Did immigrant crime really double in just THREE years?

'Never trust the Mail on crime and immigration' I said last Tuesday in response to the Mail's 'Migrant crime wave revealed: Foreign arrests have almost doubled in just THREE years' from the Mail, which was subsequently churnalised by the Telegraph and Express.

And now, thanks to hmallett in a comment on that very post, I have a bunch of links to some of the sources, which means I don't have to make a bunch of FoI requests and wait weeks for the results to find out how much shite the Mail is trying to peddle. Hurrah!

Multiple choice time! Which of the following things is wrong with the Mail's data:

A) The Mail could be merging data that isn't compatible
B) The Mail might be comparing an incomplete year's figures with a complete year
C) The way these figures are measured may have changed
D) Police may be putting greater emphasis on recording nationality
E) Some of the figures might be crappy 'typos' and not the actual figures at all

The answer is...all of the above! Seriously, Jack Doyle has managed to ignore caveats about the first four things in the list with some of the figures and include at least one buggered up number that make everything look just that bit more scary. Well done, that man.

We have the results from the Met, Durham, West Mercia, Cambridgeshire and Bedfordshire (cover letter and stats) to go on. If the Mail can use incomplete data then so can I, suckers!

Merging data that isn't compatible
Firstly, the Metropolitan Police has this to say:
Comparing this data directly with information supplied as part of another FOIA request, or with data from other sources may result in distorted and misleading figures.
It should be noted that for these reasons this force's response to your questions should not be used for comparison purposes with any other response you may receive.

Durham says:
Please note that Durham Constabulary’s response to your request is unique and should not be used as a comparison with any other Force response you receive.

I wonder how long it took after seeing this before Doyle (or whoever pulled the figures together for him to write up) thought, fuck it. Let's add them together anyway.

Comparing and incomplete year's figures with a complete year
The Met says, in the only one of their notes that have been included in bold:
It was not possible to retrieve data prior to 1 April 2008 as the previous custody system does not hold nationality data.
So that big fancy-Dan table in the middle of the page is comparing a year's worth of data with nine months' worth of data. Totally accurate though, I'll bet.

This isn't an insignificant tiny difference that might skew the figures just a little, The rise in arrests in the Met area accounts for half the rise recorded across all forces the Mail has data for. The splashy headline would be ruined without the Met's figures.

Away from the scary looking table, the original article made a fair bit out of Cambridgeshire police only arresting 27 foreign nationals in 2006 and 4,803 in 2010. It did a calculation of how many thousands of a percentage increase that would mean and everything. That bit's gone now (This could explain why there's a big picture of Cambridgeshire's MP with no quote to match it), presumably because someone pointed out the bit on Cambridgeshire's reply that says:
Information prior to 1 January 2007 is no longer retrievable.
Which might explain the low number of arrests of foreigners recorded for 2006. I wonder who it could have been who got that bit removed? Cambridgeshire's Chief Constable is on the PCC Committee by the way.

Still, the claim is still there in the Express and in the Telegraph. Wa-hey!

Changing the way figures are measured
Almost all the forces I have figures for point out that they've changed the way things are recorded in the years 2005/06/07/08. This explains the Met's partial information for 2008. It explains other things, too.

Durham Police say this:
The [custody recording] system was introduced to the force at the end of 2005 and during its first years several data quality issues were identified, including that the Nationality field was often completed as "unknown" or "not applicable". This issue has now been resolved hence increased accuracy in more recent years.
Right. A new recording system with inaccurecies in early years that have been fixed more recently. Could this possibly explain the jump from 147 arrests in 2008 to 418 in 2009? Gee, do you think?

West Mercia say:
There was a change to recording of information on WMP systems in 2007 and before this we do not have the facility to filter search criteria by nationality, therefore we are unable to provide full data for 2006 and partial data has been provided for 2007.
Ah. Do you think the new system from 2007 might have taken a while to bed in and affected the recording of 2008's figures? I wonder. No I don't.

Bedfordshire say:
The information provided only goes back to 2008, because a new custody system was introduced at teh end of 2007 and the Go Live dates throughout Bedfordshire varied therefore, 2008 is the first full years worth of data available in the custody system since it went live.
Boringly familiar now, huh?

Greater emphasis on recording nationality
What Durham Police said, above. Since the systems for recording information are changing, it most likely follows that for other forces too, natoinality was less likely to be recorded properly in the period of the system bedding in. Which conveniently happens to be the year the Mail has its first set of figures for.

Crappy typos
Big scary table records Bedfordshire as having only 39 arrests in 2008. Bedfordshire says 339.

As stupidly unreliable reporting of data about foreigners goes, this story ranks up there with the one about David Coleman's submission to the Economic Affairs Committee in 2007. Coleman included a big table of potential costs of immigraition with a massive disclaimer where the total should be saying, in effect, "Don't add these together to make a total because it would be misleading, since some of them overlap and it's not certain how much applies to immigrants" (which kind of begs the question 'why include the feckin' thing anyway?').

The Mail added the figures together, added an extra 900 million pounds to it, made the fake total the basis of the headline, amended some of the entires on the table that said 'ethnic minority' to say 'immigrant' and included the table in another article on the website without attributing it to make it look like they were part of the government's submission to the same committee.

Nice to see the paper (and its followers) staying classy.


adamknott said...

Look's like we've found April's bullshit of the month winner already..

hmallett said...

Happy to help!