Debunking the Myths Behind Nonviolent Civil Resistance

Abstract

Scholars argue that nonviolence is likelier to cause political change in comparison to other strategies, including violence. This study identifies issues throughout this literature ranging from coding procedures, observational sampling, to interpretations of phenomena. If unarmed violence, reactive violence, and omitted cases are analyzed, nonviolent success rates are worse than formerly considered. Inclusion of 19th century (1800–99) cases and previously unanalyzed cases from the 20th century reveals that nonviolent campaigns experienced a 48% rate of success, whereas campaigns that adopted unarmed violence were 61% successful, campaigns utilizing reactive unarmed violence were 60% successful, and 30% of fully violent campaigns were successful. Nonviolence is not a causal determinant of political change, but rather, its implementation falls short of a probabilistic coin toss. There is reason to presume this literature is biased toward elite interests in similar ways to how scientific inquiry on dietary and substance guidelines has historically been skewed by corporatism.

Keywords

protest revolution nonviolence social movements sociology political science

Introduction

Over the last two decades, nonviolent civil resistance (also referred to as civil resistance, nonviolence, passive resistance, people power, and peaceful protest) has garnered increased interest among social scientists. Proponents of nonviolence have emphasized that it is a realistic practice that can be used by regular people who are in search of socio-political change (Ackerman and Kruegler, 1994; Sharp, 1973a, 1973b, 1973c, 1980; Zunes, 1997). In social science, political scientists have identified positive correlations between large-scale nonviolent collective action and outcomes such as policy change, regime transition, and democratization (Ackerman and Duvall, 2000; Celestino and Gleditsch, 2013; Chenoweth and Stephan, 2008, 2011; Nepstad, 2011; Schock, 2013). Researchers in this literature contend that nonviolent civil resistance movements are more successful than violent insurgencies in obtaining major political concessions.

While early inquiry on civil resistance was based on comparative case studies of successful nonviolent revolutions (Ackerman and Kruegler, 1994; Sharp, 1973a, 1989), it was only around a decade ago when Chenoweth and Stephan’s (2008) statistical (econometric) study was published in the journal International Security. Here the first quantitative assessment of nonviolence was produced. An analysis was presented of oppositional movements featuring 323 different “nonviolent” and “violent” campaigns (1900–2006). It was discovered that nonviolent resistance is more successful (53%) than violent rebellion and insurgency (26%). Nonviolent movements were observed to be likelier to bring about policy change, less likely to result in civil war, and societies that experience a nonviolent revolution are likelier to be democratic after a transition (Bethke, 2017; Bethke and Pinckney, 2019; Celestino and Gleditsch, 2013; Kim and Kroeger, 2019). Chenoweth and Stephan’s analyses ended up being turned into a book titled Why Civil Resistance Works (2011, Columbia University Press) that won the American Political Science Association’s best book award in 2012.

Findings from the quantitative study of civil resistance have since served as a common reference point in interdisciplinary outlets and have been accepted by much of the scientific community as representing unbiased, general knowledge. Yet only recently have these findings started to receive critical evaluation. In sociologically oriented reviews, scholars have questioned the sample of cases analyzed in the foundational studies found in quantitative nonviolence inquiry literature. Through investigation of revolutions in 20th-century Latin America, Lehoucq (2016) discovered that the sample of resistance movements presented by the NAVCO (Nonviolent and Violent Conflict Outcomes) data set suffers from omitted cases of failed nonviolent movements as well as successful violent movements. Others have pointed out that the NAVCO data feature several observational and coding deficiencies pertaining to the dichotomization of violence/nonviolence (Anisin, 2018; Kadivar and Ketchley, 2018; Pressman, 2017). Many campaigns were labeled as being nonviolent when empirically, these movements actually contained a substantial degree of unarmed violence that manifested in the form of rioting, rock throwing, car and building burning, and even in the usage of Molotov cocktails.

Thus far, these significant points have been made in reference to isolated time periods of history or have been presented through analysis of a strict regional basis. If one takes these points into consideration across a wide historical and cross-national sample of cases, is nonviolence still more effective than other forms of resistance? This present study addresses this question. It undertakes a comprehensive assessment of the same series of data as originally used in the aforementioned literature featuring 323 violent and nonviolent campaigns (1900–2006). In doing so, this study first takes the occurrence of cases featuring unarmed violence into consideration as well as cases featuring reactive violence in which protesters responded to state violence with unarmed violence out of self-defense. It then carries out a statistical analysis of success rates on a more genuine sample of cases that includes omitted campaigns from the 20th century (1900–2006), which were not in the NAVCO data. Once this is done, cases stemming to the 19th century are also included—resulting in an analysis of 396 cases in total spanning the time period of 1800–2006.

The analysis reveals surprising findings: nonviolence is less effective than previously assumed, while violent insurgency is a bit more effective than assumed, and campaigns featuring reactive unarmed violence and unarmed violence are the most successful of all. To make sense of these results and their implications, subsequent sections of this study present an overview of the soundness of causal mechanisms that have been associated with nonviolent campaign success. It is demonstrated that causal mechanisms that have been linked to nonviolent strategy and campaign success are not necessarily tied to only nonviolent strategies, but are also complimentary to other forms of resistance. There also exist notable methodological perils in this literature as studies that have advanced the civil resistance argument represent an exemplary case of social scientists favoring causal inference-based argumentation over description. Above all, ideologically, there are significant biases stemming to this literature’s funding bodies and in political characteristics that underlie this widely cited argument.

Nonviolence as a Causal Determinant of Political Change

Nonviolent civil resistance is defined as the sustained usage of methods of nonviolent action by civilians (Schock, 2013: 277). In contrast to classical inquiry on protest and contentious politics, scholars studying civil resistance have conceptually differentiated nonviolence from traditional social movement research. As Schock (2013) notes, the theoretical roots of civil resistance are in religious traditions, in Gandhian philosophy and in varying forms of anarchist thought. This contrasts to the structuralist roots of social movement studies. Similarly, whereas researchers of revolution and social movements have assessed structural factors of social change, mobilization, and political context, nonviolent civil resistance scholars have focused on strategies and techniques of action as well as theorized mechanisms of nonviolent change (Schock, 2013: 280).

In the 1970s, nonviolence gained prominence as an academic topic in the study of socio-political change—with Gene Sharp’s work being pioneering in this respect. Sharp categorized 198 different tactics of nonviolent direct action that were observed through comparative case studies of successful cases of civil resistance (Sharp, 1973a; 1973b; 1973c). Sharp distinguished between symbolic nonviolent resistance, noncooperation (strikes or mass economic boycotts), and nonviolent interventions such as the occupation of public and or social places. At this point in time, the study of civil resistance was indeed a narrowly focused enterprise when compared to the interdisciplinary nature of social movement studies. Building on Sharp’s work, a number of qualitative studies on nonviolence were published in the 1990s (Ackerman and Kruegler, 1994; Zunes, 1997). By the 2000s, quantitative research on civil resistance emerged. This was made possible by the multilevel NAVCO 1.0 (Nonviolent and Violent Campaigns and Outcome) data set that was put together by political scientists during a data collection project in the Korbel School of International Studies at the University of Denver (Chenoweth, 2011).

Findings from NAVCO 1.0 were first introduced in 2008 by Chenoweth and Stephan (2008). Subsequently, data were refined with the NAVCO 2.0 data set (Chenoweth and Lewis, 2013). The arguments accompanying these data sets stem to observations of annual cross-national data and a unit of analysis that was conceptualized as a “campaign.” Campaigns represent an identifiable political group and are defined as “a series of observable, continuous, purposive mass tactics or events in pursuit of a political objective” (Chenoweth and Lewis, 2013: 2). It is important to situate this particular classification into perspective as campaigns are defined in this literature in a particular way that makes them somewhat different from what many would commonly think of as protest or collective action. While all campaigns do indeed carry out protest, they are more comparable to highly threatening social movements that sustain their mobilization for a substantial duration of time such as the course of a year or more.

Specifically, to have been included as an observation in the NAVCO data sets, the data creators conceptualized the term “maximalist” in reference to their unit of analysis: the campaign. All campaigns in this sample of cases made highly threatening demands against status quos. In detail, campaigns either sought regime change, territorial succession, or were anti-occupation based (dissenting against colonial powers or occupying foreign states). Through statistical inquiry on 323 “violent” and “nonviolent” campaigns the authors concluded that nonviolent resistance opposition movements are much larger on average than violent, are likelier to spur security force and bureaucratic defections, and most importantly have experienced greater success in winning major political conflicts.¹ Both violence and nonviolence were coded based on the “primary” resistance method used in a given campaign year (Chenoweth and Lewis, 2013). As noted by the makers of the noted datasets, “When a campaign relies primarily on nonviolent methods such as these as opposed to violent or armed tactics, the campaign can be characterized as nonviolent” (Chenoweth and Lewis, 2013: 3). Violent resistance, in contrast, is observed in campaigns that strategically “involve the use of force to physically harm or threaten to harm the opponent” (Chenoweth and Lewis, 2013: 3).

These findings and accompanying arguments in this literature have been interpreted as being objective. Articles about the effectiveness of nonviolent resistance are frequently published in outlets such as Foreign Affairs or broadly read news organizations such as the BBC. Pundits frequently draw on this literature and its unheralded findings. For example, a widely circulated piece by Dave Robinson (2019) in BBC (Future) was published in 2019 and it was titled “The 3.5% rule: How a small minority can change the world.” Its content stem to Chenoweth and Stephan’s, Why Civil Resistance Works, and it described the now well-known story of how the NAVCO data project was put together and what the seemingly miraculous results entail for real-world political struggles today.

Although broadly accepted and culturally pronounced, this literature and its findings are ridden with problems. As this study will reveal, quantitative research on civil resistance is contingent on data, but these data are based on imprecise, and in some instances, blatantly erroneous measurements of empirical phenomena. Data also do not represent an empirically valid sample of nonviolent and violent social movements/campaigns. To make matters even more grim, this literature, its results, and implications are likely driven by political bias and interests in similar ways as have been observed in explicit historical studies in the natural sciences such as those tied to the tobacco and sugar industries.

Erroneous Classifications

Quantitative research on nonviolent civil resistance has produced generalizations drawn from data that are based on inaccurate measurements of empirical reality. By imprecisely measuring key explanatory variables, this literature has, in turn, produced erroneous results. Specifically, nonviolence has been conceptualized, coded, and measured in incorrect ways. Before pinpointing these errors, let us take an example that illustrates how accurate measurements of observations of phenomena should be made. The following example is from the natural sciences. When laboratory studies on substances and their effect on the nervous system are carried out, researchers must be 100% certain that their placebo and main independent variables accurately depict the material substances (or their absence) that they believe have a potential causal association with the outcome.

In one of the hundreds of studies investigating the impact of caffeine on blood pressure, researchers (Lovallo et al., 2004) administered different capsules with caffeine contents of 0 mg, 300 mg, and 600 mg to individuals over specified periods of time in order to investigate how caffeine impacts hypertension and blood pressure. The control group did not receive any caffeine. Without going into the results of the study, the certainty underlying the administration of caffeine (i.e. it actually was caffeine) was crucial to not only the integrity of the study, but to general purpose of scientific inquiry. Researchers could not have provided their subjects with caffeine mixed with some other substance, such as sugar or fructose, and deemed their assessment to be one of weighing the influence of only caffeine on the outcome. Unfortunately, the measurement of nonviolent civil resistance does just this. It lumps together a number of resistance strategies (some of which are in no way strictly “nonviolent”) and claims to measure nonviolence.

Nonviolence is assumed to be synonymous with civil peaceful protest, yet research has been carried out on data that feature variables that were coded through what is referred to as a “primarily nonviolent” campaign. As noted in the variable of “primary method”—this “denotes the primary type of resistance method used in a campaign year,” with 0 = primarily violent campaign and 1 = primarily nonviolent campaign (Chenoweth and Lewis, 2013: 6). Primarily nonviolent campaigns contrast with primarily violent campaigns. Both campaign types lump together an array of different tactics of protest and rebellion into a dichotomous variable. What’s more, in their sample of cases, most of the “primarily” violent campaigns are partially drawn from the Correlates of War project—a series of data that is notably defined by its strict inclusion of cases of conflict based on the minimum of 1000 battle deaths per year threshold.

In justifying these classifications, Chenoweth and Lewis (2013) present several points in an attempt to narrow down a definition of nonviolence; these points, however, do not render a conceptually adequate explanation of why violence and nonviolence are coded as they are. Specifically, a major problem is unaddressed in that unarmed forms of violence are coded as nonviolent rather than being placed into the violent campaign classification or into some other median threshold. The scientists in the aforementioned study on caffeine would be in a tough position to contend that substances that contain “primarily” caffeine are what cause or do not cause hypertension. In the scope of nonviolent civil resistance however, aggregations of a variety of different strategies are dichotomized as nonviolent/violent. The latter, importantly, is based on whether a campaign featured civilians or armed militants.² Measurements of this sort are erroneous and indeed synonymous with researchers stating that they administered a substance that is “primarily caffeine” or one that was “primarily placebo” to two sets of subjects. This would not be an acceptable research strategy and would not pass peer review in any reputable scientific outlet in pharmacology. Yet in today’s social scientific milieu, the domination of causal inference-based argumentation (and data hegemony more broadly) did indeed enable such a research strategy to not only pass peer review, but to win awards and subsequently shape public knowledge of resistance and revolution in the contemporary world.

Among the first scholars to point out these inconsistencies noted that actions such as rock throwing (as observed in the first Intifada, Palestine) are considered to be nonviolent in quantitative research on civil resistance (Pressman, 2017). To overcome these shortcomings, Pressman (2017) proposed that rock throwing be viewed as unarmed violence, and for nonviolence to be conceptualized along a “spectrum” that begins with full nonviolent resistance and ranges all the way to “catastrophic violence” involving nuclear weapons. Anisin (2018) similarly proposed for scholars to observe violence/nonviolence at an event-level when analyzing dissent and repression, specifically by disaggregating different strategies according to a semi-continuous measure ranging from 0 to 1 [0 – Violence (lethal weapons, arms, knives); .33 – Lesser violence (Rocks, stones, harmful blockades); .67 – Small scale/scattered violence (physical confrontations); 1 – Nonviolence (civil resistance, no physicality)]. These points bring us to salient issues behind the noted literature’s omission of strategies including unarmed forms of violence and similar manifestations of dissent such as those that arise in the form of reactive self-defense-based unarmed violence.

Nonviolence, Unarmed Violence, and Reactive Unarmed Violence

Unarmed forms of violence explicitly feature direct action waged by participants that are not armed with firearms, rifles, shotguns, bombs, or grenades. Protests featuring unarmed violence contain lesser forms of weaponry that are most commonly used to clash with state security forces or police. Throughout recent history, unarmed violence such as the usage of stones, rioting, sticks, car burning, building burning, among other actions have not only been prevalent, but appear to have been pivotal to the success of noteworthy revolutions. Kadivar and Ketchley (2018) investigated processes of political liberalization that arose in 103 non-democracies over the span of 1990–2004 and discovered that riot occurrences in these cases were positively associated with liberalization. Likewise, their analysis of 80 different democratic transitions revealed that a high degree of unarmed violence was carried out during pro-democratic opposition movements. While these results only encompass just over two decades of history, they are telling.

For example, Kadivar and Ketchley (2018) describe the 2000 Serbian “OTPOR” revolution as a transition that featured a high degree of unarmed violence. Here the authors explain that during the fall of S. Milosevic’s regime, protesters were frequently throwing rocks and bottles at police and regime supporters. There were even several incidents leading up to the election in 2000 where protesters threw not only rocks, but Molotov cocktails at authorities. It would be difficult for one to make the claim that this revolution was not influenced by such actions. However, scholars of civil resistance do make such claims while simultaneously they gloss over such occurrences or attempt to disaggregate them into a “radical flank.” Chenoweth and Stephan (2011) describe this case as a “sustained systematic nonviolent” movement that initiated noncooperation to bring down a regime (Chenoweth and Stephan, 2011: 23). Similarly, Gene Sharp described OTPOR as being “generally of a symbolic nature, using nonviolent methods of protest and persuasion” (Sharp, 2005: 318).

What’s more, in the popular documentary Bringing Down a Dictator (2002) (which still gets shown in seminars on nonviolent civil resistance), an in-depth analysis of the Serbian revolution is presented based around activist interviews and original protest footage. In detail, the documentary offers viewers a chronology of events, video of organizational techniques, and interviews with U.S. State Department agents as well as foreign policy advocates who reveal they provided financial assistance for resources for OTPOR. These resources were logistical, but crucial to the organizational capabilities of the movement—phones, fax machines, and printers were provided. What is not mentioned or shown in the documentary is that a variety of unarmed violent actions were utilized by opposition during this revolution. These actions included stone heaving at authorities. At the height of this movement, stone throwing protesters managed to route riot police and seize the parliament building (Erlanger, 2000; Kadivar and Ketchley, 2018; The Herald Spectator, 2000, Glasgow 2000).

Perhaps the most salient point to consider here is that 2000 Serbia is not an outlier—it might actually be an exemplar. In literature on nonviolent civil resistance, scholars frequently bring up cases they believe are typical of revolution brought about by peaceful protest; cases such as 2000 Serbia, 2003 Georgia, or the 1986 Philippine revolution are drawn on as points of reference of peaceful revolution throughout different studies and books (Ackerman and Kruegler, 1994; Chenoweth and Stephan, 2008, 2011; Nepstad, 2011; Sharp, 1973; 2010). In the 1986 Philippine anti-Marcos opposition campaign, there were incidents in which protesters threw stones, wooden staves, and homemade grenades as well as Molotov cocktails at authorities (Adams, 1993: 225). This leads us to an impasse: if cases that have served as exemplars of nonviolent revolution actually were ridden by unarmed violence, could this implicate the results found in the literature?

There also have been instances in which protesters dissented nonviolently, but then turned to unarmed violence as a mode of self-defense or via a change in strategic deliberation. These particular dynamics are not accounted for in the aforementioned data sets. Reactive unarmed violence differs from the straightforward usage of unarmed violence specifically in its reactive distinction. As subsequent sections of this study will reveal, reactive unarmed violence is rarer than the utilization of fully violent action and is also rarer (slightly) than unarmed violence. Moreover, there are a number of historical cases that have featured protesters getting attacked by state security forces or police then subsequently defending themselves through unarmed violence. For example, in the 1919 Egyptian anticolonial revolution, dissent began with civil disobedience that was nationally waged by heterogeneous segments of society in both urban and rural areas; then over the course of several weeks, over 500 civilians were killed. During these mass killings, protesters started responding to state violence with rocks, sticks as well as the burning of British administrative buildings (Lees et al., 2015). Cases featuring manifestations of reactive unarmed violence are diverse and include the 1865 Jamaican rebellion, the 1965 Bahrain March Intifada, the 1992 Malawi anti-regime protests, among other oppositional challenges.

The Omission of Cases

In addition to the aforementioned concern, the NAVCO data projects (1.0; 2.0) do not actually feature a fully accurate historical sample of cases. While few data sets or projects result in the collection of a true universe of cases, the issue with the noted compilation of data is that it is adversely incomplete because a number of failed nonviolent revolutions are left out as are a plethora of successful violent revolutions. This is a fundamental problem because researchers cannot make sweeping generalizations about correlations identified in statistical analyses if data do not offer complete representations of empirical reality–especially if they present us with incomplete representations of history. The first scholar to point this out was Lehoucq (2016) who discovered that these databases do not include thirteen different campaigns (out of 26 in total) belonging to the Latin American region. Lehoucq notes that out of the thirteen overlooked (missing) cases, eleven were failures. What’s more, seven failures were nonviolent campaigns.

If we consider success rates of nonviolent/violent campaigns in Latin America, Lehoucq argues that nonviolent campaigns only achieved their goals 42% of the time (6 out of 14), which is markedly lower than Chenoweth and Stephan’s claim of a near-60%. On the other hand, 41% of violent campaigns (5 out of 12) were successful. None of these cases were included into the data sets, even though all cases occurred in years covered by the span of data. These campaigns are not obscure or understudied by scholars and historians. With reference to the accessibility of literature on these campaigns, Lehoucq notes, “None of these campaigns, I should add, requires reading obscure Spanish-language sources” (Lehoucq, 2016: 278). While proponents of nonviolent civil resistance would likely be quick to disregard Lehoucq’s claims because of their limited geographical basis (only representative of Latin America and not the entire world), Lehoucq’s findings are very relevant as they point toward a major irregularity in these findings. As it turns out, there are a number of other blatant omissions in this literature as well as controversial inclusions that have thus far not been acknowledged.

A primary example of the failure to produce and assess an accurate historical sample of campaigns can be observed in the inclusion of a campaign in the NAVCO 1.0 data set titled “Kirghiz and Kazables rebels”(1916–7) that dissented against the Tsarist regime (campaign coded as having achieved success in NAVCO 1.0). This inclusion is mind-boggling not due to the relatively understudied characteristics of this particular campaign, but due to the exclusion of the actual Bolshevik campaign that overthrew the Romanov monarchical regime in the subsequent year. Likewise, an “anti-Bolshevik” campaign (1917–21) is included in the data, without inclusion of the Bolsheviks themselves. These examples are only a few out of dozens of omissions that the authors of this present study have identified during extensive research on oppositional campaigns over the period of 1800–2006. These omissions will be unveiled in the subsequent section of this study.

Reassessing Success Rates

This section will reevaluate results produced by Chenoweth and Stephan (2008) through assessment of cases featuring unarmed violence followed by omitted campaigns as well as cases featuring reactive self-defense based unarmed violence. In total, 396 different campaigns will be analyzed. This study will assess statistical patterns of success rates of four strategic actions: nonviolent campaigns, violent campaigns, campaigns featuring unarmed violence as well as those featuring reactive unarmed violence. All data used in this study can be accessed via a public anonymous link featuring CSV files.³

Aforementioned success rates from the original studies on civil resistance are displayed in Table 1. Success is defined in the NAVCO 1.0 and 2.0 data sets (Chenoweth, 2011; Chenoweth and Lewis, 2013) through a dichotomous variable that identifies:

whether the campaign achieved 100% of its stated goals within a year of the peak of activities. In most cases, the outcome was achieved within a year of the campaign’s peak. Some campaigns’ goals were achieved years after the “peak” of the struggle in terms of membership, but the success was a direct result of campaign activities. When such a direct link can be demonstrated, these campaigns are coded as successful.

Table 1.

Original Success Rates (NAVCO 1.0).

	Success	Limited success
Violent campaign (N = 217)	26%	13%
Nonviolent campaign (N = 106)	53%	25%

Additionally, limited success is a dichotomous variable that identifies:

whether the campaign achieved some of its stated goals within a year of the peak of activities. When a regime makes concessions to the campaign or reforms short of complete campaign success, such reforms are counted as limited success.

We now turn to cases that were not included in the NAVCO 1.0 data. In Table 2, omitted cases from 1900 to 2006 are revealed. These cases were compiled by the authors of this study through qualitative inquiry of each specific context. As shown in the table, some of these omitted campaigns are comprised of incidents that are of substantial historical significance such as the 1910 Portuguese revolution or the 1930 Brazilian revolution. In total, there are 21 cases that were not included in their data but more than likely should have been. Further, in Table 3, results of success rates are presented spanning the time frame of the noted original database of 323 campaigns (1900–2006) with the inclusion of cases shown in Table 2.

Table 2.

List of Omitted Cases, 1900–2006.

Campaign	Strategy	Success	Limited Success
Thai Phu Mi Bun Rebellion 1901	Violent	No	No
Argentinean Revolution 1905	Violent	No	No
Young Turk Revolution (Ottoman) 1908	Violent	Yes	No
Portuguese Revolution 1910	Violent	Yes	No
Albanian Rebellion 1912	Violent	Yes	No
Irish Easter Uprising 1916	Violent	No	No
Irish Independence 1919–21	Violent	No	Yes
Russian October Revolution 1917	Violent	Yes	No
Egyptian Revolution 1919	Reactive Unarmed Violence	Yes	No
Polish Uprising 1919	Violent	Yes	No
Italian Fascist March 1922	Unarmed Violence	Yes	No
Georgian Uprising 1924	Violent	No	No
Brazilian Revolution 1930	Violent	Yes	No
Vietnamese Revolution 1945	Violent	Yes	No
Peasant Revolt Hyderabad, India 1951	Violent	Yes	No
Zanzibar Rebellion/Revolution 1964	Violent	Yes	No
Bahrain March Intifada 1965	Reactive Unarmed Violence	Yes	No
Peru 1980	Unarmed Violence	Yes	No
Haiti Anti-Military Resistance 1990	Unarmed Violence	No	No
Kenya pro-democratic movement 2002	Unarmed Violence	Yes	No
Niger pro-democratic movement 2004	Reactive Unarmed Violence	No	No

Table 3.

Results With Omitted Cases (1900–2006).

	Success	Limited success
Nonviolent campaign (N = 94)	48%	22%
Unarmed Violence campaign (N = 27)	56%	27%
Violent campaign (N = 232)	29%	13%

These results reveal that nonviolent campaign strategy is hardly spectacular in enabling civilians to obtain concessions from governments or enabling them to achieve major political goals. Without doubt, the 48% success rate of nonviolent campaigns is much higher than the 29% success rate of violent campaigns for the time period under attention, but at the same time, this result is less than originally reported in the literature (Table 1). Cases featuring unarmed violence are the most successful of all any of these strategies (campaigns that adopted reactive unarmed violence will be unveiled in Table 5). We now turn to an additional assessment in order to understand the outcomes that nonviolent, unarmed violent, reactive unarmed violent, and violent forms of collective action experienced over the 19th century followed by a total analysis spanning 1800–2006.⁴ In Table 4, information on campaigns from 1800–99 is presented.

Table 4.

Oppositional Campaigns, 1800–99.

Campaign	Strategy	Success	Limited success
Haitian Revolution 1804	Violent	Yes	No
Chuquisaca Uprising 1809 (Bolivia)	Unarmed Violence	Yes	No
La Paz Uprising 1809 (Bolivia)	Unarmed Violence	Yes	No
May Day Revolution 1810 (Argentina)	Violent	No	Yes
Hong Gyeong-nae 1811 (South Korea)	Violent	Yes	No
Portuguese Revolution 1820	Unarmed Violence	Yes	No
French Revolution 1830	Violent	Yes	No
Belgian Revolution 1831	Violent	Yes	No
Serbian Revolution 1835	Violent	Yes	No
Texas Revolution 1835-6	Violent	Yes	No
Separatist Movement Brazil 1840	Violent	No	No
Balaiada Uprising Brazil 1841	Violent	No	No
Republican Uprising Brazil 1845	Violent	No	No
French Revolution 1848	Violent	Yes	No
Ireland Rebellion 1848	Violent	No	Yes
Serbian Revolution 1848	Violent	Yes	No
Moldovan Revolution 1848	Violent	No	No
Wallachia Revolution 1848	Violent	Yes	No
German Revolution 1848	Violent	Yes	No
Sardinian/Italian Revolution 1848	Violent	No	Yes
Eureka Uprising 1854 (Canada)	Violent	No	No
Polish Uprising 1864	Violent	No	No
Guatemalan Revolution 1871	Violent	Yes	No
Herzegovina Rebellion 1877	Violent	No	Yes
Bulgarian Uprising 1876	Violent	No	No
Japanese Samurai Revolt 1877	Violent	No	No
Argentinean Revolution 1890	Violent	No	Yes
Argentinean Revolution 1893	Violent	No	No
Philippine Revolution 1898	Violent	Yes	No
Taiping Civil War 1850-64	Violent	Yes	No
Vukalović Uprising 1862 (Herzegovina)	Violent	No	Yes
Mexican Ayutla campaign 1855	Violent	Yes	No
Mexican Liberal campaign 1861	Violent	Yes	No
Jamaican Rebellion 1865	Reactive Unarmed Violence	Yes	No
Irish Fenian Uprising 1867	Violent	No	No
Spanish Glorious Revolution 1868	Violent	Yes	No
Puerto Rican Revolt 1868	Violent	No	Yes
Puerto Rican Yuaco Movement 1897	Violent	Yes	No

The first compelling element to bring to attention here is that throughout my time researching these cases and contention during the 19th century more broadly, not a single fully nonviolent campaign could be observed. Strict civil resistance was absent from all cases from the period 1800–99. By far, the most common method utilized during this century was full violence—with rifles, muskets, and handguns being the preferred weapons of choice by oppositional campaigns. Some of this century’s most significant revolutions featured noteworthy battles between opposition and state forces such as the numerous European revolutions of 1848. What’s more, nearly every single rebellion, insurrection, and protest was met with severe state repression and often, challenges to a governmental, monarchical, or empire status quos turned into public mass killings that were full of bloodshed.

We now turn to a comprehensive analysis of campaigns for the period 1800–2006. This assessment will enable us to investigate a much broader historical time period and in turn, will bring us closer to understanding the relationship between different strategic actions of dissent and political outcomes.

Beginning with fully nonviolent campaigns (illustrated in Table 5), success rates remain the same as noted in the previous section due to the empirical absence of any such campaign across the 19th century. In terms of the 10 cases featuring reactive self-defense based unarmed violence, 60% experienced full success and 2% saw limited success. This brings us to fully violent campaigns and a point that was briefly brought to attention earlier—there have been many more violent rebellions, insurgencies, and battle-based attempts at revolution in the history of the modern nation state than any other strategies of civilian conflict. This large historical sample of cases (1800–2006) reveals that nearly a third of all violent attempts at major political change ended up being fully successful and 14% were partially successful.

Table 5.

Complete Success Rates, 1800–2006.

	Success	Limited success
Nonviolent campaign (N = 94)	48%	22%
Unarmed campaign (N = 30)	61%	27%
Reactive unarmed violence (N = 10)	60%	2%
Violent campaign (N = 262)	30%	14%

Along with reactive unarmed violence, campaigns featuring unarmed violence had the highest success rates of all forms of resistance. A whopping 87% of these campaigns achieved some form of success (whether full or partial). These findings offer the most comprehensive glance of the effectiveness of different resistance types across a substantial portion of recent human history. The most salient point of these findings is that the strategy of nonviolent resistance is likely not a necessary or sufficient condition for socio-political change. Nonviolence as effective of a strategy as it has been made out to be in the literature under discussion. Unarmed forms of violence have experienced substantially higher rates of success. This latter result aligns with Kadivar and Ketchley’s (2018) findings in which riots and unarmed violence were found to be positively associated with transitions that resulted in liberalization. To obtain greater understanding of these results, the causal mechanisms associated with nonviolent civil resistance and success rates need to be reviewed.

Why Mechanisms of Nonviolent Change Are Obsolete

Various causal mechanisms that have been said to underpin the positive relationship between nonviolent dissent and campaign success may not be as salient as previously considered. Some of these mechanisms may even be obsolete because the trends they seek to support are not as robust as made out to be. We must keep in mind that causal mechanisms are simply scholarly constructs or theorizations that get posited in order to explain a phenomenon, its underlying dynamics, and trends. Commonly referenced variables in the literature tell us that nonviolent protest strategy enables oppositional campaigns to attract a large coalition of forces and participatory rate which in turn, makes it easier for nonviolent campaigns to win political conflicts due to the ability to apply large-scale pressure against the status quo of a given incumbent regime. Nonviolence is also said to create a socially credible environment for would-be protesters to join a campaign when compared to the challenges that civilians have in joining violent resistance campaigns.

Although it is argued that nonviolent campaigns are likelier to attract and represent diverse segments of society when compared to violent campaigns, this factor is not inclusive to nonviolent campaigns as any given campaign (whether armed, nonviolent, or those featuring unarmed/reactive violence) may need just a degree of ideological diversity to achieve their goals. Rather than a given movement needing to feature hundreds of thousands participants that contain links to all segments of a given society, several or even less than a handful of groups with shared political interests can utilize coercion and violence to attain major political concessions. Events leading up to the 1973 Chilean rebel-led coup d’état help to elucidate these points: the mechanisms of participation and movement diversity that have been proposed to underpin success rates of nonviolent civil resistance campaigns are not necessarily tied to only nonviolence.

The 1973 Chilean case paved way for the formation of a multi-year hardline dictatorship (led by Pinochet) in a country that had experienced repeated democratic elections dating back to the Great Depression period. Violent resistance was utilized by opposition and importantly, a degree of ideological diversity was present at the campaign level. In September of 1973, Salvador Allende, although popularly elected, was violently overthrown by a campaign that was comprised of an array of actors including some with links to security forces and state police. These included A. Pinochet and others that had support from US Central Intelligence Agency funding (Bizzarro, 2017: 169). In the NAVCO data set, this particular campaign (anti-Allende) was coded as being ideologically diverse. This is due to the fact that the campaign was made up of a coalition of forces that were foes prior to 1970. In the early 1970s however, the Partido Nacional aligned with the Christian Democrats as well as the organizations of Fatherland and Freedom. These actors eventually ousted Allende (Roxborough and Roddick, 1976: 120). Ideological diversity may be just as important to violent and unarmed campaigns as it has previously been hypothesized to be to nonviolent movements.

This brings us to another key causal mechanism associated with nonviolent campaign success rates: security force defection. Defection, also known as mutiny, involves military actors leaving their contracted duty of protecting a principal ruler (or organization) to join or voice support for an opposition movement. It specifically involves military actors that abrogate a basic commitment to defend their leader (Brooks, 2017). Without doubt, defections have been prevalent in a number of successful nonviolent revolutions and the author of this study does not seek to discredit the role of security force defections in such cases. However, defections have also arisen during violent campaigns. The Chilean example highlighted above featured security force defections. In addition, the Russian revolution of 1917 and subsequent civil war are notoriously associated with defection. The Tsar’s closest commanders and strategic allies literally morphed into overnight traitors when faced with the challenge of defending the monarchy against exceedingly violent Bolsheviks.

Security forces defecting to an oppositional campaign, voicing support for opposition, or engaging in mutiny are not actions that are historically inclusive to the success of nonviolent campaigns. As scholars of defection and civil resistance have pointed out, state forces that are assigned orders to repress a mass uprising face a dilemma because if they do end up repressing a campaign but repression does not quarrel mobilization and the campaign ends up overthrowing the incumbent regime, then they will have to bear the consequences of severe punishment from a new government (Pion-Berlin et al., 2014). Uncertainties of this sort, however, may even be graver if security forces are assigned orders to repress a violent opposition movement. If repression against a violent movement does not quarrel mobilization, then heightened conflict and even civil war become imminent. The latter can indeed lead to processes that tilt the balance of power in favor of insurgents and can spur defection. This entails that this crucial causal mechanism is in no way limited to contexts that only feature nonviolent campaigns.

The noted factors are among several others that cannot be touched upon in depth due to space constraints, but they do, nevertheless, pertain to the valid point that causal mechanisms associated with nonviolent civil resistance success rates are not entirely contingent on nonviolent strategy—as similar mechanisms are also relevant for violent resistance campaigns, albeit they operate through slightly different caricatures. It is indeed likely that these mentioned mechanisms operate empirically in different ways and in favor of different forms of resistance under different contextual conditions, yet the point remains, strict nonviolent strategy is not the only form of resistance that can interact with causal mechanisms which connect resistance strategies to positive outcomes.

Discussion

Favoring Causal Argumentation Over Description

Although quantitative methodology (including widely espoused Bayesian statistics) has undoubtedly become more rigorous over the last several decades and has enabled scholars to produce considerably broad explanations, what many fail to contemplate is that statistical results are entirely contingent on analyzed data. Data are built upon subjective interpretations of how events are observed and coded into variables. Among the first issues that comes to mind about problems associated with generalizations made from large-N quantitative research is that a given study may not possess enough external validity, but very rarely do scholars consider that the actual substance of what is being analyzed is erroneously interpreted, measured, and coded.

In the nonviolent civil resistance literature, the careless dichotomization of violence/nonviolence has turned out to be a terminal flaw. A likely culprit behind these errors is that descriptions of our social and political world have dropped tremendously in comparison to attempts of generalization that are supported by causal inference frameworks. This trend, as described by Gerring (2012), is so significant that principal descriptive arguments found in the total percentage of articles in the American Political Science Review have dropped from 40% (annually) in 1970 to close to 0% in the 2000s. The total percentage of articles whose principal argument is causal went from 10% in 1960 to over 80% by the 2000s (Gerring, 2012: 730).

Accurate descriptions, or at least honest attempts at accurate descriptions, should serve as the basis for quantitative research—especially research that is reliant on major data collection projects that are consequently used to make sweeping generalizations with regard to phenomena that are as important as protest and revolution. For example, describing resistance methods and political contention as being “primarily nonviolent” or “primarily violent” would not be a viable maneuver to make if one were to adequately explain a single case of contention because the isolation of different protest strategies would be necessary. As suggested by earlier referenced scholars, rather than the dichotomization of violence and nonviolence, alterations should be made in the study of different protest strategies and their causal effects on political outcomes. These can be made through disaggregation of different forms of resistance from one another as has been done in the present study. Such conceptual issues are of considerable importance for what many believe to be an end goal of social science: prediction. If empirical reality and history more broadly cannot be accurately described and accounted for, social scientists will surely continue to run into the problem of not being able to predict.

It is also plausible to believe that methodological trendiness superseded concerns about conceptual soundness, and in turn, this enabled the noted scholars to put forward a framework with several flawed components. In their justification of the nonviolence/violence dichotomy, Chenoweth and Stephan (2011) argued that since quantitative research on nonviolent civil resistance had yet to be carried out, adopting a dichotomous classification was a feasible move to make because quantitative analyses are “serious” while other modes of inquiry are not: “the serious study of strategic nonviolent action has remained something of a pariah within security studies despite decades of scholarship on the subject” (Chenoweth and Stephan, 2011: 31).

Here the favoring of causally oriented methodology over conceptual soundness and empirical accurateness is put on full display: “Despite the challenges associated with studying this subject, we argue that the theoretical and policy implications of the research questions at hand are too important to avoid” (Chenoweth and Stephan, 2011: 31).

The latter statement is especially important to deconstruct as it implies two critical points—the first being that without an in-fashion research design based around large numbers of cases and econometrics, theoretical knowledge cannot be advanced. Second, the statement also implies that certain types of policy prescriptions are to be made only inclusively through such an approach, which brings us to the key matter of ideological bias in scientific research.

Ideological Bias

Proponents of the nonviolent civil resistance argument consider the empirical potential of mass nonviolent contention to be not only positive, but ethical with relation to real-world political change. Nonviolence is advocated as the sole strategy and choice of resistance that can potentially take a population out of dictatorship and tyranny. There is ideological bias behind the universal advocation of nonviolent civil resistance as it is assumed that the exclusive way out of such conditions (and presumably forward to social progress) is through free-market institutional transition and the adoption of a capitalistic economic system.⁵ Along similar lines, this is also likely why so much subsequent research that has been published post Chenoweth and Stephan’s data set and findings have been focused on themes such as “pathways to democratization” through nonviolence or in studies assessing the effect of nonviolent civil resistance on the probability of democratic regime transition.

The nonviolent civil resistance argument has underlying political and ideological backing and this present study is not the first to point this fact out. Scholars have noted that US foreign policy started developing strategically oriented programs to fund nonviolent social movements in the 1980s (Robinson and Robinson, 1996), and enhanced its funding from 2002 onwards (Wahlrab, 2010). More recently, US institutions (National Democratic Institute; Freedom House; International Republican Institute) and even members of Congress and government (John McCain; Donald Rumsfeld, among others) were observed to either fund or voice support nonviolent campaigns in varying countries (Gan, 2013: 98).⁶ Examples of this include states such as Ukraine, Egypt, and Georgia (along with the earlier noted Serbia case).⁷ Chenoweth and Stephan (2011), however, note that foreign governments are “likelier to lend direct material support to violent resistance campaigns, and their data indicate that 35% of violent insurgencies received material support from a foreign state whereas less than 10% of nonviolent campaigns did” (Chenoweth and Stephan, 2011: 62). The problematic dynamic here is that there really is no way to verify the extent of foreign support to revolution seeking movements as this matter is submerged in clandestine and covert operations.

A range of critically oriented researchers along with activists have reviewed the aims of nonviolence in relation to US foreign policy. To little surprise, nonviolence has been affiliated with US imperialism (Gelderloos, 2007; Lakey, 2009). Gan (2013: 98) points out:

one can only wonder what the reaction would be if the United States were to discover that the Green Party was being funded by such organizations in Russia or China, or if the United States were to discover that the Tea Party were being funded by Neo-Nazi groups in Europe.

The aim of this present study is not to critique US foreign policy, but to elucidate how ideological bias can underlie and eventually determine social inquiry—inquiry that then is interpreted by the scientific community and public as being politically neutral and scientifically objective. Chenoweth and Stephan’s (2011) award-winning book was funded by the International Center for Nonviolent Conflict (ICNC), a nonprofit Washington DC-based organization that is founded and run by non-neutral figures (Gowans, 2010). The head of the ICNC is an investment banker, former director of Freedom House, and participant in funding and training dissidents in countries such as Egypt (Stolberg, 2011).

Similar institutions such as Gene Sharp’s Albert Einstein Institution have been alleged as being collaborators to the Central Intelligence Agency—especially in their promotion of anticommunist (during the Cold War Era) and antisocialist (post-Cold War) agendas (Bramball, 2012). Bramball (2012) relates Sharp’s institution to CIA activity in places such as 2011 Syria, Tunisia, Egypt, Libya, and Iran. In multiple respects, it is easy to foresee both how and why ideological and political bias can regulate the behavior of funding bodies that support research, which in turn, directly influence the type of research scholars carry out. More significantly, bias can encourage specific types of results to be produced as well as regulate what type of knowledge is unobjectionable and what is acceptable. Many have written about the influence of corporate interests on health-related research in epidemiology, pharmacology, and biology. The sugar, tobacco, and alcohol industries have historically been involved in disabling and slowing down the publication of adverse findings about their products as such findings would produce public policy recommendations that would logically lead to less consumption of their products (Glantz et al., 1998; Oreskes and Conway, 2011).

For example, in the 1960s, a sugar-industry influenced study was published in the New England Journal of Medicine. In order to play down concerns about sugar’s role in contributing to heart disease, recent information has revealed that decades ago, the sugar industry sponsored Harvard scientists to publish an argument through conducting an impartial literature review which would “discover” that previous research linking sugar to heart problems was conceptually and methodologically inadequate (Kearns et al., 2016). This pertains to the fact that over the last 50 years, the sugar industry has been influencing scientific debate about sugar and its health consequences. In a New York Times interview that was widely dissimenated across national news outlets, S. Glantz revealed that, “It was a very smart thing the sugar industry did, because review papers, especially if you get them published in a very prominent journal, tend to shape the overall scientific discussion” (NPR News, 2016). This surely rings a bell. Published research based on the NAVCO data sets is in fact, based on data that were made by scholars who are Harvard-based political scientists, and what’s more, these scholars argue that prior to their research (2008, 2011) there was no “serious” study of strategic action as mentioned in their earlier referenced quote.

As this study has demonstrated, nonviolence is not as effective of a strategy of dissent as it is made out to be, and there certainly are major forces that hold interests in preventing publics from knowing about this fact. The NAVCO data and nonviolent civil resistance argument demonstrate that ideological bias in research on political conflicts is likely analagous to other spheres that are under heavy influence of organized elite interests. Entire theories, institutions, and NGOs have been built on the premise that nonviolence is the grand causal strategy that will lead to emancipation. Yet what if we considered that an “objective” strategy entailed that combinations of nonviolence and unarmed violence were the keys to revolutionary success and to winning political conflicts more broadly? In some contexts, full violence may be the only answer if a given group is seeking pragmatic socio-political change.

In the end, these issues have real-world ramifications because in contemporary public discourse, nonviolence has become the sole method that is advocated for populations to adopt in search of social change. Nonviolence is in-style, fashionable, and discursively appropriate. As our own societies slide into populist politics and silently regress under the umbrella of Orwellian surveillance states, we are told that nonviolence will emancipate us, but we must ask ourselves when the last time a truly threatening civilian-led movement was waged against the government?⁸

Conclusion

This study has revealed that widely cited literature on nonviolent civil resistance is empirically imprecise and has produced results that are misleading. This is due to conceptual errors that went into making popular data sets that are used in the literature along with the blatant omission of cases of both successful violent campaigns and failed nonviolent campaigns. These findings have major ramifications. Not only have scholars been misled, but the general public has been misled about the empirical effectiveness of nonviolent resistance. Without doubt, nonviolent resistance campaigns have proven to bring down dictators, and the author of this present study has demonstrated in other (previous) research that strict nonviolent protest strategy has circumstantial benefits such as enabling social movements to potentially be able to overcome state repression in easier ways than violent movements in an era of mass, digitalized communication. Nevertheless, the elephant in the room remains; the global success of nonviolent civil resistance is only genuine as observed in a finite of cases and most importantly, in fewer instances than commonly acknowledged.

Many of the most frequently referenced examples of civil resistance movements by nonviolence scholars feature substantial degrees of unarmed violence as well as reactive, self-defense based unarmed violence, and hence cannot be considered to be conceptually valid with relation to the empirical phenomena they are said to characterize. Operationalizing observations from empirical reality into measurable (and eventually) testable variables dictates the types of results that get produced by any given statistical framework. When taking these facts into consideration alongside previously omitted campaigns that were left out of statistical analyses, it becomes clear that nonviolent civil resistance is not as an effective of strategy as it has been made out to be when compared to its strategic counterparts. Another relevant dynamic to consider is that the noted literature has been based on data on nonviolent campaigns that number to 94 cases which is far fewer than the 250+ violent campaigns that have arisen across the last few centuries of history. Within these 94 cases, a major issue remains unresolved (and will probably continue to be unresolved) in the absence of public knowledge on the extent to which nonviolent campaigns received aid and strategic material resources from foreign powers. As one can imagine, such help may make or break a movement in political conflicts.

Finally, this study has observed that the literature under attention is heavily influenced by political and ideological forces in ways that are not at all different from corporate influences over scientific research on substances and foods such as tobacco or sugar. With Why Civil Resistance Works being funded by non-neutral institutions and actors, the scholarly community and larger public should question and deconstruct the results produced by this literature. Emphasis must be placed on its non-neutral ideological backing. Genuine social change inspired by grassroots organization is best left disconnected from manipulative strategies waged by elites. In terms of scholarship on protest and the strategies of collective action, we still have much more to learn from the comparison of resistance types, the disaggregation(s) of resistance, and the relation of resistance types to empirical outcomes. However, in order to make such comparisons genuinely, attempts have to be made to detach political influence(s) from scientific inquiry.

Supplemental Material

appendix_reactive_violence_cases – Supplemental material for Debunking the Myths Behind Nonviolent Civil Resistance

Supplemental material, appendix_reactive_violence_cases for Debunking the Myths Behind Nonviolent Civil Resistance by Alexei Anisin in Critical Sociology

Footnotes

Acknowledgements

I would like to thank two anonymous reviewers who provided very insightful and constructive comments throughout the peer review process of this manuscript. Their comments significantly contributed to improving the empirical precision of this study.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

ORCID iD

Alexei Anisin

Supplemental material

Supplemental material for this article is available online.

Notes

References

Ackerman

DuVall

(2000) A Force More Powerful: A Century of Nonviolent Conflict. New York, NY: Palgrave Macmillan.

Ackerman

Kruegler

(1994) Strategic Nonviolent Conflict: The Dynamics of People Power in the Twentieth Century. Westport, CT: Praeger Publishers.

Adams

(1993) Theatrical Territoriality: A Geographical Enquiry Into Protest Occupations and Mass Communication (Vol. 2). Madison, WI: University of Wisconsin Press.

Anisin

(2018) Social causation and protest mobilization: why temporality and interaction matter. Territory, Politics, Governance 6(3): 279–301.

Bethke

(2017) Nonviolent resistance and peaceful turnover of power. Peace Economics, Peace Science and Public Policy 23(4): 1–5.

Bethke

Pinckney

(2019) Non-violent resistance and the quality of democracy. Conflict Management and Peace Science: 1–21. doi:10.1177/0738894219855918

Bizzarro

(2017) Historical Dictionary of Chile (Vol. 2). Lanham, MD: Rowman & Littlefield.

Bramball

(2012) 21st Century Revolution. Dr Stuart Jeanne Bramhall: Self Published.

Brooks

(2017) Military Defection and the Arab Spring. Oxford Research Encyclopedia. February 2017, doi:10.1093/acrefore/9780190228637.013.26.

10.

Celestino

Gleditsch

(2013) Fresh carnations or all thorn, no rose? Nonviolent campaigns and transitions in autocracies. Journal of Peace Research 50(3): 385–400.

11.

Chenoweth

(2011) Nonviolent and Violent Campaigns and Outcomes Dataset, v. 1.1. University of Denver, Online Repository, Korbel School of International Studies.

12.

Chenoweth

Lewis

(2013) Unpacking nonviolent campaigns: introducing the NAVCO 2.0 dataset. Journal of Peace Research 50(3): 415–423.

13.

Chenoweth

Schock

(2015) Do contemporaneous armed challenges affect the outcomes of mass nonviolent campaigns? Mobilization: An International Quarterly 20(4): 427–451.

14.

Chenoweth

Stephan

(2008) Why civil resistance works: the strategic logic of nonviolent conflict. International Security 33(1): 7–44.

15.

Chenoweth

Stephan

(2011) Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict. New York, NY: Columbia University Press.

16.

Domonoske

(2016) 50 years ago, sugar industry quietly paid scientists to point blame at fat. NPR News. 13 September. Available at: https://www.npr.org/sections/thetwo-way/2016/09/13/493739074/50-years-ago-sugar-industry-quietly-paid-scientists-to-point-blame-at-fat?t=1569249745240

17.

Gan

(2013) Violence and Nonviolence: An Introduction. Lanham, MD: Rowman & Littlefield.

18.

Gerring

(2012) Mere description. British Journal of Political Science 42(4): 721–746.

19.

Gelderloos

(2007) How Nonviolence Protects the State. Cambridge, MA: South End Press.

20.

Glantz

Bero

Slade

, et al. (1998) The Cigarette Papers. Berkeley, CA: University of California Press.

21.

Gowans

(2010) The Revolution Will Not Be Assisted By The ICNC (The Counter-Revolution Is Another Matter). Available at: https://gowans.blog/2010/03/12/the-revolution-will-not-be-assisted-by-the-icnc-the-counter-revolution-is-another-matter/

22.

Hamilton

Spectator

(2000) Riots Sparked by Police Seizure of TV Station in Serbia. Hamilton Spectator (Ontario), Print. May 18, 2000.

23.

Kadivar

Ketchley

(2018) Sticks, stones, and Molotov Cocktails: unarmed collective violence and democratization. Socius 4: 1–16. doi:10.1177/2378023118773614

24.

Kearns

Schmidt

Glantz

(2016) Sugar industry and coronary heart disease research: a historical analysis of internal industry documents. JAMA Internal Medicine 176(11): 1680–1685.

25.

Kim

Kroeger

(2019) Conquering and coercing: nonviolent anti-regime protests and the pathways to democracy. Journal of Peace Research 56(5): 650–666.

26.

Lakey

(2009) Nonviolence training and charges of western imperialism: a guide for worried activists. In: Clark

(ed.) People Power: Unarmed Resistance and Global Solidarity, 206-13. New York, NY: Pluto Press.

27.

Lees

Shin

Morales

(2015) Global Gentrifications: Uneven Development and Displacement. Bristol, UK: University of Bristol Press.

28.

Lehoucq

(2016) Does nonviolence work? Comparative Politics 48(2): 269–287.

29.

Lovallo

Wilson

Vincent

, et al. (2004) Blood pressure response to caffeine shows incomplete tolerance after short-term regular consumption. Hypertension 43(4): 760–765.

30.

Nepstad

(2011) Nonviolent Revolutions: Civil Resistance in the Late 20th Century. Oxford, UK: Oxford University Press.

31.

Oreskes

Conway

(2011) Merchants of Doubt: How a Handful of Scientists Obscured the Truth on Issues from Tobacco Smoke to Global Warming. London, UK: Bloomsbury Publishing.

32.

Pion-Berlin

Esparza

Grisham

(2014) Staying quartered: civilian uprisings and military disobedience in the Nonviolence, Unarmed Violence, and Reactive Unarmed Violence twenty-first century. Comparative Political Studies 47(2): 230–259.

33.

Pressman

(2017) Throwing stones in social science: non-violence, unarmed violence, and the first intifada. Cooperation and Conflict 52(4): 519–536.

34.

Erlanger

(2000) Protests flare as Belgrade takes control of TV station. The New York Times, May 19. Available at: https://www.nytimes.com/2000/05/19/world/protests-flare-as-belgrade-takes-control-of-tv-station.html

35.

Robinson

(1996) Promoting Polyarchy: Globalization, US Intervention, and Hegemony (Vol. 48). Cambridge, UK: Cambridge University Press.

36.

Robinson

(2019) It only takes 3.5% of people to change the world. BBC News. Available at: http://www.bbc.com/future/story/20190513-it-only-takes-35-of-people-to-change-the-world?ocid=ww.social.link.twitter

37.

Roxborough

Roddick

(1976) Chile: The State and Revolution. New York, NY: Springer.

38.

Schock

(2013) The practice and study of civil resistance. Journal of Peace Research 50(3): 277–290.

39.

Sharp

(1973a) The Politics of Nonviolent Action: Part One, the Methods of Nonviolent Action. Boston, MA: Porter Sargent.

40.

Sharp

(1973b) The Politics of Nonviolent Action: Part Two, the Methods of Nonviolent Action. Boston, MA: Porter Sargent.

41.

Sharp

(1973c) The Politics of Nonviolent Action, the Methods of Nonviolent Action: Part Three. Boston, MA: Porter Sargent.

42.

Sharp

(1980) Social Power and Political Freedom. Boston, MA: Porter Sargent.

43.

Sharp

(2005) Waging Nonviolent Struggle. Boston, MA: Porter Sargent.

44.

Stolberg

(2011) Shy U.S. intellectual created playbook used in a revolution. The New York Times February 16. Available at: https://www.nytimes.com/2011/02/17/world/middleeast/17sharp.html

45.

Associated Press, The Herald (2000) The people versus Milosevic; Serb President forced into hiding as thousands storm parliament in Belgrade. The Herald, Glasgow Print, October 6.

46.

Wahlrab

(2010) Fostering Global Security: Nonviolent Resistance and US Foreign Policy. University of Denver, Doctoral Dissertation. Electronic Theses and Dissertations. 680. https://digitalcommons.du.edu/etd/680

47.

Zunes

(1997) Nonviolent action in African liberation struggles. Social Alternatives 16(2): 23.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB