August monthly update and the strange non-impact of the riots

Just over a week ago, we uploaded the August data with some anticipation, not least because this is a time-consuming and difficult job. So thanks again to our Chief Data Architect for doing it so quickly.  We now have 9 months of historical data and UKCrimeStats is the only website where you can make comparisons over time with the crime data, right across England and Wales.

What really interested us though was the impact of the riots on this platform’s database. And our first impression was that there was no real impact. Wrongly, I reasoned to myself that at a national level, with over half a million new crimes and ASB incidents in England and Wales every month, a few thousand riot-related crimes were not going to make much of a difference.

Unfortunately, not only was I mistaken, it was much worse than that. According to this report in the Daily Mail today, it now turns out that the riot data has been officially massaged out under the guise of existing Home Office rules. Basically, if 100 people break into a shop and they all loot, this is technically recorded as one crime and so it appears that all riot crime from August has been officially deflated.

According to Peter Fahy, the chief constable of Greater Manchester Police:

‘The Home Office rules are clear and we follow them in a consistent way. We had more than 1,000 people involved in the disorder but that does not equate to 1,000 crimes. If four people break into your house then it is recorded as one crime in the same way as if one person had broken in. On the other hand if one person breaks into a house which is sub-divided into four flats then it is recorded as four crimes’.

Now I understand why, when I look at Northcote, the neighbourhood that contains Clapham Junction that was allegedly turned into some kind of post-apocolyptic war zone, all we see is a modest rise in Crime and ASB from 192 for July to 231 for August.

I can’t really see how this is a true and fair representation of what has happened.

And it raises all sorts of other questions like;

1) How many of the reported crimes are registered under this method, shouldn’t we know which ones and does it not under-represent the impact of gang crime?

2) Over time, does this not lead to a compound distortion and bias in regional crime figures, towards crimes committed by individuals?

3) Would it not conceivably create a lesser incentive for Police to investigate a scene of multiple criminal offences by a group when it is only recorded and shown to the public as a single crime?

4) Is it really so hard to report crime as one crime per criminal rather than one crime per group of criminals engaged in the same act? After all, they are charged, tried and sentenced separately on an individual basis according to the offences they’ve committed. So how do you feed that back into a clear-up rate?

5) Now that the crime data is a public good, isn’t it time to ask why do these rules exist in the first place, what is the public’s impression of them and isn’t it time for a review?

Comments as always most welcome.

4 thoughts on “August monthly update and the strange non-impact of the riots

  1. I think the main reason why crimes are recorded on a per-victim rather than per-offender basis is that the police often won’t know how many people were involved in a crime. When you go to the police station to report that you got home to find that you have been burgled, you won’t know how many offenders were involved (and if the police asked you how many were involved, you’d probably think it was a pretty stupid question).

    Sometimes you will catch the offenders and you will have some idea of how many were involved, but even then you won’t be sure. Many shop burglaries (looting) are caught on CCTV, but how do you know if the grainy blob in the picture is the same grainy blob you saw two minutes previously or a different grainy blob? And what about the accomplice who was standing just outside the view of the camera? And what if it takes two weeks to get the CCTV – how many crimes should you record in the meantime? Should we go back and change the previous month’s crime stats every time we find a new offender was involved? How about the getaway driver, or the guy down the pub who told the burglar about the delivery of new stock – should they be counted?

    Even if per-offender crime recording was possible, I don’t think it would b the right method to use. If someone breaks into a shop and steals two televisions, the loss to the shop is the same as if two people break in one after the other and take a TV each.

    As for the allegation that the figures are being “massaged”, I don’t think that’s fair. The purpose of the crime recording rules is to provide accurate, consistent figures to allow comparisons across time and space. The rules haven’t changed for several years and I’m not aware of any rule changes made to reduce the apparent impact of the riots. Anyway, why would the police want to massage the figures – does anyone imagine that people will forget the worst riots of the past decade because of the next month’s crime figures?

  2. I think the main reason why crimes are recorded on a per-victim rather than per-offender basis is that the police often won’t know how many people were involved in a crime. When you go to the police station to report that you got home to find that you have been burgled, you won’t know how many offenders were involved (and if the police asked you how many were involved, you’d probably think it was a pretty stupid question).

    Sometimes you will catch the offenders and you will have some idea of how many were involved, but even then you won’t be sure. Many shop burglaries (looting) are caught on CCTV, but how do you know if the grainy blob in the picture is the same grainy blob you saw two minutes previously or a different grainy blob? And what about the accomplice who was standing just outside the view of the camera? And what if it takes two weeks to get the CCTV – how many crimes should you record in the meantime? Should we go back and change the previous month’s crime stats every time we find a new offender was involved? How about the getaway driver, or the guy down the pub who told the burglar about the delivery of new stock – should they be counted?

    Even if per-offender crime recording was possible, I don’t think it would b the right method to use. If someone breaks into a shop and steals two televisions, the loss to the shop is the same as if two people break in one after the other and take a TV each.

    As for the allegation that the figures are being “massaged”, I don’t think that’s fair. The purpose of the crime recording rules is to provide accurate, consistent figures to allow comparisons across time and space. The rules haven’t changed for several years and I’m not aware of any rule changes made to reduce the apparent impact of the riots (although there have been cases of incorrect recording in the past). Anyway, why would the police want to massage the figures – no one is going to forget the worst riots of the past decade because of the next month’s crime figures.

  3. Thanks Matt, this is very well explained and argued. I agree my use of the word “massaged” was a bit unfair as they were only following the rules which I pointed out too. For all that though, I still think there is a distortion here because assuming next August is a normal non-riot month, then it will look much the same as August past which for those neighbourhoods it wasn’t. And August certainly wasn’t the same month as July to people living in the affected areas. Perhaps the solution lies in overlaying a separate stream of locational time-delayed offences/sentencing data?

Leave a Reply