Prefer primary sources
Where available, profiles use official police, government and national statistical sources rather than secondary summaries.
Keep the geographic unit
City proper, metropolitan area, police reporting jurisdiction, Community Safety Partnership and Local Government Area statistics remain labelled as their source geography. Similar names do not make different boundaries equivalent.
Match places conservatively
A profile uses a source geography only when the place identity can be linked unambiguously. Ambiguous or incomplete matches are not presented as public evidence.
Keep the reporting period
Every measure retains the year, month, rolling period or other period stated by the source. Different periods are not presented as though they were simultaneous.
Use period-correct boundaries for history
Historical Police.uk comparisons are shown only when the matching historical boundary information is available. A current boundary is not silently applied to an older period.
Canada metro data stays metro data
Statistics Canada census metropolitan area measures remain labelled as CMA or policing-adjusted geography. A matching city name does not turn a metro statistic into an exact municipality statistic.
Published indexes keep their meaning
Statistics Canada Crime Severity Index values remain Statistics Canada CSI values. They are not converted into a universal site safety score or treated as the same unit as incident counts.
Older official data stays historical
If a source is no longer current, its valid older observations can remain as historical context but are not relabelled as current conditions.
India reporting years stay explicit
NCRB Crime in India 2023 metropolitan-city statistics remain labelled as 2023 evidence even when the source files were published later. NCRB per-lakh rates are shown as published rather than recalculated into another denominator.
India count history stays count-only
NCRB annual count history is compared only on a like-for-like basis. Historical rates are not reconstructed from an assumed population denominator.
South Australia source geography stays explicit
SAPOL local data are published by suburb or postcode. A profile does not mix those two geography modes, and different local areas are not combined unless the source coverage is unambiguous.
South Australia privacy exclusions stay intact
Where SAPOL withholds local location information for privacy, those records are not assigned to a suburb, postcode or city.
South Australia coding changes are respected
Trend comparisons that cross a documented coding change are limited to categories the source indicates remain comparable. More detailed categories are not silently crosswalked.
South Australia trends are count-only
Like-for-like SAPOL history uses complete count windows. A rate is not inferred where the source does not publish one.
NSW rates keep the source denominator
NSW BOCSAR rates are shown only when directly published for the matching LGA and period. A source value marked not calculable remains not calculable rather than being estimated.
NSW trends are descriptive, not rankings
Published trend labels and percentage changes may provide context. LGA rank positions are excluded, and the site does not turn trends into a safest-or-most-dangerous ordering.
Victoria LGA data stays LGA data
Victoria Crime Statistics Agency statistics retain the Local Government Area and reference period stated by the source. A matching city name does not automatically make an LGA identical to a municipality.
England and Wales CSP data stays CSP data
ONS Community Safety Partnership measures remain labelled as CSP or local-area statistics. Household-denominator columns and percentage-change columns are not presented as population crime rates.
No synthetic universal score
Different crime datasets are not collapsed into an invented 0–100 safety score. A source-published index may be shown only as that source’s own measure.
Publication standard
A profile is indexable only when it has at least 3 source-backed observations across at least 2 distinct measures, at least one official, government, police or statistical source, a checked date, and no unresolved source-integrity issue.
Corrections update the evidence
When a source corrects or refreshes its data, the underlying source-backed record is updated so the public explanation and displayed evidence stay aligned.