How not to interpret crime figures

Since doing a project on violence risk assessment I have always had an amateur interest in crime statistics – nothing remotely approaching a professional interest – just what you can gather by spending 30 minutes on the Internet once a month. You would think that a professional journalist writing for a national daily paper on a regular basis, presumably with some kind of research assistance, would do better. So I was quite surprised to read this piece by Stephen Pollard which borders on the incompetent.

The results of police recorded crime in the UK and the British Crime Survey (BCS) for 2007/8 have just been released. They confirm a well-established pattern. According to the BCS most types of crime in the UK increased steadily from the 1960s until about 1995 and have been reducing steadily ever since. The recorded crime statistics support the BCS with respect to major non-violent crime; they show a mixed picture with respect to violent crime but the 2007/8 results showed an 8% reduction from 2006/7. This appears to be part of global pattern among developed countries and is particularly significant in the context of a spate of recent stories about violence and murder among the young using knives. Not surprisingly the government has expressed delight at the results and tried to claim credit for it. Pollard tries to make the case that crime is not actually reducing and that the government is misleading the public. As he puts it "the truth is that most people have an instinctive understanding that crime is a lot worse than it used to be. Although it is difficult to sort the reliable evidence from the massaged statistics, the evidence is clear. Things are worse than the official statistics suggest.".

If the evidence is clear then we might expect Pollard to give us some of it. But in fact all gives us is a wealth of irrelevant numbers and details. In particular he expends a lot of words attacking the BCS, which he describes as "woefully inadequate" because it leaves out some types of crime (for example, drug abuse and commercial crimes). This is to grossly misunderstand the importance of the BCS. There is no realistic concept of "all crimes". Every time you take a pencil from the office stationary cupboard for private use you are committing a crime. To attempt to measure all crimes would be absurd. All that any survey can do is concentrate on a well-defined set of crimes that matter. If the objective is to detect trends then the important thing is to be consistent from one year to another. The BCS does just that and does it well. The BCS includes most of the crimes that the public as potential victims most care about – violent crime, burglary, robbery, car theft.

Possibly a more serious problem with the BCS is that it omits some categories of potential victim (it is limited to adults living in private homes). Undoubtedly the BCS would be even better if, for example, it included children under 16 and adults who do not live in private homes. In 2006 the government commissioned an independent review by Professor Adrian Smith (a former president of the Royal Statistical Society). The review was supportive of the BCS describing it as "is a high-quality survey that has contributed significantly to the Home Office's and to public understanding of crime trends over the last 25 years". The review also suggested that it be extended to include under-16 year olds and the government plans to do this. But this should be put into context. It is important to know about specific groups such as children. But the trend in crime for children will only affect national trends if their experience of crime is significantly different from the groups that are included and if it comprises a significant proportion of overall crime. In fact, as Pollard himself says, 11-15 year olds will only add about 600,000 crimes to the BCS total of over 10 million. Even if their experience is quite different from adults it is unlikely to be enough to change a national trend. And Pollard presents no evidence that the trend for children is any different than the trend for adults.

 

For all its limitations the BCS provides strong evidence that the crimes that affect us most as victims have been diminishing over the last 10 years and this is backed up by the police recorded crime statistics. Pollard talks about the limitations of the BCS but offers no evidence whatsoever that this trend is false. What he does offer is ragbag of statistics showing that the risk of some crimes has increased since – well since whatever date suits his argument – 1918, 1931, 1941, 1957, 1972, 1979 – but nowhere does he offer a comparison between 1995 and the present day. It is an easy game to play – choosing specific dates and crimes to support your case. In fact I am amazed he couldn't play it a bit better. Surely he could find at least one crime that had got worse since 1995? In fact I can help him there. The recorded crime statistics show total robberies as 68,000 in 1995 and 101,000 in 2006/7. On the other hand you could compare burglary rates: 1,239,000 in 1995 as against 622,000 in 2006/7. You really can show almost any trend you like if you just pick the figures to support your case.

 

Pollard concludes: "Next time you hear a politician or someone from the liberal-Left try to dismiss fear of crime as a media panic, remember this: statistics can be used to support almost any argument. But the truth will always out." Statistics can indeed be used to support almost any argument if they are misused. However, the truth is what happened and the statistics are nothing but a record of what happened. They are closest thing to the truth that you are going to get.