## Abstract

Ideas shape human behavior in many circumstances, including those involving political violence. Yet they have usually been underplayed in studies of the causes of armed mobilization. Likewise, emotions have been overlooked in most analyses of intrastate conflict. A mixed-methods analysis of Italian resistance during the Fascist regime and the Nazi occupation (1943–45) provides the opportunity to theorize and analyze empirical evidence on the role of indignation and radical ideologies in the process of armed mobilization. These nonmaterial factors play a crucial role in the chain that leads to armed collective action. Indignation is a push factor that moves individuals away from accepting the status quo. Radical ideologies act as pull factors that provide a new set of strategies against the incumbent. More specifically, detachment caused by an emotional event disconnects the individual from acceptance of the current state of social relations, and individuals move away from the status quo. Ideologies communicated by political entrepreneurs help to rationalize the emotional shift and elaborate alternative worldviews (disenchantment), as well as possibilities for action. Finally, a radical ideological framework emphasizes normative values and the conduct of action through the “anchoring” mechanism, which can be understood as a pull factor attracting individuals to a new status.

## Introduction

In his famous essay The Rebel, Albert Camus wrote: “Not every value entails rebellion, but every rebellion tacitly invokes a value.” In a novel on the Italian resistance against the Nazi-fascist forces, Italo Calvino wrote that “all of us have a secret wound we fight to redeem.”1 Ideas shape human behavior in many circumstances, including those involving political violence. Yet, they have usually been underplayed in academic literature dealing with the causes of civil wars. Likewise, emotions have been overlooked in most analyses of intrastate conflict, despite some valuable but isolated studies on their role in ethnic violence.2 Recently, political scientists have (re)discovered the importance of introducing ideas and emotions into explanations for internal armed conflict.3

In this article, we argue that the process leading away from acceptance of the status quo toward collective armed mobilization is a much more complex issue than a simple individual evaluation of expected economic or political benefits. We analyze this process in detail and show that emotions and ideologies play essential roles in causing the outbreak of civil war as a consequence of political and economic inequality. Some emotions act as “push” factors, detaching individuals from the previous status quo. Through political entrepreneurs, such as leaders of rebel movements, ideologies act instead as “pull” factors, translating private grievances into public grievances and providing individuals with specific group identities as well as specific ways to transform the current state of affairs.

Moreover, we posit that not all emotions and ideologies are equally fungible in sustaining armed mobilization. Rather, we show that indignation and radical ideologies are crucial nonmaterial factors for violent collective action. Indignation is a powerful trigger of armed mobilization because it impinges on the links between individuals and the people around them, relying on a shared conception of what is right and what is wrong. Radical ideological networks supporting the use of violence against the status quo have a crucial role in causing armed mobilization because of the specific content of their doctrines. By contrast, other ideological networks do not achieve the same results.

In the first part of this article, we present a brief review of recent theory on armed mobilization and elaborate our own theoretical framework. In the following sections, we test our theoretical statements in a prominent historical case: the Italian resistance movement that rose against Fascist and Nazi forces, leading to a civil conflict that resulted in more than 100,000 casualties from September 1943 to April 1945.

Thanks to newly collected data, we have been able to operationalize indignation and the presence of ideological networks in Italy, performing at different levels two distinct types of statistical analysis. First, we studied the impact of indignation and different ideologies on the number of partisan groups in the Italian provinces, covering the whole country in our study. This approach gave us the opportunity to highlight a pattern among the provinces. Second, utilizing more fine-grained data, we studied the impact of those nonmaterial factors on the number of partisans who took up arms in two provinces where the resistance movement was particularly active: Cuneo and Savona. Using the records of all partisans and individuals controlled by the Fascist police in those two provinces, in this part of the analysis we focused on two neighboring northern provinces with similar war experiences but different socioeconomic backgrounds. Our findings on the impact of indignation and radical ideologies hold at both levels of analysis even when we control for spatial interdependence as well as for alternative possible explanations of armed mobilization.4 Finally, we used interviews with former partisans to provide a deeper understanding of the role of radical ideologies and indignation in the process of armed mobilization and to have a better grasp of how the mechanisms worked in real life.

## Material and Nonmaterial Sources of Mobilization

A notable part of the literature on the onset of civil wars is based on large-N cross-country analyses using a structural approach to assess the impact of material factors on the likelihood of conflict onset. Structural factors can be economic, political, or geographic features of the country,5 but they all provide opportunities to initiate a civil war for the actors who intend to reach their goals through the use of armed force. The recent strand of studies using spatially disaggregated data moved the level of analysis much closer to the individual, but most of them apply a similar approach, focusing on contextual factors that provide incentives for the perspective fighters to take up arms.6 The majority of the contextual factors highlighted in the literature constitute permissive causes for the occurrence of civil wars, and Stathis Kalyvas suggests that we should “bracket [the] question of individual motives and attitudes” in favor of theories that emphasize the more easily observable structure of opportunities and constraints facing civilians in wartime.7

Nevertheless, in this article we focus on agency and on motivations for engaging in conflict. Concerning the reasons that drive individuals to fight, the first wave of empirical studies on the causes of civil wars supported the importance of “greed” as the main driving force behind intrastate conflicts.8 Accordingly, the rebel economicus weighs the expected economic benefits of joining the risky activity represented by armed rebellion against those that follow from a safer, normal life. When rebel leaders appeal to specific ethnic, religious, or class identities, they do it to increase the cohesion of their forces in a purely instrumental way.9

Social and political grievances have been reconsidered by recent studies that abandoned the country level to focus instead on relevant political groups as the crucial units of analysis and found that grievances are important causes of civil wars. Collective identities are a decisive component for the process of armed mobilization that leads to the outbreak of civil war. Focusing on ethnonational groups, recent advanced studies fully recover the role of economic and political grievances, vis-à-vis opportunities for conflict, in the explanation for the outbreak of a civil war. In fact, horizontal inequalities between politically relevant groups and the rest of the state have been found to be the main cause of ethnonationalist conflict.10

Yet, the process that leads from horizontal inequalities (or other tangible situations) to civil conflict through the formation of grievances and collective armed action is neither automatic nor simple. On the contrary, a number of factors and steps are involved in this process. Collective identities can be taken as given for analytical purposes, but group identities may be shaped and activated. Comparisons between groups involve rational calculations, but also emotional processes.11 Finally, the possibility of solving the collective-action problem depends on various types of incentives available to the prospective rebels and, of course, on contextual opportunities. Thus, a complete theory of collective action explaining how specific groups mobilize politically and how their mobilization produces violence is a much more complex issue than a simple individual evaluation of expected economic benefits. It entails a role for nonmaterial factors such as emotions and systems of beliefs.12 These have been integrated into the study of conflicts only recently. Building on the work of Lars-Erik Cederman, Kristian Gleditsch, and Halvard Buhaug,13 our research analyzes the process leading groups from acceptance of the status quo to armed mobilization and shows the essential role played by indignation and radical ideologies.

A few studies have looked at different mechanisms that make armed mobilization possible, integrating rationalist interpretations based on the availability of material incentives with explanations that allow a role for ideational motives.14 Elisabeth Wood identifies three reasons why Salvadoran peasants decided to mobilize, none of them based on the direct comparison of material costs and benefits.15 Apparently, moral or emotional considerations led many campesinos to overcome the collective-action problem. Hence, nonmaterial motives seem able to play a powerful role, even though it is sometimes difficult to trace the precise connections between initial undesirable circumstances and the outbreak of armed violence, as well as the role of different nonmaterial elements. Among those elements, ideational factors are systems of beliefs that can be expressed as structured ideologies, general ideas of justice, or religions. In the past twenty years, sociologists have investigated the role of emotions in social movements and mass mobilization, abandoning the view of emotions and rationality as opposing alternatives.16 According to this literature, emotions may motivate individuals and help mobilization, but they may also hinder it.

Wendy Pearlman argues that emotions had a fundamental role in the mobilization of the Arab masses in 2011.17 Drawing on work by Roger Petersen, Pearlman maintains that emotions can influence how people assess information, influencing both cognition and action.18 Suggestive though not systematic evidence supports the idea that emboldening emotions can drive defiance despite strategic disincentives. Much work remains to be done, however, to clarify the role of emotions as causes, effects, or mechanisms linking contentious processes. Different forms of evidence of emotions have to be considered, and the ways emotions interact with other material and nonmaterial causal factors have to be further investigated.19 If this is true for the research on mass mobilization and protest, it is even more important for the research on armed mobilization, given that civil conflict is a much more complex activity to organize. It involves, by definition, the risk of death and usually a long time horizon.

In the literature on civil wars, emotions have been directly addressed in some studies about the causes of ethnic conflict. Fear is implicitly present and essential in Barry Posen's theory, based on the application of the security dilemma to situations of internal anarchy.20 Stuart Kaufman focuses on ethnic fears and builds a theory according to which such emotions are necessary preconditions of ethnic conflict, even though they can be purposefully created by decisions of the elites.21 Petersen develops a persuasive explanatory theory of ethnic conflict based on emotions and offers testable hypotheses linking fear, hatred, resentment, and rage to violence against specific ethnic targets in multiethnic contexts.22 In this approach, emotions are triggering mechanisms that are activated by structural changes and drive action to satisfy a pressing concern. Emotions thus help scholars understand violence against particular ethnic groups and explain how ethnic conflict can also emerge from below. Omar McDoom studies ethnic conflict in the case of the 1994 Rwandan genocide and proposes to distinguish group polarization from group violence. In his theory, fear polarizes societies, but genocidal violence does not break out in the absence of material opportunities available to individual perpetrators.23 Jeffrey Kopstein and Jason Wittenberg investigate the causes of pogroms against Jews in Poland during World War II, finding that a high degree of polarization, somehow combined with hatred and rage, is strongly correlated with the occurrence of this form of violence.24

Unlike Pearlman, we do not study the role of emotions in nonviolent mobilization, but rather in the process that leads from frustrating material circumstances to civil war.25 Moreover, our work is different from that of Petersen, McDoom, and other previous studies, because we do not focus on a case of ethnic civil war and because we study the action of both emotions and ideologies.26

As regards ideology, only a few empirical studies have tried to assess its role in civil wars. James Ron shows that Sendero Luminoso's political ideology had a crucial role in shaping the group's use of violence during the Peruvian conflict.27 Kai Thaler claims that ideology is an important element in explaining the choice between selective and indiscriminate violence by both the Mozambique Liberation Front and the People's Movement for the Liberation of Angola.28 Studying Colombia, Juan Ugarriza and Matthew Craig demonstrate that ideology influences the internal cohesion of armed groups, and that the political leaning of the family helps to predict individuals’ choice of a particular faction.29 Kristine Eck reveals that the Communist Party of Nepal (Maoist) relied heavily on a strategy of systematic indoctrination of the rural masses to build a common Maoist identity and stimulate recruitment.30 Much remains to be understood about the role of ideational resources in the process of armed mobilization. For instance, if ideology plays a role in overcoming the collective-action problem, are all ideologies equally effective in sustaining rebellion? Moreover, do ideologies and other nonmaterial resources such as emotions play the same role? Or do they enter the process in different ways?

## Emotions and Ideologies: Definitions and Role in Armed Mobilization

In line with previous literature, our basic premise is that material factors represent a necessary precondition for armed mobilization and civil war.31 Thus, the starting point of our causal chain is a material—though not necessarily economic—condition that is perceived, interpreted, and acknowledged by individuals through emotions and ideas. Such nonmaterial elements are essential because without a specific framing of the material context, a given situation would not lead to the formation of grievances and armed mobilization. Emotions are different from ideologies, but both components represent bridges between the context and individuals.

By “emotion” we mean the by-product of an event that occurred or could occur that influences a person's individual status, especially in terms of feeling or perception. Emotions are the residues of experience,32 the marks left on individuals following positive or negative shocks. In the causal chain between sheer facts (such as horizontal inequalities) and civil war onset, some emotions represent triggering mechanisms, sudden shifts that increase the saliency of a concern above others, and change the degree of “action readiness.”33 Unlike the theory of ethnic conflict presented by Petersen, ours does not require broad structural changes to generate politically relevant emotions. We agree that communities have to face a common (shocking) experience, but unlike his mass-oriented approach, we position the mechanisms of our causal chain at the individual level and focus on individual shocks that can be shared by individuals through personal connections. We focus on the links between individuals and their communities to understand how the collective-action problem of armed mobilization can be solved. As we explain below, we claim that ideational motives and political entrepreneurs have crucial roles in the aggregation process.

We adopt Ugarriza and Craig's definition of “ideology” as “a set of political beliefs that promotes a particular way of understanding the world and shapes relations between members of a group and outsiders, and among members themselves.”34 Therefore, an ideology provides a worldview, a set of beliefs, but it also implies a range of strategies to change social relationships in the world.

Emotions are shifting mechanisms that trigger “action tendencies” following sudden shocks.35 Some emotions influence the motivations to mobilize impetuously, but they do not change people's underlying values; they only clarify and activate them.36 Political activism, and even more participation in armed rebellion, is not a single great leap, but rather a flow of action in which specific emotions represent turning points and accelerating factors.37 Participation in collective action needs support, which can be derived from recurring emotional mechanisms that provide satisfaction,38 or—we argue—from ideologies. This need of support is even stronger when collective action stands for armed mobilization and participation means taking up arms in a civil conflict. Ideologies are normative and structured systems of beliefs, but also depictions of a goal, a future state. If civil war is more than a large, violent economic activity, ideational resources are essential, as they transform sheer facts into political issues. Transformative systems of beliefs act in combination with emotions, crystallizing them and thus leading from horizontal inequalities to rebellion. Ideologies constitute the backbone of the long flow of action that leads from acceptance of the status quo to collective armed mobilization. The incumbent's policies and horizontal inequalities are politically relevant for armed mobilization only insofar as they are considered illegitimate and if their existence can be attributed to a specific target, typically the government.39

Ideas and ideologies influence preferences,40 and they interact with contextual material opportunities.41 In the process of mobilization, they are powerful tools for the construction of “injustice frames.”42 Political entrepreneurs, the leaders of the rebel movement, use ideational resources to build collective-action frames and recruit new members.43 As David Snow argues, “Both ideology and its relationship to collective-action frames should be problematized and explored empirically rather than assumed or theorized in a simple, mechanistic fashion.”44

## Indignation and Armed Mobilization

We interpret emotions as explanatory factors of changes in individual conduct. A wide range of emotions exists. Interestingly, fear is one of the most cited emotions in the literature on conflict. Even within the rationalist perspective, commitment problems and security dilemmas are based on an emotional assumption of fear. Yet, “fear is far from the only emotion that influences mobilization,”45 and fear does not have a unidirectional effect on behavior. In fact, fear may trigger antagonistic actions, as assumed by many, but it may also generate passivity and submission. Furthermore, fear can push individuals to flee, which is one of the most common responses. Therefore, fear can suggest indefinite hypotheses on causal directions, especially because fear is a future-oriented emotion: individuals fear that something will happen depending on possible future actions.

Not all emotions are equally fungible; they have different effects in any particular context. We focus on indignation. This emotion is past oriented, because it is triggered by events that have already happened (or that we believe have already happened). Indignation differs from most of the emotions that have been studied in political science due to the structure of actions that trigger the emotion and the number of actors involved in those actions. For instance, anger is based on some wrong perpetrated against individuals or their kin.46 Hence, anger is an emotion about the self (extended to close relatives), where a person has directly suffered a wrongdoing and therefore reacts emotionally against the perpetrator. Indignation, however, is about a wrong done to a third party.47 Indignation emerges when an actor, B, perceives that actor/organization A has unjustly harmed an individual or group C.48 In turn, B feels indignant toward A; in Jon Elster's words, B experiences a “Cartesian indignation,” because Descartes was the first author to describe indignation in these terms.49 Although we do not exclude the possible impact of other emotions, we believe that indignation is particularly important in explaining the process of armed mobilization. Whereas anger is about direct experience, indignation is an indirect experience, where the reporting of others’ suffering triggers the emotion. This feature of indignation highlights the links between individuals and their communities, which are the focus of this study. Indignation accounts for the sense of community of individuals and underscores the relationships between each individual and the surrounding people based on shared conceptions of right and wrong. In everyday language, “resentment” and “indignation” seem to indicate the same emotion. Nonetheless, in the literature on the role of emotions in social and political phenomena, resentment refers to an implicit or explicit hierarchical relation.50 Petersen defines “resentment” as “the feeling of being politically dominated by a group that has no right to be in a superior position,” and he links it specifically to inter-ethnic relations.51 Moreover, whereas resentment so defined is a group emotion,52 because it presumes the existence of a group with a related identity, indignation is an individual emotion that can mark the birth of a new group (with a corresponding identity). In experiencing indignation, individuals detach from the status quo and realize that the current state of affairs is unjust. We argue, however, that this experience is only a first, though necessary, step toward group formation and armed collective action. Specific (radical) ideological networks are needed to take these two additional steps.

## Ideologies and Armed Mobilization

Max Weber famously wrote, “Not ideas, but material and ideal interests, directly govern men's conduct. Yet very frequently the ‘world images’ that have been created by ‘ideas’ have, like switchmen, determined the tracks along which action has been pushed by the dynamics of interest. ‘From what’ and ‘for what’ one wished to be redeemed and, let us not forget, ‘could be’ redeemed, depended upon one's image of the world.”53 We do combine emotional and ideological shifts, because emotions are preferences’ switches,54 whereas ideologies provide the Weberian “tracks” to maintain and give direction to the shift triggered by an emotional shock.

Hannah Arendt suggested that what is “crucial, then, to any understanding of revolutions in the modern age is that the idea of freedom and the experience of a new beginning should be conceded.”55 The ideas of freedom and liberty have profoundly influenced how one sees the social reality and its social relations. Arendt argued that “[Karl] Marx's transformation of the social question into a political force is contained in the term ‘exploitation,’ that is, in the notion that poverty is the result of exploitation through a ‘ruling’ class which is in the possession of the means of violence.”56 Through a Marxian lens, “poverty itself is political, not a natural phenomenon, the result of violence and violation rather than of scarcity.”57 Studying peasants in southern Italy, Antonio Gramsci noted: “They are in perpetual ferment but, as a mass, incapable of providing a centralized expression for their aspirations and their needs” because an ideological frame for action was lacking.58 Hence, ideologies provide different ways to interpret social reality and influence actions and the norms of actions beyond a mere instrumental view.59 For instance, “by reducing property relations to the old relationship which violence, rather than necessity, establishes between men, he [Marx] summoned up a spirit of rebelliousness that can spring only from being violated, not from being under the sway of necessity.”60 Later, Vladimir Lenin more explicitly elaborated this “spirit of rebelliousness,” motivating and justifying the use of violent strategies to conquer the state with ideological goals.

So, how do ideologies enter the process of armed mobilization in practice? To clarify the mechanisms, we need to distinguish between political entrepreneurs and followers. Political entrepreneurs profess an ideology, a given interpretation of facts that politicizes the current state of affairs and creates grievances. Followers join the ideological network created by the political entrepreneurs after an emotional shock that changes their action tendencies. We posit that in a given population under a given regime there are political entrepreneurs who are ready to resist and rebel. Their ideology suggesting a strategy of armed action is not enough, however, to reach a critical mass for mobilization. When an emotional shock hits the links between individuals and their communities, a larger part of the population, which was not previously captured by radical ideologies, becomes available to consider alternative views. Some emotions act as push factors, detaching individuals from the previous status quo and giving political entrepreneurs the opportunity to reach a larger group of people with their specific frame of grievance against the regime. Through political entrepreneurs, ideologies act as pull factors, translating private grievances into public grievances and connecting the individual to the aggregate level. Transformative ideas previously shared by a meager group of strong believers (“Kantian types”) can thus reach a broader audience.

In figure 1, we show the mechanisms by which emotions and ideologies may influence a change in individual actions. First, detachment caused by an emotional event disconnects the individual from the acceptance of the current state of social relations; this can be understood as a push factor, where individuals move away from the status quo. Ideologies communicated by political entrepreneurs help to rationalize the emotional shift and elaborate alternative worldviews (disenchantment), as well as possibilities for action. Finally, the alternative ideological framework provides a new base for normative values and the conduct of action through the “anchoring” mechanism, which can be understood as a pull factor attracting individuals to a new status.

Figure 1.

Nonmaterial Mechanisms for Armed Mobilization

Figure 1.

Nonmaterial Mechanisms for Armed Mobilization

The role of ideologies, however, goes beyond the framing and formation of collective grievances (disenchantment mechanism). “Ideas define the universe of possibilities for action,”61 and ideologies provide road maps, including different strategies to reach their goal through the mechanism of anchoring—for instance, the Maoist “protracted people's war.” Thus, not all ideologies are equally effective in producing armed mobilization. Framing alone does not explain armed mobilization,62 partly because a purely cognitive approach omits the role of emotions, but also because not all normative systems of beliefs are equally effective. In other words, the ideologies used by political entrepreneurs to stimulate mobilization can make a difference. The content of ideologies is strictly related to the emergence of an armed movement because different ideologies also imply different goals, different strategies, and different organizational forms to change the status quo.

## The Role of Ideological Networks: Connecting Levels and Factors

Ideological networks play an essential role in the process of collective armed action, because the aggregation process from individual discontent to collective action develops within and thanks to such networks. Moreover, they perform another crucial function, because a civil war cannot break out without organization and preparation. Individuals moved by a strong emotion and motivated by an ideology are not sufficient, as financial, organizational, and military resources are indispensable. Through ideological networks, political entrepreneurs translate ideas into practice. This point where material and nonmaterial factors come into contact distinguishes the outbreak of a civil war from a mass nonviolent mobilization, and it further highlights the different impact of the various ideologies on the process of armed mobilization. We agree with Pearlman that episodes of mass protest can be mainly driven by the spread of emboldening emotions through a frustrated population, but armed mobilization for civil war is a much more complex activity, which needs specific assets and longer periods of preparation.63

Ideological networks operate the connection between the material and nonmaterial spheres, the individuals and collective armed action, and the local and national contexts. Like Petersen, we assume that individuals who become involved in armed rebellions are usually sensitive to risk and, therefore, need to be reasonably confident that rebelling is at least practically feasible and that some tangible support exists, even though they cannot be sure about the final outcome of the rebellion.64 Thus, for ordinary people (followers), the existence of rebel networks is essential to provide not only information and political meaning to the current situation, but also practical assistance.

An underground network decreases the costs of mobilization and provides incentives to mobilize through mechanisms based on opportunity, contributing to explaining the existence of different levels of mobilization within a single conflict. Not all networks, however, are the same. Petersen highlights how different community subsets are associated with a stronger or weaker propensity to join a rebellion of their members.65 Subgroups based on economic interactions, for instance, will be associated with a low propensity for rebellion. By contrast, youth organizations could be particularly inclined to join the armed resistance. We evaluate networks not only in terms of the preferences associated with their members and for the possible constraints they can provide, but also as practical tools of rebellion. Some networks are more effective than others in sustaining an armed rebellion. As we mentioned, different ideologies imply different goals, different strategies, and different organizational forms to change the status quo. For instance, not all ideologies imply underground networks, and those that do can create different types of networks. In the case of Italy, the classical liberal individualism professed by some small parties in the early twentieth century did not promote a real ideological network, whereas Christian Democracy could count on a widespread network mainly aimed at peaceful solidarity. We focus on the underground network created by the Italian Communist Party because during the war years, the proletarian revolution still maintained a prominent role in the ideology of the party and in its political manifestoes. Its political program implied not only the defeat of Fascism, but also a more radical and deeper revolution against the structures of power and the social institutions of the state.66 The communist underground network was developed for the purpose of preparing a social and political revolution against the state through armed mobilization.

The presence of well-trained and experienced professional mobilizers facilitates the organization because, given their past experience, they can supply the know-how and logistics to set up a high-risk organization. In Italy, various individuals gained experience and training in an underground network in areas where a strong radical political organization had been present before the new regime took power. In this way they learned how to implement guerrilla and rebellion techniques, becoming professional mobilizers and rebels. Building on this theoretical work, we advance two main theoretical propositions. First, areas where individuals have experienced severe emotional shocks in terms of indignation against the incumbents will have higher levels of armed mobilization. Second, areas where radical ideological networks are stronger will have higher levels of armed mobilization

## The Italian Case, 1943–45

To test our theory empirically about the role of indignation and radical ideologies in the process of armed mobilization, we focus on the case of the Italian civil war, which was fought by partisan bands against Nazi forces and Fascist militias in 1943–45. It is a historically prominent case: the Italian civil war caused around 117,000 deaths in battle;67 about 10,000 others were victims of one-sided violence.68 Moreover, at least three features together make the Italian case a relevant and interesting one to test our theoretical propositions. First, because all previous works on the role of emotions in civil war studied cases of ethnic conflicts, we want to test our hypothesis about the role of indignation in a non-ethnic context, to show that emotions also play a crucial role in different types of civil wars. Second, the simultaneous existence of diverse ideological networks on the ground allows us to test our hypothesis on the different effectiveness of ideologies in mobilizing individuals to armed collective action. Third, because the Fascist policy that provoked the most indignation was the decision to enter World War II,69 we are able to operationalize the level of indignation in Italian localities relying on a count variable (Italian soldiers’ deaths in World War II) that represents the direct consequences of that policy. Thus, the Italian case allows us to follow the suggestions provided by several scholars,70 to empirically explore the relationship of ideologies with collective action, compare different ideologies in the same process, and look at the role of both ideologies and emotions.

Similar to other micro-comparative studies based on a single conflict, this research design raises the issue of broad generalizability. Given the present stage of theoretical understanding and empirical studies about the role of nonmaterial factors in the process of armed mobilization, however, we believe that bounding the empirical domain is an acceptable trade-off.71 In addition, our theory was not developed inductively from the data we use to test it, and we believe that it can also provide useful guidelines in other cases. Moreover, because we look at micro dynamics using disaggregated data, the possible specific features of the Italian case should not have a strong impact on our analysis.

## Research Design

We have collected new georeferenced data on the size of partisan mobilization, the local presence of the bands, Nazi violence, and Fascist control of individuals. The empirical analysis is organized in three parts: first, we perform a statistical analysis of partisan mobilization at the provincial level (between-provinces analysis) to highlight general patterns in the country; then, since center-north and southern Italy underwent different war dynamics, we perform a statistical analysis of partisan mobilization at the municipal level within two northern provinces (within-provinces analysis) so as to study the phenomena at a more fine-grained level. Through the within-provinces analysis, we reduce even more the possible omitted variable bias (e.g., the two provinces that we study underwent the same military occupation by the Nazi forces); and thanks to more disaggregated data, we can focus on the links between individuals and their communities. Finally, we provide qualitative evidence based on micro narratives and interviews with former partisans to go beyond statistical association among variables and process trace our causal mechanisms.

In the first part of our empirical analysis, the unit of analysis is the “province,” that is, the second administrative level in Italy. In 1945 there were ninety-two provinces in Italy. Therefore, we have a cross-section that can vary, given different specifications, from ninety-two to eighty-five observations. The project is based on a process of data gathering that involved collecting quantitative historical data, enabling us to create a sense of the context where the civil war took place. To work with plausible control variables, we collected data on the social, economic, and political situation of Italy in the Fascist period. In this challenging search for data, we relied on many primary and secondary sources, including available datasets, original documents, books by historians who had worked on the period in question, and historical statistical data.

### DATA: THE DEPENDENT VARIABLE

Here we use the number of partisan bands in a given Italian province as our dependent variable. The armistice and the consequent collapse of the Italian state took place in the last months of 1943. Hence, we record the presence of partisan bands in the field in 1944, as the data are more complete and reliable than those for 1943. A team of Italian historians gathered these data, and the results of this research were published in the Historical Atlas of the Italian Resistance.75 The process lasted for several years, and the items of information were collected and double-checked using a number of sources, including the archives of the Allied armed forces, the archives of the Italian military authorities, and the archives of the partisans’ associations.76 The atlas contains a series of maps showing the location of the armed bands during the years of the resistance war. We used this source to code our dependent variable, which counts the number of partisan bands in a province.77 Given the nature of our dependent variable, we have used negative binomial regressions.78

### DATA: INDEPENDENT VARIABLES

We argue that the presence of a developed ideological underground network and the severe emotional shocks experienced by the individuals who lived in a given area are two important factors explaining mobilization and armed resistance in Italy, as they indicate opportunity and motivation to rebel against the Fascist regime and the German occupiers.

We pay particular attention to the underground network created by the Italian Communist Party (PCI) because its ideology expressed a higher propensity to rebel than any other political culture in Italy at the time. Preparing an organization ready to subvert the incumbent government through the use of force was an essential part of the PCI's ideology. The strength of the communist network is operationalized through the vote share obtained by the Communist and Socialist Parties in the 1921 elections, the last democratic elections before the advent of the Fascist regime. It is also necessary to consider the votes of the Italian Socialist Party (PSI) because the Communists separated from the PSI to form the PCI only in 1921, and they did not manage to present their own lists in all the provinces. In addition, some local and national politicians who would later join the PCI were still in the PSI in 1921, and the same holds for a large proportion of the electorate. All electoral data come from the dataset of the Historical Atlas of Italian Elections, which includes the number of votes received by each party in the national elections.79

Operationalizing emotions—in our case, indignation—is extremely difficult. We needed a measure of indignation caused by the incumbent regime in a given community as a result of its policies. One of the most significant policies implemented by the Fascist regime certainly was the decision to enter World War II in support of the Nazis. Consequently, we decided to operationalize the emotional shock resulting from indignation in the local communities as the number of military casualties experienced by the soldiers from a given province before the armistice. To be clear, these soldiers did not die in those provinces, but abroad, on the war front. They had been resident in those provinces when they joined the army. We use the proportion of those soldiers dying relative to the population of the province as a whole.

The severity of the war, expressed by many thousands of deaths and the shortage of food, caused a widespread sense of injustice in the Italian population. The memories of many Italian soldiers are particularly effective in describing the atmosphere during the weeks following the armistice. Giame Pintor, soldier and then partisan, portrayed this common state: “The soldiers who in September [1943] had been travelling through Italy hungry and half-naked, wanted overall to go back home, they did not want to hear any more about war and suffering. They were those who had lost, but they were carrying inside them the true element that would lead to their [political] resumption; their resentment of offences, both inflicted and suffered; their disgust at injustice carried on in places where they had previously lived.”80

The experience of war was a possible trigger not only for the former soldiers to rebel and join the antifascist organizations. The stories that they could tell about the front were clearly understood by the Fascist regime as a dangerous element. Nuto Revelli, a lieutenant in the Italian army, survived the German-Italian defeat on the Russian front and reported that when the military trains were traveling through Italy to bring back the few survivors, police forces prevented people from talking with the soldiers and asking for information about the war or their relatives in the war. Revelli recalls that “the excuse was the fear of typhus. In reality, they just did not want any interaction between the soldiers who had survived the war and the [civilian] population.”81 The Fascist regime feared that any contact between the survivors and the population could reinforce the negative experience that the war—in different forms— had inflicted on the community life of Italians. The preoccupation was mainly caused by the socialization of the violent experience. Like Goodwin and Wood, we assume that violence and suffering can motivate individuals to rebel against the regime that is held responsible for those events.82 Witnessing the consequences of injustice and being in close contact with the victims is not less powerful than personally experiencing the violence, because the circumstances are not caused by an isolated event and they could involve anyone in the future. This is why we believe that our variable is a meaningful way of operationalizing indignation. We assume that the members of the communities who experienced particularly harsh consequences are more motivated to engage in armed rebellion to subvert the system that caused such suffering. The data on war casualties relative to each province come from the Italian National Bureau of Statistics, which prepared a detailed report for the Italian government.83

We control for the population of the provinces, and the data derive from the official census.84 We also include in our models a measure of income per capita recorded in 1951 to check whether relatively poorer provinces show higher probabilities of mobilization and rebellion.85 Additionally, we include a variable that records the number of automobiles registered in a given province relative to the population of the province.86 Even though highly correlated with income, this variable is a useful proxy to consider the development of road networks in the provinces. In fact, historical accounts and recent studies have both highlighted the road network as an important factor in explaining the location of the insurgents, because roads shape insurgent target selection.87 In the Italian case, the partisans used to attack the German convoys carrying backup troops and ammunition.88

As a result, we have two main empirical expectations. First, areas that experienced high World War II-related military casualties will have higher levels of local armed mobilization. Second, areas that had strong left-wing ideological networks will have higher levels of local armed mobilization.

## Between-Provinces Findings

In figure 2, we report our main results in a graph based on a negative binomial regression.89 Our dependent variable detects the number of partisan bands at the provincial level in 1944,90 as the data for 1944 are the most reliable and complete.

Figure 2.

Partisan Bands in the Provinces: Between-Provinces Analysis

Figure 2.

Partisan Bands in the Provinces: Between-Provinces Analysis

Figure 2 shows graphically the baseline model of our negative binomial regression, the dots representing the coefficients and the dashed lines their 95 percent confidence intervals. If a dot is on the right side of the graph, the variable increases the expected number of partisan bands. If it is on the left side of the graph, the variable tends to decrease the expected number of bands in a province. Moreover, only variables where the dashed line does not overlap the vertical line (zero impact) are statistically significant. In our case, only radical ideology (share of votes to left-wing parties in 1921), indignation (measured as military deaths before the armistice), and number of automobiles per capita are significant at 95 percent. Thus, the empirical analysis at the provincial level highlights an important role for our two nonmaterial factors in civil war. First, radical ideology does matter, because provinces with higher shares of votes for the radical left in 1921 tend to have a larger number of partisan bands in 1944. An increase of one standard deviation of radical ideology in one province increases the expected number of partisan bands in that province by 194 percent. Additionally, the role of indignation is important: our measure of military loss on the war front shows a strong positive impact on the probability of rebellion at the provincial level. In fact, the expected number of partisan bands increases by 120 percent following an increase of one standard deviation of military losses on the war front.

Furthermore, in figure 3, we have simulated quantities of interest for the expected number of partisan bands in a given province based on our estimates.91 The vertical axis reports the expected number of partisan bands, and the horizontal axis provides different possible scenarios. The dashed lines represent 95 percent confidence intervals. In a province where all variables are at mean levels, we should expect about two partisan bands. In provinces where radical ideology is one standard deviation below the mean, however, we should expect one band. Moreover, when indignation is one standard deviation below the mean, we should not expect to find any partisan bands. By contrast, in provinces where radical ideology is one standard deviation above the mean, we should expect around four bands. And in provinces that have experienced higher levels of indignation (again one standard deviation above the mean), we should expect seven bands.

Figure 3.

Expected Number of Bands in Provinces

Figure 3.

Expected Number of Bands in Provinces

Moreover, because the partisans’ presence could also be spatially clustered, we have run models that take into account spatial interdependence.92 Our two main explanatory variables, indignation and radical ideology, keep their significance and directional effects.

## Discussion

In figure 4, we show that the radical ideology variable contrasts with other ideologies in how it affects the number of partisan bands. In the graph we represent four different models, the baseline model as shown in figure 3 and then three models that control for three different alternative ideologies: Christian Democrat, liberal, and right-wing. The graph reports only the main variables; control variables are omitted for graphical purposes. Different specifications are represented with different shapes. Therefore, we represent four different point estimates for radical ideology and indignation (upper part of the figure) and show that the results do not essentially change when controlling for other ideologies. The coefficients of the alternative ideologies are represented as singular point estimates in the lower part of the graph.

Figure 4.

Partisan Bands and Ideologies: Between-Provinces Analysis

Figure 4.

Partisan Bands and Ideologies: Between-Provinces Analysis

One important contributor to the resistance was the Popular Italian Party, known as the Christian Democrats after the war. If we control for the strength of this party, we find that more votes for the Catholic party increase the expected number of partisan bands, but its effect is not significant at the 95 percent level. Moreover, the variable indicating left-wing votes is still statistically significant. Areas where the Catholics were politically strong were more likely to show resistance. In historically Catholic areas, partisan bands known as “Mauri” were formed,93 but this does not eliminate the effect of the left-wing ideological network. We would be more concerned if votes in 1921 in favor of conservative and right-wing parties had had a similar effect, because that would highlight that our electoral measurement is not capturing just a left-wing network, but some general political activism. We find that votes for the Fascist Party in the 1921 elections, however, did not affect the partisan mobilization in 1944. Neither did votes for the liberals affect the number of partisan bands. These findings also support another aspect of our theory: not all ideologies influence the process of armed mobilization in the same way. Ideologies play an essential role in creating armed mobilization in civil war, but not all of them are equally effective for this purpose. In the case of Italy, the political ideology of the Christian Democrats proposed a completely different relationship with political violence. Theirs was not a revolutionary party. So, even though it had its own network, which supported the “Catholic bands,” the mobilizing potential was much weaker than the one shown by communism. Not all ideologies transform individual resistance into collective armed action in the same way and with the same outcomes. The measurement of voting behavior for the left-wing parties in 1921 could be capturing a general level of political participation instead of just the existence of a well-organized radical network in a province. In that case, our argument for a preexisting left-wing network that consolidated during the Fascist regime would be weakened. To check for this possibility, we controlled for local turnout in 1921.94

We also control if “anger”—instead of “indignation”—can explain armed mobilization. If the Fascist regime punished radical left-wing cell members in a particular province, those cell members would feel anger toward the regime.95 The general population would not feel this anger, whereas they should respond with indignation toward the war casualties. Thus, high overall war casualties should lead to high mobilization, but high prewar radical left-wing persecutions should not affect the degree of partisans’ mobilization. For this purpose, we collected data on the individuals sent to confino or, more technically, domicilio coatto (compulsory residency) for political reasons. Compulsory residency was a practice used by the Fascist regime when someone was identified with some evidence as a possibly dangerous element for the regime. These people were arrested and deported, usually to remote Italian islands. This proxy should detect anger against the regime.96 We collected the number of individuals coming from a given province who experienced this punishment between November 1922 and September 1943 according to a dataset of the Italian National Historical Archives (Casellario Politico Centrale). Figure 5 shows, controlling for other variables, that the variable anger is not significant, whereas our two main explanatory variables keep their directional effect and are still statistically significant.97

Figure 5.

Bands at Provincial Level: Indignation and Anger

Figure 5.

Bands at Provincial Level: Indignation and Anger

Previous literature provides alternative possible explanations of armed mobilization based on the role of social ties and security-related benefits of participation in war that could be related to our main explanation based on ideologies and emotions.98 Therefore, we also add a number of control variables in our models to evaluate whether these alternative explanations can weaken our previous results.

We have robust evidence showing that our proxy for left-wing ideological networks is not picking a different political or electoral dynamic. Votes for left-wing parties, however, could capture the socioeconomic structure or social ties of a province.99 Therefore, we would not have the direct effect of the left-wing ideology and network, but only that of different social structural characteristics that could influence the pattern of mobilization. In figure 6, we report the graphical representation of four different models where we add three proxies for possible alternative explanations related to local social ties. When we control for the share of the labor force employed in the industrial sector in those years at the provincial level, however, the role of our nonmaterial explanations (ideology and indignation) is still positive and significant. Additionally, it could be argued that the socioeconomic characteristics of a province do not influence mobilization directly and that the organization of social-class interests could have facilitated resistance and mobilization. Therefore, we collected data on the density of trade-union organizations at the provincial level.100 Again our main results are robust. Finally, because combatants are usually young, and young people are considered to be more inclined to take up arms and fight,101 we include the percentage of youth in the population of the province during the partisan mobilization (1943–45).102 The results hold.

Figure 6.

Partisan Bands and Social Ties

Figure 6.

Partisan Bands and Social Ties

Finally, because one of our main results is based on a proxy of violence— the number of military deaths per capita at the provincial level before the armistice—we want to evaluate whether theories based on the security-related benefits of participation in armed mobilization could weaken our “indignation and radical ideology framework.”103 Our argument suggests that these casualties, which happened abroad, caused two types of indignation in the areas that had been home to the victims: first, the local communities that suffered more casualties experienced a stronger emotional shock as a consequence of the decision of the Fascist regime to fight the war. Second, the soldiers who survived and returned home felt greater frustration against the regime and were more willing to join the resistance. Other forms of violence, however, could have influenced armed mobilization, or there could even be some omitted variables that have made our proxy significant. Specifically, Stathis Kalyvas and Matthew Kocher show that not only rebels, but also civilians who do not join rebellions, run serious personal risks in war zones.104 Free riding can be approximately as costly as joining the armed bands because, for instance, it exposes nonparticipants to the harsh counterinsurgency of incumbents. In these cases, individuals can choose to join the rebellion, trying to increase their security. In Italy mass killings were a tool of military strategy, as the Nazis equated the population with the rebel forces. In some cases, the German forces destroyed entire villages, killing hundreds of persons in single actions.105

In figure 7, we graphically represent models where we control for the local costs of war caused both by aerial bombing and by incumbents’ violence. It could be argued that our proxy is capturing the Nazi-Fascist retaliation against civilians, because the soldiers who returned home, for instance, were reluctant to join the forces of the new Fascist puppet state. When we controlled both for the number of casualties caused by Nazi-Fascist actions per 1,000 inhabitants at the provincial level and for the number of Nazi violent events, however, we found that these factors had a negative influence on the probability of armed mobilization. Because there has been some debate in Italy on the reliability of data concerning these violent acts, we employed two different sources,106 but they produced similar results. The results also held when we controlled for aerial bombing.107

Figure 7.

Partisan Bands and Security

Figure 7.

Partisan Bands and Security

## Within-Province Statistical Analysis

The results of the within-province statistical analysis further support the findings presented above. To perform this part of our analysis, we focused on two Italian provinces: Cuneo and Savona, respectively in the Piedmont and Liguria regions. We selected these two provinces because they both experienced strong mobilization and remarkable within-province variation; moreover, they are both in northern Italy, and therefore share similar conflict dynamics with respect to World War II. They share administrative borders, but they have different geographical (mountains vs. coastline), socioeconomic (prevalence of agriculture vs. prevalence of industry), and historical features. The province of Cuneo is one of the largest in Italy, and Savona one of the smallest. In addition, because Cuneo is a mountainous area, the conscripts coming from that province usually enlisted in the army, whereas the conscripts from coastal Savona often enlisted in the navy. Moreover, in these two provinces we could benefit from high-quality original datasets collected by local researchers.108

The within-provinces analysis differs from the previous one in some respects. First, our dependent variable is no longer the presence of partisan bands in a province, but, thanks to different data sources and better measurements, we can gauge the level of local partisans’ participation. Hence, we use as a dependent variable the number of partisans in a municipality. We gathered data at the individual level on approximately 90,000 partisans in the province of Cuneo and about 7,000 partisans in the province of Savona. The data derive from the archives of the Institutes for the History of the Italian Resistance in the two provinces and concern biographical information about the partisans (e.g., date and place of birth, and residence) as well as military information (where and when the person joined the Resistance; when the person left it; and possibly the date of death). We geocoded this information and used the municipalities into which the provinces are divided as our units of analysis (68 for Savona and 205 for Cuneo). Second, we created our new variable on the number of soldiers killed on the war front before the armistice by using different data from the previous section. The data for these two provinces come from two different local sources,109 which are also different from the national data proxy source that we used for the between-province analysis.110 Using different data sources for the same phenomenon should make it less likely to find similar patterns. Third, we use a different proxy for the local presence of left-wing groups because no reliable data at municipal level exist for the 1921 elections. From the Italian State Archive, we obtained data on the persons controlled by the Fascist political police in each municipality of those two provinces. We have collected data for the period 1922–43,111 counting the number of people who were monitored by the Fascist political police because they were considered communists or socialists.112

In figure 8, we present the main model graphically. We use a negative binomial regression because our dependent variable is a count.113 Even though data gathering at such a fine-grained level was challenging, we have managed to control for variables that could be proxies for alternative explanations, as we described in our previous section on robustness.114

Figure 8.

Negative Binomial Regression within Provinces

Figure 8.

Negative Binomial Regression within Provinces

Our results show a strong relationship between the variable that we use to measure the emotional shock caused by indignation in a community and the degree of armed mobilization in the same community.

In other words, the number of military casualties occurring abroad before the armistice among soldiers who came from a given municipality correlates closely with the number of partisans who appeared in the same municipality after the armistice. Therefore, even using more disaggregated data we find that the intensity of emotions influence the local strength of armed mobilization. More precisely, increasing the number of casualties on the war front before the armistice by one standard deviation increases the number of partisans in a municipality by 107.4 percent, controlling for population. In addition, the number of partisans in a given village is also highly correlated with our proxy for radical ideology, the number of persons monitored by the Fascist political police before the armistice in the same village. Increasing the number of people monitored and labeled by the Fascist political police as “communist” or “socialist” (from 1922 to 1943) by one standard deviation increases the number of partisans mobilized in a municipality in 1944 by 50.3 percent.

Controlling for population and the local level of incumbent violence does not change the results, and the two variables are not statistically significant at standard levels. The percentage of workers employed in the industrial sector, however, is statistically significant and increases the number of participants in the local resistance, though our proxy for local ideological networks yields similar results even controlling for this salient economic feature.115

In figure 9, we provide simulations of quantities of interests within provinces. In municipalities with all variables at average levels, we could expect around thirty-five partisans; communities with less radical ideology or indignation (in the 25th percentile of the variables’ distribution), however, tend to have fewer partisans, around twenty-five. On the other hand, municipalities with higher levels of indignation or ideology (90th percentile) have more than sixty-five partisans.

Figure 9.

Quantities of Interest—Expected Number of Partisans

Figure 9.

Quantities of Interest—Expected Number of Partisans

We performed the same robustness tests on spatial interdependence that we used for the between-provinces analysis. The results for indignation and radical ideology within provinces are stable.

## Micro Narratives and Interviews

Given the unusual nature of the factors that we try to capture through quantitative methods, we complete our empirical analysis with additional evidence based on historical micro narratives, memoirs, and interviews with former partisans. We analyzed more than 100 interviews with ex-partisans who fought in Tuscany and Liguria. These interviews were recorded by the historians of the Institute for the History of Resistance in Florence and Savona. The interviews are stored in their archives and have never been used in political science analyses so far. Because they do not follow the same structure, we could not code them for a quantitative analysis.

Several leading figures of the Italian Resistance, both at the national and local levels, fought in the Spanish Civil War and brought with them the know-how and logistical skills to use weapons and organize insurgency warfare. Moreover, to survive within the security and police system of the Fascist regime, they learned how to organize an underground organization. They organized a large network based on small nodes: for instance, from various interviews and secondary sources, we found that a common mechanism in the communist organizations was that each member of the underground organization knew a maximum of only another two or three components of the organization before the outbreak of the armed mobilization. In this way, the communists could avoid betraying the rest of the network if the Fascist authorities caught and interrogated them. Moreover, this high professionalization led to a higher level of mobility for the members of the networks.119 During an interview, G.N., a former partisan, mentioned a system involving the short relocation of the members from one city to another. Frequently, the cities were no more than sixty kilometers away, but located in different regions—for instance, Piedmont and Liguria or Liguria and Emilia Romagna.120

This underground experience before and during the Fascist regime provided a well-functioning network, capable of communicating and mobilizing both goods and people. Among the several pieces of evidence, we found the creation of an alternative, underground welfare system designed to reduce the costs of armed mobilization not just for the individual partisans, but also for their families and communities. In fact, decades before Hamas and Hezbollah, the partisan movements in World War II had developed in several areas a form of welfare state to support relatives of the partisans who had been arrested or killed, or to offer financial help to the families of the partisans who were fighting. Therefore, their organization was able to both collect financial support and redistribute it, sustaining the community that was supporting the rebellion. We found several cases of this. For instance, in Genoa a Committee of Solidarity was created in May 1944. At the beginning, it sustained 165 families and in less than a year it was supporting around 1,000 families.121 The financial figures, referring just to one city, are impressive: in less than one year the disbursement to partisan-related families increased by almost 1,000 percent: from 161,150 Italian lire in June 1944 (approximately $14,375 in 2014) to 1,400,000 lire in April 1945 ($124,888 in 2014). These elements highlight how the preexisting and enduring underground network strongly and concretely increased the opportunity to mobilize, providing both meaning and material elements.

The role of political entrepreneurs in framing the emotional motivations within an ideological structure is revealed in several interviews. For instance A.B.S., a partisan in the province of Savona, stresses that he joined a band as an act of resistance against the regime, toward which his family carried a grievance, but then his action became more political once he joined the band: “We became members of the Communist Party because a Commissioner used to come and explain the motivations for political action.”132 G.S. points out another aspect of the role of ideology, the importance of its logistical network: “We decided to establish a Communist cell, not because we really knew the Communist Manifesto, but because it was the best means of struggle against Fascism.”133 In October 1943, F.R. met a couple of boys who were trying to organize a group for “doing something.”134 He decided to join them: “When I arrived on the hills, I did not even really know that we were going to fight a real guerrilla war. Then, at night, the commander came and he explained to us what it was about, and I began to understand that it was not a joke anymore, it was real armed action.”

## Conclusion

In the introduction to this article, we quote Albert Camus and Italo Calvino. Both refer to nonmaterial factors that can influence individuals’ decisions to join armed mobilization: ideas and emotions. In this article, we untangle the process that leads from material factors such as economic inequalities and political discrimination to civil war, through the formation of grievances and armed mobilization. In particular, we argue that emotions and ideologies play crucial roles in moving people from acceptance of the status quo to armed action. Emotions act as triggering mechanisms, push factors that detach individuals from the current state of affairs, modifying their preferences and making them available to consider alternative views of the world. Through the activities of political entrepreneurs, ideologies spread from small circles to a broader audience, acting as pull factors that transform private grievances into collective grievances. Ideological networks transform ideas into action, providing the organizational tools that are essential in a complex activity such as a civil war. Neither all emotions nor all ideologies have the same effect, however. Applying a threefold empirical strategy and using mixed methods to study the case of the Italian resistance movement, we showed that indignation is a powerful ignitor of armed mobilization because it impinges on the links between individuals and the people around them, relying on a shared conception of what is right and what is wrong. Moreover, we showed that radical ideologies that support the use of violence against incumbents have a crucial role in causing armed mobilization, in contrast to more peaceful ideologies and ideologies that are more in tune with the regime. Our findings are consistent across different levels of analysis and find support in interviews with former partisans. We also performed several robustness checks to account for possible alternative explanations and spatial interdependence, but the effects of indignation and radical ideologies are confirmed.

Our contribution improves current knowledge about the causes of civil war and armed mobilization, showing how specific nonmaterial factors play crucial roles in the process that leads to armed collective action. Unlike previous studies, our article emphasizes that emotions are also important in non-ethnic civil wars, and that ideologies, through political entrepreneurs and ideological networks, are essential components of the process. Moreover, we stress that it is also useful to look at the specific type of emotion and ideology. The lessons from this historical case are also relevant for current political affairs: indignation and revolutionary ideologies merit particular attention. In our case, we focused on the Italian Communist Party and on its underground network, because organizing a revolution was still an integral part of the party's ideology at the time. Needless to say, different radical political, religious, or ethnic ideologies have proposed violent strategies to change the status quo in other contexts, and noncommunist political actors have created revolutionary underground networks in other times, even nowadays.

## Acknowledgments

The authors thank Kristin Bakke, Brian Burgoon, Kristian Gleditsch, Viva La Seal, Dan Miodownik, Francesco Moro, Livia Schubiger, Julian Wucherpfennig, and participants in workshops at the University of Amsterdam, the Peace Research Institute Oslo, the University of Uppsala, and the University of Warwick for comments on previous versions of this article. They would also like to thank the anonymous reviewers for their helpful comments. Violet Benneker, Simone Pastorino Casadei, and Nicola Reineri provided great research assistance. Andrea Ruggeri acknowledges financial support from the Independent Social Research Foundation in London. The authors share equal authorship.

## Notes

1.

Albert Camus, The Rebel: An Essay on Man in Revolt (London: Vintage, 1991), p. 14; and Italo Calvino, Il Sentiero dei Nidi di Ragno [The path to the nest of spiders] (Torino: Einauldi, 1964), p. 109.

2.

Stuart J. Kaufman, Modern Hatreds: The Symbolic Politics of Ethnic War (Ithaca, N.Y.: Cornell University Press, 2001); and Roger D. Petersen, Understanding Ethnic Violence: Fear, Hatred, and Resentment in Twentieth-Century Eastern Europe (Cambridge: Cambridge University Press, 2002).

3.

Omar Shahabudin McDoom, “The Psychology of Threat in Intergroup Conflict,” International Security, Vol. 37, No. 2 (Fall 2012), pp. 119–155; and Lars-Erik Cederman, Kristian Skrede Gleditsch, and Halvard Buhaug, Inequality, Grievances, and Civil War (Cambridge: Cambridge University Press, 2013).

4.

Stathis N. Kalyvas and Matthew Adam Kocher, “How ‘Free’ Is Free Riding in Civil War? Violence, Insurgency, and the Collective Action Problem,” World Politics, Vol. 59, No. 2 (January 2007), pp. 177–216; and Roger D. Petersen, Resistance and Rebellion: Lessons from Eastern Europe (Cambridge: Cambridge University Press, 2001).

5.

Paul Collier and Anke Hoeffler, “Greed and Grievance in Civil War,” Oxford Economic Papers, Vol. 56, No. 4 (August 2004), pp. 563–595; and James D. Fearon and David Laitin, “Ethnicity, Insurgency, and Civil War,” American Political Science Review, Vol. 97, No. 1 (August 2003), pp. 75–90.

6.

Lars-Erik Cederman and Kristian Skrede Gleditsch, “Introduction to Special Issue on ‘Disaggregating Civil War,'” in Cederman and Gleditsch, eds., “Disaggregating Civil War,” special issue, Journal of Conflict Resolution, Vol. 53, No. 4 (August 2009), pp. 487–495.

7.

Stathis N. Kalyvas, The Logic of Violence in Civil War (Cambridge: Cambridge University Press, 2006), p. 101.

8.

Paul Collier, “Rebellion as a Quasi-Criminal Activity,” Journal of Conflict Resolution, Vol. 44, No. 6 (December 2000), pp. 839–853; and Collier and Hoeffler, “Greed and Grievance in Civil War.”

9.

Collier, “Rebellion as a Quasi-Criminal Activity.”

10.

Frances Stewart, ed., Horizontal Inequalities and Conflict: Understanding Group Violence in Multiethnic Societies (Houndmills, U.K.: Palgrave Macmillan, 2008); and Cederman, Gleditsch, and Buhaug, Inequality, Grievances, and Civil War.

11.

Petersen, Understanding Ethnic Violence; and Kalyvas, The Logic of Violence in Civil War.

12.

We acknowledge that most of the debate on the causes of civil war has juxtaposed emotion-based explanations against rationalist theories, whereas we stress the distinction between material and nonmaterial sources of mobilization. We believe that the first way of framing the debate tends to have a paradigmatic orientation and risks focusing too much on dichotomies, whereas emotional and rational elements can be included in the same theoretical framework, leading to a more fruitful debate.

13.

Cederman, Gleditsch, and Buhaug, Inequality, Grievances, and Civil War.

14.

Kalyvas and Kocher, “How ‘Free’ Is Free Riding in Civil War?”; and Macartan Humphreys and Jeremy M. Weinstein, “Who Fights? The Determinants of Participation in Civil War,” American Journal of Political Science, Vol. 52, No. 2 (April 2008), pp. 436–455.

15.

Elisabeth Jean Wood, Insurgent Collective Action and Civil War in El Salvador (Cambridge: Cambridge University Press, 2003).

16.

Jeff Goodwin, James M. Jasper, and Francesca Polletta, eds., Passionate Politics: Emotions and Social Movements (Chicago: University of Chicago Press, 2001); and James M. Jasper, “Emotions and Social Movements: Twenty Years of Theory and Research,” Annual Review of Sociology, Vol. 37 (August 2011), pp. 285–303.

17.

Wendy Pearlman, “Emotions and the Microfoundations of the Arab Uprisings,” Perspectives on Politics, Vol. 11, No. 2 (June 2013), pp. 387–409.

18.

Petersen, Understanding Ethnic Violence.

19.

Pearlman, “Emotions and the Microfoundations of the Arab Uprisings.”

20.

Barry R. Posen, “The Security Dilemma and Ethnic Conflict,” Survival, Vol. 35, No. 5 (Spring 1993), pp. 27–57.

21.

Kaufman, Modern Hatreds.

22.

Petersen, Understanding Ethnic Violence.

23.

McDoom, “The Psychology of Threat in Intergroup Conflict.”

24.

Jeffrey S. Kopstein and Jason Wittenberg, “Deadly Communities: Local Political Milieus and the Persecution of Jews in Occupied Poland,” Comparative Political Studies, Vol. 44, No. 3 (March 2011), pp. 259–283.

25.

Pearlman, “Emotions and the Microfoundations of the Arab Uprisings.”

26.

Petersen, Understanding Ethnic Violence; and McDoom, “The Psychology of Threat in Intergroup Conflict.”

27.

James Ron, “Ideology in Context: Explaining Sendero Luminoso's Tactical Escalation,” Journal of Peace Research, Vol. 38, No. 5 (September 2001), pp. 569–592.

28.

Kai M. Thaler, “Ideology and Violence in Civil Wars: Theory and Evidence from Mozambique and Angola,” Civil Wars, Vol. 14, No. 4 (December 2012), pp. 546–567.

29.

Juan E. Ugarriza and Matthew J. Craig, “The Relevance of Ideology to Contemporary Armed Conflict: A Quantitative Analysis of Former Combatants in Colombia,” Journal of Conflict Resolution, Vol. 57, No. 3 (June 2013), pp. 445–477.

30.

Kristine Eck, “Raising Rebels: Participation and Recruitment in Civil War,” Ph.D. dissertation, Uppsala University, 2010.

31.

The literature on material factors and armed mobilization is vast. For an early example, see Ted Robert Gurr, Why Men Rebel (Princeton, N.J.: Princeton University Press, 1970). A recent study is Cederman, Gleditsch, and Buhaug, Inequality, Grievances, and Civil War.

32.

Roger D. Petersen, Western Intervention in the Balkans: The Strategic Use of Emotions in Conflict (Cambridge: Cambridge University Press, 2011).

33.

Petersen, Understanding Ethnic Violence.

34.

Ugarriza and Craig, “The Relevance of Ideology to Contemporary Armed Conflict,” p. 450.

35.

Nico H. Frijda, The Emotions: Studies in Emotion and Social Interaction (Cambridge: Cambridge University Press, 1986).

36.

Jasper, “Emotions and Social Movements.”

37.

William A. Gamson, Talking Politics (Cambridge: Cambridge University Press, 1992).

38.

Jasper, “Emotions and Social Movements.”

39.

David A. Snow et al., “Frame Alignment Processes, Micromobilization, and Movement Participation,” American Sociological Review, Vol. 51, No. 4 (August 1986), pp. 464–481.

40.

Dani Rodrik, “When Ideas Trump Interests: Preferences, World Views, and Policy Innovations,” Journal of Economic Perspectives, Vol. 28, No. 1 (Winter 2014), pp. 189–208.

41.

Ron, “Ideology in Context.”

42.

Gamson, Talking Politics.

43.

Andrea Ruggeri, “Gramsci's Persuaders: Studying Collective Mobilization,” International Studies Review, Vol. 14, No. 4 (December 2012), pp. 677–681; and Stefano Costalli and Andrea Ruggeri, “Forging Political Entrepreneurs: Civil War Effects on Post-Conflict Politics in Italy,” Political Geography, Vol. 44, No. 1 (January 2015), pp. 40–49.

44.

David A. Snow, “Framing Processes, Ideology, and Discursive Fields,” Snow, Sarah A. Soule, and Hanspeter Kriesi, eds., The Blackwell Companion to Social Movements (Oxford: Blackwell, 2004), p. 400.

45.

Cederman, Gleditsch, and Buhaug, Inequality, Grievances, and Civil War, p. 47.

46.

Jon Elster, “Indeterminacy of Emotional Mechanisms,” in Pierre Demeulenaere, ed., Analytical Sociology and Social Mechanisms (Cambridge: Cambridge University Press, 2011).

47.

Indignation is akin to the concept of moral outrage used by Sharon Erickson Nepstad and Christian Smith, “The Social Structure of Moral Outrage in Recruitment to the U.S. Central America Peace Movement,” in Goodwin, Jasper, and Polletta, Passionate Politics, pp. 158–174. Nepstad and Smith argue that moral outrage was the foundation of the American peace movement of the 1980s. Elisabeth Wood includes our concepts of both anger and indignation in her concept of moral outrage. See Wood, Insurgent Collective Action and Civil War in El Salvador, p. 233.

48.

Elster, “Indeterminacy of Emotional Mechanisms,” p. 52.

49.

Jon Elster, “Emotions and Economic Theory,” Journal of Economic Literature, Vol. 36, No. 1 (March 1998), p. 62.

50.

Ibid., p. 105.

51.

Petersen, Understanding Ethnic Violence, p. 40.

52.

Jonathan Mercer, “Feeling Like a State: Social Emotion and Identity,” International Theory, Vol. 6, No. 3 (November 2014), pp. 515–535.

53.

Max Weber, “The Social Psychology of the World Religions,” in Hans H. Gerth and C. Wright Mills, eds., From Max Weber: Essays in Sociology (1946; repr. New York: Oxford University Press, 1958), p. 280.

54.

Elster, “Indeterminacy of Emotional Mechanisms.”

55.

Hannah Arendt, On Revolution (New York: Penguin, 1963), p. 29.

56.

Ibid., p. 62.

57.

Ibid., p. 63.

58.

E.J. Hobsbawm, Primitive Rebels: Studies in Archaic Forms of Social Movements in the 19th and 20th Centuries (New York: W.W. Norton, 1959), p. 10.

59.

Francisco Gutiérrez Sanín and Elisabeth Jean Wood, “Ideology in Civil War: Instrumental Adoption and Beyond,” Journal of Peace Research, Vol. 51, No. 2 (March 2014), pp. 213–226.

60.

Arendt, On Revolution, p. 62.

61.

Judith Goldstein and Robert O. Keohane, Ideas and Foreign Policy: Beliefs, Institutions, and Political Change (Ithaca, N.Y.: Cornell University Press, 1996), p. 8.

62.

Cederman, Gleditsch, and Buhaug, Inequality, Grievances, and Civil War.

63.

Pearlman, “Emotions and the Microfoundations of the Arab Uprisings.”

64.

Petersen, Resistance and Rebellion.

65.

Ibid.

66.

Claudio Pavone, A Civil War: A History of the Italian Resistance (London: Verso, 2014).

67.

Istituto Nazionale di Statistica, Morti e dispersi per cause belliche negli anni 1940–1945 [Dead and missing as a result of war from the years 1940–1945] (Rome: Istituto Nazionale di Statistica, 1957).

68.

Mirco Dondi, La lunga liberazione: Giustizia e violenza nel dopoguerra italiano [The long liberation: Justice and violence in the Italian postwar period] (Rome: Editori Riuniti, 1999), p. 23.

69.

Gianfranco Pasquino, “The Demise of the Fascist Regime and Italy's Transition to Democracy: 1943–1948,” in Guillermo O'Donnell, Philippe C. Schmitter, and Laurence Whitehead, eds., Transitions from Authoritarian Rule, Vol. 1: Southern Europe (Baltimore, Md.: Johns Hopkins University Press, 1986), pp. 45–70.

70.

See Snow, “Framing Processes, Ideology, and Discursive Fields”; Jasper, “Emotions and Social Movements”; and Pearlman, “Emotions and the Microfoundations of the Arab Uprisings.”

71.

Kalyvas, The Logic of Violence in Civil War, p. 248.

72.

Roberto Battaglia, Storia della Resistenza italiana [History of the Italian resistance] (Torino: Einaudi, 1953), p. 138.

73.

Mario Torsiello, Le operazioni dell unita’ italiane nel settembre-ottobre 1943 [Operations of an Italian unit from September to October 1943] (Rome: Ufficio storico dell stato maggiore, 1975), p. 643.

74.

For a thorough historical description of the Italian civil war, see Pavone, A Civil War.

75.

Luca Baldissara, Atlante Storico della Resistenza in Italia [Historical atlas of the Italian resistance] (Milan: Bruno Mondadori, 2000).

76.

The existence of a network during the war does not mean that the bands shared their members, but simply that channels for communication and coordination existed among the bands fighting in the same area. Sometimes there was also a degree of competition among the bands. Their composition was largely stable, and the main sources of fluctuation were casualties and the influx of new members who opted for armed resistance during the later stages of the war.

77.

The bands were mobile, but their area of operation was usually smaller than Italian provinces. Some of them operated across the border between two provinces, but usually in the same area. Baldissara assigned each band to the province where it usually acted. See Baldissara, Atlante Storico della Resistenza in Italia.

78.

A. Colin Cameron and Pravin K. Trivedi, Microeconometrics: Methods and Applications (Cambridge: Cambridge University Press, 2005).

79.

Piergiorgio Corbetta and Maria Serena Piretti, Atlante Storico-Elettorale d'Italia, 1861–2008 [Historical atlas of Italian elections, 1861–2008] (Bologna: Zanichelli, 2009).

80.

Giame Pintor, Il sangue d'Europa, 1939–1943 [The blood of Europe, 1939–1943] (Rome-Bari: Laterza, 1965), pp. 180–181.

81.

Nuto Revelli, Le due guerre: Guerra fascista e guerra partigiana [The two wars: Fascist war and partisan war] (Torino: Einaudi, 2003), p. 123.

82.

Jeff Goodwin, No Other Way Out: States and Revolutionary Movements 1945–1991 (Cambridge: Cambridge University Press, 2001); and Wood, Insurgent Collective Action and Civil War in El Salvador.

83.

Istituto Nazionale di Statistica, Morti e dispersi per cause belliche negli anni 1940–1945.

84.

Istituto Nazionale di Statistica, Censimento della popolazione italiana [Census of the Italian population] (Rome: Istituto Nazionale di Statistica, 1936); and Istituto Nazionale di Statistica, Censimento della popolazione italiana (Rome: Istituto Nazionale di Statistica, 1951).

85.

Istituto Tagliacarne, Il sistema camerale tra imprese e istituzioni a 150 anni dall'Unità d'Italia [The cameral system between enterprises and institutions during 150 years of Italian unification] (Rome: Edizioni Camere di commercio d'Italia, 2011).

86.

Automobile Club Italia, Serie storiche sullo sviluppo della motorizzazione e sull'incidentalità stradale in Italia negli anni 1921–2007 [Historical time series data on road accidents in Italy from the years 1921–2007] (Rome: Automobile Club Italia, n.d.), http://www.aci.it/.

87.

Yuri M. Zhukov, “Roads and the Diffusion of Insurgent Violence: The Logistics of Conflict in Russia's North Caucasus,” Political Geography, Vol. 31, No. 3 (March 2012), pp. 144—156.

88.

Giampaolo Pansa, Guerra Partigiana tra Genova e il Po [Partisan warfare between Geno and the Po] (Rome-Bari: Laterza, 1967).

89.

Figure 2 is based on table 1A in the online appendix. We provide several tables in the online appendix, but in the article we privilege graphical representations. The appendix is available online via http:dx.doi.org/10.7910/DVN/CBEDOJ.

90.

In the online appendix, we also provide models using as the dependent variable a dummy that measures whether there was any partisan mobilization in a province. The results are substantially the same.

91.

The average number of bands in a province is 4 with a standard deviation of 6.

92.

See table in the online appendix.

93.

Santo Peli, La resistenza in Italia [The resistance in Italy] (Torino: Einaudi, 2004).

94.

See table 1B in online appendix.

95.

We thank an anonymous reviewer for suggesting this robustness check.

96.

The variable “anger” has a very low correlation (0.44) with the variable used in the within-province analysis measuring the number of people under the control of the Fascist regime from 1922 to 1943.

97.

We have used the same variable for our within-province analysis, and the results are similar and robust. Tables with full models (tables D5 and D6) and a similar figure (figure D6) for the within-province analysis are in the online appendix.

98.

On the role of social ties, see Petersen, Resistance and Rebellion. See also Henrik Urdal, “A Class of Generations? Youth Bulges and Political Violence,” International Studies Quarterly, Vol. 50, No. 3 (September 2006), pp. 607–629. On security-related benefits, see Kalyvas, The Logic of Violence in Civil War; and Kalyvas and Kocher, “How ‘Free’ Is Free Riding in Civil War?”

99.

Petersen, Resistance and Rebellion; and Urdal, “A Clash of Generations?”

100.

Istituto Nazionale di Statistica, Risultati sommari del censimento della popolazione italiana 1921 [Summary of the results of the 1921 census of the Italian population] (Rome: Istituto Nazionale di Statistica, 1926).

101.

Petersen, Resistance and Rebellion; and Urdal, “A Clash of Generations?”

102.

Istituto Nazionale di Statistica, Censimento della popolazione italiana (1936); and Istituto Nazionale di Statistica, Censimento della popolazione italiana (1951).

103.

Kalyvas and Kocher, “How ‘Free’ Is Free Riding in Civil War?”

104.

Ibid.

105.

Luca Baldissara and Paolo Pezzino, Il massacro: Guerra ai civili a Monte Sole [The massacre: Warfare against civilians in Monte Sole] (Bologna: Il Mulino, 2009); and Paolo Pezzino, Sant'Anna di Stazzema: storia di una strage [Sant'Anna of Stazzema: The story of a slaughter] (Bologna: Il Mulino, 2008).

106.

Dondi, La lunga liberazione; and Baldissara, Atlante Storico della Resistenza in Italia.

107.

Data from Marco Gioannini and Giulio Massobrio, Bombardare l'Italia. Storia della guerra di distruzione aerea, 1940–1945 [The bombing of Italy: A history of warfare and aerial destruction, 1940–1945] (Milan: Rizzoli, 2007).

108.

We thank Marco Ruzzi, of the Istituto Storico della Resistenza in Cuneo, and Armando Scardaoni and Silvio Lugaro, of the Istituto Storico della Resistenza in Savona, for sharing their data on World War II deaths with us.

109.

Istituto Storico della Resistenza di Cuneo (2007) and Istituto Storico della Resistenza di Savona (2013).

110.

Istituto Nazionale di Statistica, Censimento della popolazione italiana (1951).

111.

The Fascist regime dates from 1922; 1943 is the year of the armistice and the beginning of the Italian civil war.

112.

In figure 1 of the online appendix, we show the positive correlation between these two proxies for radical ideology. In principle, it could be possible (though improbable) that the communists in areas under Fascist control were not part of a network, but simply individual, unconnected communists. Historical sources show that this was not the case, however. See, for instance, Pansa, Guerra Partigiana tra Genova e il Po; and Giorgio Gimelli, Cronache Militari della Resistenza in Liguria [Military chronicle of the resistance in Liguria] (Genoa: Istituto Storico della Resistenza Ligure, 1985).

113.

In the baseline model, we pool together all municipalities from Savona and Cuneo (N = 273). Analyzing Cuneo and Savona separately, however, and pooling the two provinces together does not provide different results. See table 4A in the online appendix.

114.

In models in the online appendix we also control for unemployment, altitude, and population density at the municipality level.

115.

See the online appendix.

116.

Pietro Secchia, Il Partito comunista italiano e la guerra di liberazione, 1943–1945 [The Italian Communist Party and the war of liberation, 1943–1945] (Milan: Feltrinelli, 1971), p. 1064.

117.

Gianpasquale Santomassimo, ed., La notte della democrazia [The night of democracy] (Milan: Saggiatore, 2003), p. 350.

118.

Ermanno Gorrieri, La Repubblica di Montefiorino. Per una storia della Resistenza in Emila [The Republic of Montefiorina: A history of the resistance in Emilia] (Bologna: Il Mulino, 1966); Pansa, Guerra Partigiana tra Genova e il Po; and Pavone, A Civil War.

119.

Gorrieri, La Repubblica di Montefiorino.

120.

G.N. (the initials indicate first name and surname of the interviewees), box 1, Interviste e Testimonianze [Interviews and testimony], Istituto Storico della Resistenza di Savona, Savona, Italy.

121.

Gimelli, Cronache Militari della Resistenza in Liguria, p. 212.

122.

B.F., box 2, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana, Florence, Italy.

123.

B.L., box 2, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.

124.

Pearlman, “Emotions and the Microfoundations of the Arab Uprisings.”

125.

Posen, “The Security Dilemma and Ethnic Conflict.”

126.

R.C., box 2, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.

127.

L.M., box 5, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.

128.

We use the term “resentment” to keep the wording of the interview. The context clarifies that the meaning of the concept is close to what we define as “indignation.”

129.

A.A., box 4, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.

130.

G.B., box 4, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.

131.

S.S., box 1, Interviste e Testimonianze, Istituto Storico della Resistenza di Savona.

132.

A.B.S., box 1, Interviste e Testimonianze, Istituto Storico della Resistenza di Savona.

133.

G.S., box 1, Interviste e Testimonianze, Istituto Storico della Resistenza di Savona.

134.

F.R., box 5, Interviste e Testimonianze, Istituto Storico della Resistenza in Toscana.