Making Sense of Crime Trends

sakhorn38

What to make of the latest crime statistics. Total crime recorded by the police is down by seven percent in a year. Over the nine years between 2002/03 and 2011/12 it dropped by 33 percent. The comparable figures for the separate Crime Survey for England and Wales are eight and 28 percent.

On a longer view the trends are even  more striking. As the figure below from yesterday’s report indicates, total crime recorded by the police rose gradually from 1981 through to 2003/04, after which it fell fairly sharply. The Crime Survey for England and Wales data (formerly the British Crime Survey) rose more sharply, reaching a peak nearly a decade earlier in 1995, followed by an equally sharp decline.

According to The Guardian this has left baffled experts ‘scurrying to rewrite their basic Criminology 101 lecture and the theory that recession leads to rising crime, particularly property crime’.

Various theories have been floated. Our homes and cars are more secure, argue some. Others suggest young people are staying at home on their smartphones and computers rather than hanging around on the streets and vandalising bus shelters.
Others query the reliability of the data. Perhaps people these days are engaging in the types of crime not adequately captured by official statistics; online fraud for instance. Maybe dodgy recording practices help explain the drop, as the The Independent and The Times suggest today.
It is difficult not to conclude that criminologists – or at least those interviewed by the media – are struggling to apply a particular set of ‘orthodox’ theoretical perspectives to a set of data that really don’t fit the theory. Policy makers, journalists and commentators likewise struggle to find a coherent explanation. So we have the spectacle of seemingly intelligent people buying into the latest cod theory linking crime rates with lead poisoning, despite its obvious flaws.
Overall crime and crime involving overalls

So what should we do? Putting to one side conceptual and definitional problems associated with crime as a term, a wide variety of harmful and less harmful acts and behaviours are marshalled under this blanket term. Homicide is a crime, as is vandalism. They share little or nothing in common apart from this legal definition. House breaking, shoplifting and sexual assaults are variously more or less harmful crimes. Their causes and consequences are varied and complex.
In short, to ask why ‘overall’ crime might be rising or falling is to ask the wrong question. It mystifies and confuses rather than elucidates and informs. Talk of overall crime rates is best reserved for the analysis of crime involving overalls.
Different crime types, different trends

Once we start looking at different types of crime, and how they have changed over time, it becomes possible to shed light on the possible causes of different crime trends. The figure below does this for four selected police recorded crimes – rape, burglary, vehicle and drug offences – setting them alongside the ‘overall’ crime rate.

To make it easier to understand how the five different categories have changed in relation to each other, their very different values have been equalised at the beginning of the data series. Changes are then expressed relative to this starting point. This allows us to understand how the different offence types have change over time, and relative to each other.

The  ‘overall’ crime trend is shown by the blue line sloping down gently from left to right. Below it, the turquoise and orange lines show the relative decline in burglary and vehicle offences. These declined at a much greater rate than the ‘overall’ crime rate: some 44 and 61 percent respectively. They also accounted for a large proportion of ‘overall’ crime: between a quarter and a third during this period.

The story of the fall in ‘overall’ crime in the period is therefore to a significant extent a story of the fall in burglary and vehicle offences. Cars have become much more difficult to break into or steal. Homes have become more secure and the resale value of stolen goods has declined. In no small part this helps to explain the fall in these crime types and, by extension, the fall in ‘overall’ crime.

Recorded drug offences and rape increased significantly during this period, by 30 and 60 percent respectively. Whether this reflects an increase in these offences in the ‘real world’ is a far more complex question to answer than first appears. I won’t attempt to do so here.

What we do know is that tackling drugs became a key police priority during this period, as did encouraging more sexual violence victims to report offences.  Police activity was therefore a key cause of the rise in recorded rape and drug offences.
The impact on ‘overall’ crime was, though, minimal. Rape and drug offences only make up a small proportion of the number of ‘overall’ crimes.. Even a large rise or fall in the number of recorded offences will have little impact on the ‘overall’ crime rate.
This final figure shows the equivalent trends for all violence offences involving injury, along with two specific categories: homicide and causing death through dangerous driving.

All violence involving injury recorded by the police rose in the first half of the period. It declined in the second half, but only to return to where it started at the beginning of the period. With the exception of a spike in 2009/10, deaths recorded as caused by dangerous driving remained constant throughout the period. The number of recorded homicides fell by a half.
There is little to be gained from trying to explain the changes in the ‘all violence’ category, covering as it does a range of different harms that should be analysed separately before considered as a whole.

On homicide we know from research by Danny Dorling, published here, that the early 1980s recession was correlated with a long-term rise in the homicide trend that affected particularly the poorest and most marginalised right through the 1980s and 1990s.
More recent research has demonstrated that economic shocks and crises are strongly correlated with a range of health and social problems, including suicide and homicide. The most recent statistics indicate that suicide is on the rise in the UK.

While the current downward trend in homicide in England and Wales is to be welcomed it is a decline from a high level. The UK is still closer to the beginning of the current depression than the end. In the coming years homicide rates could well start rising again. Meanwhile, deaths caused by dangerous driving are as prevalent now as a they were a decade ago.

Much of the current confusion about crime trends is born of the tendency to bunch together a whole range of different harms and actions under the abstract category of ‘crime’. Apart from making it more difficult to understand the causes and consequences of crime trends, it also blinds us to where the significant problems are.

Talk of ‘overall’ crime rates hinders clear thinking. We should stop doing it.

Update (Jan 27, 2013): Since posting the original blog a number of people have asked why I chose to rely on the police recorded crime data (PRC), rather than the Crime Survey for England and Wales (CSEW).

One, relatively pragmatic, answer is that though in some ways more reliable as an insight into certain crime victimisations, CSEW is also a fairly partial measure, covering a relatively narrow range of crime incidents. It does not, for instance, cover homicide or death from dangerous driving, nor does it cover sexual offences in the main dataset. PRC data, on the other hand, covers a wide range of crime types, albeit in a partial manner as it is highly dependent on what the police chose to record, or on what offences come to their attention.

But in any case, the main point I was making – that ‘overall crime’ is a wholly unhelpful and misleading category – is not dependent on the choice of dataset, whether CSEW or PRC. The point of this blog was really to argue that when crime statistics are unpacked and the problem is thought through clearly, the current downward trend – whether of CSEW or PRC – is nowhere near as baffling as some seem to think it is.

Updated (Jan 29, 2013): An analysis of some of the CSEW trends is now available here.

Don’t be baffled

This post was originally written and posted at the UK Justice and Policy Review, where you can find the original articles, and others like it.

READ RELATED ARTICLES

Undercover Pressures
The Myths of Offender Profiling.
The Formula
Numerical indigestion: how much data is really good for us?