Thursday, 23 July 2015

Think Again: David Cameron, Conveyer Belts and Non-Violent Extremism

I’ve given myself an hour [edit: ended up being 3] and no more to write this so to cut to the chase: I wanted to respond, in detail, as to why there is no causality between Islamism and terrorism. Eustonites are very fond of the idea and they have been clamouring around a heavily illiberal speech given by David Cameron – so I wanted to respond. As usual, bibliography is at the bottom. Here is what the Prime Minister said:

you don’t have to support violence to subscribe to certain intolerant ideas which create a climate in which extremists can flourish. Ideas which are hostile to basic liberal values such as democracy, freedom and sexual equality.

The root cause of the threat [of terrorism] we face is the extremist ideology itself… They [young people] are watching videos that eulogise ISIL as a pioneering state taking on the world, that makes celebrities of violent murderers… you don’t have to believe in barbaric violence to be drawn to the ideology. No-one becomes a terrorist from a standing start. It starts with a process of radicalisation. When you look in detail at the backgrounds of those convicted of terrorist offences, it is clear that many of them were first influenced by what some would call non-violent extremists.

[We must confront] groups and organisations that may not advocate violence – but which do promote other parts of the extremist narrative… We must demand that people also condemn the wild conspiracy theories, the anti-Semitism, and the sectarianism too.

The basic idea is that non-violent Islamist ideology -> violent Islamist terrorism. It’s an idea referred to as the “conveyer-belt theory of terrorism” propounded by (mostly non-academic) bodies like the Quilliam Foundation. This isn’t an unreasonable view. Back in 2011, I wrote a post advocating this idea on the basis of purported evidence provided by the Prevent review. Whilst it is not an unreasonable view, it is, as I have learnt, wrong.

Before explaining why, let’s start with definitions: an Islamist believes in the political application of Islam. A violent Islamist believes in the violent application of Islam. This is the dividing line between non-violent and violent extremism. Both are problems that should be tackled but the Quilliam view treats them as part of the same problem. Both are ideologies - which is why the idea that this isn’t an “ideological” problem is wrong, what matters is which ideology we’re talking about.
Specific Studies on the Relationship Between Islamism and Terrorism
The first study that we can refer to is Fair et al (2012) which explicitly looks at support for terrorism amongst three constituencies: (i) different degrees of practicing Muslims; (ii) Islamists (i.e., those who support political Islam); and (iii) jihadists (i.e., those who advocate a violent aspect of their ideology). To elicit whether people were Islamist, they were asked whether they supported Islamist political parties and the political implementation of Sharia. To gauge whether people were jihadists, they were asked the extent to which they supported non-state groups using violence. They found:

Supporters of [Islamist] parties were no more supportive of the militant groups than were those who supported avowedly secular parties (see column 3 of table 2) [and] even adherents of this more extreme form of sharia law are no more likely to support political violence than those who do not believe that sharia requires physical punishment (see column 4 of table 2). Further, regardless of their interpretation of sharia, respondents who thought that sharia should play a greater role in Pakistani law were no more likely to support militant groups than respondents who wanted a stricter separation of church and state (see column5).
Hence, Fair et al conclude it ‘does not appear that Islamism… is related to support for militancy.’ For those interested in more support for my last post on the lack of link between Islam and terrorism, this study also finds that religious practice is unrelated to support for terrorism. Unsurprisingly, the specific belief about violence (jihad) is related to support for terrorism. In some ways the result is unsurprising because its tautological: if you believe terrorism is justified, you support terrorism.
Next, we have Furia and Lucase (2008) who find that Islamic “consciousness” of Muslims in Egypt, Jordan, Lebanon, Kuwait, Morocco, Saudi Arabia, and UAE is unrelated to favorableness of Western countries. (In fact, somewhat interestingly, when you look at the results in Table 2 above, Islamism predicts positive attitudes toward France).
Next, Fair and Shapiro (2009) which asked people how they thoughts about the Womens Protection Act (legislation that Islamists were opposed to), intervention in curriculum of Islamist schools and ‘Talibanisation.’ They find ‘support for Islamist politics did not translate into support for militant organizations in any clean way... support for militant organizations is not positively correlated with support for “talibanization”’.
There are some nuances in the data but as should be clear from the table: ‘Overall, [their] results strongly suggest that support for Islamist politics does not predict support for Islamist militant organizations.’  David Cameron spoke about having an aversion to democracy as an extreme view. He’s right – but, again, there is no relationship between rejection of democracy and support for terrorism. The same study shows adding controls ‘removes most of the already tenuous relationship between support for democracy and support for militancy.’
Next, we have some studies which I don’t necessarily buy but mark the clear distinction between non-violent Islamists and violent Islamists. For example, Bartlett, Birdwell, and King...
...studied the biographies of 62 homegrown terrorists in Canada and Europe and compared young persons with similar political or religious convictions, of which one group was prepared to use violence whereas the other was not. What distinguished the violent from the nonviolent radicals was their longing for adventure, excitement, and a cool existence (quoted in Van San (2015)).
Studies on the Methods of Radicalisation
Next we have Githens-Mazer and Lambert (2010). After a review of a number of recent examples of radicalised individuals, Githens-Mazer and Lambert conclude:
… the mere presence of ideology, or even specific political attachment to an ideology, is not enough to explain why an individual commits a terrorist act... Approaches that emphasize specific forms of Islamic ideology or theology as causal ‘mood music’ for terrorism are, at best, existentializing red herrings that are prone to miss the point, no matter how politically faddish… [The idea that] the combination of Salafi or other specific political/theological Islamic outlooks with an Islamically inspired grievance that somehow generates Islamically inspired terrorism, also fails.
Next we have a government review which comprehensively looked at the conveyer belt theory of terrorism. For those who want the whole thing, I made a freedom of information request back in 2011 and have the document (please contact me if you want it) but the salient parts are quoted by The Telegraph:
It is sometimes argued that violent extremists have progressed to terrorism by way of a passing commitment to non-violent Islamist extremism, for example of a kind associated with al-Muhajiroun or Hizb ut Tahrir ... We do not believe that it is accurate to regard radicalisation in this country as a linear 'conveyor belt’ moving from grievance, through radicalisation, to violence … This thesis seems to both misread the radicalisation process and to give undue weight to ideological factors
Next, we have a line of papers which show there is no clear pathway to terrorism suggesting that the conveyer belt theory is false. MI5’s Behavioural Unit carried out an analysis based on ‘based on hundreds of case studies by the security service.’ They conclude that ‘are a diverse collection of individuals, fitting no single demographic profile, nor do they all follow a typical pathway to violent extremism’ (my emphasis). This research ‘also plays down the importance of radical extremist clerics.’
Again, this is all representative of the literature. Marc Sageman’s (2008) Leaderless Jihad (a study of 500 terrorists) concludes that you ‘cannot simply draw a line, put markers on it and gauge where people are along this path to see whether they are close to committing atrocities’ (p.72). This directly contradicts the conveyer belt model. And again from McCauley and Moskalenko (2010)
...these [differing] paths [to terrorism] do not include radical ideas or activism on the way to radical action, so the radicalization progression cannot be understood as an invariable set of steps or “stages” from sympathy to radicalism.
And on and on the literature goes.
The Other Side
So why did I believe in 2011 that Islamism did lead to terrorism? Or, in the Prime Minister’s vernacular it would create a culture in which terrorism could thrive? I’ve read over the old post again and again and I think the main reason was simply because I didn’t know the literature. Moreover, the Prevent Strategy stated ‘there is evidence to indicate that support for terrorism is associated with rejection of a cohesive, integrated, multi-faith society and of parliamentary democracy.’ But they don’t seem to provide it.
I was wrong to accept that so uncritically. Mea culpa. But in the spirit of epistemic humility, here’s something that could be used against me. There are two studies that could be used in support of the conveyer belt theory. The first is Silber and Bhatt (2007), otherwise known as the NYPD study. They claim there are 4 stages to radicalisation: Pre-radicalisation (being normal) -> Self-identification (exploration of Salafi Islam) -> Indoctrination (intensification of Salafi Islam) -> "Jihadisation" (they become terrorists / start planning for an attack). The study concludes ‘there is a remarkable consistency in the behaviors and trajectory of each of the plots [examined] across the stages.’
The reasons I don’t buy this study are aptly summarised by Patel (2011):
The NYPD Report uses limited data and employs faulty methodology. Not only does the report rely upon a handpicked sample to draw conclusions about a broader population, it does so based on just 10 case studies... Contrary to social science norms, the NYPD Report fails to consider whether the religious conduct and expressive activity that it characterizes as early signatures of radicalization occur with any more frequency among terrorists than among all American Muslims.107 Compounding this flaw, the innocuous nature of many of the signatures identified by the NYPD—such as growing a beard or becoming involved in community activities—means that they are likely to be found in a large segment of the American Muslim population.
Next is Gartenstein-Ross and Grossman (2009) who look at 117 domestic, homegrown Islamist terrorists. This study is far more rigorous than the NYPD one and it has the wisdom to conclude there is no fixed profile of a terrorist. It does however state that there are six ‘manifestations’ that occur in terrorists. They say that the ‘six steps differ in prevalence’ but are frequent enough to be considered significant. These six steps can be subdivided in religious (e.g. only accepting one religious authority) and political awakenings (e.g. schisms between the West and Islam).
This has the same disadvantage as the last: we’re not comparing a control group of Islamists who may have the same thing. Moreover, as Patel notes, once you look into the data, the link becomes slightly more tenuous:
...only 17.1 percent of the sample exhibited low tolerance for perceived theological deviance and only 15.4 percent of the sample attempted to impose their religious beliefs on others. The relatively low correlation between religiosity and terrorism—in a study that seemed aimed at finding such a correlation—is a strong indication that conservative religious belief may play a lesser role in radicalization than one might assume.
The overarching reason to not give weight to these studies, as opposed to the ones quoted above is because I believe in the mantra ‘follow literatures, not studies.’
The Other Things
Cameron’s speech is littered with illiberal policies and every single one rests on this idea that extremism and terrorism are linked. He could conceivably make the argument that extremism is bad per se and hence we should give Ofcom power to stop extremist channels. But he didn’t. He made an issue that should be dealt with in the market place of ideas and organic integration, an issue about people wanting to blow us all up. There’s a lot of other really weak stuff in the speech – even when Cameron is right:
So when people say “it’s because of the involvement in the Iraq War that people are attacking the West”, we should remind them: 9/11 – the biggest loss of life of British citizens in a terrorist attack – happened before the Iraq War.
At this point, its not even so much about what conclusions are reached: it’s about methodology. The simple response to Cameron is to say: yes, there was terrorism prior to Iraq but the point is that Iraq could have increased it. Now that’s false (for reasons I’ve expressed before: Western military policy reduces violence - (see hereherehereherehere and the last study here) but don’t think you’ve refuted an argument on the basis of fairly lousy reasoning.
I feel like I’m banging the same drum here: prove your statements. I don’t really mind about the result you might reach as long as you’re using the right method. Evidently, I’m talking to 2011 me here as well. Here’s a good starting point: if there is so much as a whiff of an empirical claim and your post doesn’t contain a study, you probably shouldn’t be blogging. If Rory can become mentally deranged looking at MTV VMA award nominations statistics in defense of Taylor Swift, you can read some studies. Remember Cowen’s Second Law and assess whether your post is actually utilising the countless studies available. If it isn’t, you should be ashamed of yourself.   
P.S.: The title is a play on the Twitter account that the State Department runs to try to get potential terrorists to ‘Think Again, Turn Away.’
Fair et al, ‘Faith or Doctrine? Religion and Support for Political Violence’, Public Opinion Quarterly (2012), 1
and Shapiro, ‘Why Support Islamist Militancy? Evidence From Pakistan’, Princeton Working Paper (2009) available at <>
Furia and Lucas, 'Arab Muslim Attitudes Toward the West: Cultural, Social, and Political Explanations',  International Interactions: Empirical and Theoretical Research in International Relations (2008), Vol. 34, 184
Gartenstein-Ross and Grossman, Homegrown Terrorists In The U.S. And The U.K.: An Empirical Examination Of The Radicalization Process, FDD Press (paperback edition) (2009)
Githens-Mazer and Lambert, 'Why conventional wisdom on radicalization fails: the persistence of a failed discourse', International Affairs 86: 4 (2010) 889–901
McCauley and Moskalenko, ‘Individual and group mechanisms of radicalization’ In L. Fenstermacher et al (eds.), Protecting the homeland from international and domestic security threats: Current multidisciplinary perspectives on root causes, the role of ideology, and programs for counter-radicalization and disengagement (2010).
Patel, ‘Rethinking Radicalisation’, Brenan Centre for Justice (2011) available at <>
Sageman, Leaderless Jihad: Terror Networks in the Twenty-First Century (University of Pennsylvania Press, 2008)
Van San, 'Striving in the Way of God: Justifying Jihad by Young Belgian and Dutch Muslims', Studies in Conflict & Terrorism (2015), Vol. 38, Issue 5, 328