UK Sexual Offence Density Atlas

About & limitations

• What is counted. Records in the police category Violence and sexual offences. This combines sexual offences with all other violence (assaults, harassment, stalking and more). Sexual offences are only part of it, and the source data cannot separate them.
• Reports, not prevalence. Only offences reported to and recorded by police appear. Many sexual offences are never reported. Changes over time also reflect recording practice and willingness to report, not just underlying behaviour.
• Counts and rates. By default cells show counts, and busy, populous and night-time-economy areas record more simply because more people are present. For England and Wales you can switch to a rate per 1,000 residents, or per 1,000 workday population (people counted where they work if employed, otherwise where they live), both from Census 2021. Neither is perfect: residents understate busy centres, workday population misses visitors and nightlife, and both are from 2021. A cell needs 1,500 people for a rate, and a rate is not shown where the force has data gaps. Scotland and Northern Ireland have no cell rates.
• Anonymised locations. Police publish approximate points that are snapped to the centre of a street or a place. Here they are grouped into hexagons about 4 km across, and cells with fewer than 3 records are hidden. Nothing here identifies a victim, an offender or an address, and it must not be used to try.
• Coverage gaps. Some forces did not publish for some months, and the map would otherwise read as “low”. Greater Manchester Police has almost no street-level data after June 2019. Check the Data status panel and the banner for the current list.
• Geography. Covers England, Wales, Northern Ireland and British Transport Police points. Scotland is shown differently. There is no open street-level feed, so Scotland appears as 32 council-area shapes (Scottish Government, Police Scotland returns, financial years). It uses the Scottish definition of sexual crimes, which is narrower than the combined police.uk category, so it is not comparable with the hexagons and must not be added to them or ranked against them.
• Sexual offences by partnership (optional layer). The green layer uses the Home Office table of police recorded crime by community safety partnership, which is one or more local authorities. It counts the offence group Sexual offences only, quarterly, from April 2020, and shows a rate per 1,000 residents (Census 2021). It is a different measure from the orange cells and from Scotland, so the three are never added together or ranked against each other. It leaves out British Transport Police and a few partnership names that could not be placed on a boundary. Figures are revised each quarter.
• Trend charts. Select a cell, force area, partnership or Scottish council area to see a chart of its history in the popup. It goes back as far as this atlas has records: monthly from September 2016 for cells and force areas (the police.uk archive used here), annual Home Office sexual offences for force areas from 2017-18, quarterly from April 2020 for partnerships, and annual from 1996-97 for Scotland. Charts ignore the time window you choose, and each uses its own scale. A grey mark on a cell chart means fewer than 3 records (hidden); a red mark means the force published nothing that month. The cell chart loads all monthly files the first time you open one, about 7 MB before compression.
• Cross-checks. Force totals are compared with the Home Office police recorded crime tables (violence against the person plus sexual offences, which is the same scope as the police.uk category). Where they agree, the popups say so; where a force has gaps, the popups show the Home Office total instead of guessing. The Home Office also shows that sexual offences are roughly a tenth of this combined category, though the share varies by force. That is a force-wide figure and is not applied to individual cells. Exact duplicate rows are removed, and rail points inside Scotland are left out of the cells.
• Lag. Data is published monthly, normally one to three months behind. Totals for recent months can still change.

Source code and method: github.com/z7z8nbry4t-glitch/usoda. If you or someone you know has been affected, support in England, Wales, Scotland and Northern Ireland is listed on the front page, including Rape Crisis and NHS help after rape and sexual assault.