Britain Is Safer Than You Think

Closing the Perception Gap

This is a critique of Fraser Nelson’s 5 August 2025 claim that Britain is safer than the public mood suggests, and of the counter-claims that followed. It’s part of a series I’ve been writing for months on the perception gap in crime, where a handful of high-visibility offences distort public understanding of long-term trends. I’ve coined a few tropes for this along the way, which I’ll reference here. Before the predictable lines of abuse, ‘go home’, ‘not your country’, ‘you don’t get a say’, save them. I’m not here to tell you your feelings are invalid; I’m here to show you the reality behind them.

A recurring pattern is the perception gap, due to a handful of highly visible types of crimes, such as shoplifting, phone thefts, and graffiti, which are amplified until they become emblematic of national decline. This happens even as record numbers of people report feeling safe walking alone at night. The narrative is then reinforced by framing certain images or situations, such as asylum-seeker accommodation or the government’s inability to stop irregular boat crossings, as proof that authority has collapsed. Add to this the constant drip of politically charged media content portraying Britain as broken, and the public mood drifts ever further from what long-term crime data actually shows.

The cornerstone of this critique is the Crime Survey for England and Wales (CSEW), a continuous national victimisation survey running since 1982. It measures whether people have been victims of crime, and whether those incidents were reported to the police. This matters because police-recorded crime is easily distorted by changes in law, recording practices, and reporting behaviour, whereas the CSEW’s core methodology has remained stable for decades. That stability is what allows us to compare like with like across time.

On that measure, the drop is stark. Violent crime, as measured by the CSEW, has fallen from an estimated 3.8 million incidents in 1995 to around 1.2 million in 2024, roughly a 70% decline. Burglaries have dropped from over 2.5 million a year in the early 1990s to under 700,000 today. Car thefts, once above 1.5 million annually, are now under 450,000. The total number of people experiencing any form of crime in a given year has more than halved, from nearly 20 million incidents in 1995 to fewer than 8 million today. These aren’t fluctuations, they are long-term structural changes in patterns of offending and victimisation.

The critics argue that such surveys miss entire categories of crime: shoplifting, business-related offences, and certain frauds. This is true, but it is by design, not deceit. The survey’s remit is personal victimisation. Fraud and cybercrime are measured through a separate ONS module, which estimated around 3.5 million incidents in 2024, still below the combined total of fraud and computer misuse measured in 2017, showing no hidden explosion of crime omitted from the leading figures. Business crime is tracked through other statistical channels. Criticising the CSEW for not counting these is like faulting a thermometer for not measuring rainfall. The scope is different; the stability of the trend within that scope remains.

Others focus on the issue of under-reporting, especially in sensitive crimes such as sexual offences and domestic violence. Under-reporting exists, but the survey uses anonymous self-completion modules, sections where respondents privately read and answer sensitive questions on a tablet or form without the interviewer seeing their responses. The purpose is to improve disclosure rates in areas like domestic abuse, sexual offences, or drug use. This method is specifically designed to reduce fear, shame, or social pressure, meaning that any under-reporting is consistent across decades and cannot be used to dismiss the clear long-term downward trend. If, in 1995, roughly 1 in 5 victims of domestic violence were disclosed in the survey, and today the proportion is similar, then a reduction in the survey figures still indicates a real-world drop. In fact, the CSEW shows that the proportion of women experiencing domestic abuse in the last year has fallen from around 8% in the mid-2000s to about 5% in 2024.

Some critics attempt to sidestep this by focusing on small-area crime maps, citing ‘high crime’ rates in specific neighbourhoods. These are often measured using Lower Layer Super Output Areas (LSOAs), small zones of 1,000–3,000 people, or the slightly larger Middle Layer Super Output Areas (MSOAs), which contain around 5,000–15,000 people. They’re useful for detailed analysis, but cherry-picking the highest-crime pockets from these zones and presenting them as representative of an entire town or city is deeply misleading. It’s a common scare-mongering tactic: zoom in enough on any country, and you will find somewhere that looks “dangerous,” even if the broader trend is the opposite. 

The criticism of the CSEW’s reliance on MSOAs and LSOAs, statistical zones covering hundreds of households, drawn to keep population counts consistent over time, not to pinpoint crime clusters. Crime isn’t evenly spread: some streets bear the brunt while others see little at all. But that’s the point, the survey measures trends, not micro-hotspots. If you want hyper-local data, you turn to police records. In a city of 250,000, just two LSOAs (under 2% of the population) might have burglary rates triple the national average, while the other 98% live at or below it. Holding those two pockets up as “the city” is statistical theatre. For the politically savvy, think of MSOAs and LSOAs as the crime-data equivalent of electoral boundaries, functional units that get redrawn as populations change, not moral verdicts on an area’s character.

There are also claims that by capping repeat victimisation, counting whether someone was a victim rather than the total number of times, the survey conceals the true scale of crime. In reality, capping prevents statistical distortion. Without it, a handful of heavily targeted individuals could inflate national totals by millions of ‘extra’ incidents. The uncapped figures are published separately and show the same broad downward trend: violent incidents down from over 4.2 million in 1995 to around 1.4 million today.

Take, for example, the arrests of 466 people in a single day (10th August 2025) under terrorism legislation, one of the largest mass detentions London has seen in over a decade. Many were elderly, present only with placards expressing solidarity, not conducting or planning violence. Most were released on street bail a few hours later. This moment isn’t a reflection of elevated terrorism risk; it’s a snapshot of a justice system stretched and deployed for symbolic effect. It warps statistics, generates public fear, and yet, by design, will not be reflected in personal-victimisation surveys, nor should it be. So, let’s put that fantasy to bed once and for all: Public safety isn’t under siege because of mass arrests; it’s being distorted by misapplied power and amplified narratives.

Even the temporary loss of ‘National Statistics’ status during the pandemic is sometimes used as ammunition, as though the survey’s credibility evaporated. The truth is procedural: when COVID-19 halted face-to-face interviews in 2020, the survey switched to telephone methods, which are not directly comparable to earlier data. The ONS correctly paused the designation, then restored it as in-person interviews resumed in 2022. There is no evidence that this break altered the underlying pattern; the 2021–22 telephone data still showed violent crime rates less than half those of the 1990s peak.

If we look for independent validation, we find it in NHS hospital episode statistics. In 1995, assault-related admissions were over 80,000 per year. By 2023, they had fallen to around 40,000, the lowest in 25 years. That’s not a measure of paperwork; it’s a measure of people. The downward trend perfectly matches the CSEW’s data: two completely separate systems telling the same story. Yet the stubborn belief that crime is rising not only misinterprets reality but also harms it. When public perception is so far removed from the underlying evidence, the consequences ripple through politics, policing, and society in ways that can make us less safe over time.

THE IMMEDIATE CONSEQUENCE

The most immediate impact is on policy. A public convinced it is living in a lawless country demands visible action, and politicians oblige, some more than others who have an axe to grind. That often means headline ‘crackdowns’ and quick-response measures, such as more stop-and-search, harsher sentencing, and saturation patrols, at the expense of the slower, less visible prevention initiatives that truly reduce crime over the long term. The political focus shifts to the next press conference, rather than the next decade’s trend line, which exacerbates the emerging perception gap.

The perception gap is the gulf between falling victimisation rates and the belief that crime is spiralling. Shoplifting, phone thefts, graffiti, and a few other visible offences are amplified until they become symbols of national decline. The same dynamic frames asylum-seeker hotels or boat crossings as proof of a collapsed state. Add a constant drip of politically loaded media and social content, and the mood drifts further from what the data shows. I call the political use of this gap data demagoguery.

Data Demagoguery (noun) -
The calculated manipulation of statistics or numerical data to inflame public emotion, reinforce ideological beliefs, and divert attention or scrutiny away from those in positions of power.

---

The perception gap matters because it has real-world consequences. If people believe the police and courts are incapable of dealing with crime, they are less likely to report offences, cooperate as witnesses, or share intelligence. That hampers the essential information flows crucial for effective policing, and it makes it easier for political actors to claim the system is broken and to push extreme reforms without proper scrutiny.

Recent police funding debates illustrate how this perception gap feeds on institutional strain. Despite a £1 billion increase in funding for England and Wales, taking total police budgets to around £19.5 billion, senior officers warn it’s barely enough to cover inflationary pay rises. The National Police Chiefs’ Council projects a £1.2 billion shortfall over the next two years, potentially equating to thousands of lost officers. This kind of fiscal reality, set against rising public expectations and a backdrop of high-profile incidents, makes it easier for critics to frame policing as under-resourced and ineffective, regardless of the broader, more positive crime trends Nelson highlights.

The toxicity of the perception gap spills over into social relations. Fear of crime can be easily channelled towards certain groups, young people gathering in public spaces, migrants, or residents of particular neighbourhoods, even when hard data shows no disproportionate offending. This scapegoating shifts attention away from the structural causes of crime, erodes social cohesion, and deepens mistrust between law enforcement and the very communities whose cooperation they need.

Resource allocation also suffers. When public anxiety fixates on dramatic but rare crimes, police resources and political will can be diverted from tackling the less visible but far more prevalent harms. Domestic abuse, fraud, and online exploitation now account for a significant proportion of overall victimisation, yet these offences rarely dominate headlines or social media feeds. They lack the visceral imagery of a shop robbery caught on video. In an attention economy, that means they too often get sidelined, even though addressing them would yield greater real-world safety gains.

Over time, the narrative of perpetual crisis becomes self-sustaining. People have been told year after year that their country is on the brink of lawlessness, and they have begun to internalise that belief, even if they have never been victims themselves. This perception corrodes civic engagement, weakens community bonds, and undermines the local problem-solving that genuinely improves safety. It also robs policymakers of the political space to acknowledge progress, trapping them in a posture of permanent crisis management. The cycle, media alarmism, fearful polling, and opportunistic political rhetoric feed themselves endlessly, each element reinforcing the next. In such an echo chamber, statistical reality struggles to be heard, and volume, not veracity, becomes the currency of influence.

CRISIS ACTORS PLAYING THEIR PART

This gap between reality and belief doesn’t appear spontaneously; it’s actively fed. Perception doesn’t develop in a vacuum. It is shaped, reinforced, and monetised by a network of actors who, for different reasons, benefit from the public feeling unsafe. Understanding how this system operates is crucial to bridging the gap between reality and belief.

The first actors are the traditional media. Crime stories, especially those involving violence, vulnerable victims, or brazen daylight offending, attract attention and engagement at levels few other subjects can match. Television news still leads with the most shocking cases, regardless of how rare they are statistically. Tabloids put their worst examples on the front page. Even broadsheets, when readership is at stake, are not immune to the lure of a security camera still showing a dramatic theft. In this economy, ‘drama is currency,’ and calm is worthless, and this is exactly why the gap between Britain’s relatively low levels of victimisation and the public’s sense of danger persists. Fraser’s claim that the country is safer than it feels cannot survive unscathed in an environment where the rare is treated as the routine.

The second actors are the political voices. Election cycles particularly trigger a rise in law-and-order rhetoric. Candidates and parties can rally support by promising to ‘take back control’ or reverse a supposed wave of crime. Public polling, influenced by media coverage, acts as both justification and motivation for more speeches, policy announcements, and reactive legislation. Here again, perception becomes policy, and policy, in turn, reinforces the perception, a cycle that works directly against the data-led case for a calmer view of crime trends.

The third type of crisis actors builds on the reach of the first two but operates in a unique way: social media. Where the press has deadlines and politicians have election timetables, social media works in real time, without filters, fact-checking, or editorial restraint. Platforms are engineered to reward engagement, and nothing engages like outrage or fear. A short clip of a brazen shop theft or a confrontation on public transport can reach millions within hours, stripped of context but laden with emotional punch. For viewers already primed by media headlines and political speeches to believe crime is spiralling, such footage feels like undeniable proof. Local community groups, neighbourhood ‘watch’ pages, amplify the same effect at street level, reframing even petty disputes as evidence of societal collapse. This immediacy and amplification make social media not just a mirror of the first two crisis actors, but a multiplier, one capable of burying the statistical reality Fraser points to beneath a flood of curated alarm.

Economic conditions and cognitive bias act as powerful filters. Rising living costs or visible inequality prime people to interpret boarded-up shops, rough sleepers, or litter as signs of disorder. Once convinced crime is out of control, people remember only what supports that belief and dismiss contrary evidence. Online algorithms feed this bias because outrage keeps users engaged.

Economic force - people tend to view public spaces through a lens of insecurity. A boarded-up shopfront, a rough sleeper, or litter in a high street becomes a symbol of disorder, regardless of whether crime rates are rising or falling. These signals, the ‘broken windows’ of modern Britain, influence emotions in ways no statistical chart can suppress. For the media, politicians, and social media platforms, such images are ready-made props in their preferred narratives.

Cognitive Bias force - once convinced that crime is out of control, individuals tend to seek and remember evidence that supports this view. Incidents that contradict it are dismissed as anomalies or reinterpreted as proof that authorities are hiding the truth. This ‘kindness’ towards one’s own bias is a deeply human trait; it keeps a worldview intact, even at the cost of accuracy. Online, where algorithms feed users more of what keeps them engaged, that bias is not just a tendency; it becomes a design feature, ensuring that fear, outrage, and grievance remain in constant supply.

All of this creates a self-reinforcing cycle. A visible incident triggers media coverage, which in turn fuels political rhetoric. This rhetoric encourages social media amplification, disseminating the imagery further. As audiences affirm their beliefs, polling moods shift, and media and politicians cite the polls as proof that their concerns are justified. Each participant can convincingly claim to respond to public sentiment, yet all of them are, in turn, feeding it.

The result is a perception so well-established that it seems immune to data. A decade-long decline in victimisation can be wiped from public memory by just a couple of weeks of sensational headlines and viral videos. And although these actors may not work together officially, their motivations align: attention, support, and engagement go to those who can make the country seem less safe, not more.

Which leaves the uncomfortable question: how do you break a cycle where every major participant has an incentive to keep it going? Can it be fixed at all, or is the best we can do to make its distortions visible enough that they lose their power?

HOW TO FIX THIS - OR CAN IT BE FIXED?

Having written in depth, and if you are still reading, I cannot leave this opportunity to suggest ways to fix these perception gaps, gaps that, if left unchecked, will continue to warp the public mood and drive policies untethered from reality. This is the crux of the debate sparked by Fraser Nelson’s claim that Britain is safer than the prevailing mood suggests.

But before that, a home truth needs repeating. Humans are inherently irrational and can be manipulated if the right levers are pulled. That’s an immutable fact. Deny it, and you’re already halfway lost.

The machinery of fear is self-sustaining; the challenge is to break its gears without pretending fear itself is irrational. People don’t need to be told their feelings are wrong; they need reasons, visible and concrete, to see that reality is better than the mood suggests.

That means action in three areas:

1. Give data a human scale. Telling people violent crime is down 70% since the mid-1990s is abstract; showing that their personal risk is a fraction of what it was when they were at school is tangible. Comparing Britain’s low victimisation rates internationally reframes the story, though in the UK context, the trade-off with freedoms in places like Qatar is worth stating.
   

   But in the UK context, this gets trickier. Our safety index still sits well below that of countries like Qatar. The reason? Trade-offs. Qatar’s safety comes with heavy restrictions on personal freedoms, restrictions that would never fly here. That’s the cost-benefit calculation almost nobody mentions, yet it’s central to an honest debate about what “safe” really means.

2. How institutions respond – moving from reactive optics to sustained prevention.
   Institutions have to bridge the gap between falling crime rates and public faith in justice. People tune out the statistics when they believe offenders face no real consequences. That perception isn’t plucked from thin air; charge rates for many low-level offences have indeed dropped, but it’s not proof that crime itself is rising. The fix is targeted, visible enforcement paired with equally visible problem-solving. It’s not about waving spreadsheets to ‘prove the numbers’; it’s about proving, in plain sight, that the system still works.
   

3. How communities reclaim their narrative – communities must reclaim their own narrative. This will never be fully ‘solved’; there will always be a vocal minority that thrives on chaos and division, but their influence can be blunted by starving them of an audience. The antidote is to disrupt the online and offline echo chambers that feed fear. Local authorities, police, and civic groups should seed the same networks that circulate warnings and grainy CCTV clips with context and success stories: months-long drops in burglaries, the dismantling of local gangs, or the quiet success of prevention schemes. When people see their own neighbourhoods getting safer, they rely less on national headlines to tell them how to feel.
   
   Breaking the cycle also means realigning incentives for the very actors who currently profit from the perception gap. Media outlets could choose to place crime trends in proper context as a matter of editorial standard, just as they now routinely include background in health reporting. Politicians could resist the short-term gain of fear-driven pledges in favour of taking credit for sustained, evidence-based improvement. Social media platforms could be pressured, or regulated, to give greater reach to content that informs rather than inflames.

   Britain’s greatest crime challenge isn’t reversing a rising tide; it’s convincing its citizens the tide already went out. And that holding the line depends on rejecting the stories that tell them otherwise. But when fear detaches from reality, it stops being a safeguard and becomes an obstacle to further progress. This is where Fraser Nelson’s claim, that Britain is safer than the mood suggests, deserves to be understood in full.