GROUP SIZE AND THE USE OF VIOLENCE BY RESISTANCE CAMPAIGNS: A MULTILEVEL STUDY OF RESISTANCE METHOD
Christopher J. Cyr and Michael Widmeier
ABSTRACT
We examine why some groups use violence while others use nonviolence when they push for major political change. Nonviolence can be less costly, but nonstate actors must mobilize a large number of people for it to be successful. This is less critical for violent rebellion, as successful attacks can be committed by a small number of people. This means that groups that believe that they have the potential to mobilize larger numbers of people are less likely to use violence. This potential is related to the lines along which the group mobilizes. Campaigns mobilized along ethnic or Marxist lines have fewer potential members and are most likely to use violence. Prodemocracy campaigns have a higher number of potential members and are more likely to use nonviolence. For movements against a foreign occupation, campaigns in larger countries are more likely to use nonviolence. These predictions are supported in a multilevel logit model of campaigns from 1945 to 2006. The mechanism is tested by looking at the interactive effect of democratic changes on the likelihood of nonviolence and looking at a subsample of 72 campaigns that explicitly draw from certain ethnic or religious groups.
Keywords: Civil war; nonviolent resistance; protest; rebel groups; mobilization; conflict
In the early twentieth century, citizens of India conducted a campaign of resistance against British colonial rule. The leader of these protests, Mahatma Gandhi, embraced nonviolence, saying, ānonviolence is the greatest force at the disposal of mankind, it is mightier than the mightiest weapon of destruction devised by the ingenuity of manā (Gandhi & Attenborough, 2001). Around the same time, citizens in Kenya began to oppose British rule, employing nonviolent methods of resistance as in India. Despite both movements taking place at similar times and targeting the same regime, the Indian and Kenyan opposition movements experienced different outcomes. In India, with a substantially larger population than Kenya, nonviolent protest eventually played a key role in the British withdrawal in 1950. In Kenya, however, the British government quickly squashed the protests. As nonviolent tactics had proven to be ineffective, Kenyan resistance organizations eventually turned to violent rebellion in an attempt to expel British rule.
The cases of India and Kenya raise several important questions concerning resistance movements and the choice between violence and nonviolence. Why was nonviolence effective when the Indian people targeted the British government, while it was not effective in Kenya? What factors accounted for Kenyans' decision to turn to violence?
These questions are salient as the empirical record reveals a good deal of variation in the choice of strategies that groups employ when they campaign against their government. The case of the Philippines in the 1980s is a clear illustration of this historical variance. The New People's Army, a Maoist rebellion, initiated a campaign of guerilla warfare against the government of the Philippines. A contingent of the minority ethnic Moro Islamic community also organized a rebellion against the state, launching a violent separatist movement. Within the same relative time period, the People Power movement emerged as yet another faction organized in opposition to the Philippine government. However, the People Power campaign stood in contrast to the New People's Army and the Moro insurgency, as it was a prodemocracy organization that used nonviolent tactics.
All three movements stood in opposition to the government of the Philippines, but the Moro Islamic Liberation Front and the Marxist New People's Army used violence, whereas People Power employed a nonviolent approach. While these groups targeted the same regime, their resistance strategies diverged when it came to the choice of violent versus nonviolent tactics. What accounts for this variation in tactics? Here, we explain differences in the use of violence and nonviolence by looking at how ex ante expectations of success, based on each group's potential number of supporters, impact the tactics of nonstate actors when they organize large-scale resistance efforts against a government.
Previous researchers have largely studied civil war at the state level, using state characteristics such as GDP per capita and the presence of mountainous terrain as explanatory variables that influence the likelihood of the outbreak of civil conflict (Collier & Hoeffler, 2002; Fearon & Laitin, 2003). However, research efforts that rely on country-level data do not account for variation in the features of the groups that decide to rebel. In recent years, scholarship has emerged that examines this variation (Cunningham, Gleditsch, & Salehyan, 2009). Because these studies generally only examine groups actively engaged in rebellion, they cannot consider the onset of violent conflict as a dependent variable. Here, we examine this variation by considering groups that are actively mobilized against the state, but vary on their use of violence or nonviolence.
We argue that the choice between nonviolence and violence is an instrumental decision on the part of the nonstate actor, based on which tactic is viewed as being more likely to increase the likelihood of securing concessions from the state. To this point, previous research has established that nonviolence can be more effective than violence for resistance movements attempting to gain concessions or alter their status quo (Chenoweth & Stephan, 2011). Carrying out nonviolent tactics is generally less costly to groups than the use of violence, but the number of supporters that a group mobilizes is critical to the success of nonviolent organizations. Groups that are unable to mobilize a large network of supporters are not able to hold the large rallies and general strikes that are hallmarks of successful nonviolent resistance (Cunningham, 2013). Conversely, group size and the success of mobilization efforts are less important for groups that employ violence. Terrorist attacks only require one person to inflict damage, and guerrilla groups can hide from the state to maintain a violent resistance campaign with relatively few people. Because of this, groups that anticipate having a large support base are more likely to use nonviolence due to its lower costs and greater effectiveness. Those that anticipate facing limitations on the development of their group and the size of their support base are more likely to use violence. This decision is made due to the notion that groups are aware of the need for a substantial number of followers in order to execute nonviolent approaches successfully.
We form testable hypotheses regarding the probability that a group will employ violent versus nonviolent means of resistance. We focus on the lines along which groups are mobilized, and how this can act as an ex ante indicator of the group's support base. Groups mobilized along ethnic or religious lines are generally unlikely to attract support outside of their specific ethnic or religious group. This creates a theoretical cap on the number of potential supporters that is lower than the size of the population as a whole. Furthermore, there exists many types of resistance campaigns that have difficulty gauging their level of support ex ante, such as those mobilized along Marxist lines. Groups with no inherent limitations on future organizational growth are unable to determine whether they will cease to exist on one hand or grow to include tens of thousands of cadres. These two types of campaigns, therefore, are more likely to use violence as they have no expectation that they will become large enough to effectively use nonviolence. On the other hand, groups such as prodemocracy campaigns are typically more inclusive than many other types of movements. As a result of this, a prodemocracy campaign could potentially mobilize large sections of a country and recruit a diverse set of followers. This mobilization mechanism makes these campaigns significantly more likely to use nonviolence than those with limited mobilization potential. Among campaigns designed to expel a foreign occupation, the likelihood that a campaign will use nonviolence increases as the population of the occupied country increases. Simply put, campaigns in opposition to foreign occupation in populous states will have a broader pool of potential followers to draw from than those in a smaller state, thus making the former campaigns more likely to use nonviolence.
We test our predictions using data on resistance campaigns from 1945 to 2006. We also test the mechanism behind our hypothesis in two ways. First, we examine the interactive effect of democratic change and prodemocracy mobilization on the likelihood of violence. In states that are becoming more democratic, where prodemocracy movements have a larger potential support base, the relationship between prodemocracy mobilization and nonviolence is stronger. We then examine a subsample of 72 campaigns mobilized along ethnic or religious lines. Our model implies that campaigns that represent large ethnic or religious groups, relative to the country as a whole, have potential to mobilize more supporters than ones that represent smaller groups. The former groups are therefore less likely to use violence. The results provide some initial support for this expectation, with one caveat. Five of these campaigns drew from groups that make up a majority of their state's population, which gives them significant influence over the findings. In two of these cases, the choice of tactics was driven largely by external events, thus these two cases are excluded from the analysis. When these are included, the results are significant at the 90 percent level of confidence but not the 95 percent level of confid...