Thursday, April 17, 2008

Those Damn Right-Wingers With Their Lousy Statistics

Always a bit scary, picking a scrap on stats - especially with only A-level maths and no statistics. One doesn't want to question someone's grasp of maths only to discover that it's one's own grasp that's shaky.

But the figures look OK to me and we're not talking deviations or any of that Chris Dillow stuff, so here goes. Doubtless the commentariat will pick up on any errors so let's jump off that 32-foot board :



Sunny is (again) an angry man :

This is an example of how right-wing papers feed us all b******* on immigration. The Telegraph today says:

More than one crime in five in London is now committed by a foreign national, raising fresh fears over the impact of immigration. Around a third of all sex offences and a half of all frauds in the capital are carried out by non-British citizens.



You need to read a bit slower, old boy. The Telegraph piece to which you refer came out in September 2007.

What's the problem, anyway ?

Except that, as Ministry of Truth points out:

So, for violent crimes, the top twenty list above weigh in with a total of 3812 offences, while the Met’s rolling 12 month figures give a total of 172,734 violent offences from March 07 to March 08 (down 5.3% on the previous years. As we’re only dealing a half year’s figures (and ignoring seasonal variations for ease) that gives us 86,387 violent offences of which these top twenty account for under 5%. So its not 1 in 5, its about 1 in 20.

Of course, this doesn’t include crimes committed by foreign national which aren’t solved, but if the Telegraph can compare apples and oranges to make its point, then so can I to show that their figures are a crock of ****.

Sex Offences? Full year is 8766, so a half year is 4383 and our top twenty weighs in with 263, and we have a figure of 6% and not a third as the Telegraph claims.

A crock of ****. Yeah, sounds about right. I think I’m going to complain to the PCC about this one.


"Sir ! Sir ! He's cheating, sir !"




Even though I'm politically a fair way away from him, I like Unity's blog. When he's good, he's very very good. And he's usually good. But ...

You see, he's taking these figures :



which are for SOLVED* crimes ("according to Metropolitan Police figures for solved crimes" says the Telegraph) in a six-month period - and comparing them with these figures - which are the Met's rolling totals for ALL crime in a one-year period. He's factoring down the Met numbers by 50% because they're for a full year, but he's ignoring the fact that only a small proportion of the Met numbers are solved crimes - which is what we should be comparing against.

And, tragically, he's labelling it "Lack of Basic Numeracy Skills", pointing out that "the Daily Telegraph needs a refresher course in statistics", that their stats are "a crock of poo", and calling the whole post Telegraph Can't Do Stats.

It's not a nice job. But somebody's got to do it. Let's do some statfiskics - like fisking, only with numbers.

Unity :

So, for violent crimes, the top twenty list above weigh in with a total of 3812 offences, while the Met’s rolling 12 month figures give a total of 172,734 violent offences from March 07 to March 08 (down 5.3% on the previous years. As we’re only dealing a half year’s figures (and ignoring seasonal variations for ease) that gives us 86,387 violent offences of which these top twenty account for under 5%. So its not 1 in 5, its about 1 in 20.

So the Telegraph are completely wrong ! At this point Unity does attempt some kind of exculpation for the crimes he's committing.

Of course, this doesn’t include crimes committed by foreign national which aren’t solved, but if the Telegraph can compare apples and oranges to make its point, then so can I to show that their figures are a crock of ****.

So my stats may be pants, but so are theirs, so I win ! That's probably an acceptable debating technique on a CiF comment box, but not on a post where you're flaying others for not being able to do sums. Unfortunately, he's wrong. As we shall see, the Telegraph appear to be comparing solved crimes by foreigners with all solved crimes - apples with apples. Unity is comparing solved crimes by foreigners with all crimes, solved and unsolved. Apples with lemons.

There's more :

Ah, look, this is the data for ’solved crimes’ not all crime, i.e. that committed and reported but not solved, so immediately these figures are questionable. It could be that foreign nationals are committing lots of crimes, or it could be simply that the Met is particularly good at solving crimes committed by foreign nationals, which could mean that these crimes get extra resources or that there’s just a smaller pool of potential criminals to target and that their nationality narrows the field down even further.
He's right. All these things are factors to be taken into account. Why doesn't he then ?

I think it unlikely that the police are "particularly good at solving crimes committed by foreign nationals" or that in London "there’s just a smaller pool of potential criminals to target and that their nationality narrows the field down even further". For the first I'd surmise that it's harder to solve crimes committed by foreign nationals, if only because there's likely to be less police intelligence on newcomers. Given the scale of immigration into London, where more than half of all births are to foreign-born mothers, I doubt very much that the 'small pool' holds much water either, and as for the nationality - well, how many people report that their house has been burgled by a Lithuanian ? My educated guess is that if x% of solved crimes are committed by foreign nationals, then approximately x% of all crimes will have been committed by foreign nationals.

What I think is likely is that some crimes will have a higher clear-up rate then others. Burglary, for example, is likely to have a lower "solved" rate than sexual offences, where a fair proportion of perpetrators will be known to the victim. If we knew these "solved" rates by type of crime, we could put the Telegraph and Met figures together and cook with both burners.

Alas, Unity doesn't know and the Telegraph piece doesn't say. But a quick Google DOES give an overall "solved" rate for the Met. Tony McNumpty released the figures when no one was looking, just before New Year.
Police authorities in England fared worse in the ranking than the majority of Welsh forces, with London’s Metropolitan Police being ranked the worst performer, solving just 21% of crimes – less than half of those cleared by North Wales Police.
At this point I'm going to take a huge risk and assume that Unity can add up. Let's look at his figures above again in the light of a 21% solved rate, accepting the distortions that we'll get because of different clear-up rates for different types of crime.


So, for violent crimes, the top twenty list above weigh in with a total of 3812 offences, while the Met’s rolling 12 month figures give a total of 172,734 violent offences from March 07 to March 08 (down 5.3% on the previous years. As we’re only dealing a half year’s figures (and ignoring seasonal variations for ease) and a 21% solved rate, that gives us (172,734 * 0.5 * 0.21) = 18,137 solved violent offences of which these top twenty account for over 21%. So it's not Unity's "about 1 in 20", it's 1 in 5 and the Telegraph is correct.

That was fun. Let's do it again, shall we ? Unity :

Sex Offences? Full year is 8766, so a half year is 4383 and our top twenty weighs in with 263, and we have a figure of 6% and not a third as the Telegraph claims.

And with that little 21% factor :

Sex Offences? Full year is 8766, so a half year of solved crimes is (4383 * 0.21) = 920 and our top twenty weighs in with 263, and we have a figure of 29% and not 6% as Unity claims.
Encore une fois !

Theft and handling is a little tricker - the Met gives figures for burglary and robbery but doesn’t mention handling, while the Telegraph’s copy leads on fraud, which isn’t mentioned at all in either their table or the Met’s stats. Still, we get, from our top twenty table, 3637 solved crimes while burglaries alone run to 93,894 a year, so that’s 46947 for the half year and a figure of just under 8%, which will almost certainly fall once handling offences are factored in on top.
21% ?

Still, we get, from our top twenty table, 3637 solved crimes by foreigners while burglaries alone run to 93,894 a year, so that’s (46947 * 0.21) = 9859 solved crimes for the half year and a figure of just under 37%, which will almost certainly fall once handling offences are factored in on top. Unity reckons 8%.
Management summary - Telegraph right, Unity wrong.

I hope Sunny appreciates the favour I've done him, flagging this up before he posts a CiF piece on the subject.

UPDATE - Unity in this comment seems to imply that you can't make a meaningful extrapolation from solved crimes to unsolved ones. I think you can, unless foreign criminals are uniquely prone to get caught, a proposition for which he offers no theory, let alone proof.

There is of course one crime with a very high "solved" rate - homicide. When asked for perpetrators by nationality under a Freedom of Information request, only 25 out of 43 England and Wales forces responded, covering half the homicides (I wonder who the refuseniks were ?). That showed 20% of killers as foreign nationals.

Oh, and across England and Wales, over 11,000 of the 81,000 prison population are foreigners. That's getting on for 14%.




(* “solved” - among other, better things (like convictions), the process by which some smackhead is persuaded to have another 626 offences taken into consideration in exchange for a nice steer from the police when giving evidence in the magistrates court. Our smackhead gets a lesser sentence, police “detection rates” improve and everyone’s happy except the people being burgled. If this is an inaccurate/unfair description of the process, could those police/magistrates who read the blog put me right in the comments ?)

5 comments:

Anonymous said...

I was in the police (before I fled the UK to the sanity of Oz a year ago). NO police crime statistics can be relied upon to ANY extent. They are routinely fiddled at division and force level to make each set of police managers look better (or less bad than they otherwise would). They cannot be factored in to any appreciation of foreign crime in the UK.

A far better indicator would be the number of foreigners held in prison (far more difficult to massage the prison population). Another excellent indicator is the amount spent on interpreters services.

I am reminded of a revenue and customs colleague telling me that a far more reliable indicator of the tax which should be paid by Indian and Chinese takeaways was the numbers of foil containers the meals are served in purchased rather than their declared earnings in their accounts which was almost invariably much less than the number of containers used would suggest.

On a more anecdotal and personal level the number of people I dealt with who I perceived to be foreign on the basis of declared nationality, accent or language was through the roof. I was sick of it so got out. The UK has plenty home grown scum. It doesn’t need to import any more.


kevster

Anonymous said...

"A far better indicator would be the number of foreigners held in prison (far more difficult to massage the prison population)."

Quite so

According to the DE the forces of Buiterism and their running dogs in the BBC and at the Guardian have selectively and disingenously leaked bits from the ACPO report to support their absurd contention that migration does not lead to disproportionate increase in crime


http://www.dailyexpress.co.uk/posts/view/41589

Anonymous said...

laban

I implore you to take on board the gravity of the threat we face from Buiterism

http://blogs.ft.com/maverecon/2008/04/imagine-theres-no-country/#more-178

The battle lines of conflict are being drawn now. On the dark side the Buiterists and their "allies" in the immigrant population, they are certainly stupid enough to cozy up to Islam. On the light side the champions of the indigenous population and possibly some immigrant allies.

You did great work, laban, in exposing the fraud of the Buiterist snake Brotton. But do you not see that these running dogs would be delighted to exile you, or stick you on a Scottish island? Or maybe worse, because as the Earl of Essex remarked early in the civil war "stone dead hath no fellow". This is all getting reminiscent of the last days of the Roman Republic when proscription was a common weapon against political enemies, it did for Cicero for example.

Given the ticking of the immigration time-bomb it is surely essential to do everything possible to bring the Buiterists to close action, as soon as possible. If Buiterism and all its running dogs were to be proscribed and stripped of their rights and property then this would advantage the indigenous people most prodigiously. Forces personnel discharged after years of long service to the realm have little chance of getting a decent home. Why not then promise to strip the Buiterists of their ill-gotten gains and make a present of them to these patriots? This would be a mighty blow at these snakes!

Buiter thinks that private property is sacrosanct and there is no thing as rights in the nation. Well stuff that! With his proximity to the financial trough he probably got his private property through highly disreputable means anyway. Why play to Buiter rules?

As I said yesterday we did not start this, they did, Buiter has made the nature and gravity of the threat crystal clear. We did not start it, but we must finish it, and finish them.

Anonymous said...

Most crimes are committed by males aged 18-30. Since this demographic is highly represented in the newest immigrant population (and since property crime follows the property), it's only to be expected that a young, family-less, indiscriminate immigrant population (not classified by education or skills) will feature significantly in crime statistics.

Simplemind

Anonymous said...

Re. differing rates of imprisonment as a good indicator of crime levels. This is reasonable for crimes which have a high detection rate, and probably is most reliable with homicides.

But there is some division of labour between the criminal subcultures of different ethnic groups.

Those ethnic groups whose criminals gravitate towards committing impulsive, high-visibility crimes with little organisational preparation will end up in prison more often than those following lower-risk strategies, such as fraud or cannabis cultivation. Eg Jamaicans versus Nigerians or Vietnamese.