We need a different crime survey

Professor Tim Hope
Thursday, 21 May 2015

As Richard Garside has frequently commented here, the Crime Survey for England and Wales (CSEW) is neither a survey of crime (it doesn’t cover much of it), nor a survey of victims (it doesn’t include all of them), nor a survey of victimisation (it doesn’t measure enough of it).

In fact, the Crime Survey is much better at not measuring crime than it is at measuring crime’s true extent, which is no doubt why policy-makers have come to rely upon it so much.

Where the crime survey came from

When it started out, the British Crime Survey (which morphed into the CSEW) did uncover more ‘hidden crime’ than had hitherto been recorded by the police. But that was always only intended to be the non-reported component of that which was usually reported. So, the CSEW was only ever intended as a mirror of Police Recorded Crime (PRC) and, not surprisingly, its profile has broadly similar proportions, yielding similar types of crime, by similar types of victim, to that which is usually reported to the police. It differs in the quantity rather than the kind of concerns that members of the public usually feel happy to bring to the attention of the police, which is mainly to do with the theft of their insured property.

Because of this, the CSEW/PRC mirror has been a very useful political device. It puffed-up the Government’s ‘performance’ in the War Against Crime, with Tony Blair declaring victory in his 2007 valedictorum on his premiership (although the fall in residential property crime might have been due more to the increased turnover of B&Q’s home security and double-glazing lines than political activity). Latterly, because crime is apparently dropping, the CSEW has been equally useful in cutting police numbers: ‘no need for all those Bobbies on the Beat when there’s nothing for them to do!’

Similarly, the subjective elements of the CSEW such as fear of crime, confidence in the police, etc. (i.e. all those attitudinal questions that are asked in order to fill-up the questionnaire after you’ve asked people about their victim experiences) have served very well as the Home Office’s own public opinion poll on measures of ‘satisfaction’ with criminal justice performance (didn’t we do well!). Which was why, a few years ago, there was such a flurry about the ‘reassurance gap’ - the obstinate refusal of the public to feel reassured by the decline in crime, in the same survey!

Problems with the survey

As for the CSEW/PRC mirror, it is instructive to see what has actually happened to the standing of crime statistics in Middle England. For example, H.M. Inspectorate of Constabulary (HMIC) investigated Kent Police for failing to record crime reported to them, particularly serious, sexual offences. The CSEW was of no help here; not only is it not large enough to shed any light specifically on hidden crime in Kent but even if it could, like PRC, very few respondents report serious sexual offences to the CSEW. The Kent affair led to a damning parliamentary inquiry, in turn leading to the withdrawal of UK Statistics Authority recognition for PRC. Nevertheless, the mirror itself didn’t crack: despite as similar a failure to record serious crime as PRC, the CSEW is now the only nationally-approved measure of crime.

Yet, while most of us are content to be lulled into this reassuringly utopian state of Crime Denial (crime and punishment were not even mentioned during the General Election) there is a lurking sense that not all is well below the surface of society. Maybe there is a flood-tide of violence, cruelty, appropriation, exploitation, degradation and abuse that is threatening to engulf the most vulnerable - women, children and young people, the poor and disadvantaged, the mentally ill, and all the other minorities who are not managing to scramble beneath the comfort blanket of security that officialdom would like to wrap around the core constituencies of Middle England. Who knows? Certainly, neither the CSEW, nor those who might dismiss such fears out of hand as ‘moral panics’, can give us a straight answer.

Over the years, the CSEW has got progressively better at measuring the non-occurrence of crime, and much worse at measuring crime victimisation, paradoxically because it has got better at representing the experiences of the general population and worse at representing the experiences of victims. An illustration as to why this happens is given in the table below. Extensive and sophisticated analysis of BCS/CSEW data suggests that the overall distribution of crime victimisation amongst the population is highly unequal. This reflects the well-known Pareto Principle of Inequality: that is, around 80 per cent of the population suffer only about 20 per cent of the crime (who are largely immune from victimisation) while about 20 per cent of the population take around 80 per cent of the burden of crime upon their shoulders. As such, the distribution of CSEW-measured crime victimisation in society is at least as unequal as the distribution of health, wealth and wellbeing (fact!).

An estimated distribution of crime victims amongst the population as represented
by the Crime Survey for England and Wales

Victim type

Percentage of population

Percentage of victimisation










Based on household property crime in selected sweeps of the BCS/CSEW, 1992 – 2006/7. For details see here.

Even amongst the victimised there is further inequality: while most who are vulnerable to crime may suffer only intermittently, there is a relatively small population who are chronically victimised. Even worse, the sad fact is that although Chronic Victims’ suffering is wholly disproportionate (perhaps over two hundred times more than their numbers would warrant), because they probably only amount to less than one per cent of the sample, their effect on the national crime rate is negligible. And so also is their effect on the official conscience.

While most of us – the Haves and Have-nots of Safety – put up with the differences between us, in much the same way as we do with all the other 80:20 social differences in health, wealth and well-being, the Chronic Victims are the truly victimized, not only from crime but, moreover, from official neglect. In the current CSEW, the politics of large numbers prevail: ‘we’ can relax because crime is going down for the majority of our constituents but this is only because the majority are experiencing less of what was always a rare misfortune. For the less fortunate, crime victimisation is always as it has been, an abiding part of their everyday lives. But since they don’t vote, why should it matter anyway?

Rebalancing the survey

To use a familiar rhetoric, is there a way we can rebalance the CSEW in favour of the victim? There are a few ways that can be touched on briefly here:

1. Try to select samples with a high probability of victimisation

One of the most significant changes that happened almost by stealth was that the BCS/EWCS ceased to be a crime survey and became a police performance survey instead. Earlier BCS samples had been weighted to over-select inner-city (socially deprived) areas in order to ensure there were enough victims to study. Nevertheless, during the Blair administration the sampling design was changed so that ‘inner city’ (high crime) areas were under-sampled, while suburban and rural (low crime) areas were over-sampled, ostensibly in order to provide sufficient ‘customer’ representation in every police force area.

Not only did this result in a huge sample size but clearly it also inflated the number of non-victims in the sample (and may well have reduced the count of crime accordingly). At the same time, living in a socially deprived area ceased to be a ‘risk factor’ for crime victimisation, even though it had previously been the main predictor not only of victimisation risk but also of repeat victimisation (something that the Government had once tried to do something about). In short, who knows how many more chronic victims are now being over-looked?

So, we need to rebalance the Crime Survey to better represent victims; we are not going to lose anything in terms of counting crime by reducing the number of non-victims we continue to trawl into the catch (indeed, the recent reduction of just under a third of the sample size of the EWCS did nothing to change the continuing deflationary trend of the crime rate).

2. Try to include the crime that is hard to reach

We need to find ways of encouraging victims to talk to the survey frankly, safely and openly. We should be as sensitive to interviewing in the CSEW (and find alternative ways of finding and interviewing victims) as we are now expecting the police to be, especially concerning sexual offences, domestic violence, bullying, harassment and grooming, Internet fraud, and indeed in any of the harms they experience.

3. Try to contact the people that are hard to reach

Since the start, the BCS/EWCS has always failed to contact or had refusals from around a quarter to a third of its sample. Market research and opinion polling can always write this off as they tend to deal in generalities, want to cut costs, and often do not use random selection procedures (remember, also, that it is the market research industry that has always designed and run the BCS/EWCS). But this cannot be ignored in a crime victims’ measurement survey: not only does every response count but the kinds of people who refuse or cannot be contacted are precisely the kinds of people who are most likely to be vulnerable to crime victimization (see above).

4. Try to lift the cap on the crimes we count

Finally, the counting procedures themselves need to change. Usually, an arbitrary ‘cap’ is put on the number of victimisations that victims report in order to ensure that the estimates are not affected by a very small number of respondents who report an extremely high number of incidents and which are highly variable between survey years. But it is precisely these chronic victims that we need to listen to. The consequences are dramatic. Analysis of CSEW finds that when the cap is removed there are 60% more violent crimes. The increase due to removing the cap is concentrated on violent crime against women (70% increase) rather than men (50% increase) and on violent crime by domestic relations (70% increase) and acquaintances (100% increase) rather than by strangers (20% increase), as Sylvia Walby and colleagues have shown.

Restoring the moral dimension

Obviously, much more needs to be done technically to implement a truly victim-oriented national survey. But the biggest battle is a moral one. As a member of the audience put it in the recent Election Question Time broadcast, ‘politicians always talk like accountants and never talk about the moral purpose of their policies’. And this issue goes to the very origins of ‘statistics’ themselves – are they rules for calculated governance or are they guides to the health of society? Statistics on crime were some of the first to be collected in the spirit of the latter; and we need very much to restore the moral dimension to our contemporary measurement of crime.

Dr Tim Hope is Professor in Criminology, University of Salford, Manchester