1987). It is not difficult, for example, to see how time spent in public places by young, single people, students or the unemployed differs as compared with married people with family responsibilities or the elderly, and how this may affect risk of personal victimisation.
Household crime
A number of household-level variables are important correlates of household crime (Table 4). For example:
- rates of theft from motor vehicles is lowest for lowest income households;
- rates of all household crimes with the exception of burglary tend to be lower for those who have lived many years at their current postcode;
- rates of bicycle theft are higher for those who speak only English at home; and
- rates of household crime do not vary significantly according to type of dwelling or Indigenous status.
Does risk of victimisation vary by community type?
Recently, attention has been focused on the effect of neighbourhood composition and organisation on victimisation risk. The idea that crime is concentrated in certain geographic areas is not new. Early in the last century, criminologists established that crime occurs predominantly in areas characteristics by poverty, unemployment and single parent households. However, more recent research suggests that it is not disadvantage per se but the social organisation of disadvantaged areas that affects their vulnerability to crime (Morenoff et al 2001; Sampson et al 1997). This social organisation is referred to as 'collective efficacy' or 'community cohesion' which is characterised by mutual trust among neighbours, willingness to intervene, supervise young people and help maintain public order. In poor neighbourhoods that have low crime rates, collective efficacy has been determined to be an important factor in the ability of residents to enforce collective norms and exert social control over community members (Sampson et al 1997).
The ICVS does not question respondents in detail about the characteristics of their neighbourhoods; however a census of the population, which is conducted at five-year intervals in Australia, can be utilized together with victimisation survey data to better understand the community-level context in which crime occurs. The Australian Bureau of Statistics has developed an analytical tool which ranks geographic areas according to their relative social and economic wellbeing. The Socio-Economic Indexes for Areas (SEIFA) encompasses four indexes, each summarising a different aspect of the socio-economic conditions of an area (ABS 2001). Each index has been derived from a range of questions on the 2001 Census of Population and Housing:
- Index of Relative Socio-Economic Disadvantage (derived from variables such as low income, low educational attainment, high unemployment, jobs in relatively unskilled occupations);
- Index of Relative Socio-Economic Advantage/Disadvantage (takes into account variables relating to income, education, occupation, wealth and living conditions) is a continuum of advantage to disadvantage;
- Index of Economic Resources (variables relating to income, expenditure and assets of families, rent paid, mortgage payments, dwelling size, family structure); and
- Index of Education and Occupation (including higher qualifications and employment in skilled occupations).
The indexes were derived through a process of principal components analysis which is a method used to summarise information from a variety of variables (ABS 2001). A value is provided for each Index for a wide range of geographic areas which enables researchers to link these data to other datasets by way of the geographic identifiers. The SEIFA was linked to the 2004 ICVS by way of postcodes of participating households (postcode data were missing from 165 household or 2.4% of the sample).
These Indexes are all highly correlated with one another (between 0.82 and 0.97, p < .05) which indicates there is significant overlap in the information contained in each one. Table 5 shows the mean values of each Index for those who indicated they or their households had been victims of any crime in the previous year compared with those who were not victimised. All Indexes have been constructed so that relatively disadvantaged areas have low Index values (high values equal lack of disadvantage) and are standardised to have a mean value of 1000. As shown, persons and households who reported being the victim of any crime in the previous 12 months tended to live in postal areas with higher mean scores compared with non-victims. This indicates lower relative disadvantage in victims' postal areas compared with the postal areas of non-victims. Differences were statistically significant for all Indexes with the exception of Relative Socio-Economic Disadvantage. Differences in mean scores were non-significant for individual types of personal crimes due to small numbers reporting victimisation in the one-year period.
Combining statistical data about personal, household and neighbourhood characteristics can help broaden our understanding of the factors associated with risk of victimisation. Through this, governments and communities can better understand the social distribution of crime. This is important information that can contribute to the development of crime reduction activities and help more accurately target community and police resources. The ways in which household-level and community-level characteristics interact to affect risk of victimisation is a topic for future study using multi-level modelling techniques.
What are the most important risk factors for victimisation?
Many of the factors that are associated with higher rates of personal or household victimisation are inter-rated. For example, young people are more likely than older people to be single, to be students or looking for work, and to be active outside the home in the evenings. Using logistic regression, the most important predictors of personal and household victimisation can be identified, while holding constant the effects of the others.
For ease of interpretation, the predictors in the regression were dichotomised and the group with higher rates of victimisation in the bi-variate analysis (Tables 3 and 4) were assigned a value of 1 while those with lower rates were assigned a value of 0 as the following shows:
- Age:
- 1 = young people aged 16 to 24
- 0 = people 25 years and over
- Marital status:
- 1 = single, divorced, separated people and those living in de facto relationships
- 0 = married or widowed
- Income:
- 1 = household income less than $400 per week
- 0 = household income $400 per week or more
- Time at current postcode:
- 1 = less than one year
- 0 = one year or longer
- Main activity:
- 1 = unemployed
- 0 = employed, keeping house, student, retired, on a pension
- Evening activities
- 1 = evenings out almost daily
- 0 = evenings out once a week or less
A value of 1 was also given for those who speak a language other than English at home and for Indigenous people while the reference categories were given a value of 0. Males were assigned a value of 1 and females 0.
As shown in Table 6, five factors were significant risk factors for personal victimisation in the previous year while controlling for the effects of others:
- marital status: those who were single, separated or divorced, or living in a de facto relationship had higher odds of personal crime;
- income: persons in lower income households (under $400 per week) had reduced odds of personal crime;
- residential stability: persons who were living at their current postcode less than one year had significantly higher odds of personal crime;
- main activity: unemployed persons had higher odds of personal crime; and
- night time activities: those who participated in recreational activities outside the home almost every evening had heightened odds of personal victimisation.
Both higher income and unemployment were separate risk factors for personal victimisation, in addition to marital status (other than married or widowed) and having an active lifestyle outside the home in the evenings. While young people reported higher rates of personal victimisation in the bivariate analysis, age lost its predictive power once the effects of other variables were partialed out. Speaking a language other than English at home and Indigenous status were also non-significant when the effects of other variables were controlled.
Overall personal crime includes assault or threat, robbery (theft of property with violence) and personal theft (theft of property without violence or contact with the offender). Examining the violent offence of assault/threat separately, the model is somewhat different. Like in the model predicting all personal crime, marital status, income, residential stability, unemployment and evenings out are significant predictors (Table 7). However, for assaults or threats, being unemployed raised the odds more than threefold net of the effects of others in the model. In addition, two other variables are significant predictors of assault/ threat: language spoken at home, and Indigenous status. Risk of assault or threat is highest for those who are unmarried, unemployed, not living in the lowest income households, living at the current postcode less than one year, routinely spending evenings outside the home, speaking only English at home, and Indigenous, regardless of gender or age.
With respect to household crimes, household income remained significant with lower income households showing reduced odds of household crime, net of the effects of other predictors. Living at the current postcode for less than one year also predicts higher odds of household victimisation (the same result was found at the other extreme of residential stability where those living at the current postcode for 10 years or more had significantly lower odds). Language spoken at home and Indigenous status were both non-significant predictors of household crime.
How frequent is repeat victimisation?
Repeat victimisation is an important area of study because incidents of crime repeated by the same offenders, or repeated against the same victims, contribute so substantially to the overall crime rate. With knowledge about the factors that lead to repeat victimisation, police can help citizens identify and eliminate vulnerabilities that can lead to a repeat experience of the same or other type of crime. A substantial proportion of victims in the ICVS reported more than one experience of crime within the reference periods. Overall, 52 per cent of the Australian sample reported experiencing at least one incident of the crime types covered by this survey within the previous five years. Twenty-eight per cent of the sample reported one crime type, 14 per cent reported two, and 10 per cent three or more (Figure 3). Multiple types of household crimes were more common than multiple personal crimes: 12 per cent of all households experienced more than one type of household crime while six per cent of all persons reported more than one personal crime. Looking just at those persons and household who were victims of crime, almost half (45%) of crime victims experienced multiple crime types, including 32 per cent of victims of household crimes and 22 per cent of victims of personal crimes.
Figure 3: Number of different types of crime victimisations, past five years (per cent)
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
Within crime types, many victims also reported multiple victimisations. Those who reported being victimised were asked how often the crime had occurred within the previous year (2003). Over this one-year period, 68 per cent of all victims reported experiencing one incident of crime, 19 per cent reported two and 13 per cent experienced three or more. The crime most likely to be repeated against the same victim was assaults or threats: 19 per cent of victims experienced three or more assaults or threats within the one-year period (Table 9).
Who is at highest risk of repeat victimisation?
Many of the characteristics which leave individuals and households vulnerable to victimisation are also associated with repeat victimisation. However, some researchers have argued that the single best predictor of victimisation is previous victimisation (Pease 1998), and that the probability of further victimisation increases with each subsequent victimisation (Ellingworth et al 1995). This may be due to enduring vulnerabilities of the crime target (eg. prevention activities are not undertaken; continued proximity to a violent partner), or that the success of the crime provides encouragement to the offender to repeat it. Clear-up rates for many crimes are low (on top of low reporting rates by victims for some types of crimes); therefore successful completion of a crime actually boosts the chances of a repetition (Pease 1998). Repetitions potentially involve the same rewards as the first victimisation but less effort and lower risk (Farrell et al 1995).
The ICVS data show some support for the contention that repeat victims are very similar in traits to those victimised once. Table 10 contains the results of logistic regression analysis predicting repeat personal victimisation in 2003. The dependent variable is dichotomous: experienced one personal victimisation during the year (0); experienced more than one (1). The model is a poor fit to the data with a non significant chi square and only one variable predicting repeat victimisation. Males are more likely than females to be victimised more than once during the one-year period, holding other factors constant. All other predictors were non significant indicating that together they do little to explain repeat victimisation. These results indicate that victims of one personal crime are not differentiated in any way but gender from repeat crime victims.
Similar results were produced by a logistic regression model predicting repeat household victimisation (Table 11). Although income and time at current postcode were significant predictors of household victimisation overall (see Table 8), none of these variables differentiated households victimised once during the year from those victimised repeatedly.
Personal crime: why is it important?
Personal victimisation can have far-reaching effects on victims and those around them. Violent crimes in particular can raise fear for individual victims and their communities, and can have a greater psychological impact as compared to crimes involving the loss of property and no contact between victims and offenders. Very often, assaults occur between people known to each other which can increase the emotional harm to victims.
How do assaults occur?
Assaults and threats were studied in greater detail on the ICVS relative to other crimes, in recognition of the potential seriousness and consequences of this crime and the fact that it is the most common personal crime. Results show that half of all assault/threats involved strangers and half involved offenders known to the victim (Figure 4). Approximately one quarter of assault offenders were friends or acquaintances (including neighbours, colleagues, close friends and others), one in ten offenders were known by sight only, five per cent were partners or ex-partners and five per cent were other relatives.
Assaults and threats occur in different contexts according to the gender of victims. Women are more likely than men to be assaulted within the context of intimate relationships (9% compared with 2%), while men are more likely to report assaults by strangers (53% compared with 47%). However, it is well known that traditional crime victimisation surveys covering a wide range of crime types, like the ICVS, tend to under-estimate the level of partner violence as the methodology or question wording is not designed specifically to measure sensitive experiences that victims may be reluctant to discuss (Johnson 1996). Many countries have developed a specialised approach to interviewing on sensitive topics and the results of these surveys should be consulted for a more in-depth and accurate assessment of partner violence and other forms of violence against women (ABS 1996; Johnson 1996; Tjaden & Thoennes 2000; Heiskanen & Piispa 1998; Lundgren et al 2001). The International Violence Against Women Survey (IVAWS), also being coordinated through the United Nations, is an example of a specialised survey and approach with aims similar to the ICVS: to provide internationally comparable estimates of sexual and physical violence against women in countries around the world (see Mouzos & Makkai 2004 for results for Australia). The IVAWS estimates that 12 per cent of Australian women aged 18 to 69 were victims of assault or threat by a partner in the five years prior to the survey. This is many times higher than the estimate produced by the ICVS where less than two per cent of women reported assaults or threats by partners over the five-year period.
Figure 4: Relationship of victim to offender in assault/threat incidents
Differences between male and female victims are statistically significant for strangers and partners only (p < 0.5).
Stranger includes person not known and person not seen at the time of the offence.
Friend/acquaintance includes neighbour, colleague, close friend and other known person.
Figures do not add to 100% due to multiple offenders.
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
Assaults and threats involving multiple offenders can be particularly fear-inducing and traumatic for victims. One fifth of assault/threats reported to the ICVS involved three or more offenders and 16 per cent involved two offenders (Figure 5). Male victims were about twice as likely as females to be confronted by three or more offenders (27% compared with 14%).
Figure 5: Number of offenders in assault/threat incidents
Difference between male and female victims is statistically significant at p < .05.
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
How serious are they?
The definition of assault in this survey includes both physical attacks and threats that the victim found frightening (see Box Measuring crime victimisation in Chapter 2). Sixty-one per cent were described as physical attacks and 36 per cent as threats (2% did not reply to this question). The attacks reported to this survey involved somewhat different contexts and dynamics as compared to threats. For example;
- multiple offenders of three or more attacked their victims in 50 per cent of cases while single offenders attacked in just 34 per cent of cases and threatened their victims in 66 per cent;
- strangers were somewhat more likely than offenders known to the victim to use threats only (65% compared with 60%);
- a higher proportion of partners attacked their victims (66% compared with 35% of other perpetrators); and
- a higher proportion of male assault victims reported being attacked (42% compared with 32% of females) while female victims were more likely to report threats (68% compared with 58%).
Weapons were used by offenders in 21 per cent of assaults, including 28 per cent of attacks and 18 per cent of threats. These were most often objects used as weapons; just two per cent of assaults involved guns and six per cent involved knives (Figure 6). Male assault victims were more likely than females to report being confronted with a weapon (26% of males compared with 17% of females).
Figure 6: Weapons used in assault/threat incidents
Differences between male and female victims are statistically significant at p < .05.
Source: Australian Institute of Criminology, International Crime Victimisation Survey, 2004 [computer file]
Overall, 53 per cent of attack victims were injured and 24 per cent received medical attention for their injuries. Higher proportions of men who were attacked suffered physical injuries (60% compared with 45% of women) and received medical attention (29% compared with 18% of women). However, there were no statistically significant differences in rates of injury or medical attention by relationship to offenders.
Differences in the nature of assaults reported by men and women may be a reflection of gender differences in perceptions of what constitutes an assault or threat worthy of reporting to a crime victimisation survey. Female respondents may be more willing than males to consider a threat as something that 'really frightened' them, as per the question wording used to capture assaults and threats (see Box Measuring crime victimisation). Some men may be unwilling to report an attack or threat to a survey about crime unless it contains at least one element of seriousness, such as an attack, multiple offenders, a weapon, or physical injury.
Conclusion
The key findings in this chapter show that nine per cent of respondents were victims of personal crimes and 11 per cent of households were victims of household crimes in 2003. There were significant declines in rates of most crime types in 2003 compared with those reported to the previous cycle of the ICVS in 1999. Risk of personal victimisation is associated with having moved postcodes recently, and with routine activities that regularly place people outside the home and reduce guardianship over personal safety. Additionally, assaults and threats are predicted by speaking only English at home and Indigenous status. Household victimisation is predicted by relatively higher income and having moved postcodes recently. Those who had experienced personal or household crime in the previous year tended to live in areas with lower disadvantage relative to non-victims. Almost half of all victims had experienced more than one crime type over the five-year period; one third experienced multiple incidents of the same crime type within one year. Only one variable - gender - differentiated repeat victims of personal crime from one-time victims and none of the variables tested predicted repeat household victimisation. This lends support to Pease (1998) and others who contend that the single best predictor of victimisation is previous victimisation.
- Next section: 3. Perceptions of crime and safety
- Previous section: 1. Background to the ICVS
- Contents