Seen your local crime map yet? Crime maps map crime. Crime maps pinpoint crime hotspots. Crime maps are smart and multi-coloured. Crime maps enhance national information capacity. Crime maps empower civilians through spatial dissemination of localised statistics. Crime maps make the populace feel safer.
Or perhaps not. Crime maps are actually statistical drivel, presented in a pretty way, to keep idiots happy. Let's see how.
As of this week, crime maps are now available for the whole of England and Wales. Not just Brixton, but Bristol, Bodmin and Bangor too. Enticing, isn't it? Quick, type <your local police force> and "crime map" into Google and see what's going down round your way. But don't expect little flags marking every burglary, knifing and embezzlement, because that would be an infringement of personal liberty and act like a magnet to potential criminals. Instead you'll notice that the map is shaded, and the shading tells you whether crime in an area is high, low or merely average. And it's probably average. Most things in life are roughly average, and crime statistics are nothing different. So expect to see a map coloured by indistinguishable mediocrity. Not terribly informative, sorry.
Beware of what average actually means. Average here means average for your particular police force, not average for the nation as a whole. So the western tip of Cornwall has average crime, and so does Tower Hamlets, at least according to the relevant crime maps. But Lands' End's crime rate is 2.8 crimes per 1000 people, which is average-ish for Devon and Cornwall, whereas Tower Hamlets' crime rate is 10.4 crimes per 1000 people, which is average for London. In truth living round my way is four times more crime-ridden than the Cornish fringe. But apparently not in mapland, where we're all averagely normal.
And beware the broad brush categorisation of averageness. On a crime map, "average" means "within one standard deviation of the mean". And this is where most of the data lies. On the latest Londoncrime map, this leads to a vast yellow sea around a tiny central island of darker colours. All but three London boroughs are categorised as average, even Hackney and Southwark, because their figures lie sort-of-close-ish to the London mean. This just leaves Westminster, Camden and Islington as central hotspots, with Westminster the only London borough whose crime is 'High'. Heavens it must be awful there, please stay away. Or maybe not.
Crime maps are compiled using monthly crime figures, despite the fact that months are desperately irregular units of time. Three extra days this month compared to next? Never mind. Five weekends this month but only four last month? Maybe it's not surprising that crimes related to drunken behaviour appear to be increasing. And it was pretty damned cold in December, bur rather warmer the month before. Should we expect decreases in temperature or reductions in daylight to have an effect on underlying crime statistics, or do we just conveniently ignore this and look at the headline figures instead?
Click on your map, be it Barnet or Berwick, and you can dig down to view the data at sub-district or tiny-ward level. But beware, because the further down you go the fewer the number of crimes being used to judge what colour the map should be. And statistics based on infrequent occurrences tend to be unreliable, insignificant and subject to random variation. Not that you'd guess this from the maps. There were only 76 reported crimes in Bromley-by-Bow in November, for example, down from 82 in October. Wow, an 8% decrease in crime in just one calendar month, how marvellous! Whereas in neighbouring Bow East the number of crimes rose 9% from 111 to 121. How terrible, how scandalous, how scary! Or more likely just a reflection of random fluctuation and entirely meaningless.
You can dig down even further to study different categories of crime. Take residential burglaries, for example. Now there's a crime to fear, and there's plenty of data here for the concerned Neighbourhood Watcher to get their teeth into. So what does the crime map tell me about my local area? In the ward to the south of Bow Church there was one residential burglary in November. Sounds good, but apparently one burglary is coloured "Average". And in the ward to the north of Bow Church there were no burglaries in November. Now I call that "Lucky", but the crime map colours it "Low". At this level the map's like a toddler's painting, covered with random-hued splotches, representing insignificance.
And beware the empty pointlessness of computer-generated statistics. You can play this game anywhere in the country. Violent crime in Rutland is up 40.2% on last year (it's almost one a day now). Anti-social behaviour in southern Peak District villages has increased 70% since 2007 (from very little to not much). And robbery in Bowness-on-Windermere has decreased by 100% since this time last year (down from one theft every three months to none at all). These are tales told by an idiot, full of sound and fury, signifying nothing.
And does your average UK citizen understand the underlying mathematics behind all this? Do they hell. Can they interpret these statistics with the caution they deserve? Not likely. Can they resist taking random percentage changes at face value? Of course not. Or do they just see a big red splodge on a map and go "Oh my God, it must be really scary to live round here, and it's getting worse, I shall never venture out again"? I fear so.
This is freedom of information in action. This is the new auto-statistics future. This is geospatial data churned out with invalid justification. This is the government producing online maps because they can, not because they're useful. This is tickbox social policy. This is meaningless statistical bollocks. And if there's any crime here, it's that anybody's taking these maps seriously.
