Hard evidence: how accurate are British crime statistics?

Smash-and-grab: statistics only give us a fractured picture of crime. raleighwoman

The Office for National Statistics released the latest crime statistics last week, and the headline findings are really rather remarkable. Although approximately 8 million offences were recorded by the Crime Survey for England and Wales (CSEW) for the year ending September 2013, this figure was 10% lower than the equivalent a year earlier, 20% lower than the estimate five years, and less than half the peak level of crime reached in 1995.

Good news, it would seem – and yet, not everyone is convinced. The Financial Times headline was “Recession-linked crime rises in UK as spending cuts bite”. “The ‘3% fall’ in crime that may really be an 8% surge” worried the Daily Mail. “Crime figures at all-time low amid claims of police ‘fiddling’” was the Daily Telegraph’s take.

The newspapers are far from alone in their scepticism. Crime survey data have consistently shown a majority of the adult population thinks crime is rising nationally, and that a sizeable minority also think it is rising locally – all this despite what crime figures might say.

Two data sources

Establishing the true rate of crime is difficult for a number of reasons. First, there isn’t a single source of data on crime trends, but at least two that feature regularly in the press. The first, generally regarded as the most authoritative, is the Crime Survey for England and Wales (originally called the British Crime Survey or BCS). This is a large household sample survey asking people about their experiences of crime and criminal justice over the preceding 12 months. Originally conducted every two or three years, it is now an annual rolling survey and is widely seen as our best measure of trends in crime. It is the crime survey results from ONS that were reported at the beginning of this article, and which led them to talk of a 10% drop in crime.

The second major source of crime data is what is referred to as “police-recorded crime”. Until 1981, when the BCS came into being, this was the primary source of information about crime. It is essentially administrative data, reflecting what comes to the attention of the police – and then, within that, what they choose to record. During 2012/13 the police recorded a total of 3.7 million offences, a decrease of 3% on the previous year. It is these figures that were the basis of the Mail and Telegraph stories quoted above.

So, we have two very different sources of data. Again, however, there is some good news: both sources are pointing in similar directions where trends in crime are concerned.

Both the CSEW and police-recorded crime indicate that crime rose through the 1980s and has been declining for at least the past decade. They differ in the middle of the period, in particular for around five to six years from around 1997 or thereabouts. This is largely a product of changes to what are called the “counting rules” and the introduction of the National Crime Recording Standard, both of which have affected rates of police-recorded crime. The aim of this standard was to improve the accuracy of police-recorded crime – but its real effect, recognised by the Home Office, was to artificially inflate estimates of the increase in the number of crimes recorded by the police.

Allowing for these changes, it appears that our two sources indicate broadly similar trends: crime rising through the 1980s to a peak somewhere in the mid-to-late 1990s, and with falling levels since that point. On this basis we might assume that the answer to our question is “yes”. But how trustworthy, and how accurate, are our two main sources of information?

Juking the stats

Recent revelations have cast very considerable doubt on police-recorded crime data; the House of Commons Public Administration Select Committee has been scrutinising crime statistics, and has heard evidence from a series of witnesses, including the former Commissioner of the Metropolitan Police, all revealing a variety of long-standing techniques used by the police to manipulate the recording of crime.

So serious are the concerns that the UK Statistics Authority recently took the decision to withdraw the status of “national statistics” from police-recorded crime, effectively indicating that for the time being, they should not be trusted at all.

Missing crimes

This leaves us with the Crime Survey for England and Wales. It clearly suggests crime is in long-term decline in all major offence categories – with the exception of “theft from the person”, which rose by 2%. The most recent CSEW measured declines over the past year range from 3% and 4% (for burglary and theft from vehicles respectively) to 13% for violence.

Might it be missing something? The answer is yes. The CSEW only covers people living in private households, and consequently excludes individuals living in institutions (university accommodation, prisons, care homes and so on). It is a personal survey, and therefore doesn’t include businesses. Until recently it didn’t include people under the age of 16; consequently, its measurement of long-term trends has little to say about crime affecting this sizeable group. And since it’s based on victims’ experiences of crime, the CSEW doesn’t include crimes where there is no victim who can be interviewed, such as homicide or drug possession offences.

What concerns observers most about the CSEW these days is the impact of the internet, and the possibility that there exists a world of crime the survey simply fails to capture. This would include everything from identity theft, a wide variety (and quite likely large amounts) of fraud, and the distribution of stolen goods (“fencing”) through web-based sale and exchange sites. Given this, it is perfectly possible that the crime survey is significantly over-estimating the fall in crime.

But is this over-estimate significant enough to wipe out the crime decline entirely? Is crime simply mutating rather than dropping? Without new forms of measurement, it is simply impossible to tell with any precision. But the scale of the measured crime drop over the past two decades is so sizeable that it is highly unlikely to be entirely a fiction.

So, yes, on the basis of our best measures it does indeed appear that crime is in decline. The big question, of course, is why.

Hard Evidence is a series of articles in which academics use research evidence to tackle the trickiest public policy questions.

Want to write?

Write an article and join a growing community of more than 95,500 academics and researchers from 3,107 institutions.

Register now