Using crime statistics

This page contains general information about using crime statistics. It includes information about what to be mindful of when using crime data as well as information about what crime statistics are, which crimes or offences BOCSAR has information on, how we count 'crime' as well as how we conduct statistical trend testing.

Words of caution about crime data and its pitfalls

There are a few cautionary notes that are important to know before using BOCSAR's crime statistics. Irrespective of how trends in recorded crime are presented, their interpretation is a difficult task.

Recorded crime statistics for some offence categories DO NOT accurately reflect the actual level of crime in the community. This is because the number of incidents recorded may be affected by extraneous factors which are not easily measured. In particular:

Public willingness to report crime - Many crimes which occur are not reported to police and will therefore not be recorded - for example, a large number of assaults, sexual assaults and robberies are not reported to police.
Shifts in policing policy - Recording of those offences which are detected by, rather than reported to police, are strongly affected by policing practices - examples of these are drug offences, drink driving offences, offensive behaviour and receiving stolen goods. Recorded rates for such offences do not accurately reflect actual rates.
High visitor populations compared to residential populations - Sydney Local Government Area has high recorded crime rates because, compared with other regions, the resident population is small relative to the number of people in the area. In other words the area has a high user population which is not reflected in the denominator of the rate calculation. Areas with small residential populations - Recorded crime rates in Local Government Areas with small population sizes are not always a good indicator of offending. Crime rates in areas with populations under 3,000 may be unreliable and should be interpreted with caution.

Due to such extraneous factors, in many instances it is simply impossible to state with any assurance why a particular crime trend has appeared or disappeared.

What are crime statistics?

BOCSAR's crime data consist of criminal incidents reported to, or detected by, police and recorded on the NSW Police Force's Computerised Operational Policing System (COPS). While this system is used for record-keeping for all police operations, not just for criminal matters, BOCSAR only reports on criminal incidents. BOCSAR's crime statistics therefore do not capture crimes that are not recorded on COPS. The Australian Bureau of Statistics conducts regular crime and victim surveys which attempt to capture a snapshot of both reported and unreported crimes.

What offences are included?

BOCSAR's standard recorded crime publications report on 13 major offence categories, which include the most serious personal violence and property offences such as assault, robbery, burglary and malicious damage. In addition to these offences, we also release data for 60 offences on our interactive Crime Mapping Tool. For more information about offences, see the table showing the NSW Police Force incident categories comprising BOCSAR's offence categories. For definitions of specific offences, please see our definitions page.

What is being counted?

When a crime is reported to or detected by the NSW Police they record it as a criminal event on COPS. The event date is the date that the crime was reported. Unless otherwise stated, the date BOCSAR attaches to a crime is the event date. One criminal event can contain multiple criminal incidents. BOCSAR's default counting unit for crime data is criminal incidents. Criminal incidents are activities detected by or reported to police which involve the same offender(s) and victim(s); occur at the one location during one uninterrupted period of time; fall into one offence category and one incident type.

BOCSAR also reports on victims who have come to the attention of the NSW Police Force either because they reported a crime against them or the crime was otherwise detected. Victim information is only provided for crimes against the person (not property crimes or crimes without a typical victim such as drug offences). Within BOCSAR's annual and quarterly crime reports the default counting unit used for murder and manslaughter is victims (rather than incidents).

BOCSAR also reports on alleged offenders or 'Persons of Interest' (POIs) who have had a legal action commenced against them by the NSW Police Force. Within the police recorded crime data POIs may not have been convicted in court, however they have been proceeded against in some way or other by the police.

For more detail about incidents, victims and alleged offenders, please see our definitions page.

What are crime trends and how are they calculated?

In general, BOCSAR relies upon the Kendall Rank Order Correlation statistical test (or Kendall's tau-b) to determine whether a series is trending upwards, downwards or is stable (p<.05). The statistical test is applied to the monthly data over the specified period. Where a significant trend is found, the average annual percentage change over the time period is reported. If any of the 12-monthly totals in the series have a value of less than 20, a trend test is not conducted. Within the Crime Mapping Tool trend tests are calculated on monthly rates per 100, 000 population (rather than on the number of incidents/victims/offenders as is done within our annual and quarterly crime reports) to adjust for changes in levels of residential populations for specific geographic locations over time. More details about calculating trend tests are provided within the trend test definition.