23/05/2013 13:42 BST | Updated 23/07/2013 06:12 BST

On Terrorist Probabilities


Why aren't the Scots doing more to combat their culture of violence? Why aren't its community leaders doing more to rein in their violent minority? Don't we need tougher laws to protect us from the threat posed by men of Scottish appearance?

You might think I've lost my mind. From a statistical point of view, though, I haven't.

According to table 1.18 of this pdf, there were 89 Muslims in prison for terrorist offences in March 2012 . As there are 2.7 million Muslims, this is 33 per million.

By contrast, there were at the latest count 2382 people imprisoned in Scotland for assault or murder. That's 449 per million.

The prevalence of violent people in Scotland is therefore 13.6 times that of terrorists among Muslims. In fact, this is an understatement. A Scotsman who has committed a violent crime is more likely to be out of prison than a Muslim terrorist, either because he's more likely to have gotten away with his crime or because he's more likely to have finished his sentence.

If it is reasonable to speak of a problem of Muslim terrorism, it is therefore an order of magnitude more reasonable to speak of a problem of Scottish violence.

I anticipate two objections here.

One is that terrorism is nastier than street fights. True. But we must distinguish between the cost of an event and it's probability; I'm talking about the latter.

Secondly, the prevalence of terrorism is greater among Muslims than non-Muslims; Nick Cohen is correct to point out that popular culture misrepresents terrorism in this regard. We can quantify this. There are 29 non-Muslims in prison for terrorist offences. That's a rate of 0.5 per million non-Muslims in the population generally. So, Muslims are more likely to be terrorists than non-Muslims. Coincidentally, the difference between the two groups is about the same (pdf) as that between the prevalence of assault in Scotland and New Zealand - so my analogy holds.

But of course, P(A|B) is not the same as P(B|A). The fact that a terrorist is likely to be a Muslim does not mean a Muslim is likely to be a terrorist. Even if we assume that there are ten terrorists walking the streets for every one inside, then 99.988% of Muslims are not terrorists. To put this another way, there's only around a one in 8000 chance of a Muslim being a terrorist; it's 16 times more likely that the Duke and Duchess of Cambridge will name their child Wayne.

Given all this, why does anyone think terrorism is a Muslim problem? There are several biases at work:

1. The out-group homogeneity bias. Will Davies tweeted that "in my day, a bloke who stabs a random stranger then rants about the government was called 'insane'". This is true if the bloke is one of "us". If, however, he is one of "them" then he is more representative of "them".

2. The narrative fallacy. We love to tell stories and read messages into things, and thus underweight the significance of random nuttery. Frank Furedi says:

"One problem with the construction of the random fanatic, is that virtually any form of incomprehensible act of violence - a school shooting, a crazed knife attack - can be redefined as an act of political terrorism. That is why far too many people cannot resist the temptation of defining the tragedy in Woolwich as an act of political terrorism."

3. The availability heuristic. Dramatic events weigh heavily upon our mind, and this causes us to overweight their probability.

4. Many groups have an incentive to exaggerate the significance of terrorism, and to reframe insane violence as "terrorism." For the police, such attacks give them a chance to further inflate their sense of self-importance and to seize more powers. And politicians can use the image of grave danger and an evil foe to appear Churchillian.

Some things, however, are less significant than they seem.

This blog was first published on Chris Dillow's blog, Stumbling and Mumbling