copious-git's picture
Card: link each file to its rendered page, and add the press contact
8dba88e verified
|
Raw
History Blame Contribute Delete
6.06 kB
---
license: cc-by-4.0
language:
- en
tags:
- police
- criminal-justice
- open-data
- united-kingdom
- united-states
- australia
pretty_name: Police accountability data (UK, US, Australia)
size_categories:
- n<1K
---
# Police accountability data: United Kingdom, United States and Australia
Nine derived, machine-readable tables on police complaints, stop and search, and deaths following
police contact. 728 rows. Every figure is computed from official open data by code and published
with its source, its period and its denominator attached. **No figure was written, estimated or
rounded by a language model.**
- **Documentation and downloads:** https://www.policecomplaint.com/data/
- **Column reference (data dictionary):** https://police-accountability-data.readthedocs.io/
- **Usage handbook, which calculations are wrong:** https://copious-co.gitbook.io/copious.co-docs
- **Source repository:** https://github.com/copious-git/police-accountability-data
- **Cite the concept DOI:** https://doi.org/10.5281/zenodo.22165476
## Read this before you use it
- **A disparity ratio is a measured difference in outcomes.** It is not proof that anyone acted
unlawfully and it does not establish a cause.
- **A recorded complaint is an allegation, not a finding.** A force that records complaints readily
logs more of them.
- **The UK ethnicity figures use two different bases.** Searches carry *officer-defined* ethnicity;
Census denominators are *self-defined*. People recorded as Mixed are 0.5% of searches against
2.9% of residents, a gap too large to be real. Treat the direction as solid and the magnitude as
uncertain.
- **Every rate ships with its own numerator and denominator.** Dividing by the row total instead
understates the rate on 21 of 41 forces, by up to 24.5 percentage points.
## Contents
Each file has a rendered, human-readable counterpart. If you want to check a figure by eye before
using the CSV, or cite a page rather than a file, use the third column.
| File | Rows | One row is | Rendered as |
|---|---|---|---|
| `uk/stop-and-search-outcomes-by-force.csv` | 44 | one police force | [Every UK force, ranked](https://www.policecomplaint.com/uk/forces/) |
| `uk/stop-search-ethnic-disparity-by-force.csv` | 205 | one force, one ethnic group | [Who gets stopped and searched](https://www.policecomplaint.com/uk/reports/who-gets-stopped-and-searched/) |
| `uk/stop-search-ethnic-disparity-national.csv` | 5 | one ethnic group, England and Wales | [Who gets stopped and searched](https://www.policecomplaint.com/uk/reports/who-gets-stopped-and-searched/) |
| `uk/complaint-review-outcomes-by-force.csv` | 44 | one force | [Every UK force, ranked](https://www.policecomplaint.com/uk/forces/) |
| `uk/deaths-following-police-contact-by-year.csv` | 21 | one financial year | [Deaths during or following police contact](https://www.policecomplaint.com/uk/reports/deaths-following-police-contact/) |
| `uk/deaths-following-police-contact-by-category.csv` | 6 | one IOPC category, plus a TOTAL row | [Deaths during or following police contact](https://www.policecomplaint.com/uk/reports/deaths-following-police-contact/) |
| `us/police-killings-by-state.csv` | 51 | one state, including DC | [People killed by police, by US state](https://www.policecomplaint.com/states/) |
| `us/police-killings-by-agency.csv` | 208 | one department | [US departments, ranked](https://www.policecomplaint.com/states/agencies/) |
| `australia/complaints-and-deaths-in-custody-by-state.csv` | 144 | one state, one year | [Australia, by state and territory](https://www.policecomplaint.com/australia/) |
To put two UK forces side by side on the same measures without writing any code, there is a
[force comparison tool](https://www.policecomplaint.com/uk/tools/compare/).
## Headline figures
**United Kingdom.** 1,397,874 recorded stop and searches across 41 of 44 forces. Nationally, Black
people accounted for 18.2% of searches with a recorded ethnicity against 4.0% of residents, a ratio
of 4.5, over 1,151,285 searches in 36 publishable forces. 24.5% of completed complaint reviews found
the force's own handling was not reasonable and proportionate. 3,666 deaths during or following
police contact, 2004/05 to 2024/25.
**United States.** 15,550 people killed by police 2013 to 2026. 267 cases (1.7%) led to a charge,
91 to a conviction or guilty plea, 40 to an acquittal.
**Australia.** Complaints per 100,000 people in 2024-25 ranged from 6.3 in Victoria to 97.6 in the
Northern Territory. 445 deaths in police custody, 93 of a person recorded as Aboriginal or Torres
Strait Islander.
## Coverage gaps, which are real
Greater Manchester, Gwent and British Transport Police publish no stop-and-search data at all.
Five further forces record ethnicity on too few searches to support a ratio and are marked
`publishable=false` with the reason. Scotland and Northern Ireland are absent because Police
Scotland and the PSNI do not report into these datasets. Ireland has no dataset: we checked all
12,985 Central Statistics Office tables.
## Sources and licences
data.police.uk, the Independent Office for Police Conduct, the Home Office and the Office for
National Statistics under the Open Government Licence v3.0; the Productivity Commission's Report on
Government Services 2026 under CC BY 4.0; Mapping Police Violence under the publisher's own terms;
US Census Bureau population estimates in the public domain. Only aggregates are redistributed; no
source's record-level data is republished.
## Who publishes this
Copious Ltd (England and Wales, company no. 11437826), which publishes
[PoliceComplaint.com](https://www.policecomplaint.com/) and also operates a claims referral service.
The data is published openly under CC BY 4.0 regardless.
**Journalists and researchers.** Figures, methodology notes and a named contact are on the
[press page](https://www.policecomplaint.com/press/). If a number here looks wrong, say so and it
will be corrected in public with the correction recorded, as two already have been.