Stolen car, stolen tools or attempted break-in? Northern Ireland's vehicle crime figures are not interchangeable

A single phrase—'vehicle crime'—can conceal several profoundly different events. A car driven away without permission, a work van entered for its tools and a vehicle interfered with during an attempted theft all belong in conversations about motoring security. They do not, however, generate the same police offence code or cause the same disruption for the person affected.

That distinction has become especially important as drivers encounter more searchable public data. PureCar's new vehicle-crime explorer presents 2,152 police-mapped vehicle-crime records for September 2025–August 2026. Separately, a PSNI freedom-of-information disclosure for the 2024/25 financial year lists 1,022 theft or unauthorised-taking offences, 186 aggravated vehicle-taking offences and 843 theft-from-vehicle offences across Northern Ireland. All four numbers are valid descriptions of their sources. They are not four ways of measuring the same thing.

The temptation is to use the figure that sounds most striking and call it 'cars stolen'. That would misinform readers. The 2,152 figure is drawn from a broad public map category in one time period, while the three PSNI numbers come from specified offence codes in another time period. Even adding the three PSNI counts gives 2,051 offence entries, not an equivalent mapped series and certainly not a verified number of unique vehicles stolen. Different definitions, geographies and publication processes mean such a calculation cannot replace a like-for-like comparison.

What is counted when a vehicle itself is taken?

The PSNI's category 'Theft or unauthorised taking of a motor vehicle' describes offences involving a vehicle taken without consent. It includes circumstances where the vehicle might later be recovered. It should not be understood automatically as a tally of thefts with permanent loss, nor as a breakdown of vehicle make, age or value. In 2024/25 the district totals in FOI/15673 sum to 1,022 offences of this type.

The separate category 'Aggravated vehicle taking' accounts for a different legal and recording classification. The 2024/25 total was 186. In ordinary language, both groups can be described in the public discussion of vehicle-taking crime. In accurate statistical reporting, they should either be labelled separately or combined only with an explanation of what is being added and why. Calling either total simply 'the number of stolen cars' leaves important questions unanswered.

The later calendar-year PSNI disclosure, FOI/16599, records 911 theft or unauthorised-taking offences in 2025, down from 1,015 in 2024, and 154 aggravated vehicle-taking offences in 2025, down from 200. These figures are not necessarily in tension with 2024/25 totals because a calendar year and a financial year include different months. FOI/16599 was published in February 2026 and specifically described its 2025 entries as provisional at that time.

What changes when someone steals from a vehicle?

A theft-from-vehicle offence describes property targeted in or on a vehicle rather than the taking of the vehicle itself. It may concern belongings or equipment and, under the relevant official recording definition, can include attempted theft. The PSNI FOI count fell from 982 in 2023/24 to 843 in 2024/25. The year-on-year reduction should not be presented as a fall in the number of cars stolen, because these incidents are different.

The damage to a victim can nevertheless be extensive. A work vehicle may remain in the driveway but be unusable for the following day's appointments because specialist tools have gone. A window may be broken even where the property taken is of limited value. The official table does not quantify the financial loss, downtime or insurance consequences. Attaching an arbitrary pound value per recorded offence would produce a fabricated economic total.

There is also a distinct recording category involving vehicle interference. An attempt to interfere with a car or its contents may be classified differently depending on the evidence of what the offender intended. The Police.uk broad vehicle-crime category includes theft of or from vehicles and interference with a vehicle, which makes it more inclusive than a story about complete vehicle theft alone. That broad classification is useful for examining public mapped patterns, but less useful if the precise question is how many vehicles were taken away.

Why the map shows approximate locations

The Police.uk data used in public crime mapping deliberately substitutes anonymised map points for real incident addresses. Its public documentation explains that locations are snapped to suitable preselected points, typically representing an area rather than a door, and not every incident can be displayed. Those protections are necessary for victim privacy. They also mean that a reader cannot count marks around a specific house and credibly conclude that all the crimes occurred at that address.

PureCar's archive has 36 months of records from September 2023 to August 2026, with reporting organised around Northern Ireland's 11 council areas. Its latest mapped total fell from 2,357 in September 2024–August 2025 to 2,152 in September 2025–August 2026, an 8.7% reduction. That finding belongs to the map's broad category and matched 12-month periods. It is not a fall in car thefts alone, nor a direct comparison with a financial-year FOI response.

The distinction is more than academic. Imagine that recorded thefts from vehicles declined while the theft of whole vehicles increased. A combined vehicle-crime headline might move little, concealing a very important change for drivers and investigators. The opposite could also happen. Tracking categories separately is essential whenever the question involves particular prevention methods, claims, stolen property or policing priorities.

The date trap that can distort a headline

Official datasets in Northern Ireland use different calendars for different purposes. Some are published monthly, some over January–December and others across April–March financial years. The phrase 'latest year' is not sufficient on its own. Readers need a start date, an end date, and a description of what counted as an offence.

For example, the 1,022 figure covers April 2024–March 2025 and describes theft or unauthorised taking. The 911 figure covers January–December 2025 and describes the same offence category in a different window. The 2,152 figure covers September 2025–August 2026 and describes the wider public mapped vehicle-crime category. None of those figures can be subtracted from another to produce a meaningful trend. Doing so would manufacture a change by switching both measurement and period.

There is a further problem with incomplete and anonymised location data. A force's official count of offences in a given category may not match the number that can be shown on the public map. Incidents without an appropriate mappable location may be absent from its displayed layer. PureCar's methodology page explains why map counts should be described as map counts rather than substituted for all official police totals.

The questions the public actually wants answered

Different citizens ask different questions when they read about vehicle crime. A driver buying a used car may want to know about security features and how to protect keys. A tradesperson may care most about theft of equipment from vans. Residents may want to know whether an area has experienced a sustained rise rather than a one-month fluctuation. Insurers may examine different underlying exposures again.

No one number in the released data answers all of those questions. The most useful reporting combines clear definitions with context, not maximum numerical impact. It separates what police have counted from what an analyst has inferred and leaves gaps visible when the underlying data is not detailed enough to resolve them.

Specific protective advice should also follow the risk involved. PSNI recommends protecting keyless entry by storing keys safely, including signal-blocking storage where appropriate, and using physical security devices. Thefts from vehicles raise different practical issues, such as keeping valuable belongings out of sight, checking the doors are locked and planning the secure storage of tools. Both require prompt reporting to police if an incident occurs.

A worked example of how a misleading comparison happens

Consider a headline writer who sees the PSNI's 1,022 theft or unauthorised-taking offences in 2024/25 and PureCar's 2,152 mapped vehicle-crime records for September 2025–August 2026. Subtracting 1,022 from 2,152 produces 1,130; dividing it by 1,022 produces an apparent increase of roughly 111%. That calculation is mathematically possible but statistically meaningless. It compares different categories, periods and publication systems. It does not show that car theft doubled.

A second writer might combine PSNI's 1,022 theft/unauthorised-taking offences, 186 aggravated taking offences and 843 theft-from-vehicle offences into 2,051, then compare this total with PureCar's 2,152. That apparent increase of 101 is also not evidence of a rise. The 2024/25 offence categories are more specific than the map's broad vehicle-crime category, and the windows do not match. The map may also include vehicle interference that is outside the three displayed FOI lines. The sum is not a corrected version of the map total.

A third writer might use the annual fall in mapped vehicle crime—8.7%—and say that an owner's chance of having a car stolen fell by 8.7%. This is a different error. The comparison is valid inside the mapped series, but the conclusion extends it beyond the population it measures. The category is wider than complete vehicle theft, and the map does not provide a per-vehicle exposure rate. The words used to translate data into a headline matter as much as the arithmetic.

A careful account makes the unit explicit: 'offences of theft or unauthorised taking'; 'aggravated vehicle-taking offences'; 'theft-from-vehicle offences'; or 'police-mapped vehicle-crime records'. Those descriptions are longer, but they preserve the meaning of the figures. Readers should not have to discover after publication that a report headed 'cars stolen' included attempts to take items from vehicles.

Why victims' experiences do not fit neatly into a single row

The police counting system serves a necessary administrative purpose. It classifies reports so offences can be recorded, compared and investigated consistently. But a person harmed by vehicle crime may experience consequences that cross several practical categories. If a vehicle is taken and later recovered damaged, the initial offence, property recovery and repair costs may all form part of one lived experience. The publicly released aggregate table is not a case history of those consequences.

The same is true of a vehicle that is damaged during an attempted break-in. Whether the incident appears as interference, attempted theft, theft or criminal damage can depend on evidence and recording rules. A broken window on its own does not allow a reporter to establish intent. This is why it is unsafe to look at a photograph of damage and assume which offence code police used.

The data cannot establish the number of prosecutions or convictions without separate outcomes information. Nor can it show what proportion of incidents involved organised crime, car-key theft, electronic attack or simple opportunism. These are valid investigative subjects, but they require evidence from other records, case investigations or named sources. Treating broad police statistics as proof would be a shortcut that misleads the public.

How better labelling strengthens a local newsroom

A serious local news operation should make it easy for readers to reproduce its main comparisons. For each reported figure, that means naming the source, geographical area, offence category, start and end dates and any limitations such as provisional data or anonymised locations. It also means not presenting annual financial-year statistics as if they describe the most recent 12 months of a live map.

The practice is especially important as articles are republished in snippets, search results and social previews. A qualification hidden deep in the body may disappear when the headline is shared. A precise headline does not have to be lifeless; it can be compelling because it reveals a genuine contrast: Belfast's share increased while Northern Ireland's theft-from-vehicle total decreased, or two districts recorded more vehicle-taking offences while the regional line fell.

Good statistical writing also resists dramatic claims about particular communities that the evidence cannot sustain. The publication of police-mapped locations is intended to inform the public while reducing the chance that victims can be identified. A newsroom that reverses that protection through overconfident place names does a disservice to its readers and those depicted by the data.

What the official figures can still be used for

Despite these limits, the figures are not worthless. The FOI disclosures allow meaningful comparisons within specified offence classes and matching years. The 2024/25 theft-from-vehicle total of 843 can be compared with 982 in 2023/24 because the measure and year structure match. The 2025 theft or unauthorised-taking count of 911 can be compared with 1,015 in 2024 for the same reason, subject to the provisional status of 2025 at the time of publication.

PureCar's rolling-year mapped comparison is equally valid inside its own definitions. Its 2,152 records and the earlier 2,357 arise from two corresponding 12-month windows. Investigators can ask whether that broad trend is reflected in particular councils or whether the local pattern differs. The correct outcome is not to discard either source, but to use each to answer the question it was built to answer.

For a reader worried about personal security, the figures are a context for sensible precautions, not a personalised forecast. For a councillor or policing partnership, they are a starting point for questioning local trends. For a journalist, they are a constraint on claims: the evidence must be strong enough for the particular statement being made.

Editors can also improve accuracy through small but meaningful presentation choices. A comparison table should place the dates in its heading, keep offence names visible on a mobile screen and avoid graphs that crop the axis to exaggerate a small difference. Figures should be accompanied by the underlying counts as well as percentage changes. If a total is provisional or mapped only, the qualification belongs beside the number rather than in a footnote few readers will reach. These measures do not reduce the story’s impact. They make the impact credible by keeping the evidence legible.

A comparison can therefore be powerful without being misleading. Rather than declaring that car theft is up or down because one large total changed, the writer should identify the exact series and explain how much it changed over a matching period. That is especially important for a national audience arriving from search results, where headlines are often read without the tables beneath them.

A better standard for Northern Ireland crime reporting

The purpose of PureCar's new vehicle-crime pages is to make the geography and timing of recorded incidents easier to explore. Their value depends on being precise about the limits of that exploration. A site that publishes the total without the category and reporting window may generate a more dramatic headline, but it gives readers a less reliable understanding of the problem.

The essential rule is straightforward: a mapped vehicle-crime record is not automatically a stolen car; an offence count is not automatically a unique vehicle; and two annual periods cannot be compared unless their dates and definitions align. Northern Ireland now has several useful sources of vehicle-crime information. The strongest reporting makes them complementary, rather than forcing them into a false single statistic.

Primary evidence: PSNI FOI/15673, financial years 2022/23–2024/25; PSNI FOI/16599, calendar years 2023–2025; Police.uk's publicly documented vehicle-crime category and anonymised location methodology; and PureCar's separately reconciled mapped series through August 2026.