Whilst we wait for www.police.uk, a government-sponsored website competing in and crowding out a supposedly emerging open data marketplace, to have a full run of publicity – The Sunday Times yesterday, press today and tomorrow – pending a late release of monthly and additional data that we as 3rd party developers were not consulted on – I wanted to highlight one of the few crimes to be growing significantly on an annual calendar basis – bike theft.
According to the Home Office British Crime Survey figures (see table 1), over the calendar years 2009 – 2010 bike theft increased from 477,000 to 533,000 – a 12% increase. That’s over 5% of all crimes estimated to have been committed by the British Crime Survey out of 9.7 million in 2010.
Bike theft is such a problem because according to research by Halfords, in 2010 of the estimated 533,000 bikes stolen, only 115,147 were actually reported to the Police.
It’s not just that victims don’t believe the Police can do much about it. There has to be an easier way of registering crimes like these with the Police to close the gap between reported and actual theft. It also seems reasonable to get some more granularity here. The offence of “Theft of a pedal cycle” is currently classified under “Other” crimes on our website and Police.uk. Would it really be so terrible to share with us – with the consent of the owners – where those bikes were actually stolen from?
Victims of bike theft would have far more incentive to report the loss if they knew that everyone could see where it happened on a map and at what time. We could then quickly work out where the hotspots were and owners could take additional security measures while Police Forces and local communities could start to take preventative action or at least ask some searching questions about the apparent lack of progress.
Matt says:
It would certainly be interesting to get information about more-specific categories of crime, but I don’t think (as you seem to suggest) that it would help people estimate the risk that they face in particular locations, because it wouldn’t provide any information about the number of available targets.
British Transport Police (BTP), unlike the other forces in the UK, can release information about specific addresses, because all the addresses it polices are public places like railway stations. Their crime mapping system still keeps to the national categories, though. Their stats page for Cambridge station shows that there were 79 ‘other’ crimes at that station in the past 12 months. If BTP released more specific crime types, it might turn out that all these offences were bike thefts (although it’s not likely). 79 thefts sounds like a lot of bike thefts (one every three working days), so (on your argument) people might decide not to park their bike there. It sounds like if you leave a bike there every weekday it’s likely to get nicked every three days! What the crime statistics will not tell you, and cannot tell you, is that there are several hundred bikes parked at Cambridge station. If you assume that there are 500 bikes there every day, then the apparent frequency with which your bike will be stolen falls from once every three working days to once every six-and-a-half working years. Even this, however, doesn’t tell you much about your own risk of being a victim, since not every bike has an equal chance of being stolen: bikes without locks, bikes that are particularly desirable, bikes in areas not covered by CCTV, bikes that are parked where there aren’t many people, etc., may well all be more likely to be stolen than others.
The point is that raw numbers of offences aren’t all that helpful, especially when they are aggregated into groups. The violent crime group is probably the worst, because it covers everything from murder to common assaults where there is no physical contact between the suspect and the victim. For example, I know that there was one violent crime on my road last month, but that doesn’t tell me anything useful about my level of risk. I don’t know if it was a murder or a push on the shoulder. I don’t know if it happened out in the street or in someone’s home. I don’t know if it was a gang crime or drug related or a stranger attack or two drunks fighting. The number tells me nothing about how likely I am to be a victim of violent crime.
I’m not suggesting that it isn’t possible to communicate risk about crime to the public, but I’m not sure that crime numbers are a good way to do it. That doesn’t take away from the value of these numbers for holding public authorities to account, though: they could (with more detail) probably be very valuable for citizens attending police neighbourhood meetings.
Dan says:
Hi Matt, very interesting comment, thankyou. You’re right of course that to look at crime statistics from the point of view of the crime, rather than the victim or even the criminal gives you a limited perspective. Being able to cite “the number of available targets” is only partially reported at present through static population data which deflates the crime rate but not through, say, the number of parked cars in a given area to reflect vehicle crime or the number of households relative to the quantity of burglary. Or – as you rightly point out – the number of parked bicycles at Cambridge Station.
Are you sure though that all the bikes that are stolen from Cambridge station and it’s surrounding immediate environs are actually reported to the British Transport Police and not to Cambridgeshire Constabulary?
Personally, I’m not, or rather, I just don’t know until I can see the data. If they could release that to us, we would know. It would be great too if insurance companies could release their figures on stolen bikes to us – which I suspect would surpass officially reported tallies to all our Police Forces and give us much more precise detail – metadata – which you refer to. Greater transparency in insurance calculations based on crime would be a huge advance.
You mentioned the BTP. The problem we have is that since the new BTP crime website went live last week, they have failed to release all the underlying data that allowed them to build a platform. A terrible waste of taxpayers money when developers can do it for free. This kind of monopoly provision of crime data by the Police is a serious problem and runs against the stated coalition aim of promoting open data.
Could you just explain – i.e. detail – the arithmetic behind you calculation that the frequency with which your bike will be stolen “. . . falls from once every three working days to once every six-and-a-half working years“?
You say that ” . . . raw numbers of offences aren’t all that helpful, especially when they are aggregated into groups” – but compared to what, having even less information like we did before?
Some detail is always better than none and more is almost always better. Like you I’m all for additional granularity in the crime data which will convey a better idea of risk to the public. My sense is that over time, we will have a great deal more than you can see today.
Matt says:
I’m very sure that many (perhaps most) cycle thefts aren’t reported to either force! The BCS shows that reporting rates for cycle theft are getting worse (see page 77 of the most recent edition of Crime in England and Wales). Conversely once a crime is reported to the police, it’s fairly likely that it will end up being recorded by the correct force, because forces are quite keen to off-load crimes when they can. I don’t know whether insurance figures would be higher than police-recorded figures or not, since many people aren’t insured (or their bike is covered on their house insurance but they don’t know it), and those who are insured generally need a crime number to make a claim.
To explain my arithmetic, 79 crimes every year (every 260 weekdays) is 0.30 crimes per day, or one crime every 3.29 days. If there are 500 bikes, every bike has an equal chance of being stolen and bikes are stolen at a constant rate, each of those 500 bikes will be stolen once every 500 x 3.29 days = 1,645 days or 6.32 years worth of weekdays. Except that the number of bikes isn’t constant, bikes aren’t stolen at a constant rate and not all bikes have an equal chance of being stolen, which of course was my point.
What would be better than raw numbers would be a representation of the varying risk over time. The police use various crime-mapping techniques to do this, but to do it property requires data about the distribution of the population at risk. This is relatively easy for crimes like residential burglary, since the mapping data are available from Ordnance Survey, but much harder for crimes like cycle theft. It isn’t impossible, because you can do surveys of the numbers of bikes in particular locations, but it’s too expensive to do routinely or everywhere.
As I said, the current aggregation of crime data into very broad categories makes the information almost useless for any sort of analysis. The Home Office would undoubtedly say that they don’t provide the information for that purpose, but there are plenty of numerate people about and they would be better served if crime data were treated like some of the other government data that’s available. I personally don’t see why the type of crime can’t be specified down to the level of the Home Office crime code (perhaps with exemptions for acquaintance sex offences), especially since the location is aggregated to the street level. It might, however, be necessary to slightly increase the level of geographical aggregation in areas of low population-density.
I’m not sure that a little bit of information is better than none, because it entices people to make decisions based on insufficient information, but I’m not suggesting that the Home Office should stop providing crime maps. On the contrary I strongly believe that they should provide more detail, along with more information about the limitations of the information that they’re providing. What concerns me is the (sometimes implicit) suggestion in much discussion about police.uk that you can say anything meaningful about personal risk from knowing that there were two anti-social behaviour offences on my street in December and only one in January.