How to assess a crime-rate claim

The short answer

assess a crime-rate claim

Use comparable police or statistical data, distinguish counts from rates, check reporting changes and population trends, and avoid inferring individual risk from a broad geographic average. Crime statistics are among the most frequently cited and most frequently misrepresented numbers in Australian political debate. A politician's claim that 'crime is out of control' or that 'crime is at a record low' may both be based on real data while telling opposite stories, because the speaker has selected a particular offence category, geography, time period or measurement method that supports their argument. The Australian Bureau of Statistics publishes multiple crime-related data collections, each measuring something different. Recorded crime data from police forces counts offences reported to or detected by police, not all offences that occurred. Crime victimisation surveys ask people whether they experienced specific offences, capturing incidents that were never reported to police. Corrective services data counts people in prison or under community supervision, reflecting sentencing outcomes rather than offending rates. None of these collections tells the whole story alone, and none directly measures the thing most people care about: their personal risk of being a victim. Assessing a crime-rate claim therefore requires understanding which data source is being used, what it actually measures, how it compares to alternative sources, and what it tells you — and does not tell you — about safety.

This is a method for assessing public claims, not a verdict on a party or a direction on how to vote. Conclusions should change when the underlying law, data or implementation evidence changes.

The useful question is not only “what is the rule?” but also “who administers it, which document controls it, and when might it change?” That distinction prevents an accurate general explanation from becoming wrong advice in a particular election, chamber or policy setting.

Evidence review

Understand the different types of crime data

Australian crime statistics come from several distinct sources, and each answers a different question. Recorded crime statistics, published by the ABS from data supplied by state and territory police forces, count offences that came to the attention of police and were recorded according to national counting rules. These statistics answer the question: how many offences did police formally record? They do not answer the question: how many offences actually occurred? Many crimes — particularly sexual assault, domestic violence and minor property offences — are never reported to police. The gap between actual victimisation and police recording varies by offence type, jurisdiction and over time, so a rise in recorded crime could mean more offending, more reporting, or changes in police recording practices. The ABS publishes detailed explanatory notes with each crime release that describe these limitations, and anyone assessing a crime-rate claim should read them.

Evidence review

Counts versus rates: why population matters

The most common error in crime-rate claims is confusing a count with a rate. If a politician says 'there were 50,000 assaults in New South Wales last year,' that is a count — an absolute number. But New South Wales has the largest population of any Australian state, so a higher count of almost any offence is expected. A rate adjusts for population, typically expressed as offences per 100,000 people. If New South Wales recorded 50,000 assaults and Tasmania recorded 5,000, New South Wales has ten times as many assaults but also roughly fourteen times as many people — meaning its assault rate may actually be lower. Always check whether a claim uses counts or rates, and if it uses counts without mentioning population, the claim is incomplete regardless of whether the numbers are accurate.

Population changes over time also matter. Australia's population has grown by several million people over the past decade, so raw counts of offences are expected to rise even if the rate of offending per person remains stable or falls. A claim that 'crime is worse than ever' because the total number of offences is higher than in 2010 ignores that the 2010 population was substantially smaller. Similarly, the age structure of the population affects crime rates because offending is concentrated among younger age groups. An ageing population might show falling crime rates even if offending behaviour among younger people has not changed, simply because there are proportionally fewer people in the highest-offending age brackets. These demographic effects are well understood in criminology but rarely mentioned in political debate.

Evidence review

Reporting changes and classification differences

Changes in how police record crime can create the appearance of a crime wave or a crime drop that has nothing to do with actual offending. States and territories operate under different legislation, and an act that is a crime in one jurisdiction may not be a crime, or may be a different category of crime, in another. The ABS works to standardise crime counting across jurisdictions, but residual differences remain. A comparison of 'assault' rates across states may reflect different legal definitions of assault, different police recording thresholds and different charging practices, not different levels of violence. Within a single jurisdiction over time, a change in police recording systems, a new mandatory reporting requirement or a high-profile case that prompts more victims to come forward can all change recorded crime statistics without any change in underlying offending.

The ABS periodically updates its statistical standards, which can change how offences are classified, counted and published. When this happens, the ABS explicitly warns that data before and after the change is not directly comparable. A claim that crime 'increased' or 'decreased' across a break in statistical series, without acknowledging the break, is misleading. The ABS also cautions that small numbers should be treated with care: a 100 per cent increase from one case to two cases in a small geographic area is not a crime wave, but a statistical fluctuation that would disappear with a slightly different counting period. Political claims about crime that focus on percentage changes in small areas or short time periods almost always overstate the reliability of the underlying data.

Evidence review

Aggregation and the ecological fallacy

Crime rates are usually reported at the level of a state, a city or a police district. These are averages across large and diverse areas. A claim that 'Victoria's crime rate is X' tells you nothing about any particular street, suburb or town within Victoria, because crime is geographically concentrated. A small number of locations typically account for a disproportionate share of reported offences, and most places in the same state will have crime rates far below the state average. Using a state-level statistic to assess the safety of a specific neighbourhood commits what statisticians call the ecological fallacy: assuming that a group-level statistic applies to individuals within the group. A politician who claims that 'crime in Queensland is rising' may be accurately reporting a state-level trend that is driven entirely by increases in a handful of postcodes, with crime falling or stable everywhere else. If the claim does not specify a geographic level, ask whether the geographic unit matches the claim being made.

Similarly, crime types should not be aggregated without care. A claim about 'violent crime' may bundle homicide, assault, sexual assault and robbery into a single category, but these offences have very different characteristics, trends and risk factors. Homicide in Australia is rare — typically a few hundred incidents per year nationally — and its trend over time tells you almost nothing about the much more common assault rate. If a politician claims 'violent crime is up,' ask which specific offences are included, because the trend for homicide may be flat or down while the trend for assault may be up, and combining them into one category obscures both patterns. The ABS publishes data by detailed offence categories precisely to prevent this kind of misleading aggregation.

Evidence review

Victimisation surveys fill the reporting gap

Because many crimes are not reported to police, recorded crime statistics systematically undercount certain types of offending. The ABS Crime Victimisation Survey asks a representative sample of Australians whether they experienced specific offences in the previous twelve months, regardless of whether they reported them to police. This captures a broader picture of victimisation and reveals the size of the reporting gap. For some offences, such as physical assault and sexual assault, the survey consistently finds much higher victimisation rates than police records show. If a politician cites official crime statistics without acknowledging the difference between recorded and experienced crime, they are presenting a partial picture. The victimisation survey data is particularly important for assessing claims about trends in household crime, such as break-ins and motor vehicle theft, which have high reporting rates and therefore produce more reliable comparisons between recorded and survey data.

Evidence review

A quick checklist for assessing any crime-rate claim

When you encounter a political claim about crime, work through these questions before accepting or repeating it. First, identify the data source: is it police-recorded crime, a victimisation survey, corrective services data or something else? If the source is not named, the claim is unverifiable. Second, check whether the number is a count or a rate, and whether the denominator is the appropriate population. Third, check the comparison: what time period is being compared to what, and is the comparison period representative? Fourth, check the geography: is the claim about Australia as a whole, a single state, a city or a specific location, and does the geographic unit match the claim? Fifth, check the offence: which specific offence or offences are included, and are different offence types being aggregated? Sixth, check for known data breaks: did recording practices or statistical standards change during the comparison period? Seventh, check the victimisation survey data for the same period and offence type to see whether the recorded crime trend is consistent with victimisation trends. None of these steps requires statistical expertise; each requires only ten minutes of looking at the source data rather than accepting the politician's characterisation of it.

The ABS Crime and Justice portal and the Australian Institute of Criminology website are the starting points for all of these checks. Both organisations publish their data with methodology notes that explain what is being measured, how it was collected, and what the limitations are. The most useful habit you can develop is to read those methodology notes before reading the numbers. A data point without its methodology is like a map without a scale: it tells you there is something there, but it cannot tell you whether it matters.

Common questions

Before you rely on the answer