Methodology

Police-recorded crime

The figures on this site are police-recorded crime: incidents that were reported to, or detected by, police and subsequently recorded. This is different from the total amount of crime that actually occurs, which no dataset can measure directly. Some offences are under-reported (for example, sexual assault and domestic violence); others are well captured because they typically require a police report for an insurance claim (for example, motor vehicle theft and burglary).

Counts and rates

Each state or territory page shows a simple recorded-crime count for a period, labelled with whatever word that jurisdiction's own police or statistics agency actually uses for it - most agencies (VIC, QLD, WA, NT, ACT, TAS) call this figure offences, while two (NSW BOCSAR, SA SAPOL) call it incidents. This site never picks one word and applies it to every state; it matches the source. Rate divides that count by the resident population and expresses it per 100,000 people: rate = count ÷ population × 100,000. Rates let you compare places of different sizes; raw counts do not.

Population denominators

Where shown, population figures should be sourced from the Australian Bureau of Statistics (ABS). This product does not fabricate population estimates: if a reliable population figure is not available for a location, rates are not shown for it — only counts, with an explanation.

Rates are also suppressed below a configurable minimum population (small populations produce statistically unstable rates, where one or two incidents can swing the rate dramatically). See the crime snapshot on each suburb page for whether a rate is shown.

Reporting periods

NSW BOCSAR publishes monthly recorded-incident data. This site derives annual totals by summing the relevant twelve monthly observations, rather than storing a separate annual figure that could drift out of sync with the source. Each location page states the exact reporting period covered.

Multi-state coverage & geography levels

Different states publish their open crime data at different geographic granularities, and this site never relabels one as another. NSW BOCSAR publishes at suburb and postcode level. Victoria’s Crime Statistics Agency’s bulk workbook (“Table 03”) includes a genuine suburb/postcode breakdown alongside its LGA totals, and this site imports that suburb/postcode-level data directly — not an LGA figure relabelled as a suburb. Queensland Police Service’s bulk file, by contrast, is genuinely LGA-level only (78 local government areas, no suburb or postcode breakdown exists in the source), so QLD locations on this site are whole council areas, e.g. “Brisbane City Council”, not suburbs within them. South Australia Police publishes at suburb level. Western Australia, the Northern Territory and the ACT each publish at their own police-administrative geography (police districts, reporting regions and patrol districts respectively) rather than suburb or LGA. Each location page and this site’s comparisons always state which of these levels a figure actually represents; see the Sources page for exactly which states currently have real data loaded versus are still “coming soon”.

Tasmania is different again, and different enough that it gets its own page shape rather than a directory of locations. Tasmania Police’s only recurring public dataset, the annual Crime Statistics Supplement, publishes state-wide totals only — there is no suburb, LGA, or district breakdown anywhere in the source — across just four broad offence categories, not the finer categories (residential break-ins, motor vehicle theft, domestic assault, and so on) used for every other state. This site does not invent a geographic or category split the source doesn’t provide, so the Tasmania page shows those four categories at state level, with a real multi-year trend built from thirteen separate annual publications, and nothing more granular.

Offence categories

Every observation keeps the original offence label used by the source agency (for NSW, BOCSAR’s own category names, such as “Break and enter dwelling”) alongside an internal canonical category (such as “Residential break-ins”). The canonical layer exists so that other states, which use different terminology, can eventually be compared on the same basis — see Sources for the current mapping.

Not every BOCSAR offence subcategory has a canonical home yet — 48 subcategories (drug offences, fraud, justice-procedure offences such as breach of bail, and similar) fall outside this site’s current categories and are excluded from every total and comparison shown here. This means a figure on this site is not the same as “all crime recorded by BOCSAR” for that suburb — it’s the total across the categories this site currently tracks. Every import run logs exactly which subcategories were excluded for full traceability.

Suburb coverage

This site covers the large majority of distinct suburb names in BOCSAR’s real suburb-level dataset. BOCSAR’s data has no postcode column for suburbs and no suburb-name column for postcodes, so each suburb name has to be resolved to a postcode independently. A small number of suburbs were verified by hand against general knowledge; the remainder were resolved against a public, community-maintained postcode and locality reference dataset (not an official ABS/PSMA/Australia Post publication), using a documented algorithm that prefers exact matches and leaves genuinely ambiguous names unresolved rather than guessing. Suburb names that couldn’t be confidently resolved are not shown on this site — they are not silently dropped, just not yet coverable without a more authoritative source.

For suburbs resolved this way, the local government area (LGA) and map coordinates shown are approximate — useful for grouping nearby suburbs, but not an authoritative administrative determination (a small number of postcodes that straddle an LGA boundary may show a neighbouring LGA’s name). This does not affect the crime figures themselves: suburb name and postcode, which determine which real BOCSAR data is shown, are always matched directly against the official source files.

Data revisions

Crime agencies occasionally revise previously published figures (for example, after a reclassification or a data-quality correction). When a new dataset release is imported, this site records a new import run with its own timestamp and validation status, so revisions are traceable rather than silently overwriting history.

Geographic boundaries

Suburb boundaries are administrative and can differ slightly between agencies, and change occasionally. Nearby-suburb comparisons in Stage 1 are based on suburb centroid distance or shared local government area (LGA), not a single authoritative boundary file — a known limitation to be improved in a later stage.

Small population limitations

A rate calculated from a small population can swing dramatically from one extra or one fewer incident. This site applies a minimum-population threshold before showing a rate, and always shows the underlying count regardless.

Visitor populations

Resident population is not the same as the number of people actually present in an area. Suburbs with shopping centres, transport hubs, entertainment precincts or large commercial/industrial zones can have crime counts driven substantially by non-resident visitors, which inflates the per-resident rate without meaning residents themselves face proportionally higher risk.

Comparing jurisdictions

Different Australian states and territories, and different crime-data websites in general, use different offence definitions, counting rules and reporting periods. This is a common reason two sources can show different numbers for what looks like the same thing, even when both are accurate on their own terms.

Not a victimisation probability

A recorded-crime rate describes how many incidents were recorded relative to population — it is not a measure of any individual’s personal probability of being victimised, and this site does not present it as one (for example, it will not say “1 in X chance you will be burgled”). That kind of claim requires a dedicated victimisation survey, which is a different type of dataset entirely.

Recorded-crime comparison badge

Each location page shows a badge comparing that location’s recorded break-ins, vehicle crime and property damage against its own state’s average of the same categories (for example, an NSW suburb is compared against the NSW average; a VIC LGA against the VIC average) — never blended across states. Where a reliable population figure is loaded for the location, this badge shows a population-adjusted rate (per 100,000 residents), since a raw-count comparison is structurally biased toward larger, busier locations regardless of how safe they actually are. It falls back to comparing raw counts only when a rate genuinely cannot be computed — no population data loaded for the state yet, or the location’s own population is missing or below the suppression threshold (see “Population denominators” above) — and that fallback is stated as a caveat on the location page itself: a location’s own recorded-crime volume over time is still real, sourced data, but a raw count should not be read as a per-resident danger ranking against other locations. The full year-by-year table on each page always shows both the raw count and, where available, the population-adjusted rate per 100,000 residents. The accompanying “Protect your [suburb] home” product suggestions are derived from the same location’s own rising/highest categories, and link to a general product search — never a fabricated product, category or price.

Why different crime websites show different numbers

Beyond jurisdictional differences, sites can differ because of the offence categories they group together, the reporting period they use, whether they show counts or rates, how they define a location’s boundary, and how recently they last imported data. This site documents all of these choices on every page so its numbers can be checked against the original source.