Militant Group Alliances and Relationships (MGAR)

The MGAR dataset — a collaborative effort with Erica Chenoweth, Michael C. Horowitz, Evan Perkoski, and Philip B.K. Potter — offers an original coding of cooperative and conflictual relationships between militant groups. MGAR records the type and content of relationships among 2,613 militant groups from 1950 to 2016, documenting 7,409 dyad-years in which a relationship existed between groups — significantly expanding the set of relationships identified in prior work. MGAR also captures when each relationship began and ended, and what it entailed.

We categorize the nature of each relationship along a spectrum from cooperation to conflict: allies (the strongest form of cooperation), associates, supporters, fans (verbal but not material support), hosts (primarily states), rivals, and competitors (the strongest form of conflict). For each cooperative dyad-year we further code whether the relationship involved operational (e.g., shared membership, joint operations, tactical advising), material (i.e., arms transfers), territorial (e.g., shared bases), training, or financial (i.e., cash transfers) support.

Explore & download

Explore: MGAR database (National Security Policy Center)

Data: Replication files (IO 2022) · Replication files (JOP 2022) · Replication files (JOGSS 2023)

Time series of ongoing cooperative relationships between militant groups from 1965 to 2016, as raw counts and as a proportion of all dyads

Cooperative relationships between militant groups, 1965–2016 — total, material, and rhetorical alliances — in raw counts (left) and as a share of all group dyads (right).

Plot of the marginal effect of shared religion on the probability of material-alliance breakdown across levels of state repression

The marginal effect of shared religion on the hazard of material-alliance breakdown, across levels of state repression. As repression intensifies, shared religion increasingly holds cooperative ties together (effect falls below one).

Time series of the number of al-Qaeda alliances from 1988 to 2016, split into material and rhetorical ties

The growth of al-Qaeda's alliance network, 1988–2016, distinguishing material from rhetorical ties.

Time series of the number of Islamic State alliances from 2013 to 2016, split into material and rhetorical ties

The Islamic State's alliance network, 2013–2016, before and after its split with al-Qaeda.

Survival curves showing the probability of alliance termination when a group's founder is decapitated versus not

Militant alliances terminate far faster after a group's founding leader is killed (dashed) than when the founder survives (solid).

Coefficient plot of the association between different types of leadership removal and alliance breakdown

Different forms of leadership removal and their association with alliance breakdown — every type raises the probability that an alliance ends.

Key references

Blair, Christopher W., and Philip B.K. Potter. 2023. “The Strategic Logic of Large Militant Alliance Networks.” Journal of Global Security Studies 8(1): 1–22.
Blair, Christopher W., Michael C. Horowitz, and Philip B.K. Potter. 2022. “Leadership Targeting and Militant Alliance Breakdown.” The Journal of Politics 84(2): 923–943.
Blair, Christopher W., Erica Chenoweth, Michael C. Horowitz, Evan Perkoski, and Philip B.K. Potter. 2022. “Honor Among Thieves: Understanding Rhetorical and Material Cooperation Among Violent Non-State Actors.” International Organization 76(1): 164–203.