Islamic fighters from al-Nusra Front carry banner bearing a drawing of Jerusalem’s Dome of the Rock mosque with the Arabic slogan: “We fight in Syria… and our eyes are on Jerusalem” as they parade at the Yarmuk Palestinian refugee camp, south of Damascus, July 28, 2014. (Rami Al-Sayed/AFP/Getty Images)

The emergence of the Islamic State marked a turning point in the world’s long-running struggle with Salafi-Jihadi violence. The resulting carnage has traversed national boundaries and inspired the tactics of similar Islamist movements from the Taliban in Pakistan and Afghanistan, to al-Shabab in Somalia and Kenya, and Boko Haram in Nigeria. This pressure has sparked a renewed debate in the United States over the appropriate counterinsurgency response. Drone strikes are the least costly from the standpoint of U.S. military casualties, but have a high potential for collateral damage. Reliance on local proxies may facilitate a more selective use of force, but often creates costly new problems. Targeted air strikes – the current U.S. approach in Iraq and Syria – fall somewhere in between.

Our recent research on Salafi-Jihadi violence in Russia’s North Caucasus – published in the latest issue of the American Political Science Review – suggests that selective counterinsurgency tactics are unlikely to succeed in hurting groups like the Islamic State or to deter their continued assaults. Using unique micro-level data, we analyzed violence involving nationalist and Islamist insurgents in the North Caucasus and Moscow’s efforts to counter them. We found that Islamist and nationalist rebels respond differently to coercion. While selective attacks outperform indiscriminate ones in deterring nationalist rebels, the technology of government violence has little effect on the resolve and capabilities of Islamists. Their divergent ideologies do matter – but primarily because of how they shape the relative dependence of armed groups on local versus external sources of support.

The Caucasus case is illuminating because much like in Syria the moderates and Salafi-Jihadis have fought there against the same government, on the same terrain, at roughly the same time. Russia’s experience is relevant for understanding how Islamists behave on the battlefield, the effectiveness of government coercion in deterring their activities and the implications for designing more effective policy responses.

We therefore constructed a new dataset of 9,405 rebel attacks from 2000 to 2012. We used this dataset to compare the intensity of rebel-initiated violence following the use of indiscriminate counterinsurgency tactics such as artillery shelling and airstrikes and following the use of selective tactics such as individual arrests or targeted killings. The results were striking. We found that in very similar districts where security forces used selective rather than indiscriminate tactics, nationalist violence declined by 5 percent in the subsequent 12 weeks.

Islamist violence, however, did not change in response to selective counterinsurgent tactics. Unlike Chechen separatists and other relatively moderate forces, Salafi-Jihadi groups continued their attacks irrespective of the tactics or identity of the counterinsurgent. This result holds under a variety of definitions of “Islamist violence,” various identification strategies and model specifications. This finding challenges the very logic of coercion: When an activity becomes relatively more costly, fewer people should engage in it. Yet even where it is very costly for potential rebels to take up arms, Islamist violence continues.

What is more, in contrast to previous research that suggested that indigenous, co-ethnic authorities are more effective at counterinsurgency due their superior intelligence about the identities and whereabouts of individual insurgents, we found that local security forces have no such advantage when fighting Islamist groups. In circumstances where local and federal forces were equally likely to operate, local police were indeed more successful than federal troops at reducing nationalist violence. Yet local security services – such as pro-Moscow Chechen militia – had no such luck against Islamist insurgents. Improved intelligence about the local population, our study suggests, is of little help in deterring Salafi-Jihadi violence.

The logic behind U.S. “population-centric” counterinsurgency doctrine assumes that rebels depend on local popular support. As the benefits of cooperation with security forces increase and the costs of cooperation with insurgents become more taxing, the local population should become less likely to support the rebels. For Salafi-Jihadi groups, this assumption is problematic. In the North Caucasus, nationalist Chechen rebels drew support mainly from local clan networks inside Chechnya, while Salafist groups like the Riyad us-Saliheyn, Caucasus Front and Imarat Kavkaz recruited from across the region, from other parts of Russia and beyond.

In Syria as well, the three groups that have attracted the most foreign fighters – Ahrar al-Sham, Jabhat al-Nusra and the Islamic State – are all Salafi-Jihadi. Their recruits hail from across the Arab world, Europe (including Russia and the Caucasus) and North America. Unlike nationalists, Islamists rely more on external revenue and manpower, which enables them to continue fighting even where the population faces heavy penalties for supporting them.

Two other critical differences stand out between Islamist and nationalist insurgents. First, compared to violence by nationalists, whose fights tend to be circumscribed around ethnic homelands, Islamist violence is geographically dispersed. The average road distance from the nearest rebel attack of the same type was 126–494 km in a typical week for Islamist rebels, compared to just 50–52 km for nationalists. Whereas Chechnya bore the majority of nationalist violence between 2000 and 2012, most Islamist attacks occurred outside Chechnya’s borders.

The scope of Salafi-Jihadi goals – the rejection of modernity and Western materialism, the overthrow of “apostate” and “infidel” governments and the reestablishment of an Islamic Caliphate – transcend ethnic and state boundaries, facilitating inter-regional and transnational emulation and cooperation. Islamists are therefore able to raise their visibility and mobilize support through international networks of jihadist newsletters and websites, imams, charities and support organizations. Indeed, this was the original vision of Abdullah Azzam, the late co-founder of al-Qaeda, who sought to build a transnational jihadist infrastructure after the Soviet war in Afghanistan.

Second, because Salafi-Jihadi groups make common cause with comrades in other regions and states, they are more responsive to outside events, and their activities closely mirror international trends in jihadist violence. We found a high correlation between the timing of Islamist violence inside Russia and attacks by similar groups in other countries. During weeks with higher-than-average levels of suicide terrorism outside Russia, violence by Islamist groups inside Russia rose by 78 to 93 percent. Islamist attacks also rose by over 20 percent on dates symbolic in the Islamic calendar, while nationalist attacks did not change.

The boundaries between secular and Islamist insurgents can be often fluid, especially in places like Syria, where both types operate simultaneously. Yet, the implication for current efforts to fight violent jihadi groups is simple: Islamist insurgents require different counterinsurgency and counterterrorism strategies than secular insurgents. Because Salafi-Jihadis are less reliant on the local population, selective coercion is no more effective than indiscriminate force in stopping and deterring Islamist violence. Moreover, evidence from the Caucasus suggests that indigenous security services are no more effective than federal or occupying troops in defeating Islamists insurgents.

If external support explains the resilience of Islamist insurgents to government coercion, governments and other stakeholders should focus their efforts on degrading, blocking and eroding Islamists’ support base. This includes crippling access to financial resources accessed through licit and illicit channels, selective coercion targeted at external networks of leaders and financiers, and being mindful of symbolic dates in the Islamic calendar when attacks are likely to peak. Such tactics, of course, could also be incorporated into and bolster the broader Muslim community’s existing efforts to discredit the core of violent Salafi-Jihadism.

Monica Duffy Toft is a professor of government and public policy at the University of Oxford’s Blavatnik School of Government. Yuri M. Zhukov is an assistant professor of political science and a faculty associate at the Center for Political Studies at the University of Michigan.