And it rumbles on. Hopefully this will discourage people from continually using Brixton as an example of inner city failure...
Crimes against statistics
As a former home secretary, Michael Howard can't really believe what he is saying about violence in Brixton
Mike Hough
Thursday October 14, 2004
The Guardian
Earlier this week, Michael Howard called Polly Toynbee to task for drawing on British Crime Survey figures about falling crime. "The most reliable crime statistics - those recorded by the police - show that crime in England and Wales has risen by 850,000 in the past five years," he claimed. He referred to an increase in recorded violent crime of 83% over the past five years nationally, and, in defence of his remarks about Brixton, a rise in violent crime statistics in Lambeth of 10% over the past year.
As a former home secretary, he must be aware that this is a gross misrepresentation of crime trends. Police statistics bear little relation to the reality. The British Crime Survey (BCS) shows unequivocally that major types of crime have fallen dramatically since 1995: vehicle crime down by half, house burglary down by 47%, assault down by 43%, wounding down by 28%, vandalism down 27%. Mugging shows a small fall that is statistically not significant.
Recorded crime has gone up over the past five years because the police have changed the way that they count crime. In particular, they altered their "counting rules" in 1998, and introduced a national crime recording standard from 2002. They previously rejected victims' reports of crime if they doubted them; now, under the NCRS, these are taken at face value. Both sets of changes have inflated the police count of crime, and this inflation has been greatest for crimes of violence. That is the reason for the 83% rise in violence that Mr Howard cites.
What are we to make of Lambeth's rise of 10% in recorded crimes of violence last year? Nationally, police figures for violence rose by 14%, while the BCS figure shows a fall of 3%. If Lambeth follows the pattern elsewhere, the 10% rise is simply a consequence of the "bedding in" of the NCRS.
Statisticians have always known that only a proportion of crimes committed get reported to the police, and only a proportion of those reported find their way into police records. The BCS, which has now been in existence for 23 years, can put estimates to the "dark figure" of unrecorded crime; it interviews very large samples of the population about their experience of crime. Thus in 1981, 36% of such crimes were reported to the police, a figure which has risen to 44% today. The proportion of reported crime that is recorded has risen over this period from 62% to 77%. So the proportion of unrecorded crimes has shrunk since 1981, from almost four-fifths to around two-thirds - a very large change.
The BCS trend is simple: crime rose from 1981 until 1995, and then fell back again to near 1981 levels. The trend in recorded crime is more complicated: it rose at a faster rate than the BCS trend-line in the 1980s because as a nation we began to report more of the crimes committed against us. In the early to mid-1990s the recorded statistics showed a fall while the underlying trend was still upward - possibly a disciplined police response to targets set for them by Michael Howard. Are we to believe the BCS? It is done to high technical standards by survey companies independent of the Home Office. I trust it - but as a member of the original research team, I would say that, wouldn't I?
However, there are no persuasive technical grounds for doubting the validity of the trend information it has yielded. True, response rates have dipped by five or so percentage points since 1981. True, the estimates are subject to sampling error. But measurement error is not a plausible explanation of falls approaching 50%.
This brings me to Mr Howard's second claim about the BCS, that it has significant flaws in undercounting crime. The BCS is flawed in much the same way that my car is flawed in its failure to run on railway tracks. An index of people's personal experience of crime will not measure other sorts of crime. However, the Home Office is due to publish its survey of commercial victimisation and its "self-report" survey of offending later this year. These will plug some of the gaps left by the BCS.
The safest thing to conclude is that crimes that bear some similarity to BCS crimes will follow similar trends. I would expect commercial burglary to follow a similar pattern to house burglary, and vandalism against public property to track vandalism against personal property. There is no evidence to my knowledge that crime against the under-16s is rising. One crime type that could buck the downward trend is shoplifting - the main crime committed by dependent drug users. Violent crime involving firearms is rare but rising. E-crime is surging. But the headline trend for crimes that affect everyone's daily lives is downward.
We can hardly be surprised that opposition politicians play a spoiling game when crime goes down or when the government claims all the credit for the falls. But that is no reason to discount the very real improvements in knowledge that the BCS has brought about.
· Professor Mike Hough is director of the institute for criminal policy research, King's College London
mike.hough@kcl.ac.uk