The Telegraph's Conrad Quilty-Harper on why crime maps are rubbish
Over the last couple of days I’ve been blogging my notes from a panel about the reporting of crime statistics I attended at the “Data and news sourcing” event co-sponsored by the Media Standards Trust and the BBC College of Journalism. So far I’ve published posts looking at what was said by Michael Blastland, Andrew Trotter and Dominic Casciani.
Conrad Quilty-Harper
The final talk on the crime panel was from Conrad Quilty-Harper, data reporter for The Telegraph. He was scathing of the quality of the crime maps released by the government recently, not least because of the cost. He felt he could have done a reasonable job on the app himself given the data and £5,000, rather than the £300,000 that the government spent.
He was also concerned about the quality of the data being supplied - in his local area crime has gone up by 50%, which turns out to be from one very small number to another very small number. One of the crimes in the increase is classified as “other”, a category which Conrad explained covered a multitude of sins - including bestiality.
He drew a big contrast with the way that crime data is release in the US, showing maps of stop-and-search in New York which allowed for a deep drill-down into the data. He suspected that at some point in the delivery of the UK crime maps someone had said “We need crime maps like the one in Los Angeles” but nobody had really thought “What for?”. Conrad said that problems include that crimes lack unique IDs, the categories of crime are too broad, and as a reporter he wants to know which crimes happened yesterday, so that he can be working on the story today, not what happened three months ago.
I did disagree with Conrad on one point. He thought that the ICO was hampering the release of crime data on grounds of protecting identity which he seemed to imply wasn’t really justified. I think, however, that there is a wide range of offences like rape, domestic violence and the sexual abuse of children where putting a pin on a map to indicate the location of the offence would be a really unpleasant intrusion into the lives of people who had already been victims of traumatic crimes. I don’t see any compelling reason for the geolocation of those crimes to be anything other than “fuzzy” in public data sets.
Conrad showed murdermap.co.uk on the big screen, and said it was possible to use this as an educational tool. There had been a handful of murders near him, but the site allowed him to realise that they had all either been crimes of passion or gang-on-gang killings, and that he wasn’t likely to be on the receiving end of either.
He felt that the government and the police should be doing much more to reach out to journalists with their raw data. After all, he said, the audience usually come to us, the media, to answer their questions, and not directly to the state.
“But crime sells...”
I got to ask a question from the floor at the end of the panel’s talks, and wondered if we were all ignoring an important point. Like myself, the representatives on the panel were from news organisations that considered themselves serious, or were people with a vested interest in seeing statistics reported as accurately as possible. But salacious crime reporting and fear of crime shifts newspapers and glues eyeballs to the screen. Weren’t we all being a bit po-faced about it? Given accurate information that the risk of being a victim of crime was low and was getting lower, wouldn’t we turn readers away from our coverage?
Dominic Casciani responded by saying that we could do a better job of selling “the right fear”. Michael Blastland suggested that crime reporting was always going to be about scarcity - if there were 10,000 murders a day we wouldn’t report any of them. As he put it, more accurate data doesn’t “pour cold water on stories”, in fact the opposite is the case, it generates “more stories, better stories, and more accurate stories”.
Next...
After the session on crime stats, I dashed off to City University to watch Paul Bradshaw’s inaugural lecture there: “Is ice cream strawberry?”. I hope next week to pick up a couple of points he made and blog about them, but in the meantime I notice that Ben Whitelaw of the Wannabehacks have posted about one of the things Paul identified, the problem of ego in journalism, and how he is seeing bad habits develop at an early stage of a journalism career.
One thing I've often thought about things like crime maps, road death maps etc, is that they are a prime candidate for deanonymising mashups that take the time and location data, and then run searches over local press accident reports, court reports etc. For the tinkerer, it then becomes an ethical issue as much as anything about whether to construct such an application, because from a techie point of view, it's quite an interesting scenario...?