Held to
ACCOUNT

1.1Most Britons think crime usually has no consequence

“Nothing happens”: a common view of crime today

Fewer than one in five Britons (18%) think that people who commit crimes usually face real consequences. Half (51%) believe consequences follow only occasionally (34%) or almost never (17%) – a widely held sense that crime in Britain has become consequence-free, and one that drives mistrust in the justice system. This view is shared across divides; among no party’s supporters does a majority believe consequences are the norm. The gaps that do exist run by age, education and politics: only 11% of over-65s think offenders usually face real consequences against 35% of 25–34 year-olds, non-graduates are twelve points more pessimistic than graduates, and Labour supporters (34%) stand apart from Conservatives (18%) and Reform supporters (11%), a quarter of whom (24%) say consequences almost never follow – the highest of any group.

This perception is largely borne out by reality. Around two-thirds of crime measured by the Crime Survey for England and Wales is never reported to the police, and only 7.3% of victim-based recorded offences resulted in a charge or summons in the year to March 2026¹ – and, as we report below, only one in ten of those affected by crime in our own survey saw the perpetrator convicted. For most crime, the belief that consequences rarely follow is a realistic reading of how the system currently performs. The crisis this chapter charts is therefore as much about delivery as perception.

KEY FINDINGS

This idea that crime has become consequence-free, and that nothing happens when crime is reported, was reinforced repeatedly throughout our focus groups and 1-to-1 interviews. Voters from across the political spectrum repeatedly said, unprompted, that offenders are simply ‘getting away with it’ – particularly low-level crime like shoplifting and anti-social behaviour. Many participants claimed to have seen this happening first-hand in their local communities. Frustration was rooted both in the lack of offenders being caught, and a perception of weak punishments for those who are. Several respondents also claimed that theft of under £200 in value has been decriminalised, driving rising shoplifting – a misconception, though one rooted in a real change to how such cases were handled.² Overall, we found that the crisis of confidence in criminal justice appears to be feeding a wider sense of national decline, and eroding faith in British politics and democracy. The frequent sight of crime going unpunished risks feeding a wider sense that the social contract is breaking down.³

One Cardiff participant, who works shifts in her local Co-op, described watching this play out with a police station on the same street:

The frequent sight of crime going unpunished risks feeding a wider sense that the social contract has broken down.

Victims confirm a lack of consequence for perpetrators

To inform this report, we asked our polling sample about their direct lived experience of crime as victims – either directly, or through close friends and family.

The accounts of those actually affected by crime bear the perception out. Of those affected in the past three years, seven in ten (69%) reported the crime to the police – meaning almost a third of crimes never entered the system at all. For those who did report, only 14% – one in seven – said it ended in a conviction, and more than a quarter (27%) said nothing happened to the perpetrator at all. Taken together, fewer than one in ten of all those affected by crime saw the perpetrator convicted. These accounts sit consistently alongside the official figures: in the year to March 2026 only 7.3% of victim-based recorded offences resulted in a charge or summons – and conviction comes later still, for fewer cases again.

In our focus groups and interviews, victims described difficulty reporting crime at all, reports being ignored or “shrugged on”, and – repeatedly – hearing nothing about what happened to the perpetrator, leading them to assume nothing did. Greater communication and visibility of convictions and sentences may address a real source of mistrust here, independent of the outcomes themselves.

A perception fed by headlines as much as lived experience

Two in three Britons (64%) think crime is rising nationally, but fewer (47%) think it has risen in the place they live. The gap holds for every offence tested and narrows to single figures only for burglary, car theft and vandalism, the crimes people are most likely to encounter directly. The sense that crime goes unpunished is drawn as much from the national story people consume as from what they see on their own street – a theme we return to in the next section, because it is confidence in the system, not just perceptions of crime, that is in crisis.

The official data supports parts of this picture but not others. According to the Crime Survey for England and Wales, headline crime is 14% lower than a decade ago – an estimated 9.6 million incidents in the year to March 2026, against 11.2 million in the year to March 2017.⁴ However, on fraud and theft the public's perception tracks more of the data: fraud is up by around a third on a decade ago, theft from the person rose sharply after the pandemic, and police-recorded shoplifting (507,000 offences in the year to March 2026) remains 42% above its pre-pandemic level. The data does not, however, support public perception of rising violent crime: knife-enabled offences fell 8% in the latest year, knife homicides are at their lowest level since 1991, and homicides overall are at their lowest since current recording practices began in 2003. As such, we can see that for more 'visible' street crimes, public perception is closer to reality; while on rarer and less visible crimes like knife crime, their pessimism appears out of step with available data, perhaps due to significant media coverage of these crimes.

1.2Britons show high fear of victimhood, especially women

Most Britons expect to be victims of crime within a year

Nearly two in three Britons (63%) expect to be a victim of at least one crime in the next year. Asked how likely they are to become a victim of each of eleven crimes, the public put fraud and online scams top at 36%, followed by theft of a phone, bike or other belongings at 35%; half (49%) expect one of fraud, theft or both. A quarter (24%) say “very likely” to at least one crime, and a fifth (22%) expect to be victims of five or more of the eleven crimes listed. Most striking, a third of all adults (32%) expect to suffer a violent or sexual crime in the next twelve months, with 18% thinking they are likely to face a serious assault, 17% knife crime, and 12% a sexual assault.

Fear of violence is concentrated among the young – above all young women

By party, Green (72%) and Reform (70%) voters are the most likely to expect to be victims and Liberal Democrats (52%) the least, but the factors that really matter are age, gender and experience. Fear of fraud and theft is flat across the population: 49% of 18–24 year olds and 46% of over-65s expect one or the other. Fear of violence is much more varied. 56% of 18–24 year olds expect to suffer a violent or sexual crime in the next year, falling steadily with age to 14% of over-65s. Within that, young women and young men fear different things. Half of women under 35 (53%) expect to suffer a violent or sexual crime in the next year, against 39% of men under 35 and 18% of women over 55. More than four in ten (42%) expect sexual harassment and more than a quarter (27%) sexual assault; among women aged 18–24, the harassment figure rises to 54%. Young men are the group most likely to expect a serious assault (27%) or knife crime (22%), roughly double the rate among over-55s.

Four in five victims expect it to happen again this year

People who have already been a victim also expect to be one again: 83% of those who were a victim of crime in the past three years expect to be one in the next year, against 52% of those unaffected, and 56% expect a violent or sexual crime, against 19%. Londoners are the most fearful of violence, with 30% expecting knife crime against 17% nationally. Set against perceptions of national crime, the contrast is sharp: seven in ten (69%) think knife crime has risen nationally, while one in six expect to encounter it themselves – but among 25–44 year olds, Londoners and past victims, that one in six becomes roughly one in four. The picture of rising violence that dominates the national debate is, for most people, something happening elsewhere; for the young and the recently victimised, it is something they expect to happen to them.

Fear runs far ahead of real reported victimhood

These expectations of victimhood run roughly four times ahead of the actual rates of crime reported by the most recent Crime Survey for England and Wales (CSEW): where we found 63% expect to be victims within a year, the CSEW finds 17.1% of people – roughly one in six – actually experienced a headline crime in the past year. The gap holds crime by crime, and for some offences is far wider. Where 36% of our respondents expect to be victims of fraud in the next year, the CSEW records 7.8% of adults actually experiencing it in the past year; where a third expect a violent or sexual crime, the CSEW finds around 1% experiencing violence at the hands of a stranger or acquaintance and 2.1% experiencing sexual victimisation.⁵ The CSEW has its own limitations – it excludes crimes against businesses, under-16s and people not living in households, and sensitive crimes such as domestic and sexual abuse are widely understood to be under-reported even in its self-completion format – so the true gap is likely smaller than the raw comparison suggests. But even so, it is clear that fear of crime is running far ahead of most measured experience. The reasons for this are multiple and complex and not the subject of this report. In any case, as the rest of this chapter shows, that fear is genuine, is not totally detached from reality, and is doing real damage to faith in the system.

1.3A crisis of confidence across the justice system

A perception of failure on all core objectives

There is no core purpose of the justice system for which a majority of voters rate it as doing a good job. The system is rated least bad at punishing people who commit crime (41% say it does a good job on this measure) and protecting the public (38%). It is rated worst at the things punishment is supposed to achieve: 51% say it does a bad job of deterring people from committing crime, 51% say the same of reducing crime, and 47% of reforming and rehabilitating people. On holding offenders accountable – the primary subject of this report – 40% say the system does a bad job, against 33% who say it does a good one.

There is no function of the justice system which a majority of voters rate as doing a good job.

The verdict holds across crime types too. The crimes that the public say are being handled worst by the justice system are everyday, high-volume crimes: shoplifting (55% say the system deals with this badly), phone or bag snatching (54%) and anti-social behaviour (53%). There is no offence a majority believe the system currently handles well; the best rated is serious assault or violent crime, at 38% good.

Among different groups, the voters Labour has lost since 2024 are among the system’s harshest judges on these measures. Among those who voted Labour in 2024 but would not do so now, only 15% think the system does a good job of reducing crime, 13% of deterring crime and 19% of supporting victims – as low as, or lower than, Reform UK voters. Nor is this a story of Labour losing punitive voters to the right: the leavers have gone in every direction, to the Greens and Liberal Democrats as much as to Reform, and their verdict on the system is the same wherever they have ended up. Those who have stayed with Labour, by contrast, are the most positive group in the survey: 61% think the system does a good job of punishing offenders and 46% of reducing crime, double the national figure.

Victims of crime are also more pessimistic than others about the system’s effectiveness at protecting the public, punishing offenders, holding them accountable, and supporting victims – while they are broadly in line with non-victims on the system’s broader record on reducing crime, rehabilitation and making offenders give back.

A majority has low confidence in nearly all justice services

Out of all the main institutions of the justice system, only the police retain the confidence of a majority of the public, and narrowly at 51%. Confidence is lowest for the services responsible for offenders after conviction: 41% for prisons and 38% for probation, with a majority expressing little or no confidence in both areas. Youth justice services record the lowest confidence of any institution tested (33%) – a particular area of low trust, though one in eight say they don’t know, suggesting unfamiliarity may play a part alongside judgement. The courts (49%) and Crown Prosecution Service (47%) fall just short of majority confidence. 

Victims’ confidence in the institutions is lower across the board compared to non-victims, especially for the police (a net rating of −22 among those personally a victim of crime in the past three years, against +12 for those unaffected) and the courts (−15 against +4).

Exploring the link between ethnicity and confidence

‘Ethnic minority’ is a crude banner, combining groups whose experiences of the justice system differ sharply, and other research – including the Crime Survey for England and Wales – finds wide variation between minority communities, with confidence in the police markedly lower among some groups than others. Our own data show the same variation: confidence in the police reaches 75% among Black respondents against 55% among Asian respondents, though on bases of 71 and 76 these splits are indicative only. Our combined ethnic minority base of 199 respondents (out of 1,493) is sufficient to indicate direction, not precise magnitude, and cannot speak for any single community.

With that said, the direction we found with a limited sample size is consistent: ethnic minority respondents as a whole express more confidence than white respondents in every one of the six institutions tested – the gap is widest for youth justice services (53% against 30%) and narrowest for the police (59% against 50%) – and are more positive about every function of the system, most starkly on reducing crime, where 54% say the system does a good job against 17% of white respondents. This is despite the same respondents being far more likely to say the police treat white people more favourably (as shown later in this chapter), and it replicates our 2025 findings in Course Correction.

What might explain it? Some possible explanations are not borne out by our data: we found no clear geographical pattern by region, and no clean age gradient. It may be that comparisons drawn through family or contrasting experience of justice systems abroad set a different benchmark, but our data cannot test this. Further research, with samples large enough to speak to individual communities rather than a combined banner, is needed before firm conclusions can be drawn.

1.4The roots of mistrust: underfunding and unfairness as much as leniency

Perceived leniency is only part of the diagnosis

Perceived leniency is only part of the reason for low confidence – voters also point to underfunding, ineffective delivery and unfairness of treatment. Distrust of the police is led by unfairness: that it “treats people unequally depending on who they are” is the top reason (39%) selected, rising to 48% among those with no confidence at all. For the courts, the main complaint is leniency and delay: nearly half cite a system too soft on offenders (46%) and too slow to resolve cases (43%). But for the parts of the system the public trusts least – prisons, probation and youth justice – the leading criticisms are about failure to deliver and lack of resources. Half of prison sceptics (50%) say the system does not reduce crime or reoffending, and around four in ten cite underfunding for all three services. Across every institution, barely one in twenty say their main reason for low confidence is that the system is too harsh on offenders.

In focus groups and interviews, while perceived leniency did factor, underfunding and poor delivery were just as – if not more – important in feeding a sense of decline and pessimism about the capacity of justice services and politicians to make improvements.

Low confidence in prisons rests mainly on the judgement that prison does not stop people reoffending. Fewer than three in ten Britons (29%) think going to prison makes someone less likely to commit another crime, while 28% think it makes them more likely and 37% think it makes no difference either way – so two in three voters believe prison either fails to reduce reoffending or actively worsens it. Yet this is not read as a case against prison: 87% say keeping the public safe from dangerous offenders is a very important priority for reforming the justice system, and support for prison's use remains solid. Moreover, the most common description of prison's ineffectiveness is that it is too 'easy' and not enough of a deterrent.

People who have been inside shared their dim view of prison's effectiveness, but for different reasons. In partnership with Revolving Doors, a charity working to end the revolving door of crisis and crime, we convened a group of ten people with experience of the justice system, most with long histories of addiction and repeat offending and all now in recovery or working in peer support. Asked what prison had done for them, the most common answer was that it had given them a break from chaotic lifestyles, but a lack of support and treatment meant it did very little to promote real change.

Unfairness and 'two-tier policing'

The clearest agreement across the focus groups was that the system is unfair – above all in favour of people with money and power. Nobody in any room thought the system treated the well-connected as it treats everyone else, and the closing demand in Great Yarmouth, the most punitive room, was as much for honesty and consistency as for severity.

Our polling tested one dimension of this directly. Amid rising public debate over the existence or otherwise of ‘two-tier justice’ in the UK, we found that unequal treatment by the police on the basis of ethnicity is a mainstream view: overall, two in three Britons (66%) believe that the police do not treat all ethnicities equally. However, voters divide almost exactly in half over who is treated more favourably – 34% say ethnic minorities, while 32% say white people. Only one in five believe the police treat everybody the same by ethnicity.

Who is thought to benefit is largely a left/right split: Green and Reform voters are by far the most likely to agree that policing is unequal, in opposite directions (a net rating – white people favoured minus ethnic minorities favoured – of +49 among Greens against −48 among Reform voters). Age divides the same way, with younger voters far more likely to think white people are treated more favourably and older voters the opposite. Ethnic minority respondents are the most clearly of the view that policing is unequal by race: 50% say white people are treated more favourably, against 29% of white respondents.

Voters are frustrated with sentences that don’t mean what they say

Within our focus groups, a final strand of perceived unfairness causing significant frustration and mistrust was that sentences do not mean what they say. This is an issue that has been brought to light by recent Sentencing Act reforms reducing the release point for many custodial sentences to around a third. However, participants also voiced upset with the idea that people 'only serve half' of what is announced, with the part of the sentence technically 'served' in the community not generally understood or acknowledged. In the Great Yarmouth group, when we asked participants for one thing that would restore their trust in the justice system, six of the eight answers were about sentences being served as handed down. Across our groups, there was a greater demand for honest sentences rather than necessarily longer ones. There is significant scope, therefore, for rebuilding trust in the system through clearer and more transparent communication of sentencing decisions.

On the intersection between the debates on immigration and crime, given its prominence in national debate, it was notable how rarely participants reached for this connection. The exception was Great Yarmouth, where migration ran through the discussion from the start – anecdotes of organised shoplifting by “groups of Romanians”, and of police unwilling to act against them, were the group’s main evidence that crime goes unpunished and that enforcement is unequal. In the other groups the same complaints about shoplifting and weak enforcement were made with no reference to migrants, suggesting where immigration is politically salient, it becomes more heavily intertwined with the frustrations illustrated throughout this chapter.

A system that must be held to account too

Much of this report concerns the accountability of offenders – what the public believe it means and what it should require in practice. But the evidence of this chapter points to the importance of the flip side: accountability of the system itself. When voters describe police who never come, charges that go nowhere, institutions nobody trusts to deliver, sentences that do not mean what they say and treatment that varies with money and power, they are describing a system that is not accountable to the people it serves. The two demands are connected and shared across divides: a system that is not seen to be effective, honest or fair has little standing to hold anyone else to account. This was also a view raised in our lived experience focus group: an acceptance that personal accountability matters, but a call for the system to fulfil its part of the bargain. The chapters that follow examine what the public want offender accountability to look like – but the demand for a system that is itself accountable runs underneath all of it.

1.5The change the public want is bold – and balanced

A consistent demand for major change

The crisis of confidence charted in this chapter translates directly into appetite for reform. Almost two in three Britons (62%) say the criminal justice system needs ‘major changes’, rising to 77% among Reform supporters – almost unchanged from our 2025 research with More in Common (64%), confirming a settled public judgement rather than passing anger. Just 3% think the system needs no change at all, and majorities or near-majorities of every party’s supporters want major change, from the Liberal Democrats (50%) and Labour (51%) to the Conservatives (58%).

Punishment or prevention? Britain wants both

The general direction of change is not a simple demand for toughness. Forced to choose between a government focused on tough punishment and one focused on tackling the underlying causes of crime, the public divides three ways and the largest group sits in the middle: three in ten at the tough end of the scale, three in ten at the causes end, and four in ten in the balanced middle – with the most common single answer the exact midpoint, chosen by 28%. Every group we tested averages near the middle, with a national mean of 4.9 out of 10.

Expectations of what change must deliver are equally plural. Keeping the public safe from dangerous offenders stands alone at the top of the public’s priorities (87% rate it a top priority), but reducing reoffending (69%), proper punishment (69%) and offenders taking responsibility (68%) sit in a pack just behind – and asked where new money to cut crime should go, the public’s first choice is youth services and early intervention. The public, in short, want a system that punishes and prevents at once. What they mean when they demand the first of those – accountability for people who commit crime – is the subject of the rest of this report.

↑ Back to contents

2.1The importance of accountability

What does real accountability for crime mean?

Our previous research demonstrated the central importance of accountability to conceptions of justice – and that much frustration with the status quo is rooted in the perceived lack of accountability for offenders, made more visible by everyday crimes like shoplifting going unpunished. 

But what does accountability mean to people? How should someone who’s caused harm be held to account, and how do we know when accountability has been achieved? Until now, very little research has been done into its different potential forms and their relative importance; this report aims to fill that gap.

To structure this section of the research, we tested five different potential 'forms' of accountability that we have heard most often in our work listening to voters, victims, prison-leavers and local communities across the UK. We've called them 'the 5 Rs': repercussions, responsibility, repair, reform and reintegration.

How we tested the framework

2.2The primacy of repercussions and responsibility

Facing a punishment, taking responsibility

Asked to rate the importance of all five forms of accountability together, responsibility came out narrowly on top: 96% rate it important, including 75% very important, with repercussions a close second (at 95% and 72% respectively). Britons think all the 'five Rs' of accountability are important, with no measure scoring under 88%. Just as Britons have high demands of the justice system to achieve multiple outcomes, they also have high demands of people who commit crime: to face a punishment, to take responsibility, to engage in behaviour change, to make amends, and to become a contributing member of society.

The picture sharpens when we force a choice. Asked directly what the phrase “holding someone to account for a crime” means to them, offering descriptors of the 5 Rs and forcing a single answer closest to their view, the largest group (40%) choose “making sure they face punishment proportionate to what they did” – ahead of “making sure they understand the harm they caused and accept responsibility” (29%), “making sure they change their behaviour and do not reoffend” (16%), or “making sure they repair the harm done to victims or society” (9%). Barely one in twenty (4%) define accountability by an offender going on to become a contributing member of society. However, while punishment is the most popular answer, it does not command a majority, and nearly 60% of people select one of the other forms – nor does it achieve a majority among any party's voters: it ranks top for Reform (48%), Conservative (45%) and Liberal Democrat voters (44%), while Labour and Green voters rank responsibility above punishment.

Focus group and 1-to-1 discussions about accountability followed the same sequence. Asked cold, people's first words were most often about punishment and consequences. But as soon as they expanded on what accountability requires, responsibility came to the fore: acknowledging the harm, understanding its impact, and owning what was done. It was responsibility that people kept returning to as the thing that makes a consequence meaningful. This applied across political divides, and also in answers from our lived experience group of people formerly involved in prolific offending.

Punishment as necessary but insufficient

A better understanding of beliefs may be obtained by forcing tradeoffs between options. To achieve this, respondents were shown fifteen pairs of opposing statements about justice – three for each of the five Rs – and asked to place themselves on a 0 to 10 scale between them, where 0 means fully agreeing with the first statement, 10 means fully agreeing with the second, and 5 means sitting in the middle. All fifteen appeared in a single matrix in random order, and which statement appeared first was flipped for half of respondents, so the results are not an artefact of how the pairs were laid out.

The results show strong support for the idea that repercussions – while clearly a priority for most people when pushed – alone aren’t enough. Five of the most strongly held positions in the whole battery said the same thing: punishment on its own does not achieve accountability, and some form of taking responsibility is essential. The two most popular statements in the entire battery were that ‘Accepting responsibility is an essential part of justice, not something that stops mattering once someone is punished’, and ‘Making offenders aware of the harm they caused is essential’. These findings were replicated when we tested a real-life policy scenario relating to prolific offenders, as detailed in Chapter 3.

The chart below shows all fifteen pairs of statements tested, ordered by strength of feeling. Each figure is a mean score on the 0–10 scale between the two statements: above 5 means the public lean towards the right-hand statement, and the further from 5, the stronger the lean.

The most popular statements – making offenders aware of the harm they caused, and that accepting responsibility is essential to justice – were also notable for how many people chose the maximum response of 10 out of 10: 48% for making offenders aware of the harm, and 44% for accepting responsibility. 

Our focus groups and interviews added colour and detail to what responsibility should look like: that it must be genuine, demonstrated over time, and above all remorseful. In every focus group participants raised the importance of offenders being ‘sorry’, and the absence of remorse drew particular anger. This may explain the strength of public upset about the killers of PC Andrew Harper, raised in two of the focus groups: the case is remembered not only for the prospect of the men's early release but, as one Pendle & Clitheroe participant recalled, for their laughing in court.¹ Other specific forms of responsibility raised unprompted across groups included “pleading guilty when you are guilty”, which was also raised in the lived experience group.

‘Hollow punishment’: a served sentence settles nothing on its own

The options in our battery experiment that described punishment as sufficient in itself performed universally poorly. Only around one in ten leaned towards the view that a sentence has done its job once it is served (i.e. only 11% selected 0–4 on the scale favouring that position), that an offender’s debt to victims and society is paid once they have been punished (12%), or that serving a sentence means an offender has been held to account (12%) – and only around one in twenty endorsed any of those positions outright with a 0/10 score. The view that a served sentence settles the account has almost no consistent constituency in Britain. For a large majority, what is missing is for the perpetrator to take responsibility for the harm, repair it, or make something different of their life afterwards.

As one Reform supporter described in a 1-to-1 interview, punishment alone can be “hollow”:

Only around 1 in 10 voters lean towards the view that a served sentence is enough for accountability.

2.3Responsibility is the clearest common ground

Responsibility unites where repercussions divides

If repercussions is where most Britons tend to start, we also found it creates most variance across political divides. On the other hand, a central role for responsibility is where the public converge – it is the clearest common ground. As above, this unity came to the fore in our focus groups and 1-to-1 interviews. It can be further confirmed by observing responses to the 15-question battery by voting group. 

Taking two of the questions that captured responsibility most clearly, we can see all parties averaging around 8/10 in favour, with marginal variance.

Moreover, when plotting all the battery results by voting intention, we find that responsibility-based statements are not only the most strongly held on average, they are also the most unifying across political divides, with a spread of just 0.6 points between the party average responses. This compares with a much more divided response to statements about the role of repercussions. The consensus between Labour, Conservative, Reform and Lib Dem voters on responsibility is particularly striking: a tenth of a point separates them. As across this experiment, Green voters are the notable outliers – however, they are even more in favour of responsibility-coded statements than the other parties.

Strength of feeling on repair (0.8-point spread), reintegration (0.9) and reform (1.1) is also relatively consistent, with Green voters again the outliers. Repercussions is clearly the most divisive theme tested, with a partisan spread of 1.6 – and unlike the others, this one does not depend on the Greens: Reform UK and Lib Dem voters alone are 1.5 points apart.

Undecided voters (not shown above) are statistically indistinguishable from the national average on four of the five values, sitting within 0.06 points on responsibility, repair, reintegration and reform. Repercussions is the only value where they diverge at all, sitting 0.35 points below average, though on a base of 106 that gap is within the margin of error. Non-voters, however, show more variation from the average: those who say they would not vote sit 0.3 to 0.4 points below it on every value except repercussions – including responsibility (−0.41), the value on which every other group agrees – while sitting slightly above average (+0.19) on repercussions itself.

The same general patterns hold when we add splits by education, gender and age. On responsibility-based statements, no demographic group sits more than 0.27 points from the national average of 7.9, and graduates and non-graduates are just 0.12 apart. On statements about the importance of repercussions, the splits by education and age are far more significant: graduates sit half a point below the national average while non-graduates sit a quarter of a point above it, and 18–24s sit 0.7 points below the average against the over-45s around a quarter to a third of a point above. Gender, by contrast, makes almost no difference to views on repercussions. Victims of crime, meanwhile, stay relatively close to average responses, notably scoring slightly below the average on repercussions (−0.14) and for reform statements slightly above it (+0.24).

Lived experience: taking responsibility is about treatment and recovery for addiction

Our lived experience group, convened with Revolving Doors, described taking responsibility above all as addressing the behaviour behind the offending – engaging with treatment, being in recovery from drugs and alcohol, and getting better. Several described recovery as a harder and more genuine reckoning than any sentence they had served, and some said their accountability only became possible once treatment gave them the capacity to face what they had done. They stressed that this requires personal responsibility but also a responsibility for the system to provide the opportunity for change.

2.4Owning, paying and changing

The public's three steps to accountability

A factor analysis of the fifteen statements also tells us something about the framework itself. Looking at how people answered all fifteen statements about what accountability means, their responses grouped statistically into three distinct sets rather than our suggested five. In other words, answers to the fifteen statements rose and fell together in three clusters: if someone agreed with one statement in a cluster, they tended to agree with the rest.

Repercussions held as we hypothesised: the three statements written for it move together and nothing else joins them, which tells us punishment really is a separate idea rather than one end of a single spectrum.

Responsibility holds together too, but it also grows. Alongside its own three statements it absorbs one from each of the other three groups which move together: that punishment alone is not enough and offenders should also repair the harm (which we had labelled repair), that a sentence has only done its job if the person comes out less likely to reoffend (reform), and that what an offender does after release is part of accountability (reintegration).

The remaining six statements – two each from repair, reform and reintegration – mostly moved together rather than being seen as separate ideas. In the public’s mind they are similar ideas, which we might call rehabilitation. The chart below shows how our hypothesised five collapse into three through the lens of public opinion. 

As a result, we can see that the public mainly conceive of accountability as having three key steps: paying for the harm (repercussions), owning the harm (responsibility), and changing their ways (rehabilitation – but also clearly seen as part of taking responsibility). Repair and reintegration are not cast aside, rather they fit within these three steps in different ways.

Having established the three dimensions, we can see how firmly the public hold each of them. As shown below, the pattern reinforces the analysis above: that responsibility is most heavily supported, as we see responses clustering most clearly towards the high end of the scale, with 70% of the public scoring 7 or above on responsibility statements and 53% at 8 or above, while just 10% sit at or below the midpoint. It is the only one of the three whose weight of opinion keeps rising all the way to the top of the scale.

Repercussions is the flattest and the most genuinely divided. Notably, it carries the lowest average score of the three (6.2 – below rehabilitation's 6.4): so while punishment may be the public's most common first answer on accountability, as a strength of belief it is the weakest and most contested of the three dimensions. Just over a third of the public sit at or below the midpoint, while 17% sit at 9 or 10 – two sizeable groups rather than a moderate middle, while party supporters are also most divided here.

Rehabilitation is broadly accepted but relatively weakly held, and less consensual than responsibility. It peaks just above the midpoint, with 40% scoring 7 or above but only 23% reaching 8 – broad permission rather than conviction. Taken together, the three distributions suggest that when it comes to accountability, the public are close to unanimous that offenders must take responsibility, more divided over how much punishment is needed, and open to rehabilitation without demanding it.

The role of ‘making amends’

Of the five forms, repair was the hardest to read. On the survey measures it performs weakly: it is among the least chosen definitions of accountability (9%), and in the factor analysis it splits – understood partly as paying and partly as owning – rather than surviving as an idea of its own. However, our qualitative research told a fuller story: making amends came up often and unprompted in focus groups and interviews – ‘paying back’, ‘giving something back’, unpaid work that benefits the community – suggesting the idea carries more weight in conversation than our quantitative design captured. This suggests the weakness of repair in our survey may have been more a question of survey design.

When specifically testing the most important ways to ‘make amends for what they did’, as part of the repair dimension, we found that completing a drug or alcohol treatment programme comes out as the top answer, above compensation, unpaid work or apologising. This option was framed as applying where offending is linked to addiction, which makes it an easier choice than the others – but the ordering is still striking. It suggests the public hold a broad conception of making amends, one that includes taking steps to change behaviour. It also matches the imperative victims themselves most often describe: ensuring that the offender does not go on to harm somebody else.² Victims of crime in our sample were the group most likely to choose treatment as a way of making amends, at 46%.

Conclusion: what the public mean by accountability

This chapter set out to test what the public mean when they demand accountability. Punishment is where most people start: the most common single definition, and for many the part that must not be missing. But it is nowhere near the whole. Punishment alone, without responsibility, is widely seen as hollow – no party’s voters give it a majority, while it is the most divisive of the dimensions tested.

Responsibility, meanwhile, is where the public converge: strongly-felt, near-universal, consistent across every political and demographic divide, and specific – requiring genuine acknowledgement and remorse, not a performance for the process. All five forms of accountability – the 5 Rs – command strong support, but in the public’s mind they resolve into three key steps: owning the harm, paying for it, and changing behaviour so it does not happen again.

How these principles might be applied in practice is the subject of the next chapter, which looks at the case study area of prolific low-level offending.

In the public’s mind, accountability has three key steps: owning, paying and changing.

↑ Back to contents

3.1A topical case study: prolific offending

The need to find common ground on prolific offending

Record levels of shoplifting and other retail crime have brought prolific offending to the forefront of political debate, driving public anger, mistrust of the system and a sense of consequence-free crime (see Chapter 1). It is notable that campaigns calling for quite different solutions begin from the same starting point: that a small group of around 10% of all offenders account for around half of all offences. Analysis by Revolving Doors and Newton puts the cost of this group at £242 million a year to the criminal justice system alone, and up to £5.2 billion once the wider toll on the NHS, councils and social services is counted – around a quarter of the entire national cost of reoffending.¹ Where campaigns differ is in the response. Some, such as Crush Crime, call for stiffer punishment, such as mandatory minimum and escalating prison sentences.² Others, like Revolving Doors, push for non-custodial alternatives, and diversion out of the criminal justice system into healthcare and addiction services.

But what do the public think should be done? What does accountability look like for this group? Do their principles for accountability identified in Chapter 2 survive contact with a real-life example? And might there be common ground for an effective response that would command public confidence?

This chapter works through the evidence in order: how frequency of offending changes what people want to happen, what naming the role of addiction does to those preferences, how victims differ in their views from non-victims, and then – through a conjoint experiment – what policies actually command support when there are specific trade-offs.

Please note: What follows should be read as reflecting views on prolific low-level and non-violent crime in particular, and should not be read as the public’s view about dealing with crime in general.

3.2Frequency shifts preferences from repair to punishment

Repeat offending triples demand for punishment – but with no majority

To provide a basic baseline understanding of preferences, we asked the public what the single most important response should be for a range of different offending types and behaviours. Beginning with a low-level, non-violent offence, such as minor shoplifting or vandalism, we found that the public rank punishment as fourth most important, chosen by just 16%, while the most popular answer was that the offender should make amends to their victims or community (38%). Notably, this ordering holds right across the political spectrum: making amends is the top choice for Reform UK, Conservative, Labour, Liberal Democrat and Green voters alike, with just eight points separating the parties.

However, when asking the same question for prolific or repeat low-level offending, the response changes markedly, with support for punishment tripling to 49% and becoming the clear first choice (albeit still short of a majority), while making amends falls from 38% to 17%. While punishment will mean different things to different people, and this type of forced-choice question has obvious limitations (hiding strength of support for options not chosen), this is a notable trend, as the offence type remained static and only the description of frequency changed.

Focus group participants confirmed this evidence, with many participants saying that for a first, low-level offence the response should be to make amends and give a warning – while prison should be for repetition or more serious offences.

In every focus group, the demand for escalating consequences for repeat offending came from every part of the political spectrum, including groups like Green voters who tend to be more pro-rehabilitation. As one Liberal Democrat in the North East summed up in a 1-to-1 interview, "the previous consequences did not have an effect, so they should be scaled up.”

A split scenario shows the same shift – with repair holding steady

We found more evidence of this effect by testing two versions of a scenario-based question with two randomly assigned halves of the sample. Both halves read that James snatched a mobile phone from someone’s hand in the street and ran away. In the first version, this was James’ first offence. In the second, James is part of a group that has repeatedly snatched phones and sold them on.

When the offence is a first offence, compensation or unpaid work is by far the most popular response at 63%, with prison the least popular at 17%. However, when the offence is described as repeat and organised, support for a prison sentence more than triples to 61% – from the least popular of the five options to the most popular.

Just as important, however, is what does not change, with support for compensation or unpaid work remaining the second most popular response, falling only six points to 57%; the treatment programme and help into work barely move at all. Indeed, for the more prolific example, respondents selected slightly more responses (2.2 against 2.1 on average). For repeat offenders, the public are not swapping constructive responses for punishment – they are adding punishment while still wanting to keep the rest. This is the ‘punishment plus’ pattern of Chapter 2 visible in a concrete decision. The size of that punitive shift is also where the parties divide most: prison for the repeat phone-snatcher draws 71% of Reform UK voters against 48% of Greens.

By offering up to three responses in this question, we also gained a sense of the combination of responses favoured by the public. For a first offence, the public leaned towards a combination of compensation or unpaid work (63%), a programme addressing underlying causes (46%) and a mediated meeting with the victim (41%) – repair and reform first, with prison the least chosen. While for the repeat offending of the same crime, the favoured combination becomes a prison sentence (61%) alongside compensation or unpaid work (57%) and the same treatment programme (43%): punishment moves to the top of the package, while the constructive elements stay in it.

3.3Context shifts views towards treatment – but not at the cost of punishment

Prison or treatment? Addiction shifts the answer

Two popular potential solutions to repeat offending are either to extend prison sentences for repeat offenders or to divert them instead into intensive supervision and treatment. We offered these two options to the public plainly, while then also testing a version of the question to observe the impact of added context about the characteristics of this repeat offending group.

Asked plainly, the public split down the middle – 50% for prison, 46% for treatment – with Labour voters leaning slightly to treatment, Conservatives slightly to prison, and Reform UK voters choosing prison by 70%. However, when the same question added that many of this group have complex needs such as drug or alcohol addiction, mental health problems or homelessness, the result reverses: treatment leads 57% to 40%, and every group except the Liberal Democrats moves towards treatment. Only for Reform UK voters is there no majority for a treatment-based approach, albeit support still jumps by 17 points in favour of treatment.

How close the plainly-asked result is may reflect that most people sit between these poles and favour an approach that combines both treatment options and escalating punishments for reoffending (as section 3.4 explores). A binary choice has limitations because it makes people who would actually take both, or something in the middle,  give a binary answer. The rest of this chapter explores the nuances that may be hidden by the above findings.

We also ran a second scenario-based split, based on an offender ‘Alex’ who had repeatedly stolen from shops over many years, and had breached a court order banning him from them. One half were told that Alex is homeless and addicted to drugs, and steals to fund his drug use, the other half were not.

Once more, knowing about the addiction transforms which responses people choose – albeit support for prison remains at 43%. Support for a programme addressing underlying causes jumps 19 points to 73%, the clear winner and the highest figure recorded in either scenario. Support for most other responses falls: prison by ten points, compensation or unpaid work by fifteen, a mediated meeting with the victim slightly. The public do not excuse persistent offending once they understand what drives it; they reorder what should happen around the punishment. The one other response that rises is help into work, up eight points to 47%. 

Our lived experience focus group described how punishment through a prison sentence can interrupt the cycle of crime – but it is only treatment and recovery that can end it.

Our other local focus group participants who knew or came across people involved in repeat offending raised similar points about needing to deal with the root causes, but reflected uncertainty about what might work.

We also heard the importance of treatment being available for those who want to change, with prison as the fallback for those who don't. Participants in the lived experience group also argued that treatment will only work if the person has some motivation to change.

Treatment wins across parties – but the consequence stays

Breaking the Alex scenario down by party shows the same effect across political lines: once the addiction is named, support for a programme addressing underlying causes rises by between 17 and 28 points among every party’s voters, taking all of them above 64%. Support for a prison sentence falls too, but far more modestly – between six and sixteen points – and the parties remain divided on it, ranging from 31% of Labour intenders to 56% of Conservatives. Only Labour and Green voters fall clearly below half.

However, the fact that support for prison does not drop below 43%, even with social context, is notable. As one voter explained – the context may explain the situation and even evoke empathy, but this does not trump the importance of repercussions:

In this sense, as we have found before, voters tend to show empathy and support treatment but without giving a free pass: they still want to see punishment for wrongdoing. The preference is for treatment-plus-repercussions, not leniency, with the offender still expected to answer for the harm. Meanwhile, a hard core, concentrated among tough-on-crime and Reform/Conservative voters, rejects the context outright. They read repeat offending as proof of bad character, so a difficult backstory is an excuse rather than a reason.

The lived experience group accepted the public’s half of that bargain.

A common assumption is that punitive instincts are strongest among victims – however, our data suggests the opposite.³ Those who have been victims of crime are 20 points less likely than the unaffected to choose prison for the shoplifter with addiction issues, and a regression controlling for age, party and gender shows victimhood independently cuts the odds of choosing prison by more than half. As Chapter 1 found, victims are more critical of the justice system – but that dissatisfaction is not a straightforward demand for harsher punishment.

The rest of this chapter turns from instincts to trade-offs: which specific policy combinations for prolific offending actually command support.

3.4‘Punishment plus’: preferred policy packages to deal with prolific offending

Testing policy packages to deal with prolific offending

Standard polling questions can only get us so far in understanding where the public really stands: what people endorse when nothing is at stake is a poor guide to what they choose when a choice costs them something else. To get a clearer picture, we ran a conjoint experiment, testing the popularity of packages of policies assembled at random, with each policy tagged with one of the '5 Rs' (Repercussions, Responsibility, Repair, Reform and Reintegration) they represented most obviously.

What is a conjoint experiment?

A conjoint experiment shows people complete packages of policies rather than individual measures, and asks them to choose between two. Because the contents of each package are varied at random across thousands of repeated choices, the effect of every element – and of combinations of elements – can be separated out from whatever it appeared alongside. It tests what survives a trade-off, rather than what people endorse when nothing is at stake.

Our packages were framed as ways the government should respond to prolific and repeat offenders – “people who keep committing crimes such as shoplifting, phone snatching or anti-social behaviour”. Each was built from four elements (consequences, victim involvement, use of prison, new spending), each with five options – one written to express each of the 5 Rs. Every respondent made six choices, producing 8,958 across 1,493 people.

Reading the figures: every choice was between two packages, so a package picked at random would be chosen half the time. Fifty per cent is the neutral point, and the figures below show how far above or below it each measure sits. A policy at +4 was therefore chosen four points more often than chance when it appeared in a package; a policy at −4, four points less often. These are relative preferences, not levels of public support: +4 does not mean 54% of people back the policy.

The primacy of repercussions

To assess results, we looked both at the impact of single policies on package choice, and on combinations of policies via an interaction analysis.

Firstly, as shown below, the single policies that did most to increase the likelihood of a package being chosen were both repercussions-coded: escalating prison sentences (+4.6) and stricter prison conditions with fewer privileges (+4.1). Meanwhile the largest negative effects came from policies written to express reintegration: more housing and employment support for offenders (−5.3) and housing, jobs and mentoring before and after release (−4.6). The other clear loser was unpaid work instead of going to court (−4.7).

The narrow range between the most and least influential policies is itself a finding: most measures tested are broadly acceptable to most people.

Between voter groups, repercussions are more divisive than the national table suggests: escalating sentences run from +10.2 among Reform UK supporters to −0.7 among Greens, topping the table because the right chooses them strongly. More telling is the middle: the 40% of the public between the tough and causes poles choose escalating sentences clearly above chance, and treatment spending more strongly than the causes group – they want both. Repair is the nearest thing to common ground, chosen above chance by every party's voters except the Greens; treatment spending divides more along party lines. Age divides the public on the use of prison almost as sharply as party does (packages with stricter prison conditions: 57% among over-45s against 47% among under-35s), and people affected by crime lean away from punitive measures and towards treatment in prison, consistent with the victims’ findings above.

The winning combinations balance ‘punishment plus’ rehabilitation

Beyond observing the influence of individual policies on package selection, using an interaction analysis we can also see which pairs of policies did most for a package’s chances. Using this method, a strong theme emerges of “punishment plus”: nine of the top ten pairings contain a repercussions element – far more than chance would predict, when only around a third of all possible pairings do – and in eight of the ten it is paired with something constructive from another 'R'. A clear consequence, in other words, is the entry ticket; the 'plus' is what separates the winning packages.

As seen below, the single strongest combination pairs stricter prison conditions with mental health and addiction treatment as the top spending priority, chosen almost ten points above chance. Escalating sentences paired with visible repair (offenders repairing the damage, or day-release wages paid to victims) perform nearly as well. Two notes of caution apply. The public back stricter conditions as one element of balanced packages, not as an end in itself: nowhere in the top ten do harsher conditions win alone. And ‘stricter prison conditions, fewer privileges’ is open to interpretation – the experiment cannot tell us what regime people picture when they choose it, and further research is needed into what the public actually mean by it. This is a real effect: we found that these pairs of policies outperform the sum of their parts (for example, the top pair outperforms the sum of its two measures by nearly four points). Conversely, doubling up on punishment adds less than the sum of its parts – escalating sentences plus stricter conditions scores +7.1, below the +8.7 the two earn separately – while the weakest policy pairings in the experiment tended to combine reintegration or reform policies. Most people, then, want policy packages that contain a balance: of repercussions plus reform or repair.

3.5Balancing punishment plus rehabilitation – testing a specific policy proposal 

The ICRO: ‘punishment plus’ in practice?

In our focus groups, we also tested a concrete policy proposal: the Intensive Control and Rehabilitation Order (ICRO), proposed by the Centre for Social Justice in All Options (2025) and Hijacked High Streets (2026).⁴ We tested the ICRO partly because, on its face, it offers the balance the public seem to want – punishment and restrictions plus rehabilitation and treatment – and partly because it is a live political proposition from a prominent think tank. All five groups – including the lived experience group – were read a description of the ICRO for repeat, non-violent offenders: instead of a short prison sentence, a GPS ankle tag, an alcohol tag where drinking is an issue, compulsory drug or addiction treatment, unpaid work where the person is able, and regular progress check-ins in front of the same judge – with prison for those who break the rules.

The reception to the basic idea was warm across the political spectrum. In Winchester's show of thumbs at the end of the discussion, Liberal Democrat and Conservative supporters were mostly in favour – but more notable was Great Yarmouth, the most punitive focus group, made up mostly of Reform and Conservative voters, where a majority of participants also responded positively, despite two firm rejections. Some of the most pro-punishment voices in each group liked the idea precisely because they could see the consequences in it: it takes the offender’s time. Participants suggested that paying with time in the community did sound like a real and legitimate consequence: Saturdays given up, long hours of unpaid work, freedom restricted by the tag.

Scepticism is about delivery, not leniency

Where scepticism came, it wasn’t because the ICRO was seen as soft, but from doubts that overstretched services would fund and deliver it – doubts woven together with a generally dim view of public service capability, compounded by how little the public see or trust community-based responses today. We heard calls to “bring back” community service, an option that has existed for decades, showing that many people do not know this is already happening. Many also asked for evidence that these options were safe and effective. Members of the lived experience group, who have been through versions of every component, similarly doubted the quality of what would be delivered: as one member said, he’d had every measure in the ICRO imposed on him at some point, and none had worked until he found both the motivation to change and high-quality recovery services.

The lived experience group also repeated the caveat that mandating someone to treatment can work for some, but only motivation to change makes treatment work.

Finally, for most people, consent for the ICRO rested on its backstop. The prison consequence for non-compliance is what made the order acceptable – and several participants wanted it sharpened rather than softened.

The conditions the public attach to the ICRO are therefore the same ones this report keeps finding: consequences that are certain, punishment that is visible, and honesty about what will actually be delivered. Resourcing the order properly, delivering it well, and proving its impact publicly are not implementation details – they are what would make the public trust it.

3.6Conclusion: responses to prolific offending that the public will back 

The responses to prolific offending that the public will back are balanced, and best described as ‘punishment plus’. Repeat offending sharply raises the demand for punishment, but the rehabilitative responses – compensation, treatment, help into work – stay just as popular alongside it. Context, above all addiction, turns a narrow preference for prison into a clear majority for treatment – among every party’s voters when the offender is described – without ever collapsing support for a consequence. The same balance won in the conjoint experiment: the strongest packages paired an escalating or stricter consequence with treatment or visible repair, while the weakest pairings – combinations with no clear consequence in them – fell around ten points below the 50% chance benchmark: rejected roughly six times in ten. The ICRO proposal, a real policy built on this balance, drew support across the spectrum – the caveats were about funding, delivery and enforcement, not softness. Treatment must sit inside the consequence, not instead of it – while the public need to be convinced that programmes will be properly resourced and not done on the cheap.

On prolific offending the public are not divided between punishment and treatment – they are waiting for someone to offer both, credibly delivered.

↑ Back to contents

4.1Rebuilding trust in the justice system

Clear common ground – masked by a noisy debate

This report shows a crisis of confidence in British criminal justice. No core function of the system is rated as doing a good job by a majority of the public, fewer than one in five believe consequences usually follow crime, and most Britons now expect to be victims within a year. This feeds fear and a wider sense that the social contract is failing.

But this research also gives grounds for optimism. When we listened to the public, victims and prison-leavers – across five focus groups, 78 in-depth interviews and a national survey – the polarised “tough vs soft” debate dissolved. Our research suggests the public is less divided about what should replace the failing system than our politics often implies. Political point-scoring on crime risks obscuring the common ground that does exist: on the need for meaningful consequences, greater responsibility, support for rehabilitation and a system that ultimately keeps people safe. That common ground offers a foundation on which a more effective and durable approach to justice could be built – but only if politicians work together across parties on a serious long-term plan for change.  

Accountability that cuts both ways

What the public demand is urgent action to restore trust, built on accountability that cuts both ways: for people who commit crime to take responsibility for the harm they cause, but also for a justice system that is more honest, effective and which delivers on its promises.

The first side is that people who commit crime are held accountable: making sure they own the harm they've caused, pay for it through proportionate punishment, and change their ways. Holding people to that standard is not at odds with a recognition of the addiction, mental ill-health and disadvantage that sit behind so much offending: while punishment alone is unlikely to work, both personal responsibility and addressing root causes are essential, and neither works without the other. The other side of the coin is the system visibly doing what it promises: consequence people can see, honesty about what sentences mean, and investment that reaches the causes of crime as well as its symptoms. Here are seven recommendations stemming from this report’s findings, aimed at rebuilding trust in the justice system.

Seven recommendations for politicians, policymakers and campaigners

4.2What comes next

Public opinion alone cannot tell us which policies will work: evidence of effectiveness, professional expertise and lived experience must also shape reform. But policy and campaigns which ignore the public's underlying values will struggle to win trust or support.

Our next task is therefore to deepen and test the common ground identified here: understanding the beliefs and values beneath people's views, exploring what survives when people encounter different experiences, evidence and real trade-offs, and testing how these principles can be applied to difficult policy choices.

Over the next three years we will bring that learning together with evidence, lived experience and political engagement to develop a Common Ground Blueprint for Change: a practical framework for justice reform capable of commanding public consent across political divides.

The crisis of confidence documented in this report is serious. But the common ground we found is reason to believe it is not irreversible.

Join us on this journey by signing up at commongroundjustice.uk

↑ Back to contents