Content uploaded by Martin Binder
Author content
All content in this area was uploaded by Martin Binder on Sep 19, 2020
Content may be subject to copyright.
1
Paths to intervention: What explains the UN’s selective response
to humanitarian crises?
Martin Binder, Journal of Peace Research
2015, Vol. 52(6) 712–726
Abstract
Over the past two decades, the United Nations Security Council has responded more strongly
to some humanitarian crises than to others. This variation in Security Council action raises the
important question what factors motivate United Nations intervention. This article offers a
configurational explanation of selective Security Council intervention that integrates
explanatory variables from different theories of third-party intervention. These variables are
tested through a comparison of 31 humanitarian crises (1991–2004) using fuzzy-set Qualitative
Comparative analysis. The analysis shows that a large extent of human suffering and substantial
previous involvement in a crisis by international institutions are the key explanatory conditions
for coercive Security Council action, but only when combined with negative spillover effects
to neighboring countries (path 1) or with low capabilities of the target state (path 2). These
results are highly consistent and explain 85% of Security Council interventions after the end of
the Cold War. The findings suggest that the Council’s response to humanitarian crises is not
random, but follows specific patterns that are indicated by a limited number of causal
combinations or causal paths.
Introduction
On 17 March 2011, the United Nations Security Council (SC) adopted Resolution 1973, which
authorized the use of ‘all necessary measures’ – except an occupation force – to protect civilians
2
in Libya. Two days later, an international military operation was launched to enforce Resolution
1973. The intervention in Libya’s civil war marks the latest episode in the post-Cold War
practice of United Nations (UN) humanitarian intervention. Over the past two decades, the
Council imposed sanctions, deployed peacekeeping operations (PKOs), or even carried out (or
authorized) military interventions to respond to various humanitarian crises around the globe.
But the SC’s response remained highly selective. It has responded more strongly to some
humanitarian crises than to others. What accounts for this variation?
This question raises the important issue of what factors motivate UN intervention. The
main controversy among scholars has centered on whether these interventions are to be best
explained by humanitarian considerations and a shift of international norms, or whether geo-
strategic and economic interests provide a more plausible account. Given the complexity of
intervention decisions, however, some observers have argued that these decisions are not
determined by any single factor, but that they rather involve a mix of material and ideational
motives. More than thirty years ago, Walzer (2000 [1977]: 101) found no clear case of a
humanitarian intervention, suggesting instead that there were ‘only mixed cases where the
humanitarian motive is one among several’. More recently, Lyon & Dolan (2007: 50) have
suggested that ‘there is a mixture of motives for humanitarian involvement as the certainties of
altruistic humanitarian intervention are often blurred with self-interested power pursuits’. Also,
Mason & Wheeler (1996: 95) argue that neither material factors nor humanitarian
considerations are individually strong enough to motivate humanitarian intervention, but that
they may be ‘jointly sufficient to do so’.
Given the lack of studies that examine systematically the combinations of factors or
causal paths that lead to or block intervention, the aim of this paper is to test the ‘mixed motives’
argument. More specifically, I examine whether and which combinations of conditions are
necessary and/or sufficient for intervention or non-intervention by the SC. This is done through
3
a comparison of 31 instances of major humanitarian crises (1991–2004) using fuzzy-set
Qualitative Comparative Analysis (fs/QCA).
I demonstrate that two causal paths lead to humanitarian intervention. A large extent of
human suffering and substantial previous involvement in a crisis by international institutions
are the key explanatory conditions for UN intervention, but only when combined with either
negative spillover effects to neighboring regions (path 1) or with low capabilities of the target
state (path 2). None of these conditions is individually necessary or sufficient, but they are
jointly sufficient to explain why the SC responds more strongly to some humanitarian crises
than to others.
To develop this configurational explanation, the first part of the article presents the
existing literature on third-party intervention. The second part conceptualizes SC intervention
in humanitarian crises. Part three identifies explanatory conditions for SC intervention. The
fourth part tests these conditions through a comparison of the major humanitarian crises after
the Cold War. The fifth part presents the results that are discussed and interpreted in the sixth
part. The last part sums up and briefly considers the normative and practical implications of the
findings.
Prior research
Relatively little academic attention has been devoted to the factors that motivate international
institutions, including the UN, to intervene in conflicts and humanitarian crises. Earlier studies
that researched into the subject frequently focus on situations where intervention has occurred,
while crises that remain unaddressed go unstudied. Jakobsen (1996) examines ideational and
material determinants of UN peace enforcement operations, but his study is limited to five
positive cases after the Cold War.
1
Similarly, Neack’s (1995) study compares 18 Cold War
PKOs to test whether state participation in these operations was driven by their commitment to
1
Jakobsen does focus on combinations of factors.
4
the international community, or by self-interest. More recent studies have begun to address this
bias and analyze which instances of armed conflict the UN has selected for intervention and
why. Gilligan & Stedman (2003) find that the UN is more likely to deploy peacekeeping
missions in conflicts with a high number of deaths, and less likely to intervene in states with
large government armies. According to Fortna (2004), peacekeepers get sent to difficult cases
for peacekeeping, and shy away from militarily strong states. Mullenbach (2005) finds that UN
peacekeeping is less likely if the target state is a major power or has strong allies, and more
likely if the UN has previously been involved in a civil war. Whereas these studies remain
confined to PKOs, a recent study by Beardsley and Schmidt (2012) looks at the determinants
of different levels of UN involvement in international crises where involvement is
conceptualized as a seven-point level variable. They find that measures of the severity and
escalatory potential of a conflict are better explanations of the extent of UN intervention than
variables that measure the parochial interests of the five permanent veto-holding members (P5).
However, their study is limited to UN involvement in inter-state crises and does not examine
internal conflicts with which humanitarian crises are usually associated. In sum, while this
research improves our understanding of UN intervention in important ways, none of these
studies analyzes causal combinations to explain why the UN responds to humanitarian crises
after the Cold War with varying strength.
Humanitarian Intervention and the United Nations
This article adopts a broad conceptualization of humanitarian intervention which takes into
account that humanitarian action is not a simple matter of either intervening militarily or ‘doing
nothing’ (e.g. Crocker, 2001: 229). Rather, it is conceived as a response that varies considerably
in terms of coerciveness and intrusiveness. Building on Ramsbotham & Woodhouse (1996:
106–135), humanitarian intervention may range from forcible military intervention to coercive
non-military (e.g. economic sanctions), to non-coercive military (e.g. peacekeeping), to non-
5
coercive non-military (e.g. humanitarian assistance), to complete non-intervention as a further
and final option.
2
This conceptualization is directly applicable to SC action. Having the ‘primary
responsibility for the maintenance of international peace and security’ (Article 24 of the UN
charter) and the authority to determine whether a situation is a ‘threat to international peace and
security’ (Article 39), under Chapter VII, the SC may opt for coercive action, including the
collective use of force (Article 42) and the imposition of sanctions (Article 41). Further,
although not formally included in the UN charter, the SC has authorized the deployment of
‘robust’ PKOs, whose mandate under Chapter VII includes the (limited) use of force. At the
same time, the SC may decide to take non-coercive measures, including the deployment of
traditional consent-based PKOs, observer missions, and the provision (or the call for provision)
of humanitarian assistance as a means to restore peace and security. Finally, the SC may not
take any action at all.
It has been widely noted that the decision whether the SC responds to crises or conflicts
depends heavily on the interests of its most powerful members, most notably the veto-holding
P5. However, as Gilligan & Stedman (2003: 39–40) have pointed out, precisely because P5
acquiescence is required for any SC action, the argument that P5 interests are decisive for UN
action ‘often approaches tautology’. The more relevant question is what factors motivate these
interests and ‘which cases get great power support’. Under what conditions do the members of
the SC – including the P5 – agree to take or authorize strong action, and when can we expect
that no such measures will be taken?
2
Ramsbotham & Woodhouse (1996: 113) define humanitarian intervention as ‘cross-border action by the
international community in response to human suffering made up of “forcible humanitarian intervention” …
and “non-forcible humanitarian intervention”’.
6
Explanatory conditions for UN intervention
Existing research on third-party intervention offers a number of plausible conditions to account
for UN intervention. These explanations are informed by constructivist, realist-rationalist, and
institutionalist theories in international relations (IR). Whereas some of these factors are clearly
associated with one specific theoretical approach, others are stressed by more than one theory.
However, rather than to engage in competitive theory testing, the aim of this article is to draw
from theories of third-party intervention, to determine whether any specific factors, and, if so,
which combinations thereof, contribute to a better explanation of selective UN intervention.
The question of to what extent some specific theoretical account, if any, can make sense of the
findings will be discussed at the end of the article.
Normative considerations
Constructivist informed studies on third-party intervention point to a profound change in the
normative environment and to humanitarian considerations as drivers of outside intervention.
From this perspective, the increase in third-party intervention was largely enabled by emerging
norms of humanitarian intervention during the 1990s, which now frequently supersede the
traditional principles of sovereignty in instances of ‘supreme humanitarian emergencies’
(Wheeler, 2000: 50; Finnemore, 2003). These norms were further strengthened since the UN
World Summit in 2005 – when UN member states agreed to the ‘responsibility to protect’
principle, which stipulates that coercive intervention, including the use of military force, may
be warranted in cases of severe human rights violations (e.g. Bellamy, 2010). Against this
backdrop of normative change, a number of studies show how humanitarian considerations
have led to the imposition of economic sanctions and to military interventions in situations
where no obvious security or geostrategic interests were at stake (Klotz, 1995; Hasenclever,
2001). Regarding the SC, even skeptics agree that its members have felt a ‘humanitarian
impulse’ (Weiss, 2004: 39) when faced with cases of large-scale human suffering. While it is
7
difficult for norm-based accounts to explain why humanitarian intervention remains selective
(Boulden, 2006), constructivist research stresses that intervention decisions depend on the
extent of a crisis. Interventions occur when the level of human suffering is high, and the
resulting ‘morally motivated pressure to act’ on behalf of the victims of a crisis is strong
(Hasenclever, 2001: 211). Along these same lines, it has been argued that massive human
suffering is a necessary (but not sufficient) condition for intervention (e.g. Jakobsen, 1996:
212).
Media attention
Pressure to act in the face of human suffering might result not only from the ‘real extent’ of a
crisis but also by its perceived dimension. Analogous to the ‘CNN effect’, according to which
media coverage compels, or at least influences decision makers to act, it has been argued that
the decision to intervene depends on the level of international media attention for humanitarian
crises.
3
Whereas media attention is a frequent explanation for specific foreign policy decisions,
other observers claim that media coverage also operates at the level of the SC. Malone (2004:
12), for example, refers to CNN as the SC’s ‘sixteenth member’, and argues that ‘the media
plays an important role in the life of the Council, often producing international pressure for
action to address man-made disasters’. There are at least two paths through which media may
generate pressure to act. First, from a constructivist perspective, it is argued that, via mass media
coverage, citizens and decision makers can identify with, or develop empathy for, the victims
of conflicts and humanitarian emergencies (e.g. Barnett, 1998: 96). Second, seen from a
rationalist perspective, a broader public exerts pressure on decision makers who, in turn, fearing
reputational damage from inaction, act strategically. But other observers question the effects of
media coverage. They argue that media coverage follows foreign policy decisions and not vice
3
See Gilboa (2005) and Robinson (2011) for literature reviews.
8
versa; furthermore, that there has been non-intervention despite strong media attention; or that
media coverage might, through graphic images of dead combatants, prevent rather than promote
intervention (Jakobsen, 2000; Robinson, 2001). These arguments reinforce the notion that
media coverage is context-dependent; it does not seem to be ‘sufficient on its own to cause
intervention and is unlikely to even be a necessary factor in causing policy makers to act’
(Robinson, 2001: 942).
Spillover effects and countervailing power
Realist–rationalist accounts of third-party intervention focus on material forces and stress a
second set of potential explanatory factors. Conceiving of states as rational actors who strive
for maximizing security (Waltz, 1979) or power (Mearsheimer, 2003), intervention becomes
‘an instrument of foreign policy used to promote the interests of individual nations’ (Feste,
1992: 1). Specifically, states intervene to change or maintain the international distribution of
power, to broaden or defend their respective geopolitical spheres of influence, or to defend or
extend their respective ideologies (Morgenthau, 1967). Several studies underline the
importance of realist variables to explain UN intervention (e.g. Neack, 1995; de Jonge Oudraat,
1996). Two factors in particular may explain the degree of UN response: The first is the level
of spillover effects emanating from a crisis, and thus the extent to which it poses a threat to
international peace and security. Spillover effects may contribute to conflict diffusion through
various mechanisms such as refugee flows (Salehyan & Gleditsch, 2006), transnationally
operating rebels (Saleyhan, 2009), international terrorism, or economic downturn. Spillover
effects generate costs in that they produce negative externalities for neighboring countries,
including SC member states (de Jonge Oudraat, 1996: 518–519; Dowty & Loescher, 1996).
The second factor is the extent to which possible target states of intervention are able to generate
countervailing power against outside intervention. The strength of countervailing power, which
affects the costs and risks of an intervention and its chances of success, may depend on several
9
factors. First, countries with strong militaries are better able to resist UN responses (Doyle &
Sambanis, 2006: 86). Second, beyond military strength, target countries can also resist UN
action if they are allied with a P5 or another powerful state (or if they are a P5 themselves) that
can use its political influence in the Council to prevent UN involvement on the territory of an
ally (Mullenbach, 2005: 537). The same can be expected if a potential target state is located
within the sphere of influence of a P5 or another major power (Luard, 1984: 165).
Sunk costs and path dependence
The final explanatory variable is emphasized by studies focusing on the role of ‘sunk costs’ and
resulting path-dependence for third-party intervention. Path dependence – that is, self-
reinforcing (or ‘positive’) feedback processes – often results from material and immaterial
investments in an existing institution or policy that actors have made and are reluctant to
abandon (Stinchcombe, 1968: 120; Pierson, 2004). This explains why institutions and policies
may become path-dependent and persistent. Such previous investments or ‘sunk costs’ may not
only include money or infrastructure, but also learning, coordination, or reputation.
4
Similar
processes have been observed in third-party interventions, which are often costly undertakings
and generate substantial learning and coordination effects. Von Hippel (1996) and Boettcher &
Cobb (2009), for instance, have pointed to sunk cost effects to explain the persistence of
Morocco’s intervention against the Polisario Front in the Western Sahara conflict and the US
war in Iraq, respectively. Burg (2004) has shown how incrementally the concerns for extensive
investments in diplomatic prestige and material resources in the Bosnian crisis created an
interest for UN intervention, where no such interest existed before. Thus, depending on the
extent of previous involvement in a crisis by regional or international institutions, the wish to
secure these investments through continued or increased commitment is considered a powerful
4
While the ‘sunk cost trap’ is often used to explain seemingly irrational behavior, sunk costs do not
necessarily lead to fallacious policies (e.g. McAfee, Mialon & Mialon, 2010).
10
motive for UN action (Mullenbach, 2005: 538). Yet, while sunk costs might explain why certain
established policies and institutional arrangements may persist, they do not account for why
these particular policies were adopted in the first place.
In sum, I have identified five testable explanatory conditions for the variation in UN
intervention. These are: (1) the extent of a humanitarian crisis, (2) the level of international
media attention, (3) the strength of spillover effects, (4) the strength of countervailing power,
and (5) the level of previous institutional involvement in a crisis.
I focus on these and not on other variables because they are the most prominent reasons
given for the UN’s uneven responses to humanitarian crises. Moreover, the explanatory power
of the chosen variables is supported by previous research on third-party intervention (I will
come back to this below). In the same vein, several other potential alternative explanatory
factors were excluded from this analysis because they had been repeatedly rejected by other
previous research. These include, for instance, the type of conflict (identity, religious,
secessionist, etc.), the availability of primary commodities (e.g. oil) in the target state, or the
existence of a peace treaty between the warring parties (Gilligan & Stedman, 2003; Fortna,
2004; Mullenbach, 2005). Finally, all of those conditions that are expected to influence SC
decisions on intervention are international factors operating at the same level of analysis.
Empirical analysis
Fs/QCA is now applied to determine whether one of these conditions, or some combination
thereof, account for UN intervention. The method is particularly appropriate for the aim of this
paper; that is, the analysis of causal combinations, given medium-N situations (5 to 50 cases).
Fs/QCA, which was pioneered by Ragin (2000), is now being applied increasingly in social
science research, including IR (e.g. van der Maat, 2011; Pinfari, 2011). It is based on Boolean
algebra or set theory, allowing us to test whether conditions or combinations of conditions are
necessary or sufficient for the presence of a specific outcome. It centers on multiple
11
conjunctural causality, which means that combinations of causally relevant conditions, rather
than single causes, are the ones that produce a given outcome, and that that same outcome might
be brought about by different combinations of conditions or ‘causal paths’. In fuzzy-set logic,
conditions and outcomes are not necessarily present or absent (1 or 0), but could have partial
(i.e. ‘fuzzy’) membership in a particular set (e.g. 0, .2, .4, .6, .8, 1). This allows us to make more
fine-grained distinctions between cases.
Detailed explanations and technical applications of fs/QCA are provided elsewhere
(Rihoux & Ragin, 2009; Schneider & Wagemann, 2012). But, in brief, all fuzzy-set analyses
proceed in three steps: First, fuzzy-set membership scores must be assigned to each case under
investigation; that is, the extent to which a case is a member of the sets formed by the
explanatory or outcome conditions in which the researcher is interested must be determined (set
calibration). The second stage examines whether one or more of the conditions under
consideration are necessary for the outcome of interest. In set-theoretic terms, a condition is
necessary if its membership scores are, for all cases, greater than, or equal to, the membership
scores of the outcome. The third stage determines whether one or more conditions are sufficient
for the outcome of interest. A condition is sufficient if its membership scores are consistently
lower than, or equal to, the outcome. To check for this, the fs/QCA software
5
transforms the
fuzzy-set membership scores into a (dichotomously organized) truth table, using the direct link
between the rows of the truth table and the corners of a multidimensional vector space defined
by the fuzzy-set conditions.
6
Then, via logical deduction, the truth table is minimized; that is,
the software eliminates all of those factors that are irrelevant for the outcome.
Once we have obtained the minimized solution formula(e) – namely, the causal
condition(s) that are consistently related to the outcome – we need to consider the quality of the
5
University of Arizona < http://www.u.arizona.edu/~cragin/fs/QCA/software.shtml >, accessed 10 September
2013.
6
For details, see Rihoux & Ragin (2009: 103).
12
results. Two parameters describe the extent to which the results ‘fit’ the data (Ragin, 2008: 44–
68). Consistency indicates the extent to which the solution term fulfills the requirement of
necessity and sufficiency.
7
The coverage of a solution term indicates the degree to which the
solution formula(e) or causal path(s) explain the outcome of interest.
8
Fuzzy-set analyses are
easily replicable. The required information is given in the sections that follow. Additional
information is provided in the online Appendix.
Case selection
This analysis focuses on the SC’s response to the most severe instances of ongoing violent
humanitarian crises after the Cold War (1991–2004). To identify these crises, I rely on
Väyrynen’s operational definition of human-made humanitarian crises that encompasses a high
degree of (1) collective violence, (2) displacement, (3) hunger, and (4) disease (Väyrynen,
2000). These indicators are measured with the help of various datasets, including the Uppsala
Conflict Data Program (UCDP; for collective violence), the US Committee for Refugees and
Immigrants (USCRI; for refugees and internally displaced persons), the Food and Agriculture
Organization (FAO), and the World Health Organization (WHO; for undernourishment and
infant mortality rates). Since many parts of the world are affected, to some extent, by those
dimensions of human suffering, this analysis is confined to the major instances of humanitarian
crises. To identify these cases, I use the crisis in Kosovo as the ‘minimum threshold’. The
Kosovo crisis is an appropriate benchmark, because it is widely perceived as a relatively limited
crisis that nevertheless led to a strong response when compared to most of the humanitarian
emergencies with which we are familiar (e.g. Haass, 1999: 47; Lyon & Dolan, 2007: 69), as
7
The formula for consistency is (Xi ≤ Yi) = Σ(min(Xi,Yi))/Σ (Xi) for the analysis of sufficiency. For the
analysis of necessity it is (Xi ≥ Yi) = Σ(min(Xi,Yi))/Σ (Yi).
8
The formula for coverage is (Xi ≤ Yi) = Σ(min(Xi,Yi))/Σ (Yi) for the analysis of sufficiency. For the analysis
of necessity it is (Xi ≥ Yi) =Σ(min(Xi,Yi))/Σ (Xi).
13
well as because it often is considered to be a prime example of selective interventionism and
the application of double standards (e.g. Chomsky, 1999).
9
Applying the Kosovo crisis as a
benchmark, all those crises that are included (like Kosovo) fulfill these minimum criteria in
terms of a sustained, high level of collective violence (experiencing at least two consecutive
years of armed conflict according to the UCDP, including at least one year of ‘major armed
conflict’), plus the prevalence of internal displacement, hunger, or disease. I discuss these
criteria and thresholds in more detail elsewhere (Author, XXXX). This procedure, by which 31
major violent humanitarian crises after the Cold War are identified (Table I), makes sure that
these crises are plausible candidates for SC action. More limited and shorter crises, whose
escalation might be avoided by measures such as preventive diplomacy, are hence excluded
from the analysis.
Operationalization and Data
The outcome condition
The outcome variable in this paper is the degree of SC response to ongoing violent humanitarian
crises after the Cold War. Similar to Ramsbotham & Woodhouse (1996: 106–135), the degree
of SC action is measured, first, in terms of coerciveness or intrusiveness – that is, the extent to
which it violates the principle of sovereignty – and, second, by its costs in terms of financial
resources and personnel. With these two indicators in mind, I propose that the strength of UN
responses may range from (1) complete inaction or non-intervention to (2) humanitarian
assistance, to (3) observer missions, to (4) substantial ‘traditional’ PKOs, to (5) economic
sanctions, to (6) ’robust’ PKOs and, finally, to (7) military humanitarian intervention.
9
NATO’s military intervention in Kosovo was conducted without SC authorization. Yet, in Resolution 1199,
the SC imposed sanctions.
14
The explanatory variables
The extent of a humanitarian crisis: The extent of crises and conflicts is operationalized in terms
of the number of casualties and the number of internally displaced persons. Data is taken from
Leitenberg (2006), DeRouen & Heo (2007), and the annual World Refugee Surveys of the
USCRI (Marshall 2008). Additional information is taken from individual case studies. Detailed
sources for these and all other variables are given in the online Appendix to this article. The
number of refugees is excluded from this variable so as to be able to differentiate between crises
that remain contained within a country (internal displacement) from those that spill across
borders (see below).
International media attention: Media attention is operationalized in terms of the average
number of articles per crisis year in core print media of the societies of the liberal permanent
members of the SC. Data is taken from the Factiva database.
10
Since media attention rises
dramatically once the intervention occurs, I only measure media attention preceding such
responses.
Spillover effects: To determine the number of refugees, I rely on data from USCRI (Marshall,
2008). Whether transnationally operating rebels (Cunningham, Gleditsch & Salehyan, 2013)
contributed to regional conflict contagion (interstate conflict) is assessed with the help of
individual case studies, the UCDP Conflict Encyclopedia,
11
and information provided by
Hewitt (2003, ICB data set v.10) and Ghosn, Palmer & Bremer (2004, MID dataset v.4.0). The
10
I have examined the number of articles published in the New York Times (NYT), The Times (London), and Le
Monde. Because Le Monde has only been available digitally through Factiva since January 1995, for crises
beginning prior to that date, the analysis is limited to the NYT and The Times coverage. I have used a
combination of relevant crisis keywords, which must occur in the title, or the first paragraph of an article.
This is to ensure that only those articles are included in the analysis, which report prominently on the
respective crisis.
11
UCDP Conflict Encyclopedia < www.ucdp.uu.se/database >, accessed 8 May 2014.
15
determination of the remaining factors, such as international terrorism or illegal drug
production, is based on case knowledge (all sources are detailed in the online Appendix).
Countervailing power: The military strength of a state is operationalized by the number of
military personnel (i.e. the maximum number of military personnel during the period under
consideration). Data is taken from the Correlates of War (COW) Project’s National Material
Capabilities dataset (v.3.02). To identify whether a target state has strong allies or is located
within a global power’s sphere of influence, I draw on individual case studies and the COW
Formal Alliances data set (v.3.03).
Previous institutional involvement: This is measured by the resources committed to a crisis
(financial costs and reputation). Similar to the dependent variable, the extent of institutional
involvement decreases on a scale from prior military intervention to substantial PKOs (both
‘robust’ and ‘traditional’), economic sanctions, observation missions, humanitarian assistance,
and, finally, to no previous involvement. Data is drawn from UN sources,
12
and the SIPRI
Multilateral Peace Operations Database.
13
Set calibration of the outcome and the conditions
Once the outcome and the explanatory conditions have been operationalized, the resulting raw
data has to be translated into fuzzy-set member scores – this is the so-called set calibration. To
calibrate fuzzy-sets, the researcher may rely on theoretical or case knowledge as well as on
meaningful patterns in the data. Because the calibration of a set strongly affects the results of
the analysis, explicit justification must be given for the assignment of the scores.
For some concepts or data, multi-value fuzzy-set schemes may be used. For the outcome
condition (the degree of UN response), a seven-value fuzzy-set scheme (0, .16, .32, .48, .64, .8,
1) is assigned. ‘Strong’ responses (with membership scores higher than the crossover point .5),
12
See <http://www.un.org/Depts/dpko/dpko/>, accessed 8 May 2014.
13
See <http://conflict.sipri.org/SIPRI_Internet/>, accessed 8 May 2014.
16
and ‘weak’ responses (with membership scores lower than .5) are distinguished by whether they
are coercive (mandated under Chapter VII) or consent-based. Strong action encompasses
military intervention (1), robust peacekeeping (.8), and economic sanctions (.64). Weak
measures range from traditional PKOs (.48) to observer missions (.32) and humanitarian
assistance (.16). Finally, crises where no action was taken receive a score of 0.
14
When there is
more than one reaction, the strongest response will be taken into account.
For many explanatory conditions, such as the number of victims or refugees, interval-
scale data is available. This allows for a more formalized method (the ‘direct method’; Ragin,
2008: 89–94). With the help of the ‘qualitative anchors’ for full membership (1), full non-
membership (0), and the crossover point (.5), this method allows us to assign fuzzy-set scores
for each case according to its original value.
Finally, for the remaining dichotomous indicators, such as alliance membership, the
assignment of membership scores is straightforward. A score of 1 is assigned to cases in which
the condition is present; otherwise, they receive a 0.
Since it is not possible to discuss the calibration of all conditions within the limits of
this article, the online Appendix explains in detail and justifies the assignments of the
membership scores of all 31 cases in the sets formed by the five explanatory conditions and the
outcome condition. Also explained is the aggregation of the indicators. The resulting
membership scores for all of the 31 cases of ongoing humanitarian crises between 1991 and
2004 are reported in Table I.
Table I in here
14
I recoded the outcome variable in several non-equidistant ways. My results remain robust. Details are
reported in the online Appendix.
17
Results
In order to explain (1) when the UN took strong action (i.e. coercive action under Chapter VII
authority) to respond to humanitarian crises, and (2) under what condition no such measures
were adopted, fs/QCA was applied, respectively, to the membership scores of the 31 cases
reported in Table I. This was done using the software program fs/QCA 2.5. The program
simplifies patterns in the data and ultimately identifies (combinations of) conditions that are
necessary and/or sufficient for an outcome of interest.
The test for necessary conditions indicates that neither the presence nor the absence of
any of the five explanatory conditions can be considered to be necessary for strong or limited
UN response to violent humanitarian crises. The recommended thresholds for the consistency
and coverage of necessary conditions are .9 or higher, and at least .6, respectively (Schneider
& Wagemann, 2012: 143–147). All results of the analysis of necessary conditions are reported
in the online Appendix. The only condition that meets these criteria is the absence of media
attention (~ATTENTION), which might thus be considered to be quasi-necessary for a limited
SC response (~STRONG). Given that almost all 31 crises display scores for ~ATTENTION
that are greater than (or equal to) their scores in the outcome (~STRONG), this condition has a
high consistency score (.94). Moreover, ~ATTENTION displays a relatively high coverage
score (.61). Yet, a closer inspection of the result reveals that the condition is ‘trivial’
(Braumoeller & Goertz, 2000) in that many cases display high values for the condition
(~ATTENTION), whether or not the outcome (~STRONG) is present.
18
Explanations for strong UN action
To examine the conditions that lead to strong UN action (in terms of sufficiency), I employ two
models containing different combinations of the five explanatory conditions.
15
Regarding the
analysis of sufficiency, the researcher has some leeway in determining the thresholds for
consistency. This threshold distinguishes causal combinations that are a consistent subset of the
outcome from those that are not. This threshold is determined by significant gaps in consistency
values in the truth table. It can be lower than for the analysis of necessity, but it should be at
least .75 (Ragin, 2008: 136). In this analysis, I use conservative thresholds above .9. The results
of the analysis are reported in Table II below. Note that Table II contains the totality of causal
paths obtained by the analysis of the two models. In fuzzy-set notation, the asterisk (*) stands
for the logical operator and, the absence of a condition is indicated by the tilde (~), and the plus
sign (+) indicates the logical or.
Table II in here
The analysis yields four combinations of conditions or causal paths that are sufficient
for the outcome. According to the first causal path, the UN takes coercive action if the extent
of a humanitarian crisis is high (in terms of victims or internal displacement), international
institutions have been substantially involved in that crisis (i.e. if there are significant sunk
costs), and if that crisis produces substantial negative spillover effects in terms of refugees or
other negative externalities for neighboring countries. According to the second causal path, the
UN takes coercive action in response to large-scale humanitarian crises characterized by
substantial previous institutional involvement, which occur in a country without strong
15
The four core conditions are the extent of a humanitarian crisis, the magnitude of its spillover effects, the
strength of the countervailing power, and the level of previous institutional involvement (Model 1). These are
supplemented by media attention (Model 2).
19
countervailing power to resist outside intervention (i.e. a target state without significant
capabilities or powerful allies).
16
The third and fourth causal combinations support these findings because they all include
either the first or the second path (indicated by the bold type in Table II).
17
Moreover, causal
paths three and four indicate that media attention has indeterminate effects on the decision to
intervene. While high media attention in the fourth path contributes to strong UN response, in
the third path, indications are that UN intervention may occur despite the absence of high-level
media reporting. This supports the critics’ view according to which the impact of media
coverage is limited and indirect, depending on the presence of other additional factors
(Jakobsen, 2000; Robinson, 2001: 954). Taken together, a combination of four factors – or two
causal paths – is decisive in bringing about a strong UN response to humanitarian crises:
EXTENT*INVOLVEMENT(SPILLOVER+~COUNTPOWER)
This means that the determinants for coercive UN action are a large extent of human
suffering and substantial previous investment in a crisis, when combined with either significant
spillover effects for neighboring countries (path 1), or with military weakness, or no strong
allies, thus rendering the target state unable to generate strong countervailing power against
outside intervention (path 2). This finding demonstrates the importance of these four factors.
At the same time, none of the four explanatory conditions is individually sufficient to account
for UN intervention. Moreover, countervailing power is only part of one causal path. This
means that intervention can occur despite the military strength of a target state; namely, when
16
The consistency threshold is .96. Consistency values of causal combinations in the truth table equal to – and
above – these scores receive a 1 in the outcome column; 0 is assigned to causal combinations below these
values. See Table II in the Appendix.
17
The consistency threshold is .95.
20
a crisis involves a large number of victims, substantial sunk costs, and significant spillover
effects.
How good is the ‘fit’ between the solution formulae and the data? Recall that two
parameters are relevant in fs/QCA. The consistency scores describe the quality of a given set
relationship. In this analysis, these scores are very high, varying between .92 and .98 (see Table
II). As a result, both paths can be considered as quasi-sufficient for the occurrence of the
outcome. They are quasi-sufficient because a few cases do not fulfill the sufficiency
requirement; that is, the combined values of the causal combinations are greater, and not
smaller, than the value of the outcome (as a result, these cases are located below the main
diagonal in Figure 1). However, this applies only to very few cases that are ‘near misses.’ As
indicated by the raw coverage scores in Table II and III – a measure for the empirical
importance of a causal path – the first solution term covers 71% of the outcome to be explained,
while the second path covers 56% of the outcome. This shows that spillover carries more
importance than countervailing power (if combined with a large humanitarian extent and
substantial prior involvement in the crisis). This is also reflected by the unique coverage scores
of both paths (.18 and .04, respectively). Yet, these scores are low, meaning that a substantial
number of cases is explained by both causal paths. The only exceptions are Northern Iraq
(uniquely covered by path 1) and Liberia 2 (uniquely covered by path 2).
Table III in here
Figure 1 links the results of the analysis to the specific cases. The cases explained by
the two causal paths are indicated by bold type; they are located in the upper right corner of the
figure. These cases have high values in both the causal combinations (x-axis) and in the outcome
(y-axis). Taken together, both paths cover 11 out of the 13 post-Cold War crises (or 85%) where
the SC has taken coercive action under Chapter VII. These cases include crises to which the SC
21
has responded by employing (or authorizing) the use of force (e.g. Operation ‘Provide Comfort’
in Northern Iraq, or UNITAF in Somalia), by deploying ‘robust’ PKOs (e.g. UN Mission in the
Democratic Republic of the Congo, MONUC II), or by imposing economic sanctions (e.g. the
sanctions imposed on Angola 2 through SC Resolutions 1127 and 1173). The cases located in
the upper left corner (Kosovo and Angola 1) represent instances where the SC took coercive
measures – which remain unexplained by the results and, thus, require alternative explanations.
In the case of Kosovo, the Council imposed sanctions despite the rather limited humanitarian
extent of the crisis. In Angola 1, the Council imposed sanctions although the UNs prior
involvement in the crisis had been limited and had remained confined to a small observer
mission (70 observers) in the country (UNAVEM I).
Figure 1 in here
Explanation for limited UN action
To determine under what conditions the SC does not take coercive action under Chapter VII, I
employ the same two models as I did in the previous analysis of strong UN action. However,
the results here are more diverse; that is, there are more causal paths that lead to the outcome.
Some of these paths have very low coverage scores – they explain only a small fraction of the
outcome. Only four paths have reasonably high coverage scores (more than 15%) and, hence,
explain a substantial part of the outcome. These paths are reported in Table IV.
Table IV in here
According to the first and the second causal paths, the UN does not take coercive action
if a crisis occurs in a country that is able to generate strong countervailing power against outside
intervention, and if international institutions have not been previously involved in that crisis.
22
According to the second formula, SC response remains limited in situations of strong
countervailing power, if the extent of human suffering is rather low, and if a crisis does not
produce substantial spillover effects to neighboring countries.
18
Both causal paths are supported
by the third and fourth formula adding media attention.
19
Unlike the cases of strong SC action,
the role of media attention here is clear. Little media attention is consistently associated with a
limited SC response to a humanitarian crisis. The fact that countervailing power is part of all
solution formula highlights the explanatory power of this condition, and this allows for the
following simplification of the result.
COUNTPOWER(~INVOLVEMENT+~EXTENT*~SPILLOVER)
But again, strong countervailing power is not sufficient to explain why the UN fails to
take strong action. Only in combination with either the lack of sunk costs or a large extent of
human suffering and spillover effects does it explain why UN responses remain limited to non-
coercive measures.
Regarding the quality of the results, their consistency is very high (between .92 and .98;
see Table IV). Moreover, the coverage scores in Tables III and IV show that path 1 has clearly
more empirical weight than path 2. Figure 2 shows that the overall explanation covers a
substantial number of cases, but the coverage is somewhat lower than it was for the analysis of
the positive outcome. Of the 18 cases of limited UN response, 14 are explained by the solution
formula (78%). They are indicated by bold type, located in the upper right corner of Figure 2.
In these cases, the UN did not take any action at all (Turkey and India–Kashmir), or the UN
response remained confined to non-coercive measures, such as humanitarian assistance (e.g.
Chechnya) or consent-based peacekeeping (e.g. UN Observer Mission in Sudan, UNAMIS).
18
The consistency threshold is .92.
19
The consistency threshold is .92.
23
Figure 2 in here
In sum, fuzzy-set analysis shows that the key determinants for coercive UN action are a
large extent of human suffering and substantial previous institutional involvement in a crisis; if
the crisis produces significant spillover effects for neighboring countries, or occurs in a
militarily weak target state lacking the ability to generate strong countervailing power against
outside intervention. Limited UN response is best explained by strong countervailing power,
but only when combined with other scope conditions.
Discussion
The results of the analysis have a number of implications that deserve closer attention. First of
all, the findings show that there are clear patterns in the way the UN responds to humanitarian
crises. Although every crisis may be historically unique, the SC’s responses to these crises are
not random. Rather, the way the SC selects crises for intervention follows specific trajectories
that are indicated by a limited number of causal combinations or paths. Moreover, the paths are
highly consistent and they cover most (but not all) crises.
Second, the analysis demonstrates that the variation in UN humanitarian intervention
requires configurational explanation. None of the conditions emphasized by the different
intervention theories is individually necessary or sufficient for strong or limited UN action. In
combination, however, they provide for a powerful sufficient explanation of variation in UN
humanitarian intervention.
Third, the findings of the fuzzy-set analysis are in line with previous research in the field
of third-party intervention. This applies to the number of victims of a crisis or a conflict (e.g.
Gilligan & Stedman, 2003), the number of refugees (e.g. de Jonge Oudraat, 1996), and the
previous institutional involvement (Mullenbach, 2005), which these studies found to increase
the likelihood of outside intervention. Likewise, the military strength of a target state was found
24
to be negatively associated with the likelihood of intervention by a number of large-N analyses
(Gilligan & Stedman, 2003; Fortna, 2004; Mullenbach, 2005; Beardsley & Schmidt, 2012).
These studies lend further credibility to the results of the fuzzy-set analysis. However, it is
crucial to note that fuzzy-set analysis follows a fundamentally different causal logic and serves
a different purpose. Unlike statistical analysis, which is based on co-variation and focuses on
the average net effects of single independent variables, fuzzy-set analysis is based on set-
theoretic relationships of necessity and sufficiency, and focuses on causally relevant
combinations to explain an outcome. Accordingly, the contribution of this paper lies not in
determining the effects of individual explanatory variables, but in identifying precisely which
causal combinations or causal paths lead to strong or limited UN action respectively. These
must be evaluated and interpreted in their entirety. At most, an individual component of a causal
path can be considered as an INUS conditions (insufficient but non-redundant part of an
unnecessary but sufficient condition) or as ‘one cause within a combination of causes that are
jointly sufficient for an outcome’ (Mahoney & Goertz, 2006: 232).
Finally, how can we interpret the causal paths theoretically? As I have outlined above,
some of the factors examined in this article are stressed by more than one theoretical
perspective, which renders theory testing difficult. In turn, it is equally hard to make sense of
the empirical results from a single theoretical account. A theoretically integrative perspective,
which accepts the complexity of social phenomena, and views different logics of action ‘as
complementary rather than to assume a single dominant behavioral logic’ (March & Olsen,
2006: 702) allows us to sketch an intuitive and straightforward interpretation of the results. In
line with constructivist and institutionalist inspired theories of third-party intervention,
humanitarian concerns, and the wish to protect institutional investments (sunk costs) are
important drivers of UN humanitarian intervention. Yet, both factors seem to be highly context-
dependent and lead to intervention only when joined with material considerations, as stressed
by rationalist–realist intervention theories. In the first context, humanitarian considerations and
25
sunk cost concerns seem to be sufficient for SC intervention when, additionally, the risks and
costs of possible intervention are comparatively low because the target state is unable to
generate strong countervailing power against outside intervention. In the second context,
humanitarian concerns and concerns for sunk costs seem to lead to SC intervention when they
are complemented by material concerns over the negative spillover effects to neighboring
countries. In situations like this, where all three factors – humanitarian considerations, concerns
for sunk costs, and the wish to contain spillover effects – push towards intervention, the
constraining role of countervailing power seems to be less critical.
Conclusion
This article has employed fs/QCA to compare the SC’s response to more than 30 major
humanitarian crises after the Cold War. The findings show that no single condition accounts for
the Council’s selective use of humanitarian intervention (broadly defined). Rather, the puzzling
variation in the Council’s response to humanitarian emergencies requires a configurational
explanation that consists of a limited number of causal paths. It is important to note that, while
fs/QCA helps us to identify subset relations between explanatory conditions and an outcome
condition in terms of necessity and sufficiency, it is not a substitute for an investigation into the
causal mechanisms at work. This requires additional in-depth case study research both for the
cases covered by the results and for the outliers, which require an alternative explanation.
How well does the Council function in addressing humanitarian crises? Pointing to
‘double standards’, the SC’s uneven response to humanitarian emergencies is frequently
condemned by journalists, human rights activists, and policy makers. Likewise, scholars note
that the SC’s selective enforcement of human rights undermines not only the credibility and,
eventually, the effectiveness of humanitarian interventions, but also the legitimacy of
international order more generally (e.g. Archibugi, 2004). The current debate on the SC’s
uneven treatment of the crises in Libya and Syria is a case in point. But this holds more
26
generally. In fact, in only 13 of the 31 major humanitarian crises considered in this study did
the Council members agree on strong measures. This suggests a poor track record. At the same
time it is clearly an improvement over the Cold War, during which the Council was largely
paralyzed by superpower rivalry. After 1990, the SC has become much more active in
peacekeeping and peacemaking. It is much less often blocked by vetoes than before, while the
number of resolutions, including resolution under Chapter VII, through which ‘robust’ PKOs,
sanction regimes, and peace enforcement missions are established, have increased significantly
(Wallensteen and Johansson, forthcoming). Nevertheless, the Council’s humanitarian activity
is characterized by a substantial selectivity gap in that it responds to more – but by far not all –
crises. Ultimately, the Council is a deeply political body whose action requires an agreement
among a majority of its members and the support (or at least the acquiescence) of the P5. As
such, we cannot expect that this support is forthcoming in consistent ways. Against this
backdrop, it was the aim of this study to identify under what conditions the Council members
agree (or do not agree) to intervene.
Finally, my findings also carry important policy implications. Whether or not strong
humanitarian action is taken by the SC has immediate practical consequences for decision
makers in national and international institutions, for humanitarian organizations that operate in
crisis areas, and, of course, for the populations affected by humanitarian emergencies around
the world.
27
References
Archibugi, Daniele (2004) Cosmopolitan guidelines for humanitarian intervention.
Alternatives 29(1): 1–21.
Barnett, Michael N (1998) The limits of peacekeeping, spheres of influence, and the future of
the United Nations. In: Joseph Lepgold & Thomas G Weiss (eds) Collective Conflict
Management and Changing World Politics. Albany, NY: State University of New York
Press, 83–103.
Beardsley, Kyle & Holger Schmidt (2012) Following the flag or following the charter?
Examining the determinants of UN involvement in international crises, 1945–2002.
International Studies Quarterly 56(1): 33–49.
Bellamy, Alex J (2010) The responsibility to protect – Five years on. Ethics & International
Affairs 24(2): 143–169.
Boettcher, William A & Michael D Cobb (2009) ‘Don’t let them die in vain’: Casualty frames
and public tolerance for escalating commitment in Iraq. Journal of Conflict Resolution
53(5): 677–697.
Boulden, Jane (2006) Double standards, distance and disengagement: Collective
legitimization in the post-cold war Security Council. Security Dialogue 37(3): 409–423.
Braumoeller, Bear F & Gary Goertz (2000) The methodology of necessary conditions.
American Journal of Political Science 44(4): 844–858.
Burg, Steven L (2004) Intervention in internal conflict: The case of Bosnia. In: William J
Lahneman (ed.) Military Intervention. Cases in Context for the Twenty-First Century.
New York: Rowman & Littlefield, 47–66.
Chomsky, Noam (1999) The New Military Humanism. Lessons from Kosovo. Monroe, ME:
Common Courage Press.
Crocker, Chester A (2001) Intervention: Towards best practices and a holistic view. In:
Chester A Crocker, Fen Osler Hampson & Pamela R Aall (eds) Turbulent Peace the
Challenges of Managing International Conflict. Washington, DC: United States Institute
of Peace Press, 229–248.
Cunningham, David E; Kristian Skrede Gleditsch & Idean Salehyan (2013) Non-state actors
in civil wars: A new dataset. Conflict Management and Peace Science 30(5): 516–531.
De Jonge Oudraat, Chantal (1996) The United Nations and internal conflict. In: Michael E
Brown (ed.) The International Dimensions of Internal Conflict. Cambridge, MA: MIT
Press, 489–535.
DeRouen, Karl & Uk Heo (eds) (2007) Civil Wars of the World: Major Conflicts since World
War II. Santa Barbara, CA: ABC-Clio.
Dowty, Alan & Gil Loescher (1996) Refugee flows as grounds for international action.
International Security 21(1): 43–72.
Doyle, Michael W & Nicholas Sambanis (2006) Making war and building peace: United
Nations peace operations. Princeton, NJ: Princeton University Press.
Feste, Karen A (1992) Expanding the Frontiers: Superpower Intervention in the Cold War.
New York: Praeger.
Finnemore, Martha (2003) The Purpose of Intervention: Changing Beliefs about the Use of
Force. Ithaca, NY: Cornell University Press.
Fortna, Virginia P (2004) Does peacekeeping keep peace? International intervention and the
duration of peace after civil war. International Studies Quarterly 48(2): 269–292.
Ghosn, Faten; Glenn Palmer & Stuart A Bremer (2004) The MID3 Data Set, 1993–2001:
Procedures, coding rules, and description. Conflict Management and Peace Science 21(4):
133–154.
Gilboa, Eytan (2005) The CNN Effect: The search for a communication theory of
international relations. Political Communication 22(1): 27–44.
28
Gilligan, Michael J & Stephen J Stedman (2003) Where do the peacekeepers go?
International Studies Review 5(4): 37–54.
Haass, Richard N (1999) What to do with American primacy. Foreign Affairs 78(5): 37–49.
Hasenclever, Andreas (2001) Die Macht der Moral in der internationalen Politik [The power
of morality in international politics]. Frankfurt a.M.: Campus.
Hewitt, J Joseph (2003) Dyadic processes and international crises. Journal of Conflict
Resolution 47(5): 669–692.
Hippel, Karin von (1996) Sunk in the Sahara: the applicability of the sunk cost effect to
irredentist disputes. The Journal of North African Studies 1(1): 95–116.
Jakobsen, Peter V (1996) National interest, humanitarianism or CNN: What triggers UN
peace enforcement after the cold war? Journal of Peace Research 33(2): 205–215.
Jakobsen, Peter V (2000) Focus on the CNN effect misses the point: The real media impact on
conflict management is invisible and indirect. Journal of Peace Research 37(2): 131–143.
Klotz, Audie (1995) Norms reconstituting interests: Global racial equality and U.S. sanctions
against South Africa. International Organization 49(3): 451–478.
Leitenberg, Milton (2006) Deaths in wars and conflicts in the 20th Century. Cornell
University Peace Studies Program. Occasional Paper 29
(http://www.clingendael.nl/publications/2006/20060800_cdsp_occ_leitenberg.pdf).
Luard, Evan (1984) Collective intervention. In: Hedley Bull (ed.) Intervention in World
Politics. Oxford: Oxford University Press, 137–179.
Lyon, Alynna J & Chris Dolan (2007) American humanitarian intervention: Toward a theory
of coevolution. Foreign Policy Analysis 3(1): 46–78.
Mahoney, James & Gary Goertz (2006) A tale of two cultures: Contrasting quantitative and
qualitative research. Political Analysis 14(3): 227–249.
Malone, David M (2004) Introduction. In: David M Malone (ed.) The UN Security Council.
From the Cold War to the 21st Century. Boulder, CO: Lynne Rienner, 1–15.
March, James G & Johan P Olsen (2006) The logic of appropriateness. In: Michael Moran,
Martin Rein & Robert E Goodin (eds) The Oxford Handbook of Public Policy. Oxford:
Oxford University Press, 689–708.
Marshall, Monty G (2008) Forcibly displaced populations, 1964–2008
(http://www.systemicpeace.org/inscr/inscr.htm).
Mason, Andrew & Nicholas J Wheeler (1996) Realist objections to humanitarian intervention.
In: Barry Holden (ed.) The Ethical Dimensions of Global Change. Basingstoke: Palgrave
Macmillan, 94–110.
McAfee, R Preston; Hugo M Mialon & Sue H Mialon (2010) Do sunk costs matter?
Economic Inquiry 48(2): 323–336.
Mearsheimer, John J (2003) The Tragedy of Great Power Politics. New York: Norton.
Morgenthau, Hans Joachim (1967) To intervene or not to intervene. Foreign Affairs 45(3):
425–436.
Mullenbach, Mark J (2005) Deciding to keep peace: An analysis of international influences
on the establishment of third-party peacekeeping missions. International Studies Quarterly
49(3): 529–555.
Neack, Laura (1995) UN peace-keeping: In the interest of community or self? Journal of
Peace Research 32(2): 181–196.
Pierson, Paul (2004) Politics in Time. History, Institutions, and Social Analysis. Princeton,
NJ: Princeton University Press.
Pinfari, Marco (2011) Time to agree: Is time pressure good for peace negotiations? Journal of
Conflict Resolution 55(5): 683–709.
Ragin, Charles C (2000) Fuzzy-Set Social Science. Chicago, IL: University of Chicago Press.
Ragin, Charles C (2008) Redesigning Social Inquiry. Fuzzy Sets and Beyond. Chicago, IL:
University of Chicago Press.
29
Ramsbotham, Oliver & Tom Woodhouse (1996) Humanitarian Intervention in Contemporary
Conflict. A Reconceptualization. Cambridge: Polity Press.
Rihoux, Benoît & Charles C Ragin (eds) (2009) Configurational Comparative Methods:
Qualitative Comparative Analysis (QCA) and Related Techniques. Thousand Oaks, CA:
Sage.
Robinson, Piers (2001) Operation restore hope and the illusion of a news media driven
intervention. Political Studies 49(5): 941–956.
Robinson, Piers (2011) The CNN effect reconsidered: Mapping a research agenda for the
future. Media, War & Conflict 4(1): 3–11.
Salehyan, Idean (2009) Rebels Without Borders: Transnational Insurgencies in World
Politics. Ithaca, NY: Cornell University Press.
Salehyan, Idean & Kristian Skrede Gleditsch (2006) Refugees and the spread of civil war.
International Organization 60(2): 335–366.
Schneider, Carsten Q & Claudius Wagemann (2012) Set-Theoretic Methods for the Social
Sciences: A Guide to Qualitative Comparative Analysis. Cambridge: Cambridge
University Press.
Stinchcombe, Arthur L (1968) Constructing Social Theories. New York: Harcourt, Brace &
World.
Van der Maat, Eelco (2011) Sleeping hegemons: Third-party intervention following territorial
integrity transgressions. Journal of Peace Research 48(2): 201–215.
Väyrynen, Raimo (2000) Complex humanitarian emergencies: Concepts and issues. In: E
Wayne Nafziger, Frances Stewart & Raimo Väyrynen (eds) War, Hunger, and
Displacement. The Origins of Humanitarian Emergencies. Oxford: Oxford University
Press, 43–89.
Wallensteen, Peter & Patrik Johansson (forthcoming) The UN Security Council: Trends in
decisions and actions. In: Sebastian von Einsiedel, David M Malone & Bruno Stagno
Ugarte (eds) The Security Council during the 21st Century, Boulder, CO: Lynne Rienner.
Waltz, Kenneth N (1979) Theory of International Politics. Boston, MA: MacGraw-Hill.
Walzer, Michael (2000 [1977]) Just and Unjust Wars. A Moral Argument with Historical
Illustrations. New York: Basic Books.
Weiss, Thomas G (2004) The humanitarian impulse. In: David M Malone (ed.) The UN
Security Council. From the Cold War to the 21st Century. Boulder, CO: Lynne Rienner,
37–54.
Wheeler, Nicholas J (2000) Saving Strangers. Humanitarian Intervention in International
Society. Oxford: Oxford University Press.
30
Figure 1: Cases covered by the causal combination
Afghanistan
Ang o la 1 Angola 2
Azerbaij an (Karabakh)
Bosn ia
Buru ndi
Colombia
Congo-Brazzaville
DR Congo
Georgia (Abkhazia)
Guin ea-Bi ssau
India (Kashm ir)
Iraq (Northern Iraq)
Liberia 1
Liberia 2
Moza m b iq u e
Myan m a r
Nepal
Peru
Rus sia (Chechnya) 1
Russ ia (Chech nya) 2
Rwa nda
Sierra Leone
Somalia
Sri Lanka
Sudan
Sudan (Darfur)
Tajikis tan
Turkey
Uganda (Northern Uganda)
Yugoslavia (Kosovo)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
00,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1
Strong SC Response
EXT ENT*I NV OLV (S PI LL+ ~ COUN T)
31
Figure 2: Cases covered by the causal combination
Afg ha n is t an
Ang o la 1Ang o la 2
Azerbaijan (Karabakh)
Bosn ia
Burundi
Colom bia
Congo-Brazzaville
DR Congo 3
Georgia (Abkhazia)
Guin ea- Bissau
India (Kashmir)
Iraq (Northern Iraq)
Liberi a 1
Liberia 2
Moza m b iq u e
Myanmar
Nepa l
Peru
Russia (Chechnya) 1
Russia (Chechnya) 2
Rwanda
Sierra Leone
Soma lia
Sri Lanka
Sudan
Sudan (Darfur)
Tajikistan
Turke y
Uganda (Northern Uganda)
Yugoslavia (Kosovo)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
00,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1
Limite d SC Response
COUNT(~INVOLV+ ~EXTENT*~SPILL)
32
Table I. Fuzzy-set membership scores for 31 humanitarian crises 1991-2004
Cases
Explanatory conditions
Outcome
Years
(1)
Extent
(2)
Spillover
Effects
(3)
Countervailing
Power
(4)
Institutional
Involvement
(5)
Media
Attention
Strength of UN
response
Afghanistan
1991–2004
1
1
0.08
0.64
0.16
1
Angola 1
1991–1994
1
0.66
0.44
0.32
0.12
0.64
Angola 2
1998–2002
1
1
0.35
0.8
0.06
0.64
Azerbaijan (Karabakh)
1992–1994
0.68
0.59
1
0.16
0.09
0.16
Bosnia
1992–1995
1
1
0.38
0.8
1
1
Burundi
1993–2004
0.86
0.64
0.09
0.8
0.05
0.8
Colombia
1991–2004
1
1
1
0.16
0.11
0.16
Congo-Brazzaville
1997–1999
0.49
0
0
0.16
0
0.16
DR Congo
1996–2004
1
1
0.16
0.8
0.27
0.8
Georgia (Abkhazia)
1992–1994
0.13
0.13
1
1
0.06
0.32
Guinea-Bissau
1998–1999
0.1
0
0
1
0
0.16
India (Kashmir)
1991–2004
0.24
1
1
0.16
0.07
0
Iraq (Northern Iraq)
1991–1993
0.64
1
1
1
1
1
Liberia 1
1991–1995
1
1
0
1
0
0.64
Liberia 2
2000–2003
0.6
0.49
0
0.64
0.09
0.8
Mozambique
1991–1992
1
1
0.12
0.16
0
0.48
Myanmar
1991–2004
0.86
1
1
0.16
0
0.16
Nepal
1996–2004
0.06
0
1
0.16
0
0.16
Peru
1991–1997
0.64
1
1
0.16
0.07
0.16
Russia (Chechnya) 1
1994–1996
0.41
0
1
0.16
0.7
0.16
Russia (Chechnya) 2
1999–2004
0.86
0
1
0.16
0.37
0.16
Rwanda
1993–1994
1
1
0.06
0.8
0.22
1
Sierra Leone
1991–2002
0.82
1
0
1
0
1
Somalia
1991–1995
1
0.92
0
0.8
0.08
1
Sri Lanka
1991–2002
0.92
0.28
1
0.16
0.06
0.16
Sudan
1991–2004
1
0.72
1
0.32
0.07
0.32
Sudan (Darfur)
2003–2004
1
0.25
1
0.16
0.05
0.16
Tajikistan
1992–1997
0.32
0.3
1
1
0
0.32
Turkey
1991–2004
1
1
1
0.16
0
0
Uganda (Northern Uganda)
1994–2004
1
1
0.11
0.16
0
0.16
Yugoslavia (Kosovo)
1998–1999
0.1
1
1
1
0.87
0.64
33
Table II. Fs/QCA results: causal paths leading to strong SC response
Raw
Coverage
Unique
Coverage
Consistency
(1) EXTENT*INVOLVEMENT*SPILLOVER+
. 71
.18
.92
(2) EXTENT*INVOLVEMENT*~COUNTPOWER +
.56
.04
.96
(3) EXTENT*INVOLVEMENT*~COUNTPOWER~ATTENTION+
.51
.44
.95
(4) EXTENT* INVOLVEMENT*SPILLOVER*ATTENTION+
.22
.14
.98
34
Table III. Cases covered by the four causal paths
Causal path 1
EXTENT*INVOLVEMENT*
SPILLOVER
→ STRONG
Causal path 2
EXTENT*INVOLVMENT
*~COUNTPOWER
→STRONG
Causal path 1
COUNTPOWER*~INVOLVMENT
→~STRONG
Causal path 2
COUNTPOWER*~EXTENT
*~SPILLOVER
→~STRONG
Raw
coverage
.71
.56
.66
.21
Unique
coverage
.18
.04
.53
.08
Covered
cases1
Liberia 1, Sierra Leone,
Angola 2, DR Congo,
Bosnia, Rwanda, Somalia,
Burundi, Afghanistan,
Northern Iraq
Liberia 1, Sierra Leone,
Angola 2, DR Congo,
Bosnia, Rwanda,
Somalia, Burundi,
Afghanistan, Liberia 2
India (Kashmir), Turkey,
Azerbaijan (Karabakh),
Colombia, Myanmar, Nepal,
Peru, Russia (Chechnya) 1,
Russia (Chechnya 2), Sri Lanka,
Sudan (Darfur), Sudan
Nepal, Georgia (Abkhazia),
Tajikistan, Russia
(Chechnya 1)
Consistency
.92
.96
.96
.92
1 Cases with greater than (.5) membership in the respective path
35
Table IV. Fs/QCA results: causal paths leading to limited SC response
Raw
Coverage
Unique
Coverage
Consistency
(1) COUNTPOWER*~INVOLVEMENT+
.66
.53
.96
(2) COUNTPOWER*~EXTENT*~SPILLOVER+
.21
.08
.92
(3) COUNTPOWER*~INVOLVEMENT*~ATTENTION+
.61
.51
.98
(4) COUNTPOWER*~EXTENT*~SPILLOVER*~ATTENTION+
.19
.08
.92
Biographical statement
MARTIN BINDER, b. 1977, PhD in Political Science (Free University Berlin, 2009); senior
research fellow at the WZB Berlin Social Science Center (2009–)
Acknowledgments
I thank Mathias Koenig-Archibugi, Matthias Ecker-Ehrhardt, Gary Goertz, Autumn
Lockwood Payton, Monika Heupel, Benoît Rihoux, Scott Siegel, Georg Sørensen, Barçın
Uluışık, Lora Viola, Bernhard Zangl, and Michael Zürn for their comments. Special thanks to
Felix Braunsdorf for research assistance. Any remaining errors are mine.
Replication data
All data along with the online Appendix can be found at http://www.prio.no/jpr/datasets
