Security Studies Review Paper: Max Abrahms and Jochen Mierau (2015) Leadership Matters: The Effects of Targeted Killings on Militant Group Tactics (Terrorism and Political Violence)
Excellent research examining how leadership decapitation (targeted killing) influences militants. Interesting findings, but it relies on problematic data.
The empirical research of Abrahms and Mierau examined how targeted killings of leaders of militant organizations (i.e. the counterterrorism policy of leadership decapitation) influenced the targeted groups’ use of force temporarily afterwards. It centered on changes in attack patterns (military versus civilian targets) in cases the leaders were successfully killed. The logic was that the killing impacted the leadership, which then influenced the group’s tactical choices, which resulted in changed attack patterns. Their research covered the Israeli-Palestinian conflict area and the Afghanistan-Pakistan conflict areas. Abrahms and Mierau found that successful leadership decapitation significantly changed the attack patterns: Groups increased attacks against civilians. The authors argued and concluded, after considering some possible reasons, that the most likely explanation was that successful targeted killing of leadership figures “empower[ed] lower level members with weaker civilian restraint” which caused the increase of attacks against civilians.
Contributing Reviewer(s): Shahaf Rabi
Original review publication date: 30 August 2020
Details of the Reviewed Article
Title: Leadership Matters: The Effects of Targeted Killings on Militant Group Tactics
By: Max Abrahms & Jochen Mierau
Publication: Terrorism and Political Violence (vol. 29, issue 5)
Date: September 2015 (online publication), 2017 (print)
Summary, notes, insights & remarks:
The summary focuses on the part referring to the Israeli theatre, the reviewer’s primary field of expertise and interest with regards to targeted killings. As such, the review does not include the part under the sub-title “Taliban case”, pages 13-15 (entire paper without bibliography is 16 pages);
Overall, the research idea and the way to examine it are truly excellent. Yet the reviewer is wary of embracing the authors’ (interesting) findings because of issues concerning the data that they use in it. The findings might very well be correct, but the reviewer considers the findings as ‘highly plausible’ instead of ‘solid findings’ just because of the issues with the data:
1) The most critical weakness is that the research relies on the GTD’s data for attacks in the areas of Israel, the West Bank and the Gaza Strip between the years 2000-2004. The GTD suffers from a gigantic dearth in their data for this precise time period. As such, it is unreasonable to ignore it and to fully embrace the findings since the results are based on insufficient and problematic data;
2) Some additional qualitative work to better understand the organizational leadership in Palestinian organizations, with combination of using some variables like geolocation information and group responsible for attacks, could’ve probably contributed a lot to this research article;
3) Missing clarifications on what’s considered a leadership figure;
4) A bit of polishing of the Zussman & Zussman data for their targeted killing/leadership decapitation could’ve made another wonderful positive impact (elaboration further down the text);
5) Absent from their data is basically the other side of the coin of leadership decapitation – arrests/captures. They only focus on killing people.
The research of Abrahms and Mierau specifically focuses on leadership decapitation, a niche within targeted killing studies. The authors clarify how they discern between the two: targeted killings concern with any targeted attacks killing militants at any rank or level in the organization. Leadership decapitation, however, only relates to attacks targeting leaders. In essence, the purpose of leadership “decapitation is to reduce the threat of militant groups by degrading their leaderships.”
This empirical research examines how killing militant organizations’ leadership figures influences the targets that the organizations choose to attack – emphasizing the difference between the targeting of military and civilian targets by the group(s) experiencing decapitation. Therefore, the focus is on the organizational attack patterns following successful decapitation events (successful being that the State managed to kill the leadership figure).
Importantly, the authors note this is interesting since the groups use violence as means to achieve political goals. However, targeting civilians implies wrong or at least less effective and less strategic as well as lower quality violence than targeting military objects.
Abrahms and Mierau use the dataset of Zussman and Zussman for the targeted killing in the Israeli-Palestinian conflict. They discern between successful and attempted/ failed decapitation operations.
Reviewer note: For the purposes of the authors, who focus on leadership decapitation and not overall targeted killings, this dataset should basically be a decent fit and sufficient. However, it is important to keep in mind that the authors provide no definition as to what they consider as leadership figures. This issue is thus exacerbated because the Zussmans’ dataset is general. The dataset simply contains all leadership figures since such incidents receive attention (so no leadership figure remained out) but some in it aren’t necessarily ‘serious’ leadership figures. It is also important to bear in mind that some ‘leadership figures’ in it are explosives and munitions experts.
So it remains unclear if Abrahms and Mierau think, or take into account, that this kind of decapitation of leadership figures causes a group to alter its targeting. In fact, one should ask how well this fits with the perspective of organizational theory as a qualitative explainer for their findings. The authors don’t refer to this issue (possibly unaware to it).
In addition, it is noteworthy that the researchers focus on leadership decapitation as killing, putting aside all of the capture/arrest events of leaders.
Moreover, the researchers use the Global Terrorism Database’s (GTD) for data about attack patterns of militant organizations. As such, for the years 2000-2004, the authors have a total of 209 attacks (against both, civilian and military targets). It is also noteworthy that as they split the attacks by the years to military/civilian, as well as split by the years with groups (Fatah, Hamas, Palestinian Islamic Jihad). The number of attacks in each slot, after the splits, is usually less than 30.
Reviewer note: This is a major weakness which makes the results of the research quite unreliable – at least for Israel – since the GTD is missing a huge amount of data in their dataset for the Israeli-Palestinian conflict area in the early 2000s years (Second Intifada).
For instance, the reviewer’s personal dataset contains 333 attacks between 2000 and 2004 (five years) for the four modus operandi of suicide bombers, suicidal mass-shooters, suicide bombers in VBEDs (vehicle borne explosive device) and VBEDs. That is 333 attacks even without other attack types which constitute the very vast majority of attacks types – thousands of gunfire, grenades, IEDs, etc. attacks.
(see also: Meir Amit Intelligence and Terrorism Information Center, 2006 Second Intifada Suicide Bombers Report; and Israeli Ministry of Foreign Affair’s list of dates and fatalities “Victims of Palestinian Violence and Terrorism since September 2000”).
Finally, after the split ups by years – be it by military/civilian targets or by attacking organization – the slots are mostly filled with a relatively low number of attack events. This seems at least problematic in order to be convincing – even if one does not know of the other data issue.
Describing the Findings
The authors’ findings show a significant change for both, the Afghanistan-Pakistan and Israel-West Bank-Gaza Strip conflict zones. In both, the attacks against civilians (in comparison to military targets) is higher in the 14-days time-frame following a successful decapitation event. The authors note that this time period is precisely “when tactical decision-making is presumably most affected.”
Abrahms and Mierau present additional robustness tests and analysis of the data in order to add credibility to their findings.
The researchers checked if the data show that the decapitation of one organization’s leader also shows that other militant organizations alter their attack patterns as well. If so, it likely indicates specious results. After further testing, they did not find any indication that the killing of X-organization leader caused other organization(s) to increase the targeting of civilians.
Reviewer note: This additional examination is important. It helps strengthen the possibility that their findings might be a true lead. Yet the above-mentioned issue remains a problem not to be ignored.
Moreover, no part of the research deals with the command and control structure of the organizations. For instance, what if Israel killed the leader of Hamas’ Jenin District? Why would this influence the Hamas-led attacks in the district of Hebron, if the Hamas’ Hebron District Commander was not killed or so? After all, Jenin is at the very northern part of the West Bank while Hebron is at the very south (not to mention the Gaza Strip being separated). To the best of the reviewer’s knowledge, each of the commanders run their own personnel and are responsible for their own territory. Therefore, the research seems to be missing some qualitative understanding of the organizations. If adding that, and if adding specific variables in the attack data like geolocation information and the organizations’ responsible for each attack, that could really add a great deal of value to the research and its findings.
As another robustness check, the authors inspected the effects of decapitation on the patterns of attacks for different time frames. They describe checking the two weeks period, the three-four, and the five-six weeks periods. The 14 days time-frame best captures the tactical effects, they believe. The additional examination showed them that after the first two weeks (of post-decapitation event) the militant groups decrease their civilian targeting. As such, they believe it strengthens their primary findings.
Reviewer note: The data issues remain but this helps make the potential of their findings to actually be reliable ones if one would conduct a research with proper data.
Reviewer note: Their other robustness tests and inspections seemed unconvincing. Hence not included in the summary.
Abrahms’ and Mierau’s Qualitative Explainers for the Quantitative Results
Abrahms and Mierau describe “four theoretical mechanisms” which might explain their quantitative results:
First, simple logic that soft targets are, well, soft targets – easier to attack civilians than military, hard, targets.
Second, revenge might be the reason for altering attack patterns, targeting civilians.
Third, it might be that the state’s military forces are on high alert and preparedness after a decapitation operation. The hard targets become even harder to attack. The militants are thus further incentivized to attack soft targets, civilians, that are more available for them to hit.
The fourth explanation is very elegantly put by the authors: An “explanation within organizational theory may likewise elucidate why decapitation promotes indiscriminate organizational violence. Because replacements are overwhelmingly subordinates rather than superiors, [the] killings can create leadership deficits that empower lower level members [...]. This change in the internal composition of militant groups may affect the quality and hence selectivity of their violence. A common belief within the conflict literature is that whereas militant leaders are generally strategic actors who select tactics to optimize their political platform, lower level members disproportionately lack commensurate discipline. The principal-agent framework anticipates a disconnect between the preferences of leaders (viz. principals) and the behavior of subordinates (viz. agents), especially when the former cedes decision making to the latter.”
According to Abrahms and Mierau, their “quantitative and qualitative analyses indicate […] that leadership deficits are the best explanation [for their findings].”
Reviewer note: Legitimate and convincing explanations. It is, however, worthy to consider that the incoming leadership figures might very well wish to establish themselves, interested in ‘scoring some points’, and so they may pursue a course of action some, like the authors, consider as generally wrong and short-sighted. As such, from the perspective of the incoming leaders, targeting civilians might well be a very rational/intelligent action. Bearing this in mind, evidence to buttress the ‘leadership deficits’ explanation would be that such incoming leaders also continue targeting civilians after they secure their place as leaders. At this point I tend to believe the research does not imply such a thing. So if so, it actually weakens the leadership deficit explanation although it is a very convincing one. That noted, perhaps there is a point when this changes after X series of decapitation – though unlikely that there is sufficient case studies and data to carry out such a research (with reliable findings).
*update 19 November 2020*
Response of Jochen Mierau on Twitter for the review article: