Feeds

back to article London stab murder rate entirely normal, says top stats prof

A top statistician has thrown a bucket of cold water on the stab murder media hysteria which has gripped the UK - and especially London - during the past year. Professor David Spiegelhalter, Winton Professor of the Public Understanding of Risk at the University of Cambridge, has just published a study on the subject in …

COMMENTS

This topic is closed for new posts.

Some would say.....

.....that the press play into the governments hands in order to support further restrictions on public liberty. By making people believe that it is all "for their own safety".

Of course, I would never be that cynical.....

0
0
Thumb Up

All normal?

Bet the good Prof lives in Midsomer, so of course this murder rate would actually be a little low.

0
0
Coat

risk

Murders per sq metre is what us walking around public perceive as a measure of risk.

0
0
Stop

Brits confused by statistics shock....

If we actually bothered to teach useful, everyday maths in school, we wouldn't have generations of people who are confused by statistics and relative risk.

Sadly, that would also mean the death of the nations newspapers, as the Daily Mail alone would be about 12-pages a day short of material once it had removed the "food x causes cancer", "super-food Y cures cancer", "Being born in March makes you lucky" and the aforementioned crime stats.

0
0

@stu-AC

Which would make Somalia safer than London. Your perception's even worse than your grammar.

0
0
Silver badge

Normal?

That's "normal" as in less than one standard deviation from the mean, I hope.

"Normal" in the ethical sense should be zero, any murder is seriously abnormal.

0
0
Gold badge
Go

Evidence based policy making

We've heard of it.

0
0
oxo
Thumb Down

Just..

Just cos it's normal for London doesn't make it acceptable.

0
0
Alert

Shock ...

The biggest shock for me in this article is that someone considers anything in the londonpaper to be "news". Surely some mistake...

0
0
Anonymous Coward

The Meeja

The meeja like to jump on a FUD bandwagon and then pimp their new ride for all it's worth until they get bored or something more interesting comes along.

Remember their short lived obsession with the word Tsunami? For a few months every earth tremor on the planet was analyzed by media "experts" for it's chances of causing a Tsunami. The two problems with this being that the discussions usually happened some time after the tsunami would have occured and that the great Sumatra-Andaman earthquake which caused the 2004 boxing day tsunami was either the second largest earthquake ever recorded and as such any tsunami caused by the earthquakes under discussion would have been much, much smaller. However the meeja like to scare us, so they wanted to depict tsunami as a new threat to mankind, as such they would be unlikely to tell us that the Lisbon earthquake of 1755 caused massive tsunamis to strike the west coasts of Europe and Africa. Even Blighty was affected. After all telling us that something similar hit England 250 years ago would take the edge off their "we're all going to drown in our beds!" hysteria.

Unfortunately the tsunami, knife crime and other stories only go to show that the collective news meeja of the UK have a very short attention span, little understanding of the concept of risk and a fondness for spreading fear.

0
0
Unhappy

normal killing?

there is nothing normal about one human stabbing another.

stats shmats.

0
0
Stop

Won't someone

Oh, wait; they're the little beggars stabbing everyone... Never Mind.

0
0

sillyfellow

"there is nothing normal about one human stabbing another."

History might disagree with you.. spears, swords, bayonets, etc. Human nature and ambition poitical figureheads being what they respectively are.

Being put in prison becuase you "assault" someone trying to stab you, now that's not natural.

</psuedo-sociology>

0
0
Bronze badge

Some calculations....

Having noted that the article slagged off the BBC correspondent for claiming that 4 fatal stabbings in a day is a statistical freak and then saying that 4 murders will occur once every 3 years (note the shift in definition), then I put the relevant numbers from the professor's papers through a Poisson distribution for overall murders and fatal stabbings alone. If I take the figures for homicides of 167 in total (note, yet another definition change, although I think it rather less ambiguous to count dead bodies than whether it's manslaughter or murder) and the number of fatal stabbings of 68 (or, if you prefer, killing by a sharp instrument), then I get the following probabilities of precisely 4 such events in any one day :-

4 fatal stabbings - 0.000042 (or once every 65 years)

4 homicides - 0.001156 (or slightly more that once every 3 years)

To make these calculations you have to make certain assumptions (that all the homicides are uncorrelated and hence random). Of course it's unreasonable to do that - gang fights, revenge attacks and so on, but it would give you valid numbers for where these are truly random events.

However, I cannot blame the prof here - but I can blame the Register's reporting of it. This is what he says in his paper :-

"So although four murders in one day is not a particularly surprising event when taken over a period of time, the fact that they were all stabbings is more notable."

So in other words four stabbings in a day is unusual (on a simple random statistical basis, very unusual). Most likely the mistake (as usual) is the assumption that the types of murders are uncorrelated in a period of time. In any case, the reporting of this in the Register article is not representative of what the professor is actually saying and doesn't indicate any great depth of understanding of statistics - more a wish to give a BBC correspondent a kicking.

0
0
Boffin

regression to the mean

nuff said.

0
0
Paris Hilton

I remember all of this at the time...

...walking down the street to the tube station last year after work and every day the Standard (and the free ones) would be bleating on... knife this, stab that, record-high the other... and I very well remember thinking what a load of carp.

This is exactly why I try and stay away from the kind of gutter-journalism in every daily paper and on all of the main TV news programs, esp the BBC but others as well... I just don't bother reading any newspapers or watching any news. Sitting on the train on the way home I'm often thinking 'I want something to read'.... and I can see a screwed up copy of 'The London Paper' sitting over on the seat opposite but I really can't be bother to go and pick it up because I know it wouldn't be worth my effort! Might as well read the telephone directory as nothing of any consequence is every printed in any daily rag.

Paris: because she'd take a good stabbin...

0
0
Flame

My favourite statistics..

Are the ones that follow (Doing/Not Doing/Eating *Delete as applicable) (insert activity/item here) (Hourly/Daily/Weekly/Monthly/JUST Twice/more than 14 Times since last pancake day *Delete as appicable) will make you (Insert number between 2 and 1000) (more/less *Delete as applicable) likely to (Die/get cancer/cure cancer/Live/Win something/Lose something/increase IQ/decrease IQ *Delete as applicable).

Woupty f*ing do, I drink a caffienated drink and my risk of developing life thretennign umbongo umbongo disease has increased by 38 times from one number smaller than 1 in 10^-6 to something in the range of 1 in 10^-5 both are small enough to not actually matter but mathematically incapcitated chavs and daily mail readers declare national outcry and government crackdowns.

** Please forgive my grammer/spelling unfortunately I am rubbish at using written english

0
0
DR

if they want to reduce numbers

if they want to reduce violent crime, then surely they need to make people more aware.

this mean that on the 1st of jan you start a murder count for the year, in headlines on a paper, and you continue that death toll throughout the whole year.

it's useless to say that on the 1st of march 90 people had been stabbed because that's a breif and fleeting headline that is forgotten about the next day.

if you want to raise awareness as a way of reducing the numbers then you need to push the numbers onto people every day.

then people will not want to be the next statistic, people carrying knives, (knife carriers are more likely to get stabbed) would then surely have to think twice about carrying a knife.

(and they should just do away with those knife amnesties, all that seems to do is serve as a bin for people to put their old kitchen knives into the bin, and being as there is a home wares shop in practically every town it's not like they are genuinely reducing the amount of knives on the street as they are easily replaced.

0
0

Normal?

Yes. Acceptable? No.

Still, wha'yagonnado?

0
0

Normal?

Murder might not be nice, but it is normal, in the sense that it happens consistently, as the statistics show. The whole point of the profs work is to show that nasty horrible things do happen with annoying regularity, regardless of how much work is put in trying to stop them. That is what risk is. At some point you have to accept a certain murder rate and not continue coming up new an improved ways to reduce an already minimal risk, usually at huge expense and inconvenience. Sure you could reduce the knife crime rate by banning all sorts of knives, putting up detectors everywhere and putting millions of police on the streets. But the point is, you will never ever get rid of it completely as it is a “normal” part of the human condition. If people don’t have knives to kill each other with, then pointy sticks, broken bottles or half bricks will just have to do.

If you don’t get stabbed, run over, burnt or break your neck, then you have cancer, heart attacks, stokes and cholera instead, not to mention, asteroids, super volcanoes, escaped animals, floods etc, etc.

So please try not to concentrate your attention on one single way of dying while ignoring all the others.

0
0
Boffin

The media don't want to scare us

they just want to tell us what we want to hear.

When does 'ordinary' make the news?* Compare readerships of Independant vs The Sun.

( * Apart from this time, of course. )

0
0

SHOULD be?

"Four murders on the same day in London would be expected to occur about once every three years, and it has done," says Spiegelhalter. "Seven days without a murder should occur about six times a year, and it does."

Um, no. Seven days without a murder SHOULD occur fifty-two times per year.

0
0

Well thats ok then...

So when I get stabbed to death on the streets of London, and feel my life ebb away whilst laying on the pavement, whilst an ambulance may or may not be on the way, I can assure myself that its just a normal statistical outcome.

Makes me feel so much better.

0
0
Thumb Down

Oh ok....

that makes it alright then.

There are 10's of thousands of rapes and murders in South Africa each year, but that's acceptable as it's been like that for a while now...

Stats, gotta love them.

Just set a baseline and take that as acceptable.

0
0
Bronze badge

Inconsistencies

I wonder if the the academic is also playing with words a bit :-

*******

'Tighe said: "To have four fatal stabbings in one day could be a statistical freak."

Au contraire, says the prof. It was a normal event, to be expected in London at regular intervals.

"Four murders on the same day in London would be expected to occur about once every three years, and it has done," says Spiegelhalter. "Seven days without a murder should occur about six times a year, and it does."

********

So the prof is rubbishing the BBC correspondent and, apparently, equating 4 fatal stabbings with 4 murders. Unless all murders in London are stabbings (and all fatal stabbings are murders), and there are none by any other means then you can't blithely substitute one for the other. Not to say there isn't a lot of media hysteria over these things, but at the least this was careless with words.

0
0
Thumb Down

"since 2004"

So not exactly a large set of data to evaluate then. How about looking at the rate over the past 20 years to see if it has changed? Did Labour pay for this study by any chance? Lies, damn lies and statistics.

0
0

Martin Lyne

quote martin : "History might disagree with you.. spears, swords, bayonets, etc. Human nature and ambition poitical figureheads being what they respectively are.

Being put in prison becuase you "assault" someone trying to stab you, now that's not natural."

history you say?... and oh how wise this blood soaked 'history' of ours has proven to be! NOT. you would have thought we'd have learnt something about accepting killing as 'normal' by now...

bit i do agree with your second statement.

0
0
Happy

"umbongo"

Is that not a for of Linux?

0
0
Anonymous Coward

@justin clements

what do you want then? For your hypothetical death scenario to be not counted in statistics or for it to be counted as 'two' for reasons of special significance or for them to make a new category for the statistical reporting that records number of dead people requiring more assurance or otherwise better feelings?

0
0
Silver badge

Bah

[Linbox ] No, *useful* maths would be the sort that told you on which days the stab index would be high so you could not take your visiting relatives to see the Tower of London that day, leaving the sightseeing for a more auspicious one and allowing the indigines to get on with slitting each other up in peace (which I'm sure they'd prefer).

[Steven Jones] Well said that man.

0
0
Bronze badge
Go

88.5% of statistics are unreliable. And that's a fact.

Also, there is a direct correlation between the number of ice cream sales and the number of shark attacks. Food for thought, perhaps, for the Daily Mail.

0
0
Flame

What

Glasgow and Scotland has had a problem with knifes and stabbings for decades and London and Westminister barely seem to care, yet when London shows some sort of rise there is "hysteria".

0
0
Anonymous Coward

Normal

in this context means statistically.

Arguably if statistically X number of people are fatally knifecrimed per year than that number is itself normal. Normal does not mean right, or desirable, or moral, it means normal:

Normal; Adj. according with, constituting, or not deviating from a norm, rule, or principle b: conforming to a type, standard, or regular pattern

Norm; n. a: a set standard of development or achievement usually derived from the average or median achievement of a large group b: a pattern or trait taken to be typical in the behaviour of a social group c: a widespread or usual practice, procedure, or custom

So, hand-wringing, tree-hugging, singing-kumbaya-round-the-campfire, hippy bleatings about how we should all be less horrible to each other may be valid and relevant, but not here.

0
0
Joke

Simple solution...

Stab everyone in London! One day of statistically anomalous knife crime, followed by a massive sustained decrease thereafter.

My aren't statistics lovely :)

0
0
Coat

I may be an old fart, but

I don't remember this much street violence when I was young.

Mine's the one with the Kevlar lining.

0
0

stats

I'm pretty confident that you are 6 times more likely to die falling down the stairs than being stabbed.

Won't someone think of the stairs?

0
0
Coat

Lies, damn lies and statistics

I think Homer Simpson explains this kind of situation best.

'Oh, people can come up with statistics to prove anything, Kent. 14% of people know that."

0
0
Joke

Only in London

Still bringing knives to a gun fight...

0
0
Unhappy

sample ?

"covering crime since April 2004, there has been little change in the capital's murder rate over the past five years."

Suggesting something happens every three years based on a five year cycle appears dubious at best. I suspect the register is oversimplifying ... which would be somewhat ironic

0
0
Pirate

Chavs use knives?

I thought they ate with their hands?

0
0
Bronze badge
Paris Hilton

Me too...

I haven't been fatally stabbed in the last seven days, does that count?

0
0

Ah, statistical significance!

With a big enough sample, there can be a "statistically significant" difference in height between groups - but an actual difference of a quarter of an inch . Statistically significant, but of no real-world significance to a man who want to know what range of trouser sizes to stock in his shop.

0
0
Bronze badge

@regression to the mean

Er, no. This is tail behaviour we're talking about.... ie, what's at question is the statistical significance of rare events, not the significance of local (recording-period on recording-period) variations. I would have thought that was quite clear from both the article and the comments...

0
0
Anonymous Coward

I may be an old fart, but

> Mine's the one with the Kevlar lining.

Er, Kevlar isn't very good protection against knives. Bullets, yes, knives no.

0
0
This topic is closed for new posts.