Monday, 18 August 2014

Mr Thomas Gradgrind, Facts and Terrorism

In the last week, I have had two interactions with people who have insufficient regard for the empirical literature. By this, I do not mean that we have disagreements about what the academic literature actually says, rather I mean people who have taken the position that the empirical literature is unworthy of our time, all it does is ‘validate [our] sense of the real.’ These views are not really worthy of being directly addressed at a time when there are so many vital discussions to be had. As I’m working on my Iraq posts, I thought I’d just write this quick post in the style of Ben Southwood.

Ben Southwood is the opposite of the aforementioned unempirical individuals, his tweets are based on studies, his posts are jam-packed with references and he is far from someone who would reject data driven research. He often writes posts summarising interesting new research. That’s my aim here – but my summaries will be somewhat longer. Obviously I think these studies deserve a far wider audience but a subtle point is to show what people miss when they disregard the empirical literature.

Whither COIN

Counter-insurgency doctrine (COIN) has been described as part of the ‘new orthodoxy’. The basic idea of COIN is to win the hearts and minds of a population so we can defeat an insurgent group. The way hearts and minds are won, the doctrine tells us, is by creating safe spaces and (mostly) economic development. General Stanley McChrystal went as far as to say that ‘95 per cent of the effort should be dedicated to winning popular support and 5 per cent to defeating the insurgents themselves.’ I have advocated COIN-levels of troops, military action and working with local groups to defeat insurgents but a major element of COIN – the idea that we need to win hearts and minds to defeat insurgents – can seriously be challenged on the basis of two new studies.

1. ‘Just How Important Are ‘Hearts and Minds’ Anyway? Counterinsurgency Goes to the Polls’, Raphael S Cohen, Journal of Strategic Studies (2014). Cohen’s main finding is that winning hearts and minds is an effect of military success, not a cause of it. The study looks through three examples of COIN in action: Vietnam, Iraq and Afghanistan. The reason COIN has got things wrong is that it relies on a model whereby the majority of people are undecided and can be won over. This is a fair representation of what COIN expects:

The truth is that in these combat zones, the idea the native population is remotely amenable to terrorist organisation isn’t that well supported at all. Cohen tries to measure the percentage of the population that does actually change their minds and finds that its actually only around a third. Indeed, this seems to be overstating the extent to which a population allow their hearts and minds to be changed:

The Afghan data showed less variation: beliefs about how the national government was carrying out its responsibilities varies 13 per cent in six Asia Foundation polls between 2007 and 2012.... Similarly, Iraqi attitudes towards their local leadership changed by 11 per in seven polls conducted between 2003 and 2009... When the ABC News et al. polls asked Afghans to assess the level for support for the Taliban in their area, responses varied by only 6 per cent, while the Taliban’s favourability... In Iraq, different organizations asked about whether insurgent attacks on Coalition forces could be justified, a tacit measure for pro-insurgent sympathies... results varied only by 21 per cent over eight surveys from 2003 to 2008 (p.11-12)

These data obviously have the disadvantage that they tell us who did change their views rather than who could – but it nonetheless is a fairly persuasive measure of how small the undecided people really are. Whats more important though is when these views are changed. The data is fairly clear from all three theatres that views change after military success. Thus when we look at when the military gains of the Surge, it was only after that we saw a change in attitudes toward central government and U.S forces. As Cohen notes, attitudes to Iraqi institutions were largely unchanged until 2009 (obviously long after the military effects of the Surge). The most interesting example comes from Anbar:

In fact, Anbar did not cross the 50 per cent threshold until the October 2007 poll, after the number of incidents had declined by over 90 per cent. After that point, between 60 per cent and 80 per cent of Anbaris believed that their neighborhoods were secure. This sense of security, however, did not immediately translate into support for the Iraqi government [it took until October 2008]... Ultimately, Anbar shows an important progression: first, the insurgency is defeated, then the population feels secure and then only then, can the counterinsurgent win ‘hearts and minds’.

Cohen goes on to show how intelligence is not necessarily any better when the populations hearts and minds are with coalition forces. This is one of the mechanisms that is used to explain why the whole idea is important but, once again, its not really borne out by the data. Tips off increases and showed no relationship with polling on support for Iraqi institutions or U.S forces. The same trend is found in tip offs given in Afghanistan, look at the trends for support for U.S actions (a majority throughout the period but still declining) against tip offs received:

Finally, Cohen shows that hearts and minds are a poor predictor of stability. In Afghanistan, the U.S has enjoyed broad support for its actions but violence still rages. In Iraq, the U.S did not have support for its actions but there was a success despite that lack of support. Indeed, as shown above, the best way to get the change in hearts and minds was to win militarily. One of the disappointing things about the study is its failure to elaborate on the idea that militants do not require local support to operate successfully. Cohen does note that ‘insurgencies can survive with minimal popular support’ – but thats it. I have written a post called ‘Al Qaeda vs. The People’ which touches on this:

It is undeniable that support for Al Qaeda is dropping - advocating a Taliban-style authoritarian rule whole killing scores of civilians does that - but in terms of operational capability that does not mean there is a significant decrease in their force. Bruce Hoffman makes the point that Red Army Faction had almost zero public support and yet managed to inflict serious damage for almost three decades. The same applies to Al Qaeda; Peter Bergen goes even further saying that "Al Qaeda values one recruit more than a thousand supporters."

This study has direct relevant for what is going on in Iraq today. I don’t think its a stretch to suggest that its become part of the conventional wisdom to state that Maliki’s policies have alienated the Sunni population so as to contribute to ISIS gains. In an otherwise brilliant article, here is what Marc Lynch wrote in the Washington Post:

The Islamic State recovered steam inside of Iraq as part of a broad Sunni insurgency driven by Prime Minister Nouri al-Maliki’s bloody, ham-fisted crackdowns in Hawija and Fallujah, and more broadly because of the disaffection of key Sunni actors over Maliki’s sectarian authoritarianism.

I have argued against this view before and will do so again in a lot more detail in Part 4 of Iraq post series - but for now its worth repeating exactly what the study says: the hearts and minds of a population do not impede military success. If Maliki was able to win, it would likely be followed by support from Sunnis (leaving aside that the Anbar sheikhs are still committed to fighting ISIS). The study essentially supports the idea that this is a military problem. If the argument is that the oppression Sunnis suffered has driven them to ISIS, I have debunked that several times before. Thats not to deny that Maliki has acted in a sectarian manner – but it is to deny that ISIS gains are anything to do with discontent.

2. ‘First Steps Towards Hearts and Minds? USAID’s Countering Violent Extremism Policies in Africa’, Daniel P Eldrich, Terrorism and Political Violence (2014). This study looks specifically at the U.S efforts in counter-terrorism programmes that focus on economic development and trying to win the public’s hearts and mind. As Eldrich notes

USAID has been carrying out a number of programs since 2005 in Mali focused on strengthening the economic and social resilience of local communities in the face of messages and inducements offered by violent extremist groups such as AQIM and MOJWA (p.526)

In an attempt to control for the people who were not subject to the programmes, the study looks at two different Malian towns. They obviously also control for age, socio-economic status and even political views at the start to make the comparison is as controlled as possible. Here are the different programmes:

Not particularly surprising, Eldrich found that those populations which were subject to the programmes listened to messages of peace and participated in civil society more. This is probably because they had more information on them and so acted accordingly. The difference in terms of this should not be overstated but it is nonetheless significant:

But the most significant finding of this study is that there was no significant difference in terms of support for the U.S, support for Al Qaeda, support for extremist ideas (e.g. the idea that the U.S is at war with Islam). Eldrich notes that:

In contrast to the clear distinction in answers to questions about civic engagement and peace and tolerance program listening between Malians living in Timbuktu and Dire´, bivariate tests of the other two outcomes of interest showed no measurable difference. The Chi-squared value for the answers to the question, ‘‘Is the U.S. fighting Islam or terrorism?’’ was .397 (indicating no strong divisions by control or treatment) while the answers for ‘‘Are al Qaeda’s activities justified under Islamic law?’’ had a value of .743 (again indicating little difference between the two groups) (p.536)

The idea that we can change hearts and minds through economic development is based on a structuralist idea of economic determinism (or, at least, giving primacy to economic conditions). Aside from this study, there are two streams of research that show us this line of argument is false. First, we can look at the literature on the link between terrorism and poverty. The academic literature is clear:

...the available evidence indicates that, compared with the relevant population, members of Hezbollah’s militant wing or Palestinian suicide bombers are at least as likely to come from economically advantaged families and have a relatively high level of education as to come from the ranks of the economically disadvantaged and uneducated. Similarly, members of the Israeli Jewish Underground who terrorized Palestinian civilians in the late 1970s and early 1980s were overwhelmingly well educated and in highly regarded occupations (‘Education, Poverty and Terrorism: Is There a Causal Connection?’, Alan B. Krueger and Jitka Maleckova, Journal of Economic Perspectives (2003)).

Kruger and Maleckova go on to quote several other studies which have the exact same conclusion as theirs from around the world. But this isn’t the only literature we can cite in support of the idea that economic factors do not lead extremist views. From an entirely different context, we have the work of Alexandre Afonso which finds, if anything, richer conditions lead to voting for more ‘far right’ parties in Europe:

Both of these two studies cast doubt on a major element of COIN. First, is the idea that winning hearts and minds is important for and a cause of military success. Second, the idea that we can win hearts and minds regardless of that first point through economic development is severely weakened. Just so I can needlessly push you in the direction of my old posts: I have argued against structuralism and the primacy it gives to economic incentives in relation to crime, revolution and terrorism. For those interested on the situation in Mali and counter-terrorism military operations, see this post.

Bombing Bombers

3. ‘Suicide Bombers in Iraq, 2003–2010: Disaggregating Targets Can Reveal Insurgent Motives and Priorities’, K. R. Seifert and C. McCauley, Terrorism and Political Violence (2014). This recent study adds to the emerging consensus of academics against Robert Pape’s thesis that foreign occupation causes war. Two recent books The Missing Martyrs (reviewed here) and Myth of Martyrdom (reviewed here) are additions – but this study brings even more data against Pape. This reinforced the argument against Pape made here. But to return to the study, here is the data set and aim:

Building on Hafez’s work, we independently developed a database of 1,779 suicide bombers in Iraq from 2003 through 2010... we use our 2003–2010 data to compare trends in suicide bombing with trends in non-suicide attacks over the same time span. Finally, we compare trends in targeting against foreigners in Iraq with trends in targeting against Iraqis, including security forces, government entities, and civilians

They ‘compared the pattern of suicide bombers per month with the pattern of total insurgent attacks per month’ and they find that ‘it is clear that the relationship between total insurgent attacks and suicide bombers was weak at the beginning of the insurgency, but grew stronger beginning in 2007’ (p.7). Why was this? They explain:

Our interpretation is that, early in the insurgency, Al Qaeda-inspired groups in Iraq were competing with other militant groups in opposing coalition forces. AQI used suicide bombings to intensify a “system collapse” strategy as detailed by Hafez, whereas larger militant groups hoped to control the new system being built by coalition forces.

Al Qaeda used suicide bombing, mostly directed at Iraqis and Iraqi institutions because that is how they felt they could effectively take over. The larger militias thought their conventional arms were enough to take over, not destroy, the new institutions. Hence, there was more insurgent activity than there was suicide bombing. How we explain the correlation between the two in the latter period? Because of Anbar Awakening and surge of U.S personnel: ‘It was a “synergistic interaction” between these two initiatives that led to greater overall stability in Iraq.’ (Apologies for doing this again, but see my old post here for an elaboration).

But the most interesting findings relate to who was targeted:

What is immediately clear is that ‘Target Set 3 [Iraqis, not Western targets] was by far the most frequent target of suicide attacks, drawing 77 percent’; coalition targets accounted for only 13% of suicide bombing targets. The trends of when the different targets were hit matches different assessments of risk: as the democratic institutions were set up (2003-5), Iraqis and coalition personnel were targeted. As stated, the aim was to destroy the institutions. The attacks against the coalition forces declined because the ‘more immediate threat to AQI and other small insurgent groups was a more effective counterinsurgency conducted by both Iraqi government forces and Awakening forces’ (p.16). This has an important implication that will be familiar to those who have had to listen to me bang on about blowback for the last two years:

Similarly, Robert Pape’s thesis that suicide bombings are an extreme nationalist struggle against a foreign occupier is challenged by our results showing that far more suicide attacks occurred against Iraqis, especially Iraqi civilians, than occurred against foreigners. In fact, suicide bombers began to be dispatched in significant numbers after sovereignty was handed over to the interim Iraqi government in June 2004. Although later attacks against the Awakening Movement might be rationalized as targeting perceived “collaborators” with the occupying forces, merely broadening the definition of an “occupying force” fails to take into account the change in strategy and political motivation behind these actions (p.15)