Abstract

External interventions are one of the most important aspects of intrastate conflicts since a majority of them are significantly internationalised, especially in Africa where the interventions most often occur. Factors that lead to the military intervention remain, however, puzzling. The authors therefore apply the method of fs/QCA to understand not only  conditionsbehind intervention into African intrastate conflicts, but also to catch interactions among them. The results show high complexity of various possible combinations, mainly of high intensity, massive social dislocation or presence of raw materials in case of interventions in African internal conflicts.  

Keywords

conflict, interventions, Africa, QCA, set-theoretical approach, fuzzy-set analysis

Introduction

With a growing interconnection of states not just within a shared region but globally, internal conflicts pose a considerable threat for security of neighbouring states as well as of a whole international community. Intrastate conflicts can easily spill-over the borders, especially if they are connected with huge social dislocations. They could be also a proximate cause of massive migration to quite distant destinations, as the recent influx of immigrants from Syria or Afghanistan to Europe showed. Moreover, intrastate conflicts are usually connected with immense suffering of civilians, notably women and children. Therefore, an effort of external actors to end or to manage the conflicts is quite reasonable, especially if we add a promising opportunity to increase their power in the region or to secure various assets through the intervention. But which of all the above-mentioned reasons are crucial for the decision of a third party to intervene? Are there any significant interactions (trade-offs, synergies) among them? Those are the most important questions that we are going to address in this study.

The external military interventions have naturally become a critical part of internal conflicts, having a substantial (sometimes even crucial) influence on the length as well as on the results of the conflicts.[1] The studies, however, focus mostly on interventions’ outcomes,[2] or the probability of success in ending the violence,[3] leaving the very important issue of conditions of the interventions rather at the edge of academic interest. Of course, there are a few considerable exceptions, but they usually concern the motivations or conditions per se – without any specific geographical regard.[4]

Since it is obvious that some regions are much more affected by intrastate conflicts and subsequent interventions than others, it is important to find out whether the conditions under which interventions occur vary across the regions or not. This study aims to contribute to this dilemma as it focuses solely on the African continent, trying to compare the motivations to intervene in African conflicts with the generally assumed motivations derived from the previous research.[5] The limitation on Africa is mostly motivated by its higher rate of external interventions in comparison with other regions.[6] Consequently, there are enough cases to investigate and possibly generalise, and at the same time they share (at least to some extent) common geographical, demographical, sociological, historical, economic, political and geopolitical conditions. Of course, there are many substantial differences among African states (especially between North Africa and Sub-Saharan Africa) but they also have much in common – for example membership in the African Union and participation in its security efforts, or common sub-regional threats such as al-Qaeda in the Islamic Maghreb which threaten both Sub-Saharan and North African countries. Shortly, Rwanda is quite different from Libya but even more different from Colombia or Germany.

The importance of external interventions in Africa is far from being a recent phenomenon as interventions have played a significant role in the all of ‘post-colonial’ history. In fact, they were a central instrument of foreign policy towards Africa both for colonial powers (France, UK, Belgium) and for the two new superpowers of the Cold War.[7] Regarding recent political development (especially in the Central African Republic, the Democratic Republic of the Congo or Mali) in connection with the threat of terrorism and/or migration waves, there is no reason to consider external military interventions as bygones or unlikely events. Quite the contrary, external intervention will remain a very important part of international politics, particularly in Africa. Consequently, the causes and likelihood of external interventions in internal conflicts seem to be very important for understanding the dynamics of those conflicts as well as for predicting their courses and consequences.

In order to uncover conditions under which external military interventions occurs, our study innovatively employs the QCA method which has been used in just a few cases before.[8] A benefit of QCA rests in providing a reliable understanding to relationships among conditions (independent variables). In comparison to quantitative methods (especially to the most used linear regression), QCA requires much fewer cases and stresses inter-connections among independent variables instead of an influence of a single variable. Therefore, our approach can discover the interactions between independent variables that makes it different from more conventional statistical approaches.[9] We also assume that the different method could bring different results and/or contribute to the discussion about methodological aspects of interventions’ research. To sum it up, the main aim of our study is not to derive a new theory but to test conventional hypotheses through the non-mainstream method that can show us relations among variables from a set-theoretic perspective.

Our research question is as follows: What are the conditions of third-party military interventions in African internal conflicts? Our paper is more focused on characteristics of the conflict and states where the intervention could happen.[10] The time scope of the study is limited purely to the Post-Cold War era as we suppose that the Cold War substantially influenced behaviour of the states. The cases in our study therefore come from 1989 (the year of the symbolic fall of the Berlin Wall) to 2015.

First, we present a theoretical discussion and a short methodological framework which introduces QCA. Then we describe the results of the QCA analysis of the tested hypothesis, which we subsequently discuss from the theoretical standpoints and illustrate in examples of military interventions to Sierra Leone and Mali to draw the final conclusions. 

Theory review

Before we step to theoretical approaches toward conditions of interventions, we must define the intervention itself more precisely. Intervention is a longstanding and broadly used concept, its definition is, however, rather rigid and substantially uncontested. The reason may lie in interventions’ intrinsic connection with sovereignty which is inconceivable with an external (and unwanted) intervention on its territory.[11] The principles of sovereignty and non-intervention are fundamental norms of international relations (despite different practices) and even though there is a debate on changing sovereignty in the context of globalisation, the fundamental defining criteria of intervention remain nearly unchanged.

A pioneering definition was written up by James Rosenau who perceives intervention as ‘convention-breaking’ and ‘authority oriented’.[12] Rosenau describes ‘the behavior of one international actor toward another as interventionary whenever the form of the behavior constitutes a sharp break with then-existing forms and whenever it is directed at changing or preserving the structure of political authority in the target society’.[13] Rosenau’s definition was criticised for several aspects – for example, Richard Little drew attention to Rosenau’s neglect of intervenors’ motivations as well as excluding foreign aid as a common instrument of foreign policy.[14] Another criticism targeted the absence of any underpinning of his research in empirical evidence which consequently limited his statements to a pure theoretical position.[15] Nevertheless, Rosenau’s emphasis of a third party’s impact on conflict and power in other states remains at the centre of our common understanding of interventions. 

Most of authors have been using the narrow ‘military definition’ – they perceive intervention as an occurrence of regular soldiers or military actions of one state in another state’s conflict.[16] However, recently there have been several attempts to use a different theoretical perspective and consequently also a new operational definition. For example, to add economic aspects of interventions[17] or, as Martha Finnemore[18] did, to use a constructivist approach when she stressed the acknowledgement and self-identification of an intervenor. Patrick Mello relates military intervention to a concept of ʻmilitary participationʼ (i.e. military engagement within a ʻmultilateral military operationʼ).[19] Just a few authors are trying to empirically investigate other (non-military) forms of intervention.[20] Last but not least, there is a tradition of humanitarian military intervention scholarship that introduces different definitions and different understanding of the conditions causing them.[21]

Although the non-military forms of intervention represent an interesting field of further research, we have decided to focus solely on the military forms as defined by UCDP/PRIO. Therefore, we understand the intervention in military, state-centric terms, where the intervenor represents astate that actively participates in conflict and side with one primary actor of the conflict with its military units .[22] This definition is easily distinguishable in comparison with other, not so flagrant methods of external support or influence.[23] Moreover, we assume that conditions for the military forms may substantially differ vis-à-vis to the non-military forms, their differentiation is therefore necessary. Finally, as it is clear from the definition, we understand interventions in state-centric way without the focus on international organisations. Whereas there is a mainstream consensus on military character of interventions, the debate about the conditions of interventions is less consistent as we are going to show in the rest of the theoretical part.

During the Cold War, reasons for interventions were understood in a logic of superpowers’ competition and proxy wars as (for example) Herbert Tillema’s study showed. [24] The importance of the Cold War was also stressed in Patrick M. Regan’s influential study of unilateral military and economic interventions, according to which interventions were more likely (1) during the Cold War; (2) with a lower intensity of the conflict; and (3) with large social dislocations and humanitarian issues.[25] The Cold War and proxy-war politics was a thoroughly scrutinised and highlighted factor but how then to explain the reasons for interventions after the Cold War? There are lots of different answers – for example, David Carment and Patrick James[26] and later Carment, James and Zeynep Taydas[27] investigated the role of domestic politics and the internationalisation of ethnic conflicts. Jacob Kathman[28] focused on potential war contagion and regional motivation as reasons for intervention. Mi Yung Yoon[29] examined, quite uncommonly, interventions of African states into African conflicts. She tested several variables (geographic proximity, economic motives, ethnic affinity or security concerns, or the presence of an internal conflict in potentially intervening countries) and found that the most important factors consisted of economic motives and geographic proximity. Her findings are therefore in sharp contrast with those of Regan.[30] The inconsistency, however, could be a result of the geographical focus of her research on African states’ interventions.

Although the abovementioned authors concentrated on various supposed causes of interventions, they resigned to pay attention to possible interactions and interrelations among the independent variables. Therefore, we will firstly draw our hypotheses following the previous findings of the mentioned authors, and then we will focus on the variables’ interactions.

Hypotheses

The first hypothesis is based on Regan’s geographic proximity thesis which he operationalised through a number of neighbouring countries.[31] However, we consider the geographic proximity and the number of neighbouring countries as two different factors which may, but also may not, overlap. From the reasons stated below, we consider the number of neighbouring countries as more coherent with the theoretical assumptions and therefore we won’t confuse it with the geographic proximity. Firstly, according to Regan’s hypothesis, intervention by neighbouring states should be of greater value.[32] Similarly, Yoon (as she analysed interventions of African states in Africa) stressed the stronger interaction of states with their neighbours.[33] Secondly, states are also believed to be more prone to intervene in the internal conflicts within neighbouring countries because of the possibility of war contagion, as Regan[34] or Kathman[35] argued. This could be true especially in the case of Africa (or more generally in the ‘Global South’) where states are mainly oriented towards their closest neighbourhood in seeking for their own as well as for the regional security policy,[36] whereas ‘First World’ countries could take a safe neighbourhood for granted. Mohammed Ayoob[37] argues that this is a consequence of the ‘weak state-structure’ in the ‘Global South’ that leads to spill-over effects from internal conflicts into the nearby region. In dealing with proximity there could also be an important distinction between the conditions important for African states and their non-African counterparts, especially European states. Unfortunately, the selected method does not allow us to catch this difference because the variables must be coded both for interventions and for non-interventions, and for the latter, the ‘African vs. non-African intervenor’ variable would be pointless. Nevertheless, we will at least partially focus on this issue within the two short case studies of the UK’s intervention in Sierra Leone and France’s intervention in Mali at the end of the discussion part.   

The first hypothesis thus states that: (H1) Increasing the number of neighbouring states with a conflict-ridden country increases willingness to intervene.

The second hypothesis is based on economic interests, especially on the attraction of raw materials and other gains that could be obtained from ‘successful’ (for the intervenor) intervention.[38] According to Yoon, the pursuit of gains substantially influences the foreign policy of states. In Africa, the gain is mainly a possibility of looting raw materials.[39] Other authors point out that intervention is the result of a decision to protect economic position, trade and interests in the country.[40] However, with regard to the usually negligible rate of cross border or cross regional trade, we have decided to stress the first argument.

The second hypothesis therefore supposes that: (H2) Increasing the presence of raw materials in a conflict-ridden state increases willingness to intervene. 

Another possible and very interesting condition is the intensity of conflict, supposing that higher intensity brings higher costs of intervention, moreover with uncertain outcomes. The costs of an uncertain adventure in the case of intervention in an intense conflict could simply be too high in comparison to possible gains.[41] On the other hand, from the point of view of humanitarian military intervention one could argue by ‘just cause’ for intervention to stop human suffering.[42] However, we stick to utility and costs-benefits rationale behind the intervenor view of intense conflicts.

Thus, the third hypothesis therefore supposes that: (H3) Increasing the intensity of a conflict decreases willingness to intervene.

Our fourth hypothesis is partly connected with the problem of human suffering (humanitarian aspect) but also with the regional contagion (security aspect). Refugee flows from a warring neighbour state can destabilise domestic policy and thus cause instability.[43] However, such massive social dislocation can also cause humanitarian concerns. Although there is a conventional link between a conflict’s intensity and social dislocations, it is important to stress here, as Regan[44] shows, that the intensity of conflict itself might not be only or an automatic source of large social dislocations.[45] That’s why we have decided not to perceive dislocation just as the consequence of the conflict intensity but to focus on conflict-driven social dislocations per se as a possible determinant of interventions. As we are using qualitative comparative analyses, which allow us to catch interconnections and interactions among conditions, the whole issue of a link among intensity, social dislocation and number of neighbouring countries seems very interesting to understand.

The fourth hypothesis supposes that: (H4) Great social dislocations during the conflict increases willingness to intervene.

The fifth hypothesis is based on the notion of ethnicity. States in Africa are generally very ethnically heterogeneous. Furthermore, ethnic groups often find themselves in several different states. Therefore, killing members of an ethnic group in a conflict-ridden state can easily affect a relative group in a neighbouring state which can consequently result in killing relatives (real or alleged) of the perpetrators for revenge. It is not easy to sit on the fence when your kin are being slaughtered in a neighbouring country. That’s why intrastate conflict has substantial potential to bring ethnicity to the daylight of the political agenda and over-border ethnic links could be a condition that changes foreign policy agenda and create another rationale for intervention.[46] Ethnic affinity (especially in connection with the imperative of stopping the suffering of relatives) thus could be the reason for intervention by virtue of the demands of an intervenor’s population. On the other hand, it could also serve as a governmental rationalisation of a performed intervention without preceding popular demand and/or with other, and far less humanitarian, intentions.

Either way, we suppose that: (H5) The great number of ethnic-relative states of a conflict-ridden country increases the probability of their intervention.

The last hypothesis focuses on the history of the state where the conflict is taking place. As a former colonial power, France has an exceptional relationship with its former colonies. Military and economic cooperation with several countries led to many French interventions. The close ties between France and African states were so significant that some authors use it as an important variable for conflict occurrence in Africa.[47] The interventions were an important part of French foreign policy in Africa[48] and they still are, though now partly concealed with the EU Common Security and Defence Policy. We assume then that in former French colonies interventions occur more often as France still significantly shows its presence in African internal conflicts.

Thus, the last hypothesis presumes that: (H6) Being a former French colony increases the willingness of France (or its allies) to intervene.[49]

Methodology – Qualitative Comparative Analyses (QCA) and operational criteria

The QCA, first presented in 1987 by Charles Ragin,[50] has brought new insights into the social inquiry which are broadly discussed to this day.[51] The QCA should be able to fill the gap between qualitative and quantitative methods in moderate-sized N research.[52] It combines the simple generalisation of quantitative methods as well as the complexity of qualitative methods.[53] The results of QCA are presented as ‘conjunctions of conditions’, thus, the method presents a configuration of mutually connected conditions.[54] Thereby conditions cannot be understood as self-contained statements that are separated from each other, quite the contrary – as complex ties among conditions, no matter the sole impact of a particular condition. Thus, on one hand, the results of QCA offer unexceptionable insight into great complexity imbued with many interesting interactions among independent variables which the conventional statistical methods are unable to deliver. On the other hand, as well as all other methods, even QCA has been subjected to a lot of criticism, especially for its supposed limited usage in the social sciences.[55] For example, Simon Hug[56] mentions that QCA can suffer from measurement errors which cannot be overcome entirely even by involvement of case studies. Such criticism is very useful to realise the limits and pitfalls of the QCA, we consider its benefits for a complex view of surveyed phenomenon to be prevailing, [57] especially dealing with issues of security.[58] Besides, a similar criticism could also be targeted on the conventional statistical methods anticipating certain quality and quantity of data which is hardly reachable within the realm of social science.

The results in QCA take the form of combinations of conditions that lead to the outcome. This means that we must perceive causality as an interplay between conditions instead of viewing them in a vacuum.[59] In this study we are using fuzzy-set QCA that is based on scaling membership in sets, i.e. we calibrate degree of a condition’s presence in a given set.[60]

This allows us to understand the continual character of conditions by addressing a membership score between 0 and 1, where 1 means full membership in the set, whereas 0 means that the condition is fully out of the set. [61] A score of 0.5 refers to a point of ‘maximum fuzziness’ (point of indifference) that means that it is neither in nor out. Points in-between a score of 1-0.5 and 0.5-0 means that it is ʻmore in than outʼ and ʻmore out than inʼ, respectively.[62] Thus, one of the most important steps in analyses is the calibration of set membership.[63], [64]

Our first hypothesis is based on the number of neighbouring countries (NNC) which we operationalise through the number of states neighbouring with the state where the conflict occurs. In doing so, we suppose (similarly to Regan[65]) that it threatens the stability of the neighbouring countries which consequently tend to stop the conflict to prevent its contagion. The calibration of this condition is then based on mathematical procedure when we operate with the z-score of all cases.[66] Points of 0.95 (full membership) and 0.05 (full non-membership) match with deviation 1.645 and -1.645 and the point of maximum fuzziness is deviation 0.

For the second hypothesis, we use the presence of raw materials (MAT) in the country, supposing that the more lootable resources a conflict-ridden state has, the more attractive it is for possible intervenors. As an indicator for this condition, we use data from the World Bank Databank called total natural resources rents as a percentage of GDP.[67] To be sure that our indicator is representative and not influenced by conflict we use resources rents of a state one year before the conflict started. The calibration is based on the World Bank publication The Changing Wealth of Nations.[68] A resource-rich country is understood to be a country with resource rents of at least 5 percent of GDP according to the World Bank.[69] Therefore, we use a dichotomous variable where less than 5 percent of GDP is coded as 0, and at least 5 percent of GDP is coded as 1.

The third hypothesis focuses on the conflict’s intensity (COINT). An intervention in intense conflicts could pose very high risks and costs. The calibration is centred around the traditional understanding of conflict intensity and divide the conflicts on wars (1) and minor conflicts (0). Thus, in case the conflict reached the level of war according to UCDP/PRIO[70] we code it as 1, in another case as 0. 

The fourth hypothesis relates to the humanitarian issues and also the possibility of a domestic instability caused by large social dislocations (REF). As an indicator, we use the number of refugees of the state in conflict. As a ‘large social dislocation’, we understand the difference of refugees’ population before a conflict and the highest number of refugees during the conflict episodes[71] having the threshold set on 50,000 refugees.[72] Thus the difference of at least 50,000 refugees we code as 1, in other cases we coded 0. We use data from World Bank[73] data that are based on the UNHCR data on refugees.[74]

The fifth hypothesis assumes that a state is more likely to intervene in a conflict-ridden state if both are inhabited (at least partially) by the same ethnic group. It partly addresses the issue of an intervenor’s proximity but not absolutely because some ethnic groups are very dispersed (for example Hausa-Fulani in West Africa). Ethnic affinity (ETH) is based on the dataset of James D. Fearon.[75] To be specific, we count states where an ethnic group endangered by a conflict is the largest or the second-largest ethnic group. Calibration is again based on z-score, as mentioned above.

The last condition that operates with the French condition (FRC) is coded in a dichotomous fashion as 1 is for former French colonies and 0 for the other.[76]

As a case, we use internal conflict based on the UCDP/PRIO data set.[77] We do not focus on conflict year or event but on conflict as a whole. To be relevant for our study the start date must be after 1 January 1989. We consider this date as the mark of transition between the Cold War and the post-Cold War period as it is a transitional year when we can trace a growing convergence between Russia and the USA.[78] The dataset consequently comprises 33 conflicts of which 13 were intervened by one or more states.

Results

First of all, in fuzzy-set QCA we have to make a test for the necessity both for presence and absence of the outcome. Analyses of necessity must be made in a separate way (see Appendix 1). The necessity test of outcome presence shows no single condition with a consistency level of at least 0.9.[79] On the other hand, the results indicate that (at the consistency level of 0.95 and coverage level of 0.73) the absence of high intensity (~COINT)[80] is the necessary condition for the absence of intervention. Also, consistency level of social dislocation’s absence (0.85) seems interesting but it has not passed the threshold of 0.9.

As the next step, we construct the ‘truth table’ for the positive and negative outcome. The program fs/qca generates three possible solutions: complex, parsimonious and intermediate solutions. As the names tell us, the complex solution is based just on empirical cases and is the most intricate. The parsimonious uses logical reminders to reduce the solution. Finally, the intermediate solution lies in between as it represents a reduced solution based on previous theoretical knowledge.[81] This combination makes it the generally preferred solution for analyses including this one.

Table 1 displays solutions and all the conjunctions that lead to the positive outcome – presence of the intervention.[82] As we mentioned above, we will mainly focus on the intermediate solution.

Tab01

The intermediate solution for positive (presence) outcome contains four paths:

  1. Low number of neighbouring countries * Great social dislocation
  2. High intensity * Great social dislocation
  3. Presence of raw materials * Great Social dislocation * Former French colony
  4. High intensity * Presence of raw materials * Great number of neighbouring countries * Absence of ethnic affinity * Former French colony

Thus, we can find four possible ways to the intervention. The coverage and consistency of the solutions is promising. The intermediate solutions as the main explanation to discuss reaches coverage level of 0.51 and consistency level of 0.91. The parsimonious solution also shows relatively high consistency and coverage even if the coverage of a complex solution drops a little bit.

Table 2 displays solutions and all the conjunctions that lead to the negative (absence) outcome, thus for the absence of the intervention. Overall, the consistency for the intermediate solution again shows sufficient levels with consistency at 0.94 and coverage slightly above 0.41. The paths for negative outcome to discuss are:

  1. Low intensity * Great number of neighbouring countries * Not a French colony
  2. Absence of raw materials * Absence of ethnic affinity * Former French colony
  3. Absence of raw materials * Low number of neighbouring countries * Low social dislocation * Former French colony
  4. Low intensity * Presence of raw materials * No ethnic affinity * Low social dislocation * Not a French colony

Tab02

Discussion – Back to the theory

In the following section, we will go through every condition and how they work in different combinations to contrast our results with the previous research. We will discuss just the intermediate solutions that are of the best analytical use. We find the results important in two regards – we have re-examined influential theoretical presumptions with the use of this unusual method and show that some of them do not match. Secondly, the results indicate that the QCA method can reflect interventions as more interconnected and complex phenomena than a purely quantitative approach of our predecessors.  

Let´s take a closer look at every condition in our study while also discussing the path they are part of. The number of neighbouring countries seems not to unequivocally support previous theoretical assumptions. If we focus on its interplay with other conditions, we can find it in two positive (i.e. the presence of intervention) and in two negative paths (i.e. the absence of intervention). Within the former, it appears in the solution with great social dislocation. A possible explanation could rest in the assumption that the lower number of neighbouring states, the larger share of refugees falls on every single state. Therefore, they are motivated to intervene in order to prevent the contagion (e.g. Guinea as the only relatively stable neighbour of conflict-ridden Sierra Leone). The second path, however, presumes the opposite effect of this variable – interventions are caused in situation of a great number of neighbouring countries, high intensity and abundant raw materials in a former French colony with low ethnic affinity. Even if this solution is too complex to call, we will discuss it below. The number of neighbouring countries is ambivalent even for a negative outcome. However, this is one of the main signs of QCA that in different combinations the same condition can lead to a different outcome. A great number of neighbouring countries don’t lead to intervention if it is accompanied by low intensity in a non-French former colony. But we can see the same outcome if the number of neighbouring countries is low, accompanied by no refugees, no raw materials and being a former French colony. It could mean that these states are on the very edge of interest, for both France and states in the region due to other conditions like absence of raw materials or low intensity.

Our analysis shows that the number of states per se could be an important factor, but its rationalisation differs from the geographic proximity as is commonly understood in theory. Although we suppose that the number of neighbours is a relevant factor (especially in combination with refugees as shown above and in the subsequent case of Sierra Leone), we suggest that the operationalisation of proximity should be addressed more profoundly in future research, especially in order to cover a distance from an intervenor to the intervened conflict, and an importance of the region for overseas actors (given by for example proximity to an important sea route,  geo-strategic position and so on).

To the second variable, the economic interests and raw materials proves to be of high importance. Therefore, Yoon’s[83] finding that raw materials are important for African states’ willingness to intervene is plausible. However, it is not a solely sufficient condition for intervention. The variable is present in two out of four positive outcomes’ paths and in three out of four negative outcomes’ paths. For the positive outcome, the raw materials are important in combination with the refugee’s flux and French colonial history that is consistent with theoretical (and empirical) expectations (e.g. French intervention in Mali – see below). The second path combines raw materials with high intensity, a great number of neighbouring countries, absence of ethnic affinity and being a former French colony. The combination could be interpreted from a costs and profits perspective: the great costs of military intervention in a highly intensive conflict are compensated with possible economic gains, and also by a threat of destabilisation of the whole region resulting from hesitation. The possibility of French intervention just increases the probable success of the intervention and thus also its utility.

For the negative outcomes, the absence of raw materials appears in combination with the absence of ethnic affinity and French colonial history. French colonial history is present also in the second solution, together with a low level of social dislocation, and a low number of neighbouring states. The last solution describes the situation when even a high amount of raw materials is not perceived as a sufficient reward to balance missing security variables (high intensity, social dislocation) or ethnic affinity.

The frequent combination of raw materials and French colonial history could indicate a possible relation between these variables, both within positive (two of four) and negative (three of four) outcomes. It seems that raw materials (and connected economic interests) could be an important factor not only for African states (as Yoon suggested[84]) but also for France as we could see for example in Mali, Niger, the Central African Republic and so on.

The intensity of conflict seems to be of high importance, mainly because it contests the results of Regan’s study.[85] Namely, we found that high intensity does not bring any exorbitantly high cost for a potential intervenor; quite the contrary, intense conflicts are more likely to be intervened. Moreover, the first combination – high intensity and large social dislocation – could indicate that a potential intervenor pays more attention to the humanitarian and security concerns than to the costs. This issue is obviously complicated to understand, and it is hard to distinguish humanitarianism from political and security goals where humanitarian arguments are only a ‘veil’ of justification. Obviously, humanitarian military intervention presents a specific kind of military intervention where other conditions, like just cause, intention, etc., are also questioned. [86]

The conflict intensity seems at least equally interesting for negative outcomes, mainly because the low intensity is the (only one) necessary condition for absence of intervention. The results of the conflict intensity, both the positive and negative outcomes, mostly contest the previous research and show that the security concerns or even humanitarian issues (together with the issue of social dislocations) are important. We have already mentioned the issue of social dislocations several times. Most of our results show that social dislocations relate to humanitarian concerns of intervening states.

The issue of ethnicity seems to be much more complicated and confusing for intervenors’ motivations. Even though ethnic affinity is included in some solutions, it seems not so strong in the explanatory ways. It is important to stress here that ethnicity in Africa is a very hard issue to study and especially hard to collect relevant data about. In general, data about ethnicity are far from being rigorous and up-to-date, which could also have affected our inconclusive findings.

Last but not least, French colonial history seems to be important. In two paths of positive solutions being a former French colony was relevant to the presence of intervention. In this case, it was accompanied by great social dislocation and (in both cases) raw materials. This means that French military relations to former African colonies remained and it is enhanced with an economic significance for France, or great social dislocation may serve as justification for the intervention.[87] However, French colonial history is neither a solely necessary condition nor a solely sufficient condition.

French colonial history is important also for understanding the absence of intervention since it is present (always with absence of raw materials) in two out of four negative paths. A possible explanation could rest in a reluctance of prospective intervenors to interfere in matters of France’s ‘clients’, expecting France to intervene instead. In combination with a low level of raw materials and social dislocation, an intervention is not worthy, not even for France. In former non-French colonies, intervention does not occur even if abundant material resources are present together with low conflict intensity, low social dislocation and no ethnic affinity.[88]

To sum up, our results show great complexity of the decision to intervene or not. States must consider costs and benefits whereas every single factor added could substantially change the equation through various interconnections with the others. That’s why the conditions must be seen in the mutual interactions and not separately. However, resulting equations may then appeal rather theoretically and even artificially. That’s why we are going to show their logic in two brief examples of military intervention - Sierra Leone and Mali.

In the case of Sierra Leone, there were three intervening states - Guinea, Nigeria and later also the UK. There are two possible explanations for the intervention – interplay of (1) a low number of neighbouring countries and great social dislocation, or (2) great intensity and great social dislocation.

For Guinea - as for one of the two neighbouring countries - the fear of great social dislocation and possible contagion of the very intensive conflict played a significant role for the decision to intervene, as for example David Keen proves.[89] Moreover, Sierra Leone’s second neighbour, Liberia, was far from being a safe haven for immigrants since the whole country had been inflicted with civil war. Guinea, as the only one relatively safe neighbouring country, was therefore legitimately afraid of contagion and a great refugee influx, especially when over 140,000 refugees from Sierra Leone came in just the first year of the more than decade-long civil war (1991–2002).[90] Both explanations are therefore plausible.

The second intervenor, Nigeria, is not a neighbouring country to Sierra Leone, but it does not necessarily mean that the low number of neighbours in combination with social dislocations were not relevant factors. Nigeria could be concerned with a possible influx of immigrants from Sierra Leone (and Liberia) especially with regard to bitter experiences from clashes between Nigerians and immigrants (for example with Liberians in Oro). It is important to stress, however, that the number of Sierra Leonean refugees in Nigeria was negligible in comparison to Guinea. The threat of social dislocation in combination with a high intensity of conflict and/or with a low number of neighbouring states should be rather perceived in a context of the Nigerian hegemonic position in West Africa. Nigeria has often portrayed itself as a regional peace-keeper and as a champion of the English-speaking countries of West Africa against the threat of French interventional politics. A conflict of high intensity in an English-speaking country would notably constest Nigeria’s reputation.

Since the decision to intervene is quite multi-causal, there were many reasons beyond our analysis – especially the role of personal relationships and animosities among West African presidents.[91] Their real impact on the decision, however, could be hardly measured and incorporated into the model.  

Intervention by the UK cannot be explained by the first solution, as it was clearly not threatened by refugee flows or conflict contagion, but the second solution seems plausible. The incentives for interventions might therefore lie in an instability that threatened the UK’s investments in Sierra Leone and the broader region all together with an imperilment of the UK citizens in Sierra Leone which helped with the justification of intervention.[92] Great social dislocations played a crucial role by spilling the conflict over the borders and having a substantially affected majority of countries within the region, including Ghana - an important partner of the UK. The UK’s intervention could be therefore explained as a reaction to great instability (caused by the high intensity of the conflict and great social dislocations) that jeopardised its interests in the region.

The combination of high intensity and great social dislocations was also important in the case of the military intervention to Mali in 2013 that followed the civil war which started in 2009.[93] Mali has kept quite close trade and political ties to France, thus representing a typical example of the former French colony condition. France has traditionally been one of the biggest importers to Mali, although the export to Mali in 2012 represented just 0.079% of total French export and Malian export to France was negligible.[94] However, Mali’s importance for France is not been derived from total capacity of trade, but rather from its composition because Mali (together with Niger) are important sources of raw materials, especially uranium. Moreover, stability in the area of Malian and Nigerien Sahel is crucial for the stability of Algeria, which is an important source of crude oil and petroleum gas.[95] Also as Douglas Yates aptly writes: ‘the economic importance of Africa to France’s African policy must be understood as less about its macroeconomic importance to France as a whole than about its importance to a small predatory lobby of influential French profiteers’.[96] Therefore, the reasons why France so swiftly intervened, besides the Malian government’s request for the intervention in January 2013,[97] rested primarily in the quest for stability in the Sahelian part of Mali, Niger and Algeria. Although uranium (or raw materials in general) in Mali itself could contribute to the decision, it didn’t play a crucial role vis-à-vis the risk of contagion and spreading the conflict into the whole region. The Malian case could be therefore explained both by the combination of high intensity and great social dislocation, destabilising the region of France’s interest, and by the combination of raw materials, great social dislocation and a former French colony (the request of Malian government). In the case of Mali, the two combinations are very close to each other.    

For the minor intervening states (mostly states of West Africa), the reasons rested mostly in the imminent threat of contagion of a high intensity conflict with great social dislocation, especially after the bitter experience from the civil wars of the 1990s that had spread so quickly over the region. The possible contagion through the Sahelian and Saharan area was so dangerous as the area is very hard to control. [98]

In the beginning, we argued that intensity is more likely seen as an obstacle to intervention. In the end situation seems to be different. Great intensity of conflict together with great social dislocation seems to be important from regional stability point of view. It is not the bloody conflict alone. Conflict needs other reasons to pull the attention of the states. An interesting fact is that great intensity does not show together with great social dislocation every time in the solutions and the interplay between these two variables should be investigated. The number of neighbours or as some use it as a proxy for geographic proximity still seems to be confusing. Also, it is important to mention that the reasons for intervention probably differs for African and extra-African countries as was the French and the UK case in Mali or Sierra Leone. As we have already mentioned, ʻFrench colonial historyʼ condition is not solely a necessary nor sufficient for intervention therefore France still needs other pull factors to participate in conflict.

To sum up, our results show great complexity as the decision to intervene or not. States must consider costs and benefits whereas every single factor added could substantially change the equation through various interconnection with the others. That’s why the factors (conditions) must be seen in the mutual interactions and not separately.

Conclusion

The reasons for interventions into African internal conflicts need to be perceived as a complex and interconnected phenomenon. We pointed out this complexity using the QCA method which revealed some interesting results. Firstly, in sharp contrast to previous assumptions, we showed that the high intensity of conflict and large social dislocations led to external intervention. On the other hand, intervention was generally unlikely in the conflicts of low intensity. Secondly, occurrence of raw materials (and economic interests of the intervening states in general) played a very important role for the decision to intervene not just in the case of African states (as for example Yoon[99] presumed) but also in the case of France.

To assign appropriate motives to these factors is very complicated and not the goal of the article. Rather, we have focused on conditions under which interventions occur. Nevertheless, we could conclude  that balancing possible costs and benefits is important for intervenors even though the complexity of the solutions is high.

Despite many limitations (especially representativeness of data or static character of QCA[100]), the results and the research design bring a specific understanding of the issue of military interventions in Africa after the Cold War. With this article we contribute to the growing number of research based on a configurational approach. QCA as a method offers an important insight into the study of security in Africa as it is highly complex. Military interventions are not easily grasped by one factor. Rather we have to understand the tangled net of relations between several conditions as in our example where a mutual influence of high intensity, great social dislocation or raw materials contributed to presence of interventions.

This work is part of the project "Internal conflicts in Sub-Saharan Africa" and it was supported by grant of the Specific research projects of Philosopical Faculty of University of Hradec Kralové.

Martin Schmiedl is affiliated with the Department of Politics of the Philosophical Faculty of University of Hradec Králové and can be contacted at martin.schmied@uhk.cz.

Jan Prouza is affiliated with the Department of Politics of the Philosophical Faculty of University of Hradec Králové and can be contacted at jan.prouza@uhk.cz.

Appendix

APP01

App02

App03

 

Endnotes

[1] Ibrahim A. Elbadawi and Nicholas Sambanis (2000), External Interventions and the Duration of Civil Wars. World Bank Policy Research Working Paper.; Dylan Balch-Lindsay and Andrew J. Enterline (2000), Killing time: The world politics of civil war duration. International Studies Quarterly 44(4), pp.  615–642.; in civil wars,” Journal of Conflict Resolution 50(5), pp. 736–756.Patrick M. Regan (2002a), “Third party interventions and the duration of intrastate conflicts,” Journal of Conflict Resolution 46(1), : 55–73.; Patrick M. Regan (2010), “Interventions into civil wars: A retrospective survey with prospective ideas,” Civil Wars 12(4), pp. 456–476.; Dylan Balch-Lindsay, Andrew J. Enterline and Kyle A. Joyce (2008), “Third party intervention and Civil War Process,” Journal of Peace Research 45(3), pp. 345–363

[2] E. g. David T. Mason & Patrick J. Fett (1996), “How civil wars end,” Journal of Conflict Resolution 40(4), pp. 546–568.; Bruce Bueno de Mesquita and George W. Downs (2006), “Intervention and democracy,” International Organization 60(3), pp. 627–649.; Balch-Lindsay, Enterline and Joyce (2008), pp. 345–363.

[3] E. g. Patrick M. Regan (1996), “Conditions of successful third party intervention in intra-state conflicts,” Journal of Conflict 40(1), pp. 336–359.; David Carment and Patrick James (1996), “Two-Level games and third-party intervention: Evidence from ethnic conflict in the Balkans and South Asia.! Canadian Journal of Political Science 29(3), pp. 521–554.; David Lektzian and Patrick M. Regan (2016), “Economic sanctions, military interventions, and civil conflict outcomes,” Journal of Peace Research 53(4), pp. 554–568.

[4] Carment and James (1996), pp. 521–554.; Mi Yung Yoon (1997), “Explaining U.S. intervention in third world internal wars, 1945–1989,” The Journal of Conflict Resolution 14(4), pp. 580–602.; Patrick M.  Regan (1998), “Choosing to intervene: Outside interventions in internal conflicts,” Journal of Politics 60(3), pp. 754–779.; Patrick M. Regan (2002), Civil wars and foreign powers: Interventions and intrastate conflict. Ann Arbor: University of Michigan Press.; Jacob D. Kathman (2011), “Civil war diffusion and regional motivations for intervention,” Journal of Conflict Resolution 55(6), pp. 847–876. Jacob D. Kathman (2010), “Civil war contagion and neighbouring interventions.” International Studies Quarterly 54(4), pp. 989–1012.; Patrick A. Mello (2014), Democratic Participation in Armed Conflict. New York: Palgrave Macmillan.

[5] The same geographical focus could be found in a few earlier authors but with completely different methodological standpoints – as shown below – see Mi Yung Yoon (2005), “Internal conflicts and cross-border military interventions in Sub-Saharan Africa in the Post-Cold War era.” Journal of Political and Military Sociology 33(2), pp. 277–293; Elizabeth Schmidt (2013), Foreign Intervention in Africa: From the Cold War to the War on Terror. Cambridge: Cambridge University Press.; Irit Back (2016), Intervention and Sovereignty in Africa: Conflict Resolution and International Organizations in Darfur. London: I. B. Tauris.;

, Back 2016 or more generally Reno 2011, Young 2016

[6] Richard Jackson (2002), “Violent internal conflict and the African state: Towards a framework of analysis,” Journal of Contemporary African Studies 20(1), p. 32.

[7] Ibid, p. 33.

[8] See for example Mello (2014).

[9] Karel Kouba (2011), “Kvalitativní srovnávací analýza (QCA) a konfigurativní metody v politologii,” in Miroslav Novák (ed.) Úvod do studia politiky, Praha: Sociologické nakladatelství, pp. 473–474; Bernard Grofman and Carsten Q. Schneider (2009), “An introduction to crisp set QCA, with a comparison to binary logistic regression,” Political Research Quarterly 62(4), p. 666.

[10] This is also connected with the limits of the used method and level of analysis.

[11] Martha Finnemore (2003), The Purpose of Intervention: Changing Beliefs About the Use of Force. Ithaca: Cornell University Press, p. 7.

[12] James Rosenau (1969), “Intervention as a scientific concept,” The Journal of Conflict Resolution 13(2), p. 161.

[13] Ibid., p 167.

[14] Richard Little (1975), Intervention: External Involvement in Civil Wars. New Jersey: Rowman and Littlefield, p. 2.

[15] Karen Feste (1992), Expanding the Frontiers: Superpower Intervention in the Cold War (Boulder, CO: Praeger Publishers cited in Regan (2010), p. 458.

[16] Frederik S. Pearson (1974), “Geographic proximity and foreign military intervention,” Journal of Conflict Resolution 18(3), pp. 432–460.; Herbert K. Tillema (1989), “Foreign overt military intervention in the nuclear age,” Journal of Peace Research 25(2), pp. 179–196.; Jeffrey Pickering and Emizet F. Kisangani (2009), “The international military intervention dataset: An updated resource for conflict scholars,” Journal of Peace Research 46(4), pp. 589–599.; Stephen E. Gent (2007), “Strange bedfellows: The strategic dynamics of major power military interventions,” Journal of Politics 69(4), pp. 1089–1102.; Stephen E. Gent (2008), “Going in when it counts: Military intervention and the outcome of civil conflicts,” International Studies Quarterly 52(4), pp. 713–735.; Patricia L. Sullivan and Michael T. Koch (2009), “Military intervention by powerful states, 1945–2003,” Journal of Peace Research 46(5), pp. 707–718.

[17] Regan (1998), pp. 754–779; Elbadawi and Sambanis (2000).

[18] Finnemore (2003); Martha Finnemore (2009), “Legitimacy, hypocrisy, and the social structure of unipolarity: Why being a unipole isn´t all it´s cracked up to be,” World Politics 61(1), pp. 58–85.

[19] Mello (2014), p. 4.

[20] Patrick M. Regan and Aysegul Aydin (2006), “Diplomacy and other forms of intervention in civil wars,” Journal of Conflict Resolution 50(5), pp. 736–756.; Anke Hoeffler (2014), “Can international interventions secure the peace?,” International Area Studies Review 17(1), pp. 75–94.

[21] Taylor Seybolt (2007), Humanitarian Military Intervention: The Conditions for Success and Failure. Oxford: Oxford University Press.; John M. Kabia (2009), Humanitarian Intervention and Conflict Resolution in West Africa: From ECOMOG to ECOMIL. Burlington: Ashgate.

[22] Lotta Themnér (2016), UCDP/PRIO Armed conflict dataset codebook. Uppsala: Uppsala Conflict Data Program (UCDP).

[23] Cf. Patrick M. Regan (2002), Civil wars and foreign powers: Interventions and intrastate conflict. Ann Arbor: University of Michigan Press, pp. 9-10.

[24] Tillema (1989), pp. 179–196

[25] Regan (1998), pp. 774–777.

[26] Carment and James (1996), pp. 521–554

[27] David Carment, Patrick James & Zeynep Taydas (2009), “The Internationalization of ethnic conflict: State, society, and synthesis,” International Studies Review 11(1), pp. 63–86.

[28] Kathman (2010), pp. 989–1012; Kathman (2011), pp. 847–876.

[29] Mi Yung Yoon (2005), “Internal conflicts and cross-border military interventions in Sub-Saharan Africa in the Post-Cold War era.” Journal of Political and Military Sociology 33(2), pp. 277–293.

[30] Regan (1998), pp. 774–777.

[31] Regan (2002), p. 49.

[32] Regan (1998), p. 765.

[33] Yoon (2005), p. 280.

[34] Regan (2002), p. 49.

[35] Kathman (2011), p. 851.

[36] Mohammed Ayoob (1989), “The Third World in the System of States: Acute Schizophrenia or Growing Pains?,” International Studies Quarterly 33(1), p. 70.; Barry Buzan, and Ole Waever (2003), Regions and Powers: The Structure of International Security. Cambridge: Cambridge University Press, pp. 229–232.

[37] Ayoob (1989), p. 71.

[38] Yoon (2005), p. 280.; Michael G. Findley and Josiah F. Marineau (2015), “Lootable resources and third-party intervention into civil wars,” Conflict Management and Peace Science 32(5), pp. 465–486; Vincenzo Bove, Kristian S. Gleditsch and Petros G. Sekeris (2015), ““Oil above water”: economic interdependence and third-party intervention,” Journal of Conflict Resolution 60(7), pp. 1251–1277.

[39] Yoon (1997), p. 582.

[40] Aysegul Aydin (2012), Foreign powers and intervention in armed conflicts. Stanford, California: Stanford Security Studies.; Benjamin O. Fordham (2007), “Revisionism reconsidered: Exports and American intervention in World War I,” International Organization 61(2), pp. 277–310.; Benjamin O. Fordham (2008), “Power or plenty? Economic interests, security concerns, and American intervention,” International Studies Quarterly 52(4), pp. 737–758.

[41] Regan (1998), pp. 766–777; Regan (2002), pp. 49–50; Seybolt (2007), p. 15

[42] Seybolt (2007), p. 25.

[43]  IdeanSalehyan and Kristian Skrede Gleditsch. (2006). “Refugees and the Spread of

Civil War” International Organization 60(2), p. 360

[44] Regan (1998), p. 767.

[45] As an example Regan uses the first Shaba Crisis in Zaire, which has a small intensity but large social dislocation (Regan (1998), p. 767). A similar problem could be seemed in present day Nigeria where intensity of conflict (measured by number of deaths) is substantially declining from 2014 but the number of refugees is significantly rising from the same year (UNHCR Nigeria (2017), “Nigeria situation,” < http://data2.unhcr.org/en/situations/nigeriasituation> (accessed 25 August 2017).; Uppsala Conflict Data Program (2017), “UCDP Conflict Encyclopedia: www.ucdp.uu.se, Uppsala University,” <http://www.pcr.uu.se/research/ucdp/database/> (accessed: 25 August 2017). ,). 

[46] Carment and James (1996), p. 524; Regan (1998), pp. 755–756; Yoon (2005), p. 279; Carment. James and Taydas 2009. p. 79.

[47] Paul Collier, Anke Hoeffler and Dominic Rohner. 2009. “Beyond greed and grievance: feasibility and civil war,” Oxford Economic Papers 61(1), p. 10.

[48] Jackson (2002), p. 33.

[49] Obviously, we are limited to smaller set of conditions also due to methodological reasons. Other diverse explanations could exist. Mainly, we could mention those on the side of the intervenor or which could be derived from the character of international arena, e.g., interventions during Cold War or domestic affairs as we have mentioned earlier. In case we understand intervention as a specific kind or extension of internal conflicts we could also derive some arguments from IPE literature (we already use some of these arguments in case of condition Former French colony) and apply them to interventions - for example prices of resources which could shorten conflicts and therefore prevent intervention or ethnic fractionalisation in conflict-ridden country that could raise the costs of intervention and therefore prevent it, see Paul Collier, Anke Hoeffler and Måns Söderbom (2004), On the Duration of Civil War, Journal of Peace research 41(3), pp. 253–273. However, as mentioned we are methodologically limited and therefore decided to focus on conditions on side of conflict country,

[50] Charles Ragin (2014), The comparative method: Moving beyond qualitative and quantitative strategies: with a new introduction. Oakland: University of California Press.

[51] See Grofman and Schneider (2009), pp. 662–672.; Benoît Rihoux and Axel Marx (2013), “Qualitative comparative analysis at 25: State of play and agenda,” Political Research Quarterly 66(1), pp. 167–171; Patrick Emmenegger, Jon Kvist and Svend-Erik Skaaning (2013), “Making the most of configurational comparative analysis: An Assessment of QCA applications in comparative welfare-state research.” Political Research Quarterly 66(1), pp. 185–190.

[52] Charles Ragin and David Shulman, et al. 2003. “Complexity, generality, and qualitative comparative analysis,” Field Methods 15(4), p. 324.

[53] Ragin (2014), x–xi.

[54] Grofman and Schneider (2009), p. 666

[55] Seungyoon Sophia Lee (2013), “Fuzzy-set Methods in Comparative Social Policy: a critical introduction and review of the applications of the fuzzy-set method,” Quality & Quantity 47(4), pp. 1919–1920.

[56] Simon Hug (2013), “Qualitative Comparative Analysis: How Inductive Use and Measurement Error Lead to Problematic Inference,” Political Analysis 21(2), pp. 252–265

[57] Lee (2013), pp. 1919–1920.; Michal Mochťák (2013), “Inspiring QCA: An Alternative Approach for Security Studies, “Peace and Conflict Review 7(2), pp.

[58] Michal Mochťák (2013), “Inspiring QCA: An Alternative Approach for Security Studies, “Peace and Conflict Review 7(2), pp. 68–84.

[59] Emmenegger and Skaaning (2013), p. 186

[60] Carsten Q. Schneider and Claudius Wagemann (2012), Set-Theoretic Methods for the Social Sciences. Cambridge: Cambridge University Press., pp. 27–30.

[61] Charles Ragin and Paul Pennings (2005), “Fuzzy sets and social research,” Sociological Methods & Research 33(4), pp. 423–425.

[62] Charles Ragin (2008), Redesigning Social Inquiry: Fuzzy Sets and Beyond. Chicago, London: University of Chicago Press, pp. 29–31.

[63] Schneider and Wagemann (2012), p. 32.

[64] For calculation of results we use software fs/QCA 3.0 (Charles Ragin and Sean Davey (2016), Fuzzy-Set/Qualitative Comparative Analysis 3.0. Irvine, California: Department of Sociology, University of California.).

[65] Regan (1998), p. 49.

[66] We use mathematical procedure to standardise raw data according to z-score. For example, according to some authors it is a possible way to substitute a priori theoretical knowledge of necessary points of 0.95 and 0.05 for calibration (cf. Mello (2014) or Jaroslav Bílek (2017), “Authoritarian Repressive Strategies and Electoral Competitiveness.” Sociológia - Slovak Sociological Review 49(4), pp. 343–368).

[67] World Bank (2017), “World Bank Databank. < http://data.worldbank.org/> (accessed 7 June 2017).

[68] World Bank (2011), The Changing Wealth of Nations. Washington DC: World Bank.

[69] Ibid, p 10.

[70] UCDP/PRIO. (2016), “UCDP/PRIO Armed Conflict Dataset version 4-2016,” http://ucdp.uu.se/: http://ucdp.uu.se/downloads/ucdpprio/ucdp-prio-acd-4-2016.xlsx (accessed 9 October 2016).

[71] We count every year from the year that represents the start date in the UCDP/PRIO dataset. We count data also from the years between conflict years but only in the situation when there is no end for the episode. When there is the end to the episode, we do not count years up to the next conflict year.

[72] The calibration is based on (Regan (1998), p 771).

[73] World Bank (2017), “World Bank Databank. < http://data.worldbank.org/> (accessed 7 June 2017).

[74] In case of missing data we impute them from the UNData (UN Statistical Division (2017), “UNdata: A World of information,” <http://data.un.org/>  (accessed 25 August 2017). ). In other cases, we count it as zero as in the closest year the population of refugees is close to zero so we can expect to be as such.

[75] James D. Fearon (2003), “Ethnic and cultural diversity by country,” Journal of Economic Growth 8(2), pp. 195–222.; James D. Fearon (2003a), “Replication data for: Ethnic and Cultural Diversity by Country,” Journal of Economic Growth 8(2), https://web.stanford.edu/group/fearon-research/cgi-bin/wordpress/paperspublished/journal-articles-2/ (accessed 13 September 2016).

[76] We count Libya as French colony as France ruled Fezzan and nowadays Libya is part of French power interests. We did the same with Cameroon where half of the country was in French colonial empire.

[77] UCDP/PRIO. 2016. “UCDP/PRIO Armed Conflict Dataset version 4-2016,” http://ucdp.uu.se/: http://ucdp.uu.se/downloads/ucdpprio/ucdp-prio-acd-4-2016.xlsx  (accessed 9 October 2016).; Erik Melander, Therése Pettersson and Lotta Themnér (2016), “Organized violence, 1989-2015,” Journal of Peace Research 53(5), pp. 727–742; Nils Petter Gleditsch, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard Strand (2002), “Armed Conflict 1946-2001: A New Dataset,” Journal of Peace Research 39(5), pp. 615–637.

[78] Regan (1998), p. 771.

[79] Generally, for conditions to be necessary it should show high (above 0,9) consistency and also at least 0,5 coverage (Nicholas Legewie (2013), “An Introduction to Applied Data Analysis with Qualitative Comparative Analysis (QCA),” Forum Qualitative Sozialforschung / Forum: Qualitative Social Research 14(3), Art. 15. <online: http://nbn-resolving.de/urn:nbn:de:0114-fqs1303154> (accessed 5 September 2017).

[80] The symbol “~” means absence of the cause or outcome.

[81] Legewie (2013); Mello (2014), pp. 59–60.

[82] In case of the tied prime implicants we have made decision according to the theoretical knowledge (Legewie (2013)).

[83] Yoon (2005), p. 288.

[84] Yoon (2005)

[85] Regan (1998)

[86] Seybolt (2007), pp. 7–28.

[87] Cf. Douglas A. Yates (2018) Paradoxes of Predation in Francophone Africa. International Journal of Political Economy, 47(2), pp. 130-150,

[88] We have also tested an alternative operationalisation of the French influence using la Francophonie including also former Belgian colonies. The results, however, showed lower coverage in both positive and negative solutions, and the interplay between presence of raw materials and French colonial past could be explained in the same way as in the case of the former operationalisation. Thus, we have decided to use the former operationalisation to enhance the model’s coverage.

[89] David Keen (2005), Conflict & Collusion Sierra Leone. New York: Palgrave, p. 36.

[90] UN Statistical Division (2017), “UNdata: A World of information,” <http://data.un.org/> (accessed 25 August 2017).

[91] Some authors mention also some domestic reasons why it was useful for Abacha to take part in the Sierra Leone conflict for more information (see McGregor 1999).

[92] Michael Kargbo (2012), “International Peacebuilding in Sierra Leone: The Case of the United Kingdom,” in Tunde Zack-Williams (ed.) When the State Fails: Studies on Intervention in the Sierra Leone Civil War, London: Pluto Press, pp 65–69.; Andrew M. Dorman (2009), Blair's Successful War: British Military Intervention in Sierra Leone. Farnham, Surrey: Ashgate, 65–66.

[93] UCDP/PRIO (2016). “UCDP/PRIO Armed Conflict Dataset version 4-2016,” http://ucdp.uu.se/: http://ucdp.uu.se/downloads/ucdpprio/ucdp-prio-acd-4-2016.xlsx  (accessed 9 October 2016).; Melander, Pettersson and Themnér (2016), pp. 727–742; Gleditsch, Wallensteen, Eriksson, Sollenberg, and Strand (2002), pp. 615–637.

[94] Alex J. G. Simoes and César A. Hidalgo (2011), “The Economic Complexity Observatory: An Analytical Tool for Understanding the Dynamics of Economic Development. Workshops at the Twenty-Fifth AAAI Conference on Artificial Intelligence.” https://atlas.media.mit.edu/en/ (accessed 27 August 2017)

[95] Ibid.

[96] Yates (2018), p. 144.

[97] Baz Lecocq and Georg Klute (2013). “Tuareg separatism in Mali.” International Journal 68(3), pp. 424–434.

[98] For further discussion of Niger and its involment in intervention see for example Martin Schmiedl (2019), Důvody Nigeru k účasti na intervenci v Mali: heterarchie v oblasti Sahelu/Sahary a regionální bezpečnost. Politické vedy (22)3, pp. 164-191

[99] Yoon (2005), pp. 277–293.

[100] The static character of QCA does not allow us to understand the dynamics of decision-making, especially related to an issue of possible endogeneity of intervention and conflict intensity.