Wednesday, July 23, 2008

Pinpointing Crme

Heather Brooke in the Guardian last week suggests that the reason crime statistics are not published at a very detailed level by the Met is because they are concerned about being held to account for crime:
Knowing what crimes happen and where is important for several reasons. First, people want to know how safe (or unsafe) they are. They need accurate and detailed data if they are to form an opinion of the safety of their neighbourhood. When they know what's happening, they are in a better position to help or support the police. They are also better able to hold the police to account. This is perhaps what the police fear most...
Perhaps this is oversimplifying the issue. There has been much discussion within academic and police circles about the anonymisation of crime statistics, what precisely would constitute adequate anonymity? Is just leaving out the victims name but providing precise address details sufficient? Remember that a unit postcode in urban areas represents about 12 houses, would it be ok to pinpoint a crime to that extent? Heather Brooke talks about mapping crime to "street level" - unfortunately defining a street for this purpose may not be as easy as it sounds, how would one distinguish between a cul de sac with only 4 houses and a major thoroughfare with several hundred? There may also be some crime types which need to be anonymised to a greater extent than others to ensure that no one can deduce the identity of the victim (or alleged perpetrator) from the apparent location shown on a crime map - family violence and child abuse for example might fall into this category.

Publishing information that can be de-anonymised (horrible word) through local knowledge for example could be counter productive and might indeed breach a victims right to privacy. The potential difficulties are not insuperable and I do not believe that the reluctance to publish detailed information by the police service is part of an effort to avoid being held accountable.

Perhaps what is needed is a carefully considered model of what information should be published at what level of granularity to protect the rights of the victim whilst providing useful information to the public. One for the Jill Dando Institute?


Charles said...

All good points, Steve, especially about how anonymised a (for example) postcode would be.

But I'd take issue with this: "I do not believe that the reluctance to publish detailed information by the police service is part of an effort to avoid being held accountable."

Well, there's definitely resistance among police forces to have their crime stats revealed even to each other; Brian Paddick explained as much on the Today programme (there's a transcript on the Free Our Data blog).

Steven said...


Reading through the interview transcript at it seems that Brian Paddick is suggesting that the principal way that the crime maps would impact on crime levels is through peer review rather than public scrutiny.

If there was a national site that accumulated and where appropriate aggregated crime statistics people would not only be able to understand trends in their neighbourhood but would also be able to benchmark one force against another - would be powerful for the Home office as well.

Christopher Osborne said...

Unfortunately we're dealing with a public service that can't even introduce a nation-wide database system.

They need to sort out their base infrstructure now, and then start thinking about analysis. All police forces store their data according to different conventions, any analysis on inconsistent data is bunk.

You'd have one police force proudly displaying their reduced crime dates and another one claiming that they'd cooked the books.

Lets not even talk about the fact that the Police and Ambulance services don't even share data...

Steven said...


I think we agree.

We need a national database which drives a standard approach to recording stats and would allow meaningful comparisons. The current situation which each force deciding what to publish and at what level of granularity and currency is not ideal.

Data sharing between emergency services is another topic but i would imagine that the police would be happy to at least share the data they are willing to make public through some form of interface.

sexy11 said...