This article was downloaded by: [Max Abrahms]
On: 14 July 2014, At: 19:28
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Terrorism and Political Violence
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ftpv20
Does Terrorism Pay? An Empirical
Analysis
a
Max Abrahms & Matthew S. Gottfried
b
a
Department of Political Science , Northeastern University ,
Boston , Massachusetts , USA
b
Department of Political Science , University of California, Los
Angeles , Los Angeles , California , USA
Published online: 14 Jul 2014.
To cite this article: Max Abrahms & Matthew S. Gottfried (2014): Does Terrorism Pay? An Empirical
Analysis, Terrorism and Political Violence, DOI: 10.1080/09546553.2013.879057
To link to this article: http://dx.doi.org/10.1080/09546553.2013.879057
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Terrorism and Political Violence, 0:1–18, 2014
Copyright # Taylor & Francis Group, LLC
ISSN: 0954-6553 print=1556-1836 online
DOI: 10.1080/09546553.2013.879057
Does Terrorism Pay? An Empirical Analysis
MAX ABRAHMS
Department of Political Science, Northeastern University, Boston,
Massachusetts, USA
MATTHEW S. GOTTFRIED
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Department of Political Science, University of California, Los Angeles,
Los Angeles, California, USA
Does terrorism help perpetrators to achieve their demands? Few research questions
about terrorism generate as much controversy. This study contributes to the debate
in two main ways. First, we identify major limitations within the burgeoning literature
on the effectiveness of terrorism. Specifically, we highlight the main methodological
problems vexing empirical assessments of whether terrorism promotes government
concessions. Second, we present a research design that circumvents those recurrent
methodological shortcomings. In short, we find no empirical evidence to suggest that
terrorism pays. In fact, multiple variants of the tactic in hostage standoffs impede
the perpetrators from coercing government compliance. The negative effect of terrorism on the odds of compliance is significant and substantial across logistic and
multilevel logistic model specifications, particularly when civilians are killed or
wounded in the coercive incident. These findings have important implications for
both scholars and practitioners of counterterrorism.
Keywords bargaining, coercion, hostage crises, tactical effectiveness, terrorism
consequences
Introduction
Does terrorism help perpetrators to achieve their demands? In the 1980s and 1990s,
terrorism specialists claimed that the tactic fails to induce government concessions.1
Political scientists gravitated towards the opposite stance at the turn of the century,
maintaining that terrorism is a surprisingly effective method for coercing government compliance.2 The pendulum seems to be swinging back,3 though skeptics
continue to abound.4 This debate over the political consequences of terrorism is
not only intrinsically important; it has potentially valuable implications for the
broader study of conflict in international relations.
The following analysis contributes to this debate in five main ways. The first
section summarizes the growing literature on the political efficacy of terrorism. This
Max Abrahms is an Assistant Professor of Political Science at Northeastern University. He is
the author of ‘‘Why Terrorism Does Not Work,’’ International Security (2006) and other studies
on the consequences of terrorism. Matthew S. Gottfried is a PhD candidate in the Department of
Political Science at UCLA. All data and estimation files are available upon request.
Address correspondence to Max Abrahms, Department of Political Science, Northeastern
University, 215 F Renaissance Park, Boston, MA 02115, USA. E-mail: maxabrahms@hotmail.com
1
Downloaded by [Max Abrahms] at 19:28 14 July 2014
2
M. Abrahms and M. S. Gottfried
section highlights the main methodological limitations vexing current empirical
assessments of whether terrorism promotes government concessions. The second
section presents a research design that circumvents those recurrent methodological
shortcomings. The empirical strategy exploits variation in the tactical decisions of
international hostage-takers and their odds of winning from foreign governments
material concessions such as money or prisoners. Because a universally accepted
definition of terrorism remains elusive, our models operationalize multiple tactical
variants and alternatives based on whether the perpetrators killed or wounded civilians, military personnel, or other officials in the course of the hostage incident. This
investigative approach provides superior leverage for assessing the willingness of
governments to accommodate non-state challengers in response to various tactical
pressures within a relatively common, controlled bargaining setting. The third section interprets the statistical results. In short, the analysis does not support the position that terrorism pays. Multiple variants of the tactic impede the hostage-takers
from achieving their demands. The negative effect of terrorism on the odds of
government compliance is significant and substantial across logistic and multilevel
logistic model specifications, especially when civilian captives are physically harmed
in the confrontation. The fourth section supplies additional empirical evidence that
the negative return from harming civilians is not an artifact of selection issues
endemic to the research landscape. The fifth section explores the relevance of these
findings to active research programs on conflict within international relations.
Limitations of Extant Research Designs
Aggrieved groups issue two kinds of demands to governments: strategic demands
and redemptive demands. Strategic demands are located in the political platform
of the group. They call on governments to make political concessions, such as to
improve their human rights, withdraw troops from abroad, or adopt a new national
ideology. Redemptive demands, by contrast, appeal for governments to cede
material concessions such as money or prisoners to sustain the group.5 The coercive
logic is identical for obtaining strategic or redemptive demands. In both cases, the
challenger tries to incentivize government concessions by inflicting pain for noncompliance.6 Because coercion is a concept indigenous to the field of international
relations, however, studies on the effectiveness of terrorism have focused on its
utility for obtaining strategic demands.7 This research orientation poses two broad
sets of methodological problems for assessing terrorism’s coercive value.
The first pertains to coding the dependent variable. Historically, terrorism datasets
have neglected to code the political outcomes of asymmetric campaigns, the standard
unit for assessing the tactical value. To compensate for this lacuna, scholars testing the
effectiveness of terrorism have themselves coded the outcomes of the campaigns, inviting allegations of confirmation bias.8 Furthermore, scoring the extent to which terrorists accomplish their political ends is objectively difficult for reasons inherent to the
complex nature of the challenger. Terrorists are notorious for issuing protean, ambiguous political demands or sometimes none at all.9 Disagreement over terrorism’s effectiveness can therefore hinge on mini-empirical disputes over whether the perpetrators
accomplished their desired strategic goals. Alan Dershowitz and Robert Pape, for
example, maintain that terrorism works as illustrated by Palestinian political gains,
whereas Abrahms and Moghadam make the opposite claim, scoring Palestinian
terrorism as a failure.10 In recent years, such dissension has likewise plagued the coding
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Does Terrorism Pay? An Empirical Analysis
3
of whether al-Qaida, its affiliates, the Irish Republican Army, its splinter groups, the
Tamil Tigers, and its organizational rivals have realized their political goals,
muddying assessments of terrorism’s overall tactical value.11 Adding to the confusion
over coding the dependent variable is that terrorists are suspected of possessing unusually long time horizons.12 Lashkar-e-Taiba, for example, has thus far failed in its stated mission of spreading Islamic rule throughout India. But the group remains
operational and may one day wrest control over Indian Kashmir. Although terrorist
campaigns may persist for decades without any perceptible political return, this lengthy
timeframe is perhaps acceptable for those committed to the causes of the armed struggle. Methodologically, scholars have dealt in an ad hoc way with these ongoing campaigns by excluding them from the analysis, which artificially drives up the coercion
rate, or by including them in the analysis, driving down the coercion rate.13 Scoring
political progress is also problematic if terrorists express unrealistic demands in order
to obtain even a fraction of them.14 Bin Laden’s 1998 fatwa, for example, called on the
United States to withdraw from the Middle East. This demand remains unmet, of
course, but al-Qaida may deem the smaller concession of withdrawing from Saudi
Arabia five years later as an important victory in itself. Similarly, the Revolutionary
Armed Forces of Colombia may be content to command southeastern parts of the
country despite failing to enact Bolivarianism as the official national ideology. Indeed,
terrorist groups tend to fall short of accomplishing their strategic demands, but may
achieve some measure of progress in the form of partial government accommodation.15
An ordinal dependent variable can help capture such middling levels of bargaining
success.16 Yet weighting the political outcomes inevitably introduces an element of
subjectivity.17 To minimize such thorny methodological issues, scholars sometimes
use public opinion instead of policy outcomes as the dependent variable.18 The former
is only an indirect proxy of the latter, however; when countries are terrorized,
electorates may shift to the political left or right without altering policy.19
The second set of methodological problems pertains to coding the independent
variables, particularly terrorism versus other non-state tactics. At the conceptual
level, scholars have developed a fine-grained nomenclature to distinguish terrorism
from other forms of resistance. European and North American scholars generally
define terrorism as the use of violence by non-state actors against civilian targets
in particular.20 When military personnel, security services, and other government
officials are the ones physically harmed, the tactic is increasingly differentiated from
terrorist acts as militant, guerrilla, or insurgent attacks in ascending degrees of specificity.21 When nobody is physically harmed in the coercive incident, the tactic is
usually differentiated as nonviolent resistance, direct action, or a failed terrorist plot
in ascending degrees of extremeness.22 Yet empirical tests of terrorism’s coercive value
have been far less specified and thus struggle to isolate the independent tactical effects.
Several scholars claim that terrorism helps to coerce government compliance, but
proceed to highlight cases of asymmetric campaigns against military personnel that
spared civilians.23 To substantiate their view that ‘‘terrorism often works,’’ Kydd
and Walter note how U.S. Marines withdrew from Lebanon after their barracks
was attacked in October 1983.24 Abrahms attempts to disaggregate the tactical
effects of militant groups by comparing the political plights of Foreign Terrorist
Organizations that mainly killed civilians versus military personnel.25 Yet both the
‘‘civilian-centric groups’’ and ‘‘guerrilla groups’’ in his sample tended to harm some
mixture of both targets. Chenoweth and Stephan also lack variation on their
independent variables. They compare the strategic outcomes of 323 nonviolent and
Downloaded by [Max Abrahms] at 19:28 14 July 2014
4
M. Abrahms and M. S. Gottfried
violent resistance campaigns, but brook no distinction between attacks against
civilians, military personnel, or both sets of targets. Further, the nonviolent
campaigns in their sample are not actually violence-free because ‘‘few campaigns, historically, have been purely violent or nonviolent, and many resistance movements,
particularly protracted ones, have had violent and nonviolent periods.’’26 The
broader methodological difficulty is that non-state actors employ a hybrid of tactics,
posing challenges for pinpointing their discrete effects. To exact political concessions
from Israel, for instance, Fatah, Hamas, and Hezbollah have trained their violence on
both the population and military, while concurrently underwriting anti-Zionist civil
resistance initiatives.27 In fact, many terrorist groups have a political wing that participates democratically. Japan’s Aum Shinrikyo, for example, selected candidates to
contest local elections while dispersing sarin gas in the Tokyo subway.28 Teasing
out the political consequences of such tactics is exceedingly difficult when employed
in tandem.29
Variation in their usage is also problematic analytically, however, if the tactics are
not adopted at random since that raises concerns of selection issues driving the political outcome. If non-state actors gravitate to terrorism when the prospects of victory
appear dim, then its usage may be endogenous to political failure and thus epiphenomenal to the negative coercive return. Holding the strategic context fixed is the main
methodological challenge of any coercion study, but is especially difficult if terrorism
is a proverbial ‘‘weapon of the weak.’’30 In addition to the capabilities of non-state
challengers, their strategic demands may also correlate with the use of terrorism or
other tactics. Terrorism is an extremism of means, but its practitioners are also known
to harbor extreme ends. Terrorists are notorious for issuing maximalist political goals
that governments are loath to concede, such as for democracies to adopt communism
or radical Islam as the national ideology.31 In theory, non-state challengers should
ratchet down their strategic demands to create a bargaining space with governments.32 In practice, however, terrorists may regard their political aspirations as
sacred and thus indivisible.33 This disinclination to compromise on their maximalist
strategic demands lowers the odds of coercive success independent of the tactical
decision-making per se.34 Due to such confounds, researchers have struggled to
isolate the independent effects of terrorism relative to tactical alternatives.
In fact, most studies on the efficacy of terrorism do not even compare the instrument to alternatives.35 Coercion studies offer limited analytical value when they do
not evaluate a tactic relative to others. Without comparing terrorism to alternatives,
it is unknowable whether other tactics would fare better. Some studies do compare
terrorism to other methods, but to unrealistic ones. Robert Pape, for example, compares the political success rate of suicide terrorist campaigns to that of economic sanctions, even though sanctions are not a viable method for terrorist groups.36 Similarly,
Abrahms compares the political outcomes of non-state campaigns directed against
civilians versus military personnel, even though attacking and defeating such hardened targets is not always a viable option.37 Because effectiveness is intrinsically a
relative concept, research on terrorism would benefit by systematically comparing
the tactic to alternatives available to all perpetrators.
In sum, a profusion of empirical research assesses the utility of terrorism for
non-state challengers to obtain their strategic demands. This research program suffers
from recurrent methodological problems, impeding analytical clarity on the more
basic question of whether terrorism promotes government compliance. The next
section presents a research design that largely sidesteps those pitfalls.
Does Terrorism Pay? An Empirical Analysis
5
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Testing Terrorist Efficacy in International Hostage Crises
International Terrorism: Attributes of Terrorist Events (ITERATE) is uniquely well
suited for investigating empirically the relative effectiveness of terrorism within a
particular context. Previous studies have employed this dataset to examine bargaining
dynamics, but none has utilized it for the purpose of assessing terrorist efficacy.38
ITERATE contains fine-grained information from over a thousand international
hostage situations between 1968 and 2005 in which the perpetrators adopted a variety
of tactical approaches to pressure governments into accommodating their redemptive
demands. Although all of the organizations in the sample are deemed terrorist
groups, they employ a variety of tactics in the hostage incidents. ITERATE furnishes
the tactical data in terms of whether the hostage-takers killed or wounded foreign
civilians, foreign government officials, both kinds of captives, or nobody at all, as well
as the outcome of the coercive bargain. This information provides incomparable
leverage for analyzing empirically the coercive effects of various non-state tactics in
a relatively controlled bargaining setting.
Because coercive success depends on whether the defender complies with the
given preferences of the challenger, the dependent variable is based on whether the
hostage-takers obtained their redemptive demands. Whereas the strategic demands
of non-state challengers reflect the political aspirations of their supporters, redemptive demands are designed to sustain the perpetrators by eliciting financial and other
material concessions.39 All of the demands issued are clear and explicit, the bargaining outcomes hail directly from ITERATE, and none of the confrontations is
ongoing, all of which facilitate objective assessment of whether the government
acquiesced. Following Gaibulloev and Sandler, each confrontation is coded as either
a success or a failure depending on whether the perpetrators received any of their
redemptive demands. This is purposely a modest bar for success owing to the relative
strength of the defender.40
Due to the discrete nature of the dependent variable, a logit model is used for testing the following independent variables. The independent variables are unusually
nuanced, capturing the effects of multiple tactical approaches by the international
hostage-takers. Because consensus is still lacking over the definition of terrorism, this
variable is coded in multiple ways. Model 1 employs a variable for terrorism based on
a loose interpretation of the definition. In this model, the tactic is coded as present for
incidents in which the hostage-takers killed or wounded civilians, government officials, or both sets of captives.41 This dummy variable simultaneously captures an
alternative tactical approach—the decision to not physically harm anyone at all. As
Crenshaw and others note, the decision to use terrorism is never the only available
option because refraining from physical harm is always an alternative.42 Model 2
codes the variable of terrorism based on a narrower definition by incorporating only
incidents in which the perpetrators killed or wounded civilians in particular. Model 3
includes a separate tactical variable for incidents that killed or wounded only security
forces, military personnel, or other government officials, a distinction from terrorism
that is becoming the norm.43 This empirical analysis thus departs from previous
research designs by adding specificity to the tactical variable of terrorism and
comparing its efficacy with available alternatives in which neither the means nor ends
are under dispute.
All models include a raft of theory-informed control variables to help isolate the
independent tactical effects. First, coercion studies find that the nature of the demand
Downloaded by [Max Abrahms] at 19:28 14 July 2014
6
M. Abrahms and M. S. Gottfried
influences the success rate, with the greater the anticipated cost of accommodation,
the lower the odds of concessions.44 Several scholars hypothesize that against
non-state challengers in particular, the nature of their demands is a key determinant
in the outcome of the pressure.45 For this reason, the specific types of redemptive
demands issued are held constant, namely, whether the perpetrators demanded
financial ransom, prisoner release, or safe passage out of a location or to a specific
destination. The number of demands issued in the standoff is also controlled because
economic theories on bargaining imply that additional demands boost the chances of
locating a mutually acceptable compromise.46
Second, group capability is another potentially important tactical confound
taken into account. Predictably, most research on interstate conflict finds that the
capability of the challenger is positively related to winning concessions.47 Further,
many studies contend that terrorism works given the weakness of the perpetrators.48
Multiple measures for group capability are employed because of its potential relevance to the empirical investigation. Specifically, we use RAND data for group
membership size, a standard proxy for measuring the capabilities of non-state challengers.49 Such actors unquestionably gain power in numbers.50 Yet membership size
is nonetheless an imperfect proxy for a couple reasons. The rosters of terrorist groups
are seldom known due to their clandestine nature.51 Additionally, terrorist groups
claim as members an indeterminate number of free-riders, many of whom never actually participate in the armed struggle.52 To operationalize organizational capability,
group membership size is therefore supplemented with the number of hostages
taken—a comparatively reliable measure for analyzing the capacity of the perpetrators to inflict physical harm on the targets in the coercive incident.53
Third, the capability of the target country also merits consideration. In conflicts
between states, stronger defenders are predictably less susceptible to coercion.54 The
same causal logic potentially applies against non-state challengers. Two variables are
employed to test the fighting capability of target countries. The first is with data from
the Correlates of War Composite Index of National Capability, an aggregate indicator of military power (e.g., army size) and economic power (e.g., energy consumption). The second measure is operationalized with Polity 2 democracy-autocracy
scores from the Polity IV database because regime type may affect the ability of
governments to resist non-state challenges.55 Finally, a dummy variable controls
for coercive incidents in which the demands were issued not to governments, but to
private corporations or individuals.
Despite the richness of these data, several of the variables lack numerical values.
To account for these data, Amelia II is used to multiply impute them. This statistical
program applies a bootstrapping algorithm that imputes m values for each missing
cell in the data matrix. By doing so, data from all 1,075 reported hostage incidents
are preserved.56 In cases where the same group engages in multiple confrontations
over time, these may not be fully independent, so robust standard errors are included,
thereby correcting for heteroskedasticity and autocorrelation. The next section
analyzes the determinants of bargaining success to begin assessing the evidence of
whether terrorism pays.
Determinants of Bargaining Success
Table 1 displays summary statistics of the international hostage incidents. Table 2 reveals
that the negotiated outcome is not random, but systematically influenced by several
Does Terrorism Pay? An Empirical Analysis
7
Table 1. Summary statistics of coercive bargaining incidents
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Variable
Cow Capability Index
Polity 2 score
Group size
Hostages
Financial ransom redemptive demand
Prisoner release redemptive demand
Safe haven redemptive demand
Safe passage redemptive demand
Robin Hood ransom redemptive demand
No physical harm
Civilians harmed
Officials harmed
Negotiation success
Observations
Mean
SD
1075
1075
1075
1075
1075
1075
1075
1075
1075
1075
1075
1075
1075
0.009
1.184
2369
17.91
0.156
0.108
0.069
0.045
0.002
0.835
0.137
0.039
0.218
0.024
5.912
3693.795
57.080
0.363
0.310
0.254
0.207
0.046
0.371
0.344
0.193
0.413
Dummy variable equals 1 if the determinant is present during a hostage incident and 0
otherwise. Success equals 1 if the group received any of its demands and 0 otherwise.
factors. As expected, both the type and number of redemptive demands issued affect
the odds of concessions. Governments are significantly more likely to accommodate
the perpetrators when their demand is financial (p < .05) or packaged with other
redemptive demands (p < .001). Both the number of hostages seized and regime type
also seem to affect the likelihood of compromise; governments tend to become more
pliant as the number of captives rises (p ¼ .064), and democracies are superior at
resisting the redemptive demands of the hostage-takers (p ¼ .075). The other control
variables do not have a statistically significant effect on bargaining outcomes; in our
sample, the military resources of the target country and the membership size of the
group do not exert a statistically significant impact on the willingness of governments
to comply. Even after accounting for such factors, however, the choice of asymmetric
tactics strongly influences whether the perpetrators will receive their redemptive
demands. Across model specifications, the evidence indicates that aggrieved groups
benefit from tactical restraint at least within international hostage settings.
Model 1 estimates that the decision to kill or wound civilians, government
officials, or both kinds of captives significantly lowers the likelihood of bargaining
success (p < .05). By ensuring the physical safety of others, hostage-takers are
apparently more likely to induce government accommodation. Model 2 adds analytic
clarity by operationalizing a narrower definition of terrorism. In this model, the tactical variable of terrorism is restricted to cases in which only civilian captives are killed
or wounded. The results show that harming civilians in particular has an independent,
negative impact on the likelihood of government accommodation (< .05). By
contrast, Model 3 illustrates that the harming of military personnel and other
government officials does not have a significant effect in impeding concessions. We
thus find no empirical evidence that terrorism, as we carefully define it, is a winning
tactical decision for aggrieved groups to obtain their demands.57 On the contrary,
escalating against the captives is consistently the worst tactical option, particularly
when civilians are the target of pressure.
8
1075
1075
0.634 (0.32)
0.6813 (0.5)
0.6601 (0.32)
1075
2.786 (0.233)
0.526 (6.143)
0.031 (0.018).
0.00002 (0.00004)
0.006 (0.003).
1.05
(0.383)
0.205 (0.533)
0.27
(0.731)
1.412 (1.148)
1.271 (7.313)
1.03
(0.16)
1.035 (0.295)
2.795 (0.233)
0.176 (6.032)
0.033 (0.0169).
0.00002 (0.00005)
0.006 (0.003).
1.063 (0.383)
0.242 (0.532)
0.288 (0.725)
1.413 (1.32)
1.507 (7.479)
1.018 (0.159)
1.049 (0.293)
3.444 (0.335)
0.3956 (6.057)
0.032 (0.017).
0.00002 (0.00005)
0.006 (0.003).
1.0567 (0.383)
0.217 (0.531)
0.27
(0.727)
1.378 (1.132)
1.328 (7.35)
1.029 (0.16)
1.036 (0.294)
0.6558 (0.282)
Model 3
Model 2
Model 1
0.687
0.773
0.524
(0.33)
(0.517)
(0.718)
1075
2.789 (0.228)
0.562 (6.132)
0.032 (0.018).
0.00002 (0.00005)
0.006 (0.003).
1.0358 (0.395)
0.193 (0.519)
0.26
(0.733)
1.437 (1.156)
1.31
(7.333)
1.031 (0.161)
1.033 (0.293)
Model 4
Note: Each column reports the estimated coefficients of a separate logit regression in which the dependent variable is negotiation success, which equals
1 if the group received any demands and 0 otherwise. Robust standard errors appear in parentheses. significant at .001; significant at .01; significant
at .05; .significant at .1.
Constant
COW Capability Index
Polity 2 score
Group size
Hostages
Financial ransom redemptive demand
Prisoner release redemptive demand
Safe haven redemptive demand
Safe passage redemptive demand
Robin hood ransom redemptive demand
Number of demands
Non-government target
No physical harm
Civilians harmed
Officials harmed
Negotiation harm
Observations
Explanatory variables
Table 2. Determinants of coercive success
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Does Terrorism Pay? An Empirical Analysis
9
Table 3. Estimated effect of expected first differences on negotiation success
Model 1
D Hostages
D Terrorism
Model 2
D Hostages
D Terrorism
D Prob
0.0677
6.74
0.066
6.89
success % (0.0023 – 0.224) (22.24 – 0.375) (0.0006 – 0.22) (23.08 – 0.25)
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Note: Each column reports the estimated effect of first differences on the probability of
negotiation success (%). The parentheses contain a 95% confidence interval for each expected
first difference. The change in the number of hostages is 1 to 2 hostages captured. The terrorism variables change from 0 (not present) to 1 (present). The tactical variable of terrorism in
Model 1 includes incidents in which civilians and=or government officials are physically
harmed whereas terrorism in Model 2 refers only to incidents in which civilians are harmed.
To estimate the substantive impact of terrorism on government compliance, we
calculate the discrete marginal effects of employing this tactic holding all other
variables at their mean. Table 3 displays the predicted probabilities of negotiation
success and the 95% confidence intervals for each point estimate. Compared to
refraining from physically harming any of the hostages, engaging in terrorist violence
lowers the chances of bargaining success by 6 to 7 percentage points, depending on
whether civilians are the only ones killed or wounded in the coercive incident. The
marginal effect of capturing additional hostages is also calculated to put this level
of impact into greater perspective. This is a useful empirical comparison because
many scholars theorize that the success of non-state actors depends at least as much
on their capabilities as on their tactical decisions.58 Yet refraining from killing or
wounding civilians actually has a relatively large substantive impact, the equivalent
of the perpetrators seizing 99 hostages per incident, roughly six times the mean and
twice the standard deviation. For achieving the most testable type of non-state
demands, the decision to escalate is ostensibly quite folly.
Robust Results
Numerous tests were performed to ensure that the tactical results are not a quirk of
our econometric specifications.59 Skeptics might wonder, for example, whether the
findings are due to reverse causation in which the perpetrators escalate to terrorism
upon governmental refusal to accommodate their demands. To address this potential
objection, Model 4 in Table 2 includes an important dummy variable called
‘‘Negotiation Harm’’ that controls for whether the terrorist violence was perpetrated
during the bargaining process.60 This variable is not statistically significant; killing or
wounding civilians is evidently exogenous to bargaining failure, consistently lowering
the odds of government compliance regardless of when the escalation occurs in the
course of the standoff. This finding holds across other model specifications, including
one that uses the ‘‘Negotiation Harm’’ variable as the only indicator for violence
against hostages. We also considered whether the length of bargaining time had
any impact on the probability of terrorist success. While there is a very weak positive
relationship between harming civilians and the duration of hostage incidents—a
Pearson’s product-moment correlation equal to 0.08 (p ¼ .07)—we find no evidence
that duration has any impact on negotiation success in any of the model specifications.
Downloaded by [Max Abrahms] at 19:28 14 July 2014
10
M. Abrahms and M. S. Gottfried
An original dataset was created for groups that engaged in multiple hostage
incidents to allay concerns of a potential selection issue. The purpose of the new
dataset was to determine whether the challengers tend to escalate to terrorism over
time, particularly in response to previous negotiation failures. Such evidence would
invite concern that the decision to employ terrorism is a reflection rather than a cause
of government intransigence and thus spurious to the negative bargaining outcome.
The data offer no support for that causal story, however; groups are apparently as
likely to employ un-harmful tactics as they are harmful tactics irrespective of success
or failure in the preceding standoff with the government; statistically, the frequency of
adopting these tactical decisions is identical. Even more important, groups are more
likely to become less harmful upon failing with terrorism than they are to escalate with
this tactic upon failing with more restrained tactics; the Pearson’s product-moment
correlation is 0.691 (p < .0001) and 0.1268 (p < .05), respectively. Together, these
robustness checks indicate that non-state challengers do not adopt terrorism out of
desperation. Although groups seem to apply a modest learning curve by changing their
tactics if they previously failed, the perpetrators are more likely to de-escalate than to
escalate against subsequent captives. The regression analysis controls for the capabilities of both the non-state challengers and government defenders in a highly specified
bargaining context. Because terrorism is often described as a weapon of the weak,
however, we also inspected the bivariate relationships between the group capability
measures and the decision to kill or wound civilians. Neither the membership size of
the group, nor the number of captives seized, is related to the use of terrorism. One
may therefore rule out the possibility that the groups escalated to compensate for
deficient membership or hostage seizures, strong proxies for capability.
Conceivably, terrorist tactics in hostage settings matter more for certain types of
redemptive demands than others. For this reason, we also interacted our tactical
variables with each redemptive demand. Consistent with our earlier findings, terrorism decreases the likelihood of success when the perpetrators request a safe haven
(p < .05), and has no effect on financial ransoms, so-called Robin Hood ransoms,
or the release of prisoners. The data does suggest, however, that violence may
increase the chances of terrorists obtaining safe passage to a new location, but even
this modest concession is not statistically significant at the .05 level. Thus, we still
find that extremists are better off not harming anyone during the course of a hostage
incident irrespective of their redemptive demand.
A final concern to address is whether one group or a few groups are driving the
statistical results. To attend to this potential issue, all of the models have been rerun
using multilevel logistic regression taking into account individual group factors.
These models use the previous covariates as fixed effects and allow each of the intercepts to vary according to the perpetrator (i.e., group random effects). As Table 4
demonstrates, there is virtually no change in the coefficients for the tactical variables.
The negative effects of terrorism are thus not due to a single group or a small number
of groups. Compared to refraining from harming any of the captives or killing or
wounding only non-civilian ones, terrorism consistently decreases the probability
of government concessions regardless of the specific hostage-takers.61
Research Implications
The forgoing analysis contributes to numerous research programs within the burgeoning study of conflict. First, the empirical investigation enhances the evidence
11
1075
1075
0.635 (0.321)
0.679 (0.527)
0.66 (0.321)
1075
2.751 (0.204)
0.764 (6.0682)
0.031 (0.0173).
0.069 (0.177)
0.006 (0.003).
1.426 (1.002)
0.785 (0.978)
0.031 (0.459)
0.327 (0.738)
1.164 (7.397)
1.04 (0.15)
1.04 (0.296)
2.76 (0.206)
0.442 (5.974)
0.033 (0.0174).
0.072 (0.174)
0.006 (0.003).
1.431 (0.992)
0.788 (1.004)
0.057 (0.463)
0.339 (0.736)
1.39 (7.547)
1.029 (0.147)
1.044 (0.295)
3.408 (0.33)
0.643 (5.994)
0.032 (0.017).
0.072 (0.175)
0.006 (0.003).
1.402 (0.985)
0.789 (0.986)
0.04 (0.459)
0.319 (0.73)
1.218 (7.442)
1.875 (0.497)
1.032 (0.297)
0.656 (0.287)
Model 3
Model 2
Model 1
0.686 (0.335)
0.77 (0.529)
0.514 (0.773)
1075
2.753 (0.201)
0.791 (6.049)
0.032 (0.0173).
0.0733 (0.177)
0.006 (0.003).
1.431 (1.02)
0.787 (0.951)
0.0324 (0.453)
0.335 (0.75)
1.208 (7.407)
1.042 (0.15)
1.03 (0.295)
Model 4
Note: Each column reports the estimated coefficients of a separate multilevel logistic regression in which the dependent variable is negotiation success,
which equals 1 if the group received any demands and 0 otherwise. Standard errors appear in parentheses. significant at .001; significant at .01;
significant at .05; .significant at .1.
Constant
COW Capability Index
Polity 2 score
Group size
Hostages
Financial ransom redemptive demand
Prisoner release redemptive demand
Safe haven redemptive demand
Safe passage redemptive demand
Robin hood ransom redemptive demand
Number of demands
Non-government target
No physical harm
Civilians harmed
Officials harmed
Negotiations
Observations
Explanatory variable
Table 4. Determinants of coercive success controlling for group effects
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Downloaded by [Max Abrahms] at 19:28 14 July 2014
12
M. Abrahms and M. S. Gottfried
that tactical decision making matters. Although coercion studies have traditionally
evaluated the instruments of states, a growing body of research finds that the tactical
choices of non-state actors also affect their political success.62 Specifically, the results
bolster the proposition that terrorism is an unprofitable tactic at inducing government concessions.63 Of course, all studies have their limitations: our sample is restricted to a certain bargaining context—hostage crises in which the perpetrators and
victims are of different nationalities (viz. international hostage crises) and the
demands are redemptive in nature. It is therefore indeterminate from this analysis
whether escalating against civilians impedes government concessions outside of hostage contexts, when perpetrated by domestic groups, or for larger demands. Despite
such limitations, however, this study forges new ground in establishing that the poor
return from harming civilians is not an artifact of the weakness of the non-state challengers, the strength of government opposition, or the extremeness of their aims.
Second, the results complement the emerging literature on nonviolent resistance.64 Admittedly, none of the groups in our sample was entirely nonviolent since
each seized captives against their will. Our study provides additional evidence,
though, that tactical restraint can aid aggrieved groups in achieving their demands.
Unlike previous scholarship, the estimation strategy isolates the tactics employed in
each coercive incident, compares terrorist tactics to less violent alternatives available
to the perpetrators, and assesses the relative utility with over a thousand cases in a
fairly steady bargaining context.
Third, this study accords with findings that terrorism tends to increase electoral
support for politically intransigent politicians.65 Although our focus is on governmental responses to terrorism, Table 2 shows that private corporations and individuals are significantly more likely to comply (p < .01). These results invite the
question of why governments are less likely than private citizens to appease terrorism.
A potential explanation from bargaining theory is that governments face higher audience costs than private citizens for accommodating the perpetrators. Unlike in the
private sector, national leaders are loath to comply because their electorates may
punish them at the polls, the nation may lose standing in the international community, or even incentivize more political violence.66 Clearly, additional research
should scrutinize the mechanism of precisely why terrorism appears to lower the odds
of government accommodation.67
Fourth, our findings vitiate the conventional wisdom that liberal democracies
are uniquely prone to appeasing terrorism. Liberal democracies are supposedly less
resilient to terrorism because their commitment to civil liberties inhibits them from
adopting sufficiently harsh countermeasures and their low civilian cost tolerance
limits the capacity to withstand attacks.68 Across models (Tables 2 and 4), however,
liberal countries are actually negatively associated with government concessions
albeit only at the .10 level of statistical significance. Future research should further
test how regime type affects the odds of terrorist appeasement.
Fifth, the results accord with the civilian victimization literature, which generally
finds that state challengers also fail to benefit at the bargaining table by harming civilians.69 Punishing civilians appears to be no more effective as a coercive tool for hostagetakers than indiscriminate aerial bombings, scorched-earth campaigns, or sieges are for
state challengers. An unexplored question is whether the political utility of civilian
targeting has declined over time due to normative changes in the international system.
Sixth, the analysis provides an empirical basis for assessing the aphorism of
terrorism as a weapon of the weak. This assumption has been a staple of terrorism
Does Terrorism Pay? An Empirical Analysis
13
Downloaded by [Max Abrahms] at 19:28 14 July 2014
studies since the early 1960s.70 As Fortna notes, however: ‘‘Almost no empirical
work has tested this conventional wisdom directly.’’71 In our study, group capability
is uncorrelated with the tactical choices of the perpetrators. Contrary to the conventional wisdom, the decision of hostage-takers to harm their captives is unrelated to
the numerical strength of either party. Moreover, the robustness checks indicate that
non-state challengers may be more likely to de-escalate than to escalate tactically
after negotiation failures.
Finally, this study raises new questions about terrorist motives. Political scientists often assume that groups adopt terrorism because of its effectiveness in coercing
government concessions.72 If harming civilians is politically counterproductive, however, then its practitioners may be motived by an alternative incentive structure.73
Future research should continue to explore why aggrieved groups harm civilians
given the mounting empirical evidence that such indiscriminate violence does
not pay.
Acknowledgments
The authors wish to thank Steven Pinker, Robert Trager, and Ron Gurantz for
excellent research suggestions.
Funding
Max Abrahms wishes to thank the philanthropist Jason Karp for his generous
financial support. Matthew S. Gottfried would like to thank the Institute on Global
Conflict and Cooperation and the Lynde and Harry Bradley Foundation for their
financial support as well.
Notes
1. Bonnie Cordes, Bruce Hoffman, and Brian Michael Jenkins, Trends in International
Terrorism, 1982 and 1983 (Santa Monica: RAND Corporation, 1984); Martha Crenshaw,
‘‘The Logic of Terrorism: Terrorist Behavior as a Product of Strategic Choice,’’ in Walter
Reich, ed., Origins of Terrorism: Psychologies, Ideologies, Theologies, States of Mind
(Washington, DC: Woodrow Wilson Center Press, 1988), 7–24; Virginia Held, ‘‘Terrorism,
Rights, and Political Goals,’’ in R. G. Frey and Christopher W. Morris, eds., Violence,
Terrorism, and Justice (Cambridge: Cambridge University Press, 1991), 59–85; Paul
Wilkinson, Terrorism and the Liberal State (New York: New York University Press, 1986).
2. Andrew Kydd and Barbara F. Walter, ‘‘The Strategies of Terrorism,’’ International
Security 31, no. 1 (Summer 2006): 49–80; David A. Lake, ‘‘Rational Extremism: Understanding Terrorism in the Twenty-First Century,’’ Dialogue-IO 1, no. 1 (January 2002): 15–29;
Robert A. Pape, ‘‘The Strategic Logic of Suicide Terrorism,’’ American Political Science
Review 97, no. 3 (August 2003): 243–361; Sprinzak Ehud, ‘‘Rational Fanatics,’’ Foreign Policy,
no. 120 (September=October 2000): 66–73.
3. Max Abrahms, ‘‘Why Terrorism Does Not Work,’’ International Security 31, no. 2
(Fall 2006): 42–78; Max Abrahms, ‘‘The Political Effectiveness of Terrorism Revisited,’’
Comparative Political Studies 45, no. 3 (February 2012): 366–393; Max Abrahms, ‘‘Does
Terrorism Really Work? Evolution in the Conventional Wisdom since 9=11,’’ Defence and Peace
Economics 22, no. 6 (December 2011): 583–594; Max Abrahms, ‘‘Al Qaeda’s Miscommunication
War: The Terrorism Paradox,’’ Terrorism and Political Violence 17, no. 4 (Spring 2005):
529–549; Max Abrahms, ‘‘Al Qaeda’s Scorecard: A Progress Report on Al Qaeda’s Objectives,’’
Studies in Conflict & Terrorism 29, no. 5 (2006): 509–529; Max Abrahms, ‘‘The Credibility
Paradox: Violence as a Double-Edged Sword in International Politics,’’ International Studies
Downloaded by [Max Abrahms] at 19:28 14 July 2014
14
M. Abrahms and M. S. Gottfried
Quarterly 57, no. 4 (December 2013): 600–671; Max Abrahms and Karolina Lula, ‘‘Why
Terrorists Overestimate the Odds of Victory,’’ Perspectives on Terrorism 6, nos. 4–5 (October
2012); Audrey Kurth Cronin, How Terrorism Ends (Princeton: Princeton University Press,
2009); Page Fortna, ‘‘Do Terrorists Win? Rebels’ Use of Terrorism and Civil War Outcomes’’
(unpublished manuscript, Department of Political Science, Columbia University, 2011); Anna
Getmansky and Tolga Sinmazdemir, ‘‘Success Breeds Failure: The Effect of Terrorist Attacks
on Land Control in the Israeli-Palestinian Conflict’’ APSA 2012 Annual Meeting Paper
(2012); Anna Getmansky and Thomas Zeitzoff, ‘‘Terrorism and Voting: The Effect of Exposure
to Rockets on Voting in Israeli Elections’’ forthcoming in the American Political Science Review
(2013); Seth G. Jones and Martin C. Libicki, How Terrorist Groups End: Lessons for Countering
al Qa’ida (Santa Monica: RAND Corporation, 2008).
4. Eric D. Gould and Esteban F. Klor, ‘‘Does Terrorism Work?,’’ The Quarterly Journal
of Economics 125, no. 4 (November 2010): 1459–1510; Bruce Hoffman, ‘‘The Rationality of
Terrorism and Other Forms of Political Violence: Lessons from the Jewish Campaign in
Palestine, 1939–1947,’’ Small Wars & Insurgencies 22, no. 2 (May 2011): 258–272; Sharon
L. Caudle, ‘‘Review of Debating Terrorism and Counterterrorism: Conflicting Perspectives
on Causes, Contexts, and Responses,’’ Journal of Homeland Security and Emergency Management 7, no. 1 (2010); Jakana L. Thomas, ‘‘Rewarding Bad Behavior: How Governments
Respond to Rebel Group Tactics in African Civil Wars’’ (paper presented at the annual
conference of the ISA, San Diego, California, April 4, 2012); Peter Krause, ‘‘The Political
Effectiveness of Non-State Violence: A Two-Level Framework to Transform a Deceptive
Debate,’’ Security Studies 22, no. 2 (April 2013): 259–294; William Rose, Rysia Murphy,
and Max Abrahms, ‘‘Does Terrorism Ever Work? The 2004 Madrid Train Bombings,’’ International Security 32, no. 1 (Summer 2007): 185–192; Max Abrahms, H-Diplo=ISSF Response
to Peter Krause, ‘‘The Political Effectiveness of Non-State Violence: A Two-Level Framework
to Transform a Deceptive Debate,’’ Security Studies 22, no. 2 (2013): 259–294; Erica Chenoweth, Nicholas Miller, Elizabeth McClellan, Hillel Frisch, Paul Staniland, and Max Abrahms,
‘‘What Makes Terrorists Tick,’’ International Security 33, no. 4 (Spring 2009): 180–202; Aaron
Edwards, ‘‘When Terrorism as Strategy Fails: Dissident Irish Republicans and the Threat to
British Security,’’ Studies in Conflict & Terrorism 34, no. 4 (2011): 318–336; Sarah V. Marsden,
‘‘How Terrorism Ends: Understanding the Outcomes of Violent Political Contestation’’ (PhD
diss. University of St Andrews, 2013); Ami-Jacques Rapin, ‘‘Why Is It So Difficult to Evaluate
the Political Impact of Terrorism?,’’ in Samuel Justin Sinclair and Daniel Antonius, eds., The
Political Psychology of Terrorism Fears (New York: Oxford University Press, 2013).
5. Gary C. Gambill, ‘‘The Balance of Terror: War by Other Means in the Contemporary
Middle East,’’ Journal of Palestine Studies 28, no. 1 (Autumn 1998): 51–66.
6. Branislav Slantchev, ‘‘The Power to Hurt: Costly Conflict with Completely Informed
States,’’ American Political Science Review 97, no. 1 (February 2003): 123–133.
7. Abrahms, ‘‘Al Qaeda’s Scorecard’’ and ‘‘The Political Effectiveness of Terrorism
Revisited’’ (see note 3 above); Cronin, How Terrorism Ends (see note 3 above); Getmansky
and Sinmazdemir, ‘‘Success Breeds Failure’’ (see note 3 above); Jones and Libicki, How
Terrorist Groups End (see note 3 above); Hoffman, ‘‘The Rationality of Terrorism and
Other Forms of Political Violence’’ (see note 4 above); Pape, ‘‘The Strategic Logic of Suicide
Terrorism’’ (see note 2 above); Thomas, ‘‘Rewarding Bad Behavior’’ (see note 4 above).
8. Krause, ‘‘The Political Effectiveness of Non-State Violence’’ (see note 4 above); Rose,
Murphy, and Abrahms, ‘‘Does Terrorism Ever Work?’’ (see note 4 above); Chenoweth et al.,
‘‘What Makes Terrorists Tick’’ (see note 4 above); Assaf Moghadam, ‘‘Suicide Terrorism,
Occupation, and the Globalization of Martyrdom: A Critique of Dying to Win,’’ Studies in
Conflict & Terrorism 29, no. 8 (2006): 707–729.
9. Crenshaw, ‘‘The Logic of Terrorism’’ (see note 1 above); Thomas C. Schelling, ‘‘What
Purposes Can International Terrorism Serve?,’’ in Raymond Gillespie Frey and Christopher W.
Morris, eds., Violence, Terrorism, and Justice (New York: Cambridge University Press, 1991),
21–32; Jessica Stern, ‘‘The Protean Enemy,’’ Foreign Affairs 82, no. 4 (July=August 2003): 27–40.
10. Abrahms, ‘‘Why Terrorism Does Not Work’’ (see note 3 above); Moghadam,
‘‘Suicide Terrorism, Occupation, and the Globalization of Martyrdom’’ (see note 8 above);
Pape, ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above); Alan Dershowitz,
Why Terrorism Works: Understanding the Threat, Responding to the Challenge (New Haven:
Yale University Press, 2002).
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Does Terrorism Pay? An Empirical Analysis
15
11. Chenoweth et al., ‘‘What Makes Terrorists Tick’’ (see note 4 above); Edwards,
‘‘When Terrorism as Strategy Fails’’ (see note 4 above); Rose, Murphy, and Abrahms, ‘‘Does
Terrorism Ever Work?’’ (see note 4 above).
12. Lake, ‘‘Rational Extremism’’ (see note 2 above); Beth Roy, John Burdick, and Louis
Kriesberg, ‘‘A Conversation between Conflict Resolution and Social Movement Scholars,’’
Conflict Resolution Quarterly 27, no. 4 (Summer 2010): 347–368.
13. Abrahms, ‘‘Al Qaeda’s Scorecard’’ (see note 3 above); Pape, ‘‘The Strategic Logic of
Suicide Terrorism’’ (see note 2 above).
14. Ahmer Tarar and Bahar Leventoğlu, ‘‘Public Commitment in Crisis Bargaining,’’
International Studies Quarterly 53, no. 3 (September 2009): 817–839.
15. Jones and Libicki, How Terrorist Groups End (see note 3 above).
16. Cronin, How Terrorism Ends (see note 3 above); Thomas, ‘‘Rewarding Bad Behavior’’
(see note 4 above).
17. Rose, Murphy, and Abrahms, ‘‘Does Terrorism Ever Work?’’ (see note 4 above).
18. Claude Berrebi and Esteban F. Klor, ‘‘Are Voters Sensitive to Terrorism: Direct
Evidence from the Israeli Electorate,’’ American Political Science Review 102, no. 3 (August
2008): 279–301; Claude Berrebi and Esteban F. Klor, ‘‘On Terrorism and Electoral Outcomes:
Theory and Evidence from the Israeli–Palestinian Conflict,’’ Journal of Conflict Resolution 50,
no. 6 (December 2006): 899–925; Eric D. Gould and Esteban F. Klor, ‘‘Does Terrorism
Work?,’’ The Quarterly Journal of Economics 125, no. 4 (November 2010): 1459–1510;
Matthew S. Gottfried, ‘‘The Domestic Politics of Coercive Terrorism’’ (paper presented at
the annual meeting for the American Political Science Association, Chicago, Illinois, August
29–September 1, 2013); Matthew S. Gottfried, ‘‘The Origins and Consequences of Public
Opinion in Coercive Terrorist Crises’’ (PhD diss. University of California, Los Angeles,
2014); Anna Getmansky and Thomas Zeitzoff, ‘‘Terrorism and Voting: The Effect of
Exposure to Rockets on Voting in Israeli Elections’’ (working paper, New York University,
2013).
19. See Neil Malhotra and Alexander G. Kuo, ‘‘Attributing Blame: The Public’s
Response to Hurricane Katrina,’’ Journal of Politics 70, no. 1 (January 2008): 120–135.
20. Audrey Kurth Cronin, ‘‘Rethinking Sovereignty: American Strategy in the Age of
Terrorism,’’ Survival 44, no. 2 (Summer 2002): 119–139; Boaz Ganor, ‘‘Defining Terrorism:
Is One Man’s Terrorist Another Man’s Freedom Fighter?,’’ Police Practice and Research:
An International Journal 3, no. 2 (2002): 287–304; Karolina Lula, ‘‘Why Defining Terrorism
Matters,’’ Monkey Cage, May 28, 2013; Brian Michael Jenkins, The Study of Terrorism: Definitional Problems (Santa Monica: RAND Corporation, 1980); Nicholas Sambanis, ‘‘Terrorism
and Civil War,’’ in Philip Keefer and Norman Loayza, eds., Terrorism, Economic Development
and Political Openness (New York: Cambridge University Press, 2008), 174–208; Michael Walzer, ‘‘Five Questions about Terrorism,’’ Dissent 49, no. 1 (Winter 2002): 5–10; Joseph K.
Young and Michael G. Findley, ‘‘Promise and Pitfalls of Terrorism Research,’’ International
Studies Review 13, no. 3 (September 2011): 411–431.
21. Daniel Byman, A High Price: The Triumphs and Failures of Israeli Counterterrorism
(New York: Oxford University Press, 2011); Jeff Goodwin, ‘‘A Theory of Categorical
Terrorism,’’ Social Forces 84, no. 4 (June 2006): 2027–2046; David C. Rapoport, ‘‘Terrorism,’’
in Mary Hawkesworth and Maurice Kogan, eds., Routledge Encyclopedia of Government and
Politics, Vol. 2 (London: Routledge, 1992), 111–127; Alex P. Schmid and Albert J. Jongman,
Political Terrorism: A Research Guide to Concepts, Theories, Data Bases and Literature (New
Brunswick: North-Holland, 1984); Paul Wilkinson, Terrorism and the Liberal State, 2nd ed.
(New York: New York University Press, 1986).
22. Maria J. Stephan and Erica Chenoweth, ‘‘Why Civil Resistance Works: The Strategic
Logic of Nonviolent Conflict,’’ International Security 33, no. 1 (Summer 2008): 7–44; Bron
Taylor, ‘‘Religion, Violence and Radical Environmentalism: From Earth First! to the
Unabomber to the Earth Liberation Front,’’ Terrorism and Political Violence 10, no. 4 (Winter
1998): 1–42.
23. Gordon H. McCormick and Lindsay Fritz, ‘‘Is Suicide Terrorism an Effective
Tactic?,’’ in Stuart Gottlieb, ed., Debating Terrorism and Counterterrorism: Conflicting
Perspectives on Causes, Contexts, and Responses (Washington, DC: CQ Press, 2010), Ch. 5;
Pape, ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above).
24. Kydd and Walter, ‘‘The Strategies of Terrorism’’ (see note 2 above), 49.
Downloaded by [Max Abrahms] at 19:28 14 July 2014
16
M. Abrahms and M. S. Gottfried
25. Abrahms, ‘‘Al Qaeda’s Scorecard’’ (see note 3 above).
26. Erica Chenoweth and Maria J. Stephan, Why Civil Resistance Works: The Strategic
Logic of Nonviolent Conflict (New York: Columbia University Press, 2011), 16.
27. Byman, A High Price (see note 21 above).
28. Leonard Weinberg and Ami Pedahzur, Political Parties and Terrorist Groups
(London: Routledge, 2003).
29. Daniel Byman, Peter Chalk, Bruce Hoffman, William Rosenau, and David Brannan,
Trends in Outside Support for Insurgent Movements (Santa Monica: RAND Corporation,
2001); McCormick and Fritz, ‘‘Is Suicide Terrorism an Effective Tactic?’’ (see note 23 above).
30. On this methodological challenge, see Daniel W. Drezner, ‘‘The Hidden Hand of
Economic Coercion,’’ International Organization 57, no. 3 (Summer 2003): 643–659.
31. Abrahms, ‘‘The Political Effectiveness of Terrorism Revisited’’ (see note 3 above).
32. James D. Fearon, ‘‘Rationalist Explanations for War,’’ International Organization 49,
no. 3 (Summer 1995): 379–414.
33. Ron E. Hassner, ‘‘To Halve and to Hold: Conflicts over Sacred Space and the Problem
of Indivisibility,’’ Security Studies 12, no. 4 (Summer 2003): 1–33; Paul R. Pillar, Negotiating
Peace: War Termination as a Bargaining Process (Princeton: Princeton University Press, 1983).
34. Thomas, ‘‘Rewarding Bad Behavior’’ (see note 4 above).
35. See, for example, Cordes et al., Trends in International Terrorism, 1982 and 1983 (see
note 1 above); Crenshaw, ‘‘The Logic of Terrorism’’ (see note 1 above); Held, ‘‘Terrorism,
Rights, and Political Goals’’ (see note 1 above); Walter Laqueur, ‘‘The Futility of Terrorism,’’
Harper’s 252, no. 1510 (March 1976): 99–105; Schelling, ‘‘What Purposes Can International
Terrorism Serve?’’ (see note 9 above).
36. Pape, ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above).
37. Abrahms, ‘‘The Political Effectiveness of Terrorism Revisited’’ (see note 3 above).
38. John A. Call, ‘‘Negotiating Crises: The Evolution of Hostage=Barricade Crisis
Negotiation,’’ in Harold V. Hall, ed., Terrorism: Strategies for Intervention (Binghamton:
Haworth, 2003), 69–94; Khusrav Gaibulloev and Todd Sandler, ‘‘Hostage Taking: Determinants of Terrorist Logistical and Negotiation Success,’’ Journal of Peace Research 46, no. 6
(November 2009): 739–756; Kent L. Oots, ‘‘Bargaining with Terrorists: Organizational
Considerations,’’ Terrorism 13, no. 2 (1990): 145–158.
39. Gambill, ‘‘The Balance of Terror’’ (see note 5 above).
40. Gaibulloev and Sandler, ‘‘Hostage Taking’’ (see note 38 above).
41. This is a minority interpretation of the definition, but not without adherents. See
Pape, ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above). On the complexities
of coding the tactical variable for terrorism, see Alex P. Schmid and Albert J. Jongman, Political Terrorism: A New Guide to Actors, Authors, Concepts, Data Bases, Theories, and Literature
(New Brunswick: Transaction, 2005).
42. Martha Crenshaw, ‘‘Decisions to Use Terrorism: Psychological Constraints on
Instrumental Reasoning,’’ International Social Movements Research 4, no. 1 (1992); James
DeNardo, Power in Numbers: The Political Strategy of Protest and Rebellion (Princeton:
Princeton University Press, 1985).
43. Max Abrahms and Philip B. K. Potter, ‘‘Explaining Terrorism: Leadership Deficits
and Militant Group Tactics,’’ forthcoming in International Organization (2015); Ganor,
‘‘Defining Terrorism’’ (see note 20 above); Goodwin, ‘‘A Theory of Categorical Terrorism’’
(see note 21 above).
44. Daniel Byman and Matthew Waxman, The Dynamics of Coercion: American Foreign
Policy and the Limits of Military Might (New York: Cambridge University Press, 2002);
Alexander George and William Simons, The Limits of Coercive Diplomacy (Boulder: Westview
Press, 1994).
45. Jones and Libicki, How Terrorist Groups End (see note 3 above); Lake, ‘‘Rational
Extremism’’ (see note 2 above).
46. Todd Sandler and John L. Scott, ‘‘Terrorist Success in Hostage-Taking Incidents: An
Empirical Study,’’ Journal of Conflict Resolution 31, no. 1 (March 1987): 35–53.
47. Robert J. Art and Patrick M. Cronin, eds., The United States and Coercive Diplomacy
(Washington, DC: United States Institute of Peace Press, 2003).
48. Hoffman, ‘‘The Rationality of Terrorism and Other Forms of Political Violence’’ (see note
4 above); McCormick and Fritz, ‘‘Is Suicide Terrorism an Effective Tactic?’’ (see note 23 above).
Downloaded by [Max Abrahms] at 19:28 14 July 2014
Does Terrorism Pay? An Empirical Analysis
17
49. Jones and Libicki, How Terrorist Groups End (see note 3 above).
50. DeNardo, Power in Numbers (see note 42 above).
51. Alex Schmid and Albert Jongman, Political Terrorism, 2nd ed. (Oxford: NorthHolland, 1988).
52. Abrahms and Potter, ‘‘Explaining Terrorism’’ (see note 43 above); Jacob N. Shapiro
and David A. Siegel, ‘‘Moral Hazard, Discipline, and the Management of Terrorist Organizations,’’ World Politics 64, no. 1 (January 2012): 39–78.
53. On the logic of this coding decision, see Adam Dolnik, ‘‘Contrasting Dynamics of
Crisis Negotiations: Barricade versus Kidnapping Incidents,’’ International Negotiation: A
Journal of Theory and Practice 8, no. 3 (2003): 495–526.
54. Daniel W. Drezner, ‘‘Conflict Expectations and the Paradox of Economic Coercion,’’
International Studies Quarterly 42, no. 4 (December 1998): 709–731.
55. Max Abrahms, ‘‘Why Democracies Make Superior Counterterrorists,’’ Security
Studies 16, no. 2 (April-June 2007): 223–253; Jason Lyall, ‘‘Do Democracies Make Inferior
Counterinsurgents? Reassessing Democracy’s Impact on War Outcomes and Duration,’’
International Organization 64, no. 1 (January 2010): 167–192; William Eubank and Leonard
Weinberg, ‘‘Does Democracy Encourage Terrorism?,’’ Terrorism and Political Violence 6,
no. 4 (Winter 1994): 417–443; Gil Merom, How Democracies Lose Small Wars: State, Society
and the Failures of France in Algeria, Israel in Lebanon, and the United States in Vietnam (New
York: Cambridge University Press, 2003).
56. This approach is superior to the common practice of list-wise deletion, which can jettison valuable information. As a robustness check, we also reran the regressions using list-wise
deletion, which does not affect the coefficient estimates in the next section. On list-wise
deletion, see James Honaker and Gary King, ‘‘What to Do about Missing Values in
Time-Series Cross-Sectional Data,’’ American Journal of Political Science 54, no. 2 (April
2010): 561–581.
57. We also investigated the effect of violence on the probability of the terrorist group
obtaining all of its demands. The relationship between our harm variables and terrorist success
substantially increases in magnitude, all in the expected direction. Yet, the different specifications of violence and non-violence in these models are not statistically significant, an artifact
that results from the very few incidences of terrorist groups obtaining all of their demands (less
than 5% of the time). This non-finding, however, still confirms that violence during hostage
negotiations fails to coerce governments into granting concessions.
58. DeNardo, Power in Numbers (see note 42 above); Lake, ‘‘Rational Extremism’’ (see
note 2 above).
59. Drezner, ‘‘The Hidden Hand of Economic Coercion’’ (see note 30 above).
60. An interesting extension to this analysis would be to look at how violence against
hostages (before the bargaining process begins) affects the probability of negotiation success.
ITERATE provides information on whether a victim was killed while attempting to escape
after their initial capture, perhaps capturing this dynamic, but it only records two observations
of this nature in the entire dataset. This data limitation prevents us from investigating the
effect of violence on negotiations before the beginning of the bargaining process. We do
believe, however, that this is an important question for future research.
61. To our knowledge, there is no statistical package in R for running multilevel logistic
regression on the imputed datasets. We therefore wrote our own statistical program, which
runs separate multilevel logistic regression on each imputed dataset and then combines the
results with Rubin’s Rule. See Donald B. Rubin, Multiple Imputation for Nonresponse in
Surveys (New York: Wiley, 1987).
62. On state coercion, see George and Simons, The Limits of Coercive Diplomacy (see
note 44 above); Robert Anthony Pape, Bombing to Win: Air Power and Coercion in War
(Ithaca: Cornell University Press, 1996); Alastair Smith, ‘‘The Success and Use of Economic
Sanctions,’’ International Interactions 21, no. 3 (1995): 229–245.
63. Abrahms, ‘‘Al Qaeda’s Scorecard’’ and ‘‘The Political Effectiveness of Terrorism
Revisited’’ (see note 3 above); Cronin, How Terrorism Ends (see note 3 above); Edwards,
‘‘When Terrorism as Strategy Fails’’ (see note 4 above); Jones and Libicki, How Terrorist
Groups End (see note 3 above); Getmansky and Sinmazdemir, ‘‘Success Breeds Failure’’ (see
note 3 above).
Downloaded by [Max Abrahms] at 19:28 14 July 2014
18
M. Abrahms and M. S. Gottfried
64. Chenoweth and Stephan, ‘‘Why Civil Resistance Works’’ (see note 22 above); Brian
Martin, ‘‘Nonviolence and Communication,’’ Journal of Peace Research 40, no. 2 (March
2003): 213–232; Stephen Zunes, ‘‘The Role of Nonviolent Action in the Downfall of
Apartheid,’’ Journal of Modern African Studies 37, no. 1 (March 1999): 137–169.
65. Berrebi and Klor, ‘‘On Terrorism and Electoral Outcomes’’ and ‘‘Are Voters
Sensitive to Terrorism’’ (see note 18 above); Gottfried, ‘‘The Domestic Politics of Coercive
Terrorism’’ and ‘‘The Origins and Consequences of Public Opinion in Coercive Terrorist
Crises’’ (see note 18 above).
66. James D. Fearon, ‘‘Domestic Political Audiences and the Escalation of International
Disputes,’’ American Political Science Review 88, no. 3 (September 1994): 577–592; James D.
Fearon, ‘‘Signaling Foreign Policy Interests: Tying Hands versus Sinking Costs,’’ Journal of
Conflict Resolution 41, no. 1 (February 1997): 68–90; James D. Morrow, ‘‘How Could Trade
Affect Conflict?,’’ Journal of Peace Research 36, no. 4 (July 1999): 481–489; Kenneth Schultz,
‘‘Domestic Opposition and Signaling in International Crises,’’ American Political Science
Review 92, no. 4 (December 1998): 829–844; Alastair Smith, ‘‘International Crises and
Domestic Politics,’’ American Political Science Review 92, no. 3 (September 1998): 623–638.
67. For a recent study on this question, see Abrahms, ‘‘The Credibility Paradox’’ (see
note 3 above).
68. Abrahms, ‘‘Does Terrorism Ever Work?’’ (see note 4 above); Kydd and Walter, ‘‘The
Strategies of Terrorism’’ (see note 2 above); Merom, How Democracies Lose Small Wars (see
note 55 above); Pape, ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above).
69. Abrahms, ‘‘The Credibility Paradox’’ (see note 3 above); Byman and Waxman, The
Dynamics of Coercion (see note 44 above); Caleb Carr, The Lessons of Terror: A History of
Warfare against Civilians (New York: Random House Digital, 2003); Michael Horowitz
and Dan Reiter, ‘‘When Does Aerial Bombing Work? Quantitative Empirical Tests,
1917–1999,’’ Journal of Conflict Resolution 45, no. 2 (April 2001): 147–173; Pape, Bombing
to Win (see note 62 above).
70. Martha Crenshaw, ‘‘The Causes of Terrorism,’’ Comparative Politics 13, no. 4 (July
1981): 379–399; Brian Crozier, The Rebels: A Study of Post-War Insurrections (New York:
Chatto and Windus, 1960); Alex P. Schmid, ‘‘Terrorism and the Use of Weapons of Mass
Destruction: From Where the Risk?,’’ Terrorism and Political Violence 11, no. 4 (1999):
106–132; Thomas Perry Thornton, ‘‘Terror as a Weapon of Political Agitation,’’ TerrorismCritical Concepts in Political Science 3, no. 1 (1964): 41–64; Stephen M. Walt, ‘‘Taming
American Power,’’ Foreign Affairs 84, no. 5 (2005): 105–120; Reed M. Wood, ‘‘Rebel
Capability and Strategic Violence against Civilians,’’ Journal of Peace Research 47, no. 5
(September 2010): 601–614; Dustin Ells Howes, ‘‘Terror in and out of Power,’’ European
Journal of Political Theory 11, no. 1 (January 2012): 25–58.
71. Fortna, ‘‘Do Terrorists Win?’’ (see note 3 above), 25.
72. Hoffman, ‘‘The Rationality of Terrorism and Other Forms of Political Violence’’ (see
note 4 above); Kydd and Walter, ‘‘The Strategies of Terrorism’’ (see note 2 above); Lake,
‘‘Rational Extremism’’ (see note 2 above); Gordon H. McCormick, ‘‘Terrorist Decision
Making,’’ Annual Review of Political Science 6, no. 1 (June 2003): 473–507; Pape ‘‘The Strategic Logic of Suicide Terrorism’’ (see note 2 above).
73. Max Abrahms, ‘‘What Terrorists Really Want: Terrorist Motives and Counterterrorism Strategy,’’ International Security 32, no. 4 (Spring 2008): 78–105; Abrahms and Potter,
‘‘Explaining Terrorism: Leadership Deficits and Militant Group Tactics’’ (see note 43 above).
Studies in Conflict & Terrorism, 29:707–729, 2006
Copyright © Taylor & Francis Group, LLC
ISSN: 1057-610X print / 1521-0731 online
DOI: 10.1080/10576100600561907
Suicide Terrorism, Occupation, and the
Globalization of Martyrdom: A Critique
of Dying to Win
ASSAF MOGHADAM
John M. Olin Institute for Strategic Studies
Harvard University
Cambridge, Massachusetts, USA
This article offers a three-pronged critique of Robert A. Pape’s book Dying to Win.
The first section of the article highlights problems related to the book’s definition of key
concepts, its assessment of existing research on suicide terrorism, and its presentation
of data. The next section challenges the book’s argument that suicide attacks have a high
success rate of 54 percent. The alternative analysis offered here arrives at a significantly
lower success rate of 24 percent. The last section argues that Pape exaggerates the link
between occupation and suicide terrorism, especially with regard to the case of Al
Qaeda. In this context, a distinction between traditional (localized) and contemporary
(globalized) patterns of suicide attacks is introduced. It is argued that the occupation
thesis may help explain the traditional (localized) pattern of suicide attacks, but falls
short of illuminating the causes of the contemporary “globalization of martyrdom.”
The growing interest in suicide terrorism in recent years, and particularly since
11 September 2001, has generated a steep rise in the number of books that address a
topic that is inherently fascinating—a mode of operations that requires the death of its
perpetrator to ensure its success. Since 2001, when the first book on suicide terrorism
was published,1 journalists, terrorism experts, and political scientists have examined this
phenomenon from a variety of angles, including in-depth interviews with suicide bombers,
comparative studies, historical accounts, or a combination of those.2 Notable in nearly all
of these studies was the absence of statistical data—a problem that terrorism analysts
have long decried. In Dying to Win: The Strategic Logic of Suicide Terrorism,3 Robert A.
Pape, Professor of Political Science at the University of Chicago, demonstrates that it is
indeed possible to assemble statistical data about terrorist incidents, and that empiricism
can help shed new light on the study of the etiology of terrorism, and specifically suicide
terrorism.
Pape, who also directs the Chicago Project on Suicide Terrorism, offers a series of
provocative arguments that address the origins of suicide terrorism. Some of his findings
Received 27 December 2005; accepted 27 December 2005.
The author is indebted to Bruce Hoffman, Sean Lynn-Jones and Monica Duffy Toft for their
helpful comments.
Address correspondence to Assaf Moghadam, Olin Institute for Strategic Studies, Harvard
University, 1737 Cambridge Street, Cambridge, Massachusetts 02138, USA. E-mail: assafm@
hotmail.com
707
708
A. Moghadam
deviate from the more conventional explanations scholars of this subject have enounced to
date. Pape’s key conclusion is that religion is rarely the root cause of suicide terrorism, and
that the main goal of suicide terrorist attacks is “to compel modern democracies to withdraw
military forces from territory that the terrorists consider to be their homeland” (4). Pape’s
main findings are supported by three general patterns he finds in an extensive database he
compiled and that he describes as “the first complete universe of suicide terrorist attacks
worldwide” (3). The first pattern is that suicide terrorist attacks occur as part of organized
campaigns; second, democratic states are uniquely vulnerable to suicide terrorists; and
third, these campaigns are directed toward a strategic objective, namely to establish or
maintain political self-determination. Even Al Qaeda, Pape argues, fits this pattern. He
adds that the increase in suicide terrorism in recent years is due to the fact that terrorists
have learned that the tactic works.
In collecting the largest dataset of suicide attacks available at the time of the publication
of Dying to Win4—a total of 315 attacks—and in gathering sociological data on the 462
suicide attackers, Pape has done an immense service to students of suicide terrorism. His
data also provides some important insights into the context of this modus operandi. Pape
shows, for example, that countries that have been plagued by suicide terrorism between 1980
and 2001 are far from being the poorest in the world. Many are “middle income societies”
with life expectancies not significantly lower than those in the United States (18). “Poverty
remains a poor indicator of suicide terrorism,” (19) he notes, thus confirming the findings of
several other studies conducted on the correlation between poverty and terrorism in general.5
Some of the strengths of the book lie in the methodical way in which Pape goes about
conducting social scientific inquiry. Pape lays out his hypotheses clearly, then identifies
observable implications to test these hypotheses, and finally examines the data against
his initial assumptions. Dying to Win, in fact, is a textbook case of how social science
methods can be applied to the study of terrorism while remaining accessible to the general
readership.
Dying to Win is also strong in many of its case studies, most of which are well written
and well researched. The description of suicide terrorist acts conducted by Hizballah
(129–138), for example, is an enlightening overview in which Pape shows that the
majority of suicide attackers in South Lebanon were not Islamic fundamentalists, but
rather communists and Christians. Here, as in most other parts of his book, he bases his
findings on his dataset, which is included as an appendix to the book. Another excellent
case study conducted by Pape is that of the Sikh independence movement, a genuinely
understudied topic within the study of suicide terrorism (154–162).
Yet, despite the availability of this large dataset, which is sure to benefit many current
and future students of suicide terrorism, and despite offering a number of illuminating
arguments, Dying to Win suffers from a number of shortcomings. The aim of this review
article is to highlight some of the main weaknesses of Dying to Win.
The review is structured as follows. The first part addresses several problems related to
how some key concepts are defined in Dying to Win, as well as related problems concerning
Pape’s data-collection. It argues that a more rigorous definition of key terms and concepts
could have resulted in a radically different interpretation of Pape’s key findings. This section
will also examine the fairness of Pape’s description of the overall state of the study of suicide
terrorism.
The second part of the article offers a critique of Pape’s contention that there has been
an increase in the use of suicide attacks mainly because terrorists have learned that the
tactic works. It argues that Pape has exaggerated the success of suicide terrorism, and that
A Critique of Dying to Win
709
a closer look at past and ongoing campaigns significantly reduces the 50-percent success
rate Pape ascribes to this tactic.
The third part of the review article examines Pape’s insistence that suicide terrorism
is mainly a response to foreign occupation. Although a correlation between occupation
and suicide attacks indeed exists, Pape exaggerates the link between the two. The case
of Al Qaeda in particular does not fit Pape’s occupation theory as well as he maintains,
partly because he fails to appreciate the importance of religion as a driving force of Al
Qaeda. Pape’s insistence that occupation is the main cause of suicide terrorism—if it was
ever the case—is increasingly losing its relevance. Pape fails to acknowledge that the
pattern of suicide attacks has changed from a predominantly localized to a predominantly
globalized one, with important implications as to the causes of this tactic. Pape’s book
provides some important insights into the traditional form of suicide attacks, but it
fails to provide a convincing explanation of what may be termed the “globalization of
martyrdom.”6
Suicide Terrorism, Suicide Attacks, and the “Conventional Wisdom”
on Suicide Terrorism Research
How Conventional is the Conventional Wisdom?
At the outset of Dying to Win, Pape cites what he calls the “conventional wisdom” on suicide
terrorism, a phenomenon he believes—with little basis in the opinion of this reviewer—to
be “significantly different” (9) from non-suicidal terrorism.7 Pape writes that “the small
number of studies that explicitly address suicide terrorism tend to focus on the irrationality
of the act of suicide from the perspective of the individual attacker” (16). This argument
was baseless when it first appeared in article format in 2003 in the prestigious American
Political Science Review (APSR) that formed the foundation for Dying to Win,8 and it was
flawed in 2005, the year of the book’s publication. To support his claim that the few studies
on suicide terrorism focus on the irrationality of the act, Pape cites three works of which
none, upon closer inspection, solidifies his case.9 At the same time, he fails to refer to a
large and growing body of work on suicide terrorism that explicitly rejects the notion that
suicide terrorists are irrational.10 The first work cited by Pape in support of his argument is
Martin Kramer’s “The Moral Logic of Hizbullah.”11 As Kramer himself, however, pointed
out in a response to Pape’s APSR article, Kramer argues precisely the opposite of what Pape
ascribes to him, namely that suicide attacks are indeed a rational choice of the organization.
These attacks, Kramer writes in “The Moral Logic of Hizballah,” “enjoyed such stunning
success that leading Shi’te clerics were prepared to bend their interpretation of Islamic law
to sanction it.”12 The second article Pape cites to support his claim that the “conventional
wisdom” in the study of suicide terrorism focuses on the irrationality of the act is “The
Readiness to Kill and Die” by Ariel Merari, a leading expert on terrorist psychology and
a longtime observer of suicide terrorism.13 In that article, Merari indeed labeled suicide
terrorism as an irrational act. However, since that article’s first publication in 1990, Merari
has fundamentally changed that initial conclusion. In numerous speeches and papers Merari
has delivered since, he stressed that suicide terrorism is not an individual, but instead an
organizational, phenomenon. Pape does not cite these latter studies.14 The third and last
article cited by Pape with regard to the “conventional wisdom” is an article by Jerrold Post
entitled “Terrorist Psycho-Logic.”15 Post, a leading political psychologist, indeed argues in
that article that terrorists are driven to violence as a product of psychological forces. The
710
A. Moghadam
subject of Post’s article, however, are terrorists in general, not suicide terrorists, which he
does not mention even once. Given that Pape views terrorism as “significantly different”
from suicide terrorism, it is puzzling why he would cite an article on non-suicidal terrorism
to support his view of the “conventional wisdom” of suicide terrorism.
Pape’s failure to provide evidence that suicide terrorism researchers emphasize the
irrationality element of this tactic is not surprising. The reason is that the vast majority of
researchers have gone to great lengths to stress that suicide terrorists are not irrational, as
an exhaustive summary of research conducted on suicide terrorism by a U.S. Department
of Homeland Security researcher has shown.16 Of the 47 articles cited in that compilation,
not a single one refers to suicide terrorism as an outcome of irrational behavior.
Pape is thus clearly mistaken that the conventional wisdom on suicide terrorism
research focuses on the irrationality of the act. Needless to say, this does not mean that the
study of suicide terrorism has reached theoretical and analytical saturation. Neither does it
mean that Pape’s work is not a serious contribution to the study of suicide terrorism, which
it most certainly is. However, in attempting to set his mark in the field, and by placing his
own research in the larger context of the study of suicide terrorism, it is unfortunate that
Pape has not provided a fairer, more accurate, and more respectful account of the current
state of research on suicide terrorism.
What Constitutes Suicide Terrorism?
One of the terms more commonly used to describe the modus operandi of suicide attacks
is “suicide terrorism,” a label adopted by Pape throughout his book.17 Pape’s choice of the
term suicide terrorism, rather than suicide attacks, suicide missions, or suicide operations,
has important implications. In fact, several of his conclusions depend on his particular
definition of suicide “terrorism.” Pape does not abide by the most common and most
widely accepted definition of suicide terrorism. Had Pape done so, then the important role
he ascribes to the Liberation Tigers of Tamil Eelam (LTTE)—and thus to secular suicide
terrorist organizations in general—would be called into question. Given that Pape describes
the LTTE as a secular organization, this, in turn, would have undermined his argument that
the role of religion in suicide terrorism is marginal.
The reason why Pape’s use of the term suicide terrorism is problematic is that terrorism
is frequently understood as attacks against noncombatants,18 whereas many suicide attacks
listed in Pape’s database are attacks against military targets. He labels the 315 suicide attacks
he identified between 1980 and 2003 as “suicide terrorist attacks” even though many of
these attacks have been targeted against uniformed soldiers that were on duty at the time
the attacks took place.19 Bombings, shootings, kidnappings, and other attacks employed
against armed and uniformed personnel, however, are generally labeled acts of guerrilla
warfare, insurgency, or low-intensity warfare if they are directed against uniformed men
and women on duty.
One of the arguments Pape uses to support his thesis that there is little correlation
between suicide terrorism and religion is that the LTTE—a secular group, according to
Pape20—is the organization that has staged most suicide terrorist attacks. Of the 75 suicide
attacks carried out by the LTTE identified by Pape, however, half were conducted against
military targets. Only 37 attacks listed by Pape are suicide “terrorist” attacks if one applies
the most widely accepted definition of terrorism—a number that heavily reduces the relative
weight of attacks by LTTE versus other organizations, and thus of “secular” organizations
versus “religious” organizations.
A Critique of Dying to Win
711
Table 1
Number of suicide attacks by LTTE and Palestinian groups by target∗
Campaign
4: LTTE vs. Sri Lanka
5: LTTE vs. Sri Lanka
6: Hamas vs. Israel
7: Hamas/PIJ vs. Israel
9: Hamas vs. Israel
10: Hamas vs. Israel
13: LTTE vs. Sri Lanka
17: Hamas / PIJ vs. Israel
Date
Military
Civilian
Jul ’90–Oct ‘94
Apr ’95–Oct ‘00
April 1994
Oct ’94–Aug ‘95
Feb ’96–Mar ‘96
Mar ’97–Sep ‘97
Jul ’01–Nov ‘01
Oct ‘00 (ongoing)
10
26
0
3
0
0
2
15
5
28
2
6
4
3
4
77
∗
Source: Robert Pape, Dying to Win.
Under a strict definition of terrorism as attacks against noncombatant targets, however,
Pape’s argument no longer holds. If the suicide attacks perpetrated by the LTTE are
juxtaposed with those of the Palestinian group Hamas, and these groups’ attacks are sorted
according to military and civilian targets,21 it becomes clear that Hamas (a predominantly
religious organization) is responsible for more suicide “terrorist” attacks than the LTTE, if
“terrorism” is defined as attacks against civilians. Whereas the LTTE is responsible for 37
suicide attacks against civilians in its three campaigns against Sri Lanka identified by Pape
(see Table 1), Hamas has planned and executed 48 suicide attacks against civilians22 (see
Table 2).
It is of course not argued here that suicide attacks against military, or combatant,
targets are irrelevant, nor that the LTTE is an organization that does not merit a great
deal of scholarly attention. On the contrary, and as Pape correctly points out, the LTTE is
among the most sophisticated terrorist and insurgent organizations, and the effectiveness
of its suicide cadres, the “Black Tigers,” is second to none. Rather, the point in making the
differentiation between military and civilian targets of suicide attacks is to highlight the
importance of terminology and definitions applied, particularly when social scientists use
large amounts of data to reach important conclusions. Pape could have easily evaded this
definitional pitfall by labeling the attacks in his database in more neutral terms like “suicide
attacks” or “suicide missions,” rather than using the term “suicide terrorism.”23
What Constitutes a Suicide Attack?
Pape’s understanding of what constitutes a suicide “attack” is another example of how
his particular definition of a key concept has a direct bearing on his interpretation of
Table 2
Summary of suicide attacks by LTTE and Hamas by target∗
Organizations (all campaigns)
Military targets
Civilian targets
38
6
37
48
LTTE
Hamas
∗
Source: Robert Pape, Dying to Win.
712
A. Moghadam
data, and thus on some of his main arguments. The following section will show how
a slightly different—and arguably more accurate—understanding of what constitutes a
suicide “attack” could affect Pape’s data, and could thus have altered the conclusions he
reaches.
Suicide attacks are traditionally defined as attacks whose success is contingent upon
the death of the perpetrator.24 According to Schweitzer, for example, a suicide attack is “a
politically motivated violent attack perpetrated by a self-aware individual (or individuals)
who actively and purposely causes his own death through blowing himself up along with
his chosen target. The perpetrator’s ensured death is a precondition for the success of his
mission.”25 Boaz Ganor adds that “the terrorist is fully aware that if he does not kill himself,
the planned attack will not be implemented.”26
What this classic definition of suicide attacks suggests—and Pape himself adopts this
definition in Dying to Win27—is that a suicide attack should be counted as a single attack if
an individual bomber killed him or herself along with the target. By the same token, suicide
attacks should be counted as multiple attacks if multiple individuals killed themselves along
with their chosen target. Nevertheless, Pape lists as single attacks even those suicide attacks
that were perpetrated by multiple individuals or against several targets, if they took place
nearly simultaneously.28 The 16 May 2003 attacks in Casablanca, for instance, were aimed
at a total of five targets, including two restaurants, a cemetery, a hotel, and a community
center. Rather than to count the Casablanca bombings as five separate attacks, however,
Pape lists them as one. He also counts the four separate attacks conducted on 11 September
2001 as a single attack, as he does with the dual bombings of the U.S. embassies in Kenya
and Tanzania on 7 August 1998—even though the embassy bombings took place in two
separate countries. A better description for multiple, simultaneous attacks occurring at the
same place would have been to label them as an “incident.”29 For example, labeling the
9/11 attacks as one incident consisting of four separate attacks would have been a more
precise labeling of the data.
A closer look at the attacks claimed by Al Qaeda, which often uses multiple suicide
bombings as its signature modus operandi, shows that the 21 attacks claimed by Al Qaeda
according to Pape’s database are in fact 21 incidents consisting of 39 separate attacks, as
Table 3 summarizes.30 Here too, the way in which Pape counts his data could potentially
impact his interpretation of it. If, for example, religious organizations frequently use
multiple suicide bombers as their signature mode of attack—which, in fact, is the case
with Al Qaeda—then the relative weight of Al Qaeda’s attacks, as well as those of religious
groups more generally, could be downplayed by counting multiple attacks as one attack.
This would be particularly grave if terrorist incidents by secular organizations would
rarely consist of multiple suicide attacks. This, however, is not the case because the LTTE
frequently employs multiple suicide bombers for its incidents as well.33
To conclude this section, it is important to note that in order to arrive at the most
accurate count, and especially for the purposes of cross-data comparison among terrorism
researchers, it is important for scholars to be as meticulous in their counts as possible.
Pape has certainly done an impressive job in collecting a wealth of data on suicide
terrorism in his database, but by grouping several independent attacks (such as the four
9/11 attacks) as single attacks, he does not provide the most accurate picture of the
numbers of suicide attacks, and the relative weight between attacks by religious and secular
organizations. Future terrorism scholars who will collect data on suicide terrorism—datasets
that will likely integrate Pape’s data—would be wise to compartmentalize the data to the
largest extent possible in order to provide the most accurate snapshot of suicide attack
incidents.
A Critique of Dying to Win
713
Table 3
Attacks by Al Qaeda against the United States
Date
11/13/1995
6/25/1996
8/7/1998
10/12/2000
9/9/2001
9/11/2001
4/11/2001
5/8/2002
6/16/2002
10/6/2002
10/12/2002
11/28/2002
5/12/2003
5/16/2003
6/7/2003
8/5/2003
11/8/2003
11/15/2003
11/20/2003
12/25/2003
12/28/2003
Description of attack
Incident31
U.S. Military base, Riyadh
U.S. Military base, Dhahran
U.S. Embassies Kenya/Tanzania
U.S.S. Cole, Yemen
Ahmed Shah Massoud, Afghanistan
9/11 attacks, United States
Synagogue, Djerba, Tunisia
Car bomb, Sheraton, Karachi
Car bomb, US consulate, Karachi
French oil tanker, Yemen
Bali nightclub bombing
Paradise hotel, Mombasa
Car bombs, Riyadh
Casablanca bombings
German military bus, Kabul
Car bomb, Jakarta
Car bomb, Riyadh
Synagogue bombings, Istanbul
Istanbul bombings
PM Musharraf, Rawalpindi
Car bomb, Kabul Airport
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Total number of attacks
21
Actual attacks∗
1
1
2
132
2
4
1
1
1
1
2
3
4
5
1
1
1
2
2
2
1
39
∗
Source: MIPT Terrorism Knowledge Base.
The Strategic Illogic of Suicide Terrorism
Of all the arguments put forward in Dying to Win, Pape’s insistence that “the main reason
that suicide terrorism is growing is that terrorists have learned that it works” (61) is the
most problematic. This part calls into question Pape’s notion that suicide terrorism works.
By including “ongoing campaigns” left out in Pape’s count, and by challenging Pape’s
assessment of three particular cases of suicide terrorist campaigns, I will argue that Pape’s
claim of a 54-percent success rate of suicide terrorism is exaggerated. By providing an
extensive contextual description and alternative explanation, this part of the critique will
argue that suicide terrorism has been “successful” in only 24-percent of the cases. Contrary
to Pape’s contention, suicide attacks more often have resulted in failure than in success.
To begin with, of the 18 “suicide terrorist campaigns” since 1981 identified by Pape, he
analyzes the success or failure of only those 13 campaigns that he describes as “completed”
by December 2003. Pape omits from his examination of success or failure five suicide
terrorist campaigns that he describes as “ongoing” at the time of the book’s writing.34 Yet,
it seems particularly arbitrary on Pape’s part to exclude ongoing campaigns, especially given
that their duration is, on average, three times longer than that of completed campaigns.35 In
measuring the success (or failure) of suicide terrorist campaigns, it would appear to be fair
to include at least those ongoing campaigns in the count that, on average, have lasted longer
714
A. Moghadam
than the average duration of completed campaigns, which is how this section proceeds.36
Based on this count, conclusions on the success or failure can be reached about 17 of the
18 campaigns identified by Pape, rather than only about the 13 completed campaigns.
Of these 17 campaigns, Pape maintains that “seven correlate with significant policy
changes by the target state toward the terrorists’ major political goal” (64). As noted earlier,
the outcome of 5 campaigns (corresponding to 29 percent of the 17 campaigns) was still
to be determined,37 and their success or failure thus unclear. There was, according to Pape,
“no change” at all in 6 campaigns (35 percent).38
Pape claims that suicide terrorism has led to a successful outcome in seven cases,39 but
admits that of those seven, “three concessions, or arguable concessions, are less clear-cut”
(65). In fact, not only are those “arguable concessions” less clear-cut, but calling them a
“success” requires a considerable stretch of the imagination.
In the first of the three cases of “arguable concessions”—the release of Hamas’ spiritual
leader, Sheikh Ahmed Yassin40—Pape makes the rather unconvincing case that it was
Hamas’s suicide terrorist campaign against Israel from March to September 1997 that led
to the release of the wheelchair-bound Yassin. In fact, the Hamas leader’s release had little,
if anything, to do with suicide terrorism, a possibility that Pape himself acknowledges
elsewhere (66). Instead, it was the direct result of the diplomatic embarrassment suffered
by the Netanyahu government in the aftermath of Israel’s failed attempt on 25 September
1997, to assassinate Khaled Meshal, an exiled leader of Hamas in Jordan. Meshal was
attacked by Mossad agents, who failed to kill their target with a lethal injection, and were
subsequently captured by Jordanian security officials. Following the botched assassination
attempt, an enraged King Hussein of Jordan said that the captured foreign agents would
be put on trial unless Israel released Sheikh Yassin along with dozens of Hamas prisoners.
Israeli prime minister Benjamin Netanyahu initially resisted the Jordanian demand, but the
United States pressed for a quick end to the crisis, persuading the Israeli prime minister to
release Yassin along with dozens of jailed Hamas militants.41
Pape’s suggestion that Israel released Sheikh Yassin due to the “coercive pressure”
(63) of Palestinan suicide terrorism is entirely unrealistic for at least two reasons. First, as
the New York Times reported on 23 September 1997—two days before the assassination
attempt—Jordan relayed an offer to Israel indicating the Arab monarchy’s willingness to
mediate between Israel and Hamas to achieve a moratorium on suicide attacks.42 Had Israel
really wanted to put an end to suicide terrorism at all costs—including, as Pape’s argument
suggests, by appeasing Hamas via the release of its leader—then Israel would have been
expected to at least consider the Jordanian king’s offer to mediate such an outcome. Israel,
however, rejected the Jordanian offer.
Second, and more importantly, it has been Israel’s long-standing policy not to acquiesce
to terrorist demands.43 On the contrary, Israel has demonstrated on countless occasions that
it is willing to respond forcefully to the use of terror, and one would certainly expect this
policy to be upheld by an Israeli prime minister who himself authored a book recommending
that Western countries adopt strict non-concessionary strategies in their fight against terror.44
Furthermore, there is an ironic twist in Pape’s claim because the very incident that led to
the release of Yassin, namely the attempted elimination of Khaled Meshal, was the very
result of Israel’s relentless policy of pursuing terrorists wherever they are—precisely the
opposite of Pape’s claim that Israel acquiesced to Hamas’s “coercive pressure.” The reason
Israel attempted to eliminate Meshal was his apparent authorization of a terrorist attack in
Jerusalem on 30 July 1997.45
Two other suicide terrorist campaigns that Pape labels successful are not only
“arguable,” as he admits, but rather far-fetched (see Campaigns 6 and 7 in Table 4). Pape
A Critique of Dying to Win
715
Table 4
Success and failure of the 17 suicide terrorism campaig...
Purchase answer to see full
attachment