An alarming claim

Driving to work the other day, my ear was caught by an advertisement by ADT for burglar alarms. It claimed that 84 per cent of burglars avoid homes with alarms. In some ways this cheered me up, as I constantly pay huge sums to Messrs Chubb to maintain our alarm. If the claim is right, I am getting value.
 
But how could ADT know? Were these in fact figures for the effect of burglar alarms, with 84 per cent fewer burglaries at homes with them? Or had ADT somehow contacted a sample of burglars and asked them? If that was how it had happened, it seemed to me that there was something intrinsically dodgy about basing a claim on burglars’ responses since by definition they were dishonest people. I felt a complaint to the Advertising Standards Authority, on behalf of Straight Statistics, coming on.
 
Straight Statistics’ director, Nigel Hawkes, turned his eagle eye to the ad and found something that gave me pause. On ADT’s website, the claim was referenced to the Home Office. ADT could hardly be blamed if they were using statistics from the government. So I put down a parliamentary question to the Home Office.
 
Baroness Neville-Jones, a government minister, answered on 8 December referring me to a 2004 Home Office publication. “The findings were based on interviews of {sic} a small sample of burglars in southern England who responded to questions regarding a range of deterrent factors against burglary, “ she said. “The use of these findings in an advertisement does not constitute an endorsement by the Home Office of a particular company’s products or services.”
 
I should hope not, because the study by the Criminal Policy Research Unit at South Bank University is quite extraordinary. The sample was drawn in the following way. Convicted burglars were identified by police and probation services and they in turn were asked to identify other burglars for the survey. They managed a total of 82 interviews. There is no way of assessing whether this sample was or was not representative of burglars as a whole.
 
The researchers themselves pointed out another weakness in the work: the degree of credence that can be given to offenders. “Some will under-report their offending; others will exaggerate or over-rationalise their behaviour in order to appear daring or sophisticated.” Others of course might simply take the piss. Nevertheless they felt able to report that among “deterrents rated with high frequency” the “presence of alarms outside property” was mentioned by 84 per cent of burglars. On that fact, ADT base their claim.
 
What of course a homeowner wants to know is not the result of some half-cocked survey of what a few randomly-chosen burglars say. They want to know if they are likely to be burgled less often if they fit an alarm. For this, they should consult the authoritative British Crime Survey.
 
The good news is that while homes with no basic home security had a 5.8 per cent chance of being burgled in 2009-10, those with enhanced security had a 0.6 per cent chance – an odds ratio of one to ten.
However, the bad news is that homes with basic home security – defined as window locks and double deadlocks - had a 0.9 per cent chance of having been burgled. This suggests that most of the benefit comes from basic security. By fitting and maintaining an alarm, at enormous expense, a householder will typically avoid one burglary roughly every 300 years.
 
Curiously, ADT do not point this out in their advertisement.