Conditional probability watch

by Chris Bertram on August 3, 2005

Eve Garrard at Normblog :

The statistics suggest that the chances of a Muslim man being killed by the police are considerably less than the chances of a Muslim man being killed by suicide bombers, given that the latter make no effort to avoid killing Muslims. So assuming that these policies do indeed prevent some successful bombing attempts, then people who reject them in favour of ones which don’t impinge more on Muslims than on others are actually prioritizing policies which will save fewer Muslim lives over ones which will save more Muslim lives.

I suppose the conclusion might be true …. and I don’t suppose we actually have any statistics that would allow us to estimate the chance of a Muslim man being killed by the police. There are about 1.3 million male Muslims in the country, and Garrard takes the chances of any person from the whole population being killed by being a suicide bomber as relevant to their chances of being killed in that way: 1 in a million? Is their prospect of being killed by police marksmen more remote than that? Anyway, “statistics suggest” that they are surely safer either than men carrying table legs in a suspicious manner in a public place or Brazilian electricians boarding tube trains.

{ 18 comments }

1

dsquared 08.03.05 at 2:55 am

The last crop of suicide bombers did, in fact, try to avoid civilian casualties – or at least they say they did, and the fact that their explosions went off at end-of-line stations at lunchtime suggests that they certainly weren’t trying to maximise.

2

Chris Bertram 08.03.05 at 3:15 am

I don’t think, Daniel, that the somewhat contradictory reports from the Italian press are a good guide to anything, and whether or not they were trying to maximize they were certainly trying to kill (those close enough to them to have noticed their disappoinment are lucky to be alive). So there’s no grounds for your assertion that “in fact” they were trying to avoid civilian casualties.

(BTW I meant my post to be about EG’s dodgy reasoning …)

3

Matthew 08.03.05 at 3:48 am

There aren’t 1.3 million male muslims in the country. In 2001 there were 1.59m male and female muslims.

4

Chris Bertram 08.03.05 at 3:50 am

I think I got my figure from the CIA World Factbook. Anyway, a lower figure makes Eve’s argument even worse.

5

Brendan 08.03.05 at 5:33 am

Increasingly you feel that the pro-invasion mentality can only be investigated from a psychological point of view. The old Freudian concept of ‘Projection’ (whereby you project onto an external object/person what you are unable to face up to about your own position) seems to be the only explanation for statements like this:

‘the multiculturalist agenda seems to favour form over substance, the appearance of protection over its reality, the preservation of the trappings of ‘respect’ over the preservation of lives’, accusations, all of which, with a reasonable degree of accuracy, might be levelled at the pro-invasion position.

Seriously, though, what are we to make of the incoherent farrogo of half-truths, quarter truths, and sloppy argumentation that make up ‘Eve Gerrard’s’ argument?

To take one example.

‘The statistics suggest that the chances of a Muslim man being killed by the police are considerably less than the chances of a Muslim man being killed by suicide bombers, given that the latter make no effort to avoid killing Muslims.’.

This is of course true, given that the guy the police shot was a Brazilian. But the point is not that the police now seem to have a ‘shoot-to-kill policy towards those suspected of intending to detonate bombs’. The point is that the police now apparently have a shoot to kill policy against people who no sane person would have thought would have an intention to detonate a bomb.

The key point is not that the police are killing a ‘disproportionate’ amount of Muslims: the point is that the police seem to be shooting innocent people per se. Frankly I don’t care how many of these people are Muslim. I don’t want the police to shoot me .

Lest we forget:

‘De Menezes, an electrician, was travelling to north London to fix a fire alarm.

He was not wearing what witnesses called a “black bomber jacket”, but a denim jacket. It was about 17C and his clothing would not have been out of the ordinary.

He did not vault a ticket barrier, as claimed. He used a travelcard to pass through the station in the normal way. His family believes that he may have started to run simply because he heard the train pulling in — something Londoners do every day. Indeed, a train was at the platform when he got there.

Police clearly believed that de Menezes might have been a suicide bomber, even though he was not carrying a rucksack . This raises a key question: why was de Menezes allowed to board a bus in Tulse Hill and travel to Stockwell, if officers thought that his body might be rigged with explosives? It also raises questions about the new shoot-to-kill protocol. The protocol — which is specific to individual targets — can be put into force only when police have reason to believe that a suspect may be carrying a bomb. The order can be issued only by a “gold commander” at Scotland Yard.

The order, once given, clears officers to shoot the suspect in the head if they believe that he is about to activate the bomb. The idea is to give the individual no time to react. Police do not have to shout a warning before they act: to do so would negate the effect of the head shot.

Some witnesses say that de Menezes was given no chance to give himself up. They say that once on the train he was pinned to the ground and shot.’

Of course (and this is, yet again, a pro-invasion straw man argument) no one will deny that the police should have the right to kill someone, even by shooting them in the head without warning if they have reasonable grounds to think that he is a suicide bomber .

But the key point here is: did the police actually have such reasonable grounds? To put it even more bluntly, did they have any grounds at all?

To go even further: how, precisely, will shooting innocent Brazilians with no bombs cut the risk of terrorism? Let us not forget: number of suicide bombers currently having been shot by the police: zero.

I’m going to skirt over the issue as to how Tunku Varadarajan begins by defending ‘ethnic profiling’ (which is contentious enough), an argument that Gerrard uses before gliding effortlessly to the far more controversial ‘shoot to kill’ policy as though Varadarajan was defending that (he wasnt).

We should also not forget that it is a matter of public record that the Met is ‘institutionally racist’ and that decisions as to how it ought to act should be considered in that context.

6

Anthony 08.03.05 at 5:35 am

The risks of both situations also depend on numerous other factors, such as where you are in the country, your mode of transport, if you are exiting a watched house, and thousands of other possible differences.

Perhaps it is better to look at this in a different way. We give vaccines to the population in order to prevent a loss of life from disease, but all vaccines have side effects which certain members of the population will suffer (and some may have a greater propensity to these effects).

In the same way we are all subject to security checks to some extent, but the burden may fall on some groups that fit the profile more than others. However, the point is that this is undertaken with a good intent – to prevent loss of life in the general population (including that of innocent Muslim men sat on buses).

It seems to me that this debate is as much about a low-risk society suddenly coming to terms with the fact that risk is not avoidable. Risk is not equally distributed – and never will be. Concerns about profiling and police action as a risk seem to be predicated on the mistaken foundation that a no-risk situation can be engineered, when in fact outside agents are bringing the risks in the form of bombings. The fact that a risk is attached to preventing bombings, does not mean that such prevention is not necessary.

In absolute figures the damage of the bombings far outweighs the damage from Police action so far – and that includes the deaths of Muslims.

7

Brendan 08.03.05 at 6:13 am

‘In absolute figures the damage of the bombings far outweighs the damage from Police action so far – and that includes the deaths of Muslims.’

Ipso facto, given that the police have not yet shot any suicide bombers, the benefits from the ‘shoot to kill’ policy are currently non-existent. Justification of such a policy requires the belief that at some point in the future this policy will actually have some benefit, which may be true or false or half true but at the moment is merely a belief with no objective evidence to back it up. This is where the comparison with vaccines breaks down.

Does any reader now feel any safer knowing that apart from the current risks from terrorism, they now face as well the risks of being shot for no reason by the police?

8

Wrong 08.03.05 at 6:38 am

Leaving aside the statistics (and the fact that racial profiling is useless), Eve’s argument is bad moral philosophy, too. She writes:

The only reason I can think of for doing this is that the most important thing is felt to be the avoidance of any policy which bears more heavily on Muslims than on other groups, even if that avoidance actually harms Muslims more than the alternative. So what appears to matter most isn’t saving Muslim lives, but ensuring that there’s no disadvantaging of Muslims compared to non-Muslims, even if a disadvantaging policy would actually be better for Muslims.

With respect to these policies, the multiculturalist agenda seems to favour form over substance, the appearance of protection over its reality, the preservation of the trappings of ‘respect’ over the preservation of lives.

But unless you endorse the simplest form of utilitarianism, it’s perfectly sensible to reject a policy which discriminates against Muslims, even if that leads to worse outcomes for Muslims. (Non-utilitarian) ethics is generally about acts, and about qualities which they have independent of their outcomes: are they courageous or cowardly; do they respect rights or violate them; are they racist or not? So it may be wrong to perform an act which would prevent some bad thing occouring. The fact that doing the morally right thing may carry risks is not usually felt to be an argument against the morality of it.

Perhaps Eve thinks that all rights-based and virtue-based ethical discourses are just ‘form over substance’; perhaps she thinks the only thing at stake in an anti-racist ethics is “the trappings of ‘respect'”. But if she does, she’s going to need to argue for those positions.

9

James Wimberley 08.03.05 at 9:16 am

The argument in the link is a straw man. Given that there is a higher prevalence of a given crime in a given social group, and that is all the police know, then it makes sense for them to target that group. This is true even if the difference is small. If 5% of all young males commit street crimes, and 6% of black young men, a short-run rational random policing policy concentrates on blacks. Of course this is wrong on discrimination grounds (since ex hypothesi there are large numbers of white street criminals as well) and likely to be counter-productive. (If the police have additional knowledge about frequent offenders, as in the straw-man counter-argument, then they should change the policy following Bayesian rationality.)

Now suppose the crime is much more concentrated. Northern Irish sectarian terrorism is – was – committed by white men (and a very few women) with Ulster accents. Basque terrorism is, wait for it, committed by Basques and nobody else. Jihadi terrorism is committed by young Muslim men of a variety of ethnic origins. Since on the street there is no such thing as a “Muslim” appearance, this means young men of Middle Eastern or South Asian or Afro-Caribbean appearance. This is a very large group. Latinos like the unfortunate Mr De Menezes overlap with this group in appearance so are likey to be included in the target group for rational street-level policing.
Is it really discrimination to focus on the smallest superset potentially including all the criminals? Or should the police be blind to common sense as well as colour, and check the handbags of old white and black and Asian women for explosives?

10

Brendan 08.03.05 at 9:50 am

Could I point out, yet again, that the worthless article by Eve Gerrard that began this thread effortlessly and wrongly elided between arguments for racial profiling and arguments for the shoot to kill policy, as though to agree to one commits you to another? The key point is not about racial profiling (well it is a point but not, in my opinion, the key point). The point is, should the police be allowed to kill innocent people, and if so, how can we stop them?

By this meaningless sentence: ‘The statistics suggest that the chances of a Muslim man being killed by the police are considerably less than the chances of a Muslim man being killed by suicide bombers, given that the latter make no effort to avoid killing Muslims.’ Gerrard attempts to argue that giving the police the right to shoot (effectively) anyone they want will make us all safer.

How?

11

Brian 08.03.05 at 9:50 am

Of course if you really want to get the targetting accurate, you should focus on searching people who are known to terrorise the English, such as this guy.

More seriously, the outbreak we’ve seen of more or less unqualified utilitarianism in posts like Eve’s is rather odd. The England of better to let 100 guilty go free than convict (kill) one innocent seems like a wistful memory. Now it’s just a matter of having fewer improper killings than proper ones. (Not that we’ve met that standard either.)

12

Brendan 08.03.05 at 9:57 am

‘The point is, should the police be allowed to kill innocent people, and if so, how can we stop them?’

This should of course have read ‘and if NOT, how can we stop them’.

13

Sebastian holsclaw 08.03.05 at 11:03 am

“The England of better to let 100 guilty go free than convict (kill) one innocent seems like a wistful memory.”

Isn’t the traditional number 10 guilty?

14

dearieme 08.03.05 at 11:53 am

Matthew, do you really believe that any of us knows the number of Muslims in the country accuate to three significant figures?

15

Kevin Donoghue 08.03.05 at 12:03 pm

Isn’t the traditional number 10 guilty?

Volokh did a lengthy piece on this:

http://www.law.ucla.edu/volokh/guilty.htm

16

Daniel 08.03.05 at 12:59 pm

Matthew, do you really believe that any of us knows the number of Muslims in the country accuate to three significant figures?

Since there is a census taken every now and then and Matthew is quoting from the 2001 one, I would say “yes”.

17

abb1 08.03.05 at 2:51 pm

Probabilities: apparently the US Department of Justice projects that 32% of African-American men born in 2001 will spend time in prison. Interesting.

18

bryan 08.04.05 at 4:25 pm

‘If 5% of all young males commit street crimes, and 6% of black young men, a short-run rational random policing policy concentrates on blacks’
I would appreciate a breakdown of what percentage of police manpower should be expended on targetting the black youths contra what percentage should be expended on white youths in order to maximize arrests of both groups. you will of course also note how too high a focus on the black group, even if that group is the one with 1 percent higher a rate of criminality, could result in a greater percentage than 1 of the white escapng justice. please include your extensive datasets, this should be amusing.

Comments on this entry are closed.