Selected coverage
Preliminary stance distribution
A neutral headline can conceal a strongly evaluative article. These labels use only the RSS headline and any summary the publisher supplies.
Tool archive
Media perception monitor
Coverage of the police is measurable, but not with one persuasive-looking sentiment score. This monitor keeps publisher, article stance, keywords, public opinion and significant events separate, then lets readers compare them with the source evidence still attached.
Members · free
Sign in to unlock the media perception monitor
Sign-in takes under a minute: create a password or use Google, and you’re in. One free sign-up unlocks everything members get:
- Save articles and practical summaries to your library
- Download cited research briefings
- Use the Research Assistant — free
Live evidence monitor
Media coverage, public confidence and the events between them
Police-related reporting from across the UK media is collected every six hours, classified by its stance towards policing and retained with its source evidence. Public-opinion measures and dated milestones are shown alongside it, but never treated as proof that coverage caused a change in trust.
Last feed refreshLoading…
Articles retained—
Healthy feeds—
Coverage begins—
What is recurring
Keywords and topics
Loading the latest feed data…
Select a term to filter the evidence records that produced it.
Most recent
Articles in the selected corpus
No article currently matches these filters.
Parallel trends
Coverage records and public opinion
Monthly stance balance
The balance is favourable minus critical articles as a share of classifiable coverage. Collection starts at launch; no historical media values are fabricated.
CSEW public perceptions
Percentages use the survey’s own measures. The 2021 and 2022 gap records the pandemic interruption; the line does not bridge it.
Dated context
Events that may help frame a hypothesis
Black Lives Matter protests
George Floyd’s murder and the protests that followed changed the volume and focus of policing coverage in the UK.
Open primary or official source ↗Sarah Everard and Wayne Couzens
Separate markers are needed for the disappearance, the policing of the Clapham Common vigil, the guilty plea and sentencing.
Open primary or official source ↗CSEW interruption
Comparable annual perception estimates are missing during the pandemic. The chart must show the gap rather than interpolate it.
Open primary or official source ↗Elon Musk agrees and completes purchase of Twitter
Record the agreement and completion separately when testing changes in the social-media environment.
Open primary or official source ↗David Carrick and the Casey Review
A criminal case and an institutional review are distinct events and should not be collapsed into one marker.
Open primary or official source ↗Southport attack and subsequent disorder
Track the attack, misinformation and disorder as separate linked events across England and Northern Ireland.
Open primary or official source ↗Audit trail
Every visible trend resolves to article records
| Date / source | Article | Role | Stance | Confidence | Record |
|---|
Public opinion
Separate series for separate questions and jurisdictions
Local police doing a good or excellent job
49%Year ending March 2025
Question and limits
Taking everything into account, how good a job do you think the police in this area are doing?
The 2026 dataset was released on 23 July 2026; this displayed point remains the last fully checked value in the monitor.
Confidence in local police
67%Year ending March 2025
Question and limits
Composite confidence measure reported by the CSEW.
Question wording and comparability notes must travel with the series.
Trust in the Metropolitan Police Service
66%Q3 2025/26
Question and limits
Believes the MPS is an organisation they can trust.
MOPAC surveys 19,200 London residents per year; the displayed quarterly result is not a GB measure.
Police do a good job locally
43%Q3 2025/26
Question and limits
Believes the police do a good job in their local area.
Keep separate from trust in the MPS.
Public perceptions of policing
Tracked2024/25 release
Question and limits
Measures and denominators vary across the SCJS tables.
Shown as a tracked source until a like-for-like measure is selected and checked.
Policing Plan Survey
Tracked2025
Question and limits
Use the published survey wording and base for each selected measure.
Kept jurisdictionally separate from England and Wales, Scotland and GB polling.
Trust in the police to tell the truth
Tracked2025
Question and limits
Trust to tell the truth is not the same construct as local police confidence.
Displayed separately; never spliced into the CSEW series.
Source register
Publishers are not assigned political labels by this tool
Where a publisher describes itself as liberal, progressive, left, socialist or politically non-aligned, the register records and links that wording. Otherwise it says no left/right self-description was found. That is evidence of how a publisher presents itself, not an independent finding about political position.
What is stored
Canonical URL, publisher and feed, timestamps, jurisdiction, relevance role, preliminary stance, confidence, triggering phrases, topics, keywords and a concise rationale. Full article bodies are not copied.
How stance works
Rules look for evaluative phrases attached to a policing target. Favourable and critical evidence in the same feed text produces a balanced label. No trigger produces neutral, not favourable.
What can fail
Feeds can disappear, summaries can be thin, duplicated stories can survive and a headline can misrepresent the full article. Feed failures remain visible and low-confidence labels need review.
What the comparison cannot prove
Coverage and polling can move at the same time for many reasons. A causal claim would need a separate design with exposure, timing and plausible alternative explanations tested.
Historical baseline and method note
The earlier 2000–2025 series remains the historical baseline in the linked data story. It is largely Guardian-led, uses a headline lexicon and calculates source diversity from a capped provenance sample. It must not be presented as the same measure as this multi-source, target-aware monitor.
Read the historical data story →