Belfast, Newry and Derry lead the vehicle-crime map — but the pins do not show crime scenes

A vehicle-crime map offers a seductive promise: zoom in, find a cluster of dots and discover where a car is most likely to be targeted. In practice, responsible analysis requires far more restraint. The public maps are designed to show patterns while protecting the location of victims. Treating a point as the position of a stolen car can misidentify an innocent address and turn a useful public dataset into a source of false allegations.

PureCar's Vehicle Crime Northern Ireland explorer contains 36 months of police-mapped vehicle-crime data, from September 2023 through August 2026, spread across 11 council areas. Its latest rolling year includes 2,152 mapped records, compared with 2,357 in the preceding 12 months. Across the imported archive, PureCar has reconciled 4,015 map-cell rows. Those rows are grouped data records within the mapping system, not 4,015 verified distinct crime scenes or street addresses.

At council level, Belfast, Newry, Mourne and Down and Derry City and Strabane are the leading areas in the site's latest-year mapped comparison. That is an important finding about where the public mapped offences are concentrated in absolute numbers. It is not a ranking of the probability that any one driver will be victimised, and it provides no basis for declaring a named street or postcode the most dangerous place to park.

What a map point actually represents

The Police.uk public-data system has a deliberate anonymisation procedure. When incident information is prepared for its public map, the recorded location is replaced by a nearby preselected anonymous point. Official documentation says these map points are placed at street centres or suitable public and commercial places and selected so their catchment area contains at least eight postal addresses or no postal addresses at all. The original incident coordinates are not released as the dot that a member of the public sees.

The system also acknowledges that some incidents cannot be shown in a suitable place. A victim may not know exactly where a theft occurred, or a reliable mapped point may not be available within the permitted distance. Police.uk says incidents can be omitted from the map in those circumstances. It follows that map totals and map clusters should not be described as a complete, address-accurate inventory of offending.

This protection is not merely a legal formality. Suppose a public dot sits outside a shop, a small workplace or a house. It does not follow that the premises was a victim, an offender's location or a place where multiple vehicles were targeted. The mark may stand for offences occurring elsewhere in its catchment. Precision-looking graphics can create certainty that the source data does not support.

The 4,015 map-cell-row figure needs the same care. It is a count of grouped records across PureCar's three-year archive. Without examining how cells repeat across months and how offences aggregate within them, it would be wrong to call it the number of individual hotspots. The site retains monthly and council breakdowns precisely so the aggregation rules remain visible rather than being disguised by the design of the map.

The value of comparing councils—and the limits

Absolute counts tell the public where the largest number of mapped records appears in the selected period. For public-service reporting, that matters: it can identify where larger volumes of recorded incidents may justify additional prevention attention. But counts alone do not measure comparable exposure. A district with a major city, dense car parking and large flows of commuters has different circumstances from one with dispersed villages and fewer parked vehicles.

The presence of Belfast, Newry, Mourne and Down, and Derry City and Strabane near the top of the absolute mapped totals should therefore prompt questions, not a simplistic risk league table. How many vehicles are registered or routinely parked in each area? How many are left in publicly accessible car parks? How long do they remain unattended? How much visiting traffic does each council attract? What proportion of the mapped offences involves entire vehicle theft as opposed to theft from or interference with vehicles?

PureCar's public mapped dataset does not supply all those denominators or detailed offence splits. A responsible assessment cannot divide the number of mapped crimes by resident population and call it a personal vehicle-theft risk rate without considering where vehicles are actually exposed. Population rates can help answer some research questions, but they do not solve this exposure problem by themselves.

Three important places, three different questions

For Belfast, the most useful investigative question is about concentration: are observed mapped incidents spread across the city or concentrated in changing groups of approximate locations, and how does that pattern compare between months? In the separate official offence series, Belfast accounted for 451 of Northern Ireland's 843 theft-from-vehicle offences in 2024/25. That historic finding adds a reason to study the city but is not itself part of the later mapped annual total.

For Newry, Mourne and Down, the geographical size and mix of settlements make it especially important to avoid treating a district-level position as a statement about every town in the area. The new pages enable users to follow the mapped monthly pattern without assuming that the whole council area has an equal distribution of offences.

For Derry City and Strabane, a headline about the wider district must likewise be careful not to imply that all recorded incidents occurred within the city. The council comparison covers a named administrative geography, not just the best-known urban centre inside it. The precise anonymised map placements should not be converted into identifiable allegations about particular businesses or residences.

These are questions that the maps help organise; they are not invented findings about specific streets. The ability to move between months and councils is valuable because it reveals whether a broad district headline holds across time. The danger comes when journalists or readers use the same interface to claim a level of location certainty the underlying source intentionally removed.

What three years can show that one year cannot

A three-year archive is useful for testing whether a busy month is an exception or part of a more persistent run. PureCar's latest-year mapped count, for example, fell 8.7%, from 2,357 to 2,152, while the month of August 2026 still reached 228 records, the highest month of that latest 12-month period. Either fact alone could be turned into a different headline. The 36-month context encourages readers to put both into the same assessment.

Even so, three years do not establish a causal explanation. A movement in police-recorded crime could reflect changes in offending, reporting, recording, police activity or data revisions. Month-by-month patterns deserve careful comparison, especially where small numbers can change percentages sharply. The map cannot reveal an offender's motive or prove that a new deterrent caused a fall.

The archive's reconciled monthly and council rows matter for a second reason: they give readers a consistent route to test a claim. PureCar's 2026 data report presents the 36-month series and the 11 council comparison in one place, rather than relying on screenshots of a map whose selection period might not be obvious.

Where the data cannot go

A public mapped vehicle-crime count does not establish whether a car was recovered, whether a criminal was identified, the value of property taken or the amount of damage. It does not verify a particular model was targeted. It does not document CCTV footage, victim experiences or the effectiveness of a security measure. Those questions require other sources and, in many cases, interviews and additional police records.

It would be just as misleading to compare PureCar's mapped total directly with an official theft-of-vehicle statistic from a different period. Police.uk's category encompasses theft of or from vehicles and interference, whereas PSNI publications also provide narrower offence codes. Both systems can be informative if the dates and definitions are made explicit.

Why a hotspot can move even if a neighbourhood does not

Public map points can change their apparent importance from one month to the next because the underlying count changes. But the apparent movement of a cluster is not proof that offenders moved from one named street to another. The coordinates reflect anonymised catchment locations, not a trail of exact scenes. A cluster may describe several events distributed within an area; the number of symbol placements is not automatically the number of individual victim addresses.

The system is designed this way for a reason. Police.uk says its master list of anonymous points is selected so each point's catchment includes at least eight postal addresses or none, and the coordinates of an actual incident are replaced by those of a selected point. That protects people who might otherwise be identified in a quiet street where a single incident would be obvious. Removing that protection would not make the statistics more truthful; it would change the kind of information released and create privacy risks.

There is also a practical effect on comparisons. A hotspot visible at the centre of an urban street or near a public facility is a representation of a wider reporting area. It may draw a viewer's eye to a familiar landmark that had nothing to do with the incidents. A local article should describe the broader mapped area, not accuse the point's host business or nearest resident of being involved.

Some people will naturally use a map while deciding where to park, buy property or operate a business. Those decisions deserve better evidence than a single symbol. Security features, lighting, the type of parking, CCTV arrangements, vehicle exposure and recent locally verified information may all be relevant. The map contributes one piece of historical context, and even that piece is intentionally approximate.

How the archive was checked

The value of a 36-month dataset rests partly on whether the internal totals reconcile. PureCar's build checks compared the archive across its 11 council areas and monthly records; the project reported zero reconciliation mismatches across the imported series. The retained structure includes 36 monthly positions and 4,015 map-cell rows. Those checks do not independently audit PSNI's underlying crime-recording decisions, but they reduce the risk that PureCar's own displayed totals fail to agree with its imported dataset.

An internal reconciliation test is different from external fact-checking. The former can verify that a database total matches the sum of rows attributed to that database. It cannot verify that an incident was correctly classified when originally recorded, nor that every crime is present in a public mapping product. The police source and the publication's transformation processes each have their own limitations.

The value of exposing those checks to readers is straightforward. When a story says 2,152, the reader should be able to see the same number in the latest-year summary and the corresponding monthly total. When it compares 11 council areas, the relevant period and category should be the same. When it moves to earlier data, the published history should preserve the months being compared rather than silently replace them.

Why a council ranking needs three accompanying questions

First, does the ranking measure total mapped records or a rate? The present PureCar leaders refer to counts, not incidents per vehicle or per parking hour. Second, does the council boundary match the place name used in the headline? 'Derry City and Strabane' is not a synonym for Derry city alone, just as 'Newry, Mourne and Down' covers much more than Newry. Third, does the map category match what the headline describes? Theft from vehicles is not the same as theft of vehicles.

A fourth issue is time. An annual council comparison may be shaped by months that are no longer representative of the latest local experience. The monthly controls let a reader inspect that variation, but one month with a low baseline should not be presented as definitive evidence of improvement or deterioration. Where the figures cannot resolve a question, the most accurate answer is that the public series does not show it.

This matters to local businesses whose reputations may be affected by descriptions of 'dangerous' areas. A large mapped count near a commercial zone is not an indictment of its security, its staff or its customers. It may reflect anonymisation choices, reporting geography, traffic exposure or a range of other factors the map cannot distinguish. Fair local reporting should avoid drawing allegations from the proximity of a dot.

What an evidence-led local investigation could add

A genuinely deeper investigation would combine the public map with properly sourced interviews, more specific official offence counts and relevant measures of vehicle exposure. It might request comparable theft-from-vehicle figures for several financial years, examine whether a specific broad area has experienced a persistent change, or ask policing partnerships about the effectiveness of particular prevention measures. Any publication of location-level detail should continue to protect victims.

An interviewer could seek testimony from motorists or businesses, but those voices must be real and attributable or carefully anonymised with consent. A statistical feature should not invent a victim or a police explanation to make a narrative feel complete. The current public records support a clear account of trends and geography; they do not license fabricated scenes or direct quotations.

Until more is available, the strongest version of a 'hotspots' report uses restrained language: leading council totals, defined periods, anonymised points and clear sourcing. That is more informative than a colourful ranking of streets that the underlying map never measured.

There is a final distinction between identifying a pattern and diagnosing a solution. A map can point investigators towards questions about parking facilities, changing reporting levels or types of victimisation, but it cannot determine which intervention works. Claims that a gate, camera, streetlight or patrol reduced offending require separate evidence about the relevant place and comparable periods. Treating a map as the beginning of that inquiry, rather than its conclusion, protects against both misplaced blame and false promises of safety.

The standard PureCar should apply to every local headline

The map offers real public value when its limits are visible. A reader can examine patterns across districts, see which months are changing and use that knowledge to ask better questions of officials. The map becomes harmful when a reporter takes an anonymised symbol and declares a home, street or small business a crime hotspot without supporting evidence.

For future coverage, the strongest standard is to label the window, define the crime category, identify the geographic unit and explain the approximation. A statement such as 'Belfast is among the leading council areas for publicly mapped vehicle-crime records in September 2025–August 2026' is defensible. A statement such as 'this Belfast address had the most car thefts' is not supported by these records.

The investigation's central conclusion is about interpretation: Northern Ireland now has a clearer public view of vehicle-crime patterns across 11 councils, but it does not have a public map of the exact addresses where every crime occurred. Preserving that distinction is what allows the data to inform local debate without putting unsupported blame on anyone's doorstep.

Methodology: PureCar's reconciled Police.uk/PSNI mapped vehicle-crime archive, September 2023–August 2026, including 4,015 map-cell rows and latest-year 2,152 records. Council leaders refer to absolute mapped volumes. Police.uk locations are anonymised and approximate, not verified offence addresses. Further details: PureCar vehicle-crime methodology.