Erica Chenoweth and the Shoddiness of Nonviolence Research

Below is an excerpt from Peter Gelderloos’ The Failure of Nonviolence, the first four chapters of which are available on The Anarchist Library. This passage is from chapter 3, the entirety of which is very much worth reading as a counterpoint to nonviolence absolutists, many of whom are ensconced in academia, the non-profits, influential foundations, or the State Department (or a combination thereof).

Also worth reading is the recent critique of Extinction Rebellion, which takes aim at Chenoweth’s & Stephan’s research and their soundbite-friendly “3.5%” claim, and this now-classic critique of Gene Sharp’s work and his relationship with the state.

Social scientists Erica Chenoweth and Maria Stephan are the authors of a study that is among the only statistical analyses of the effectiveness of nonviolence. Like many social scientists before them, they use statistics to obscure more complex truths. They claim to have compiled a list of 323 major nonviolent campaigns or violent conflicts from 1900 to 2006, and then superficially rate these as “successful”, “partially successful”, or “failed”. They do not use revolutionary criteria for success, and in their mind the “Color Revolutions” and many other reformist, dead-end, or self-betraying movements were successful. Although they rate campaigns as objectively violent or nonviolent, they do not define violence, and they also uncritically use loaded terms like “the international community”. They credit nonviolence with victory in cases where international peacekeeping forces, i.e. armies, had to be called in to protect peaceful protesters, as in East Timor, and they define victory simply as the achievement of a movement’s goals, as though movements ever had a consensus on their goals.

They do not publish the list of campaigns and conflicts with their original study, and after extensive searching I was unable to find it. They explain that the list of major nonviolent campaigns was provided to them by “experts in nonviolent conflict”, in other words, people who are almost exclusively proponents of nonviolence. Given widespread manipulation by such “experts,” who frequently describe heterogeneous struggles as “nonviolent,” such as the independence movements in South Africa and India, the Civil Rights movement, or the uprisings of the Arab Spring, we can only assume that many of successful nonviolent campaigns on the list included armed and combative elements. The violent conflicts that they include in their study come from a completely different source: lists of armed conflicts with over 1,000 combatant deaths. In other words, wars. They are comparing apples and oranges, lining social movements up against wars, as though these different kinds of conflicts arose in the same circumstances and were merely a product of the choices of their participants.

One methodological weakness they do admit to, in a footnote, is that by focusing on “major” nonviolent campaigns, they weed out the many ineffective nonviolent campaigns that never assumed large proportions. But none of the measures they took, ostensibly to correct that bias, could possibly have any effect. Circulating “the data among leading authorities on nonviolent movements to make sure we accounted for failed movements” is useless since there is no objective distinction between major and minor campaigns, and the biggest failures never become major campaigns. Running “multiple tests both across nonviolent and violent cases and within nonviolent cases alone to ensure robustness on all results” is worthless if the study sample is stacked from the start.[21]

Their entire method is superficial to the point of being useless. They are using statistics to obscure complex realities. But even in this flawed endeavor, they have to manipulate the statistics in order to affirm their preconceived conclusions. Most of their paper centers on a detailed explanation of their hypotheses, and pseudo-logical arguments for why their hypotheses must be correct. For example, they cite psychological studies on individual decision-making, with the unspoken assumption that complex social conflicts between institutions and heterogeneous populations will follow the same patterns.[22] They provide no evidence for key arguments like “the public is more likely to support a nonviolent campaign” (p. 13) nor do they interrogate the figure of “the public”. They also make convenient use of non sequiturs, as in the following paragraph:

Second, when violent insurgents threaten the lives of regime members and security forces, they greatly reduce the possibility of loyalty shifts. Abrahms finds that terrorist groups targeting civilians lose public support compared with groups that limit their targets to the military or police.[footnote removed] Surrendering or defecting to a violent movement […] [p. 13]

All the subsequent arguments in the paragraph, which are rhetorical arguments lacking any documentation or data, refer to the topic sentence of the paragraph. All of them are intended to convince readers that so-called violent movements are less effective at provoking defection or “loyalty shifts” among state forces. The only sentence that makes any reference to evidence is the second one, quoted above. But notice how the study cited actually has nothing to do with the topic sentence, no bearing on the question of defection nor the variable violence/nonviolence (Abrahms’ study only addresses violent groups, distinguishing between those that do and do not target civilians).

Elsewhere in the study, the authors ambiguously admit that the statistics do not reveal more defections in the face of nonviolent movements, but they structure the entire article to hide that inconvenience and advance their preconceived arguments.

Such operational successes occur among violent campaigns occasionally, but nonviolent campaigns are more likely to produce loyalty shifts. Although in the quantitative study these findings are qualified by data constraints, our case studies reveal that three violent campaigns were unable to produce meaningful loyalty shifts among opponent elites, whereas such shifts did occur as a result of nonviolent action in the Philippines and East Timor. [p. 42]

To put it more plainly, these “data constraints” are a lack of data supporting their argument, or “insignificant effects” as they admit on page 20. The three case studies they call in to save the day are three examples cherry-picked to prove the point they are trying to make. We can do better: the Vietnam War, the Soviet invasion of Afghanistan, partisan resistance during World War II in Yugoslavia and in Italy, and the anarchist resistance in Ukraine during the Russian Civil War. Five examples of armed movements provoking major defections among the armies sent to crush them, all of them more definitive and on a higher scale than the “loyalty shifts” provoked in the Philippines and East Timor.

In one paragraph summing up her research, Chenoweth acknowledges that the impact of a “violent wing” on the success rates of a movement is “not statistically significant” and then in the next paragraph say that “the most troubling possibility is that the armed wing will reduce the movement’s chances of success.” Later, she commits the most basic error in statistics, confusing correlation with causation, to say that “an armed wing can reduce popular participation [her emphasis]” even though her own data do not support this assertion.[23]

It is significant that mention of this study made the rounds on a number of nonviolent websites. From what I saw, the nonviolence advocates who used the statistics to prove the superiority of their method never linked directly to the study. They probably never even read it.

In order to evaluate the successes and failures of the major uprisings of the last twenty-odd years since the end of the Cold War, we need a fair and sensible set of criteria. We can set aside the superficial question of “who won?”, given that nobody has won, except for those who continue to rule us.

We should also avoid the criterion of whether or not a movement leads to increased repression. I can remember countless arguments in which supporters of nonviolence have tried to paint a struggle as a failure on the grounds that it was heavily repressed. The semi-effective nonviolent movements of the past all provoked an increase in government repression whenever they could encourage widespread disobedience. The belief of modern pacifists, which was not shared by King or Gandhi, that peaceful struggle can avoid brutal consequences at the hands of police and military, has been effectively used as a selling point to flood the ranks of nonviolent movements with opportunists, weekenders, fair-weather friends, cowards, careerists, and naïve citizens who think that changing the world can be easy and hassle-free. Repression is inevitable in any struggle against authority. It is important to be able to survive this repression, but in the worst case, a struggle that is completely crushed by repression is still more effective — because it can inspire us today — than a struggle that allows itself to be recuperated for fear of repression, as happens with many nonviolent movements. Therefore, because the long-term effects of repression still remain to be seen, we will not include this as a criterion, but we will note if a particular rebellion was successfully defeated by repression or recuperation, so that readers will notice a pattern if the combative movements truly are unable to cope with repression, as their critics claim, or if nonviolent movements are frequently recuperated, as we claim.

One criterion of the utmost importance is whether a movement succeeds in seizing space in which new relations can be put in practice. New relations mean: do people share communally and enjoy direct access to their means of survival, or is the social wealth alienated; are people able to organize their own lives, activity, and surroundings, or is decision-making authority monopolized by government structures; do women, trans, and queer people enjoy means of self-defense and self-determination, or are they fully exposed to the violence of patriarchy; do people of color and indigenous people have means of self-defense and autonomy, or are they at the mercy of colonial structures like the market and the police? While the forms are different, the social relations are fundamentally the same between one capitalist state and another, whereas there is a marked difference in the social relations in a stateless commune or an independent indigenous territory. Even though autonomous space will usually be reconquered by the State, we take the experiences of self-organization away with us. The more of these experiences we win, the more powerful our struggles become, the greater our capacity for self-organization on a higher level, and the more people there are who know that obedience to the existing system is not the only option.

This suggests a second criterion: to what extent a movement spreads awareness of its ideas. And this, in turn, needs to be evaluated in terms of whether those ideas are spread as passive information, or whether they are communicated as ideas worth fighting for (or in the case of the nonviolent, taking action and making sacrifices for).

Because of the importance of recuperation in defeating social movements, one important criterion is whether a movement has elite support. If a part of the elite supports a movement, it is much more likely that the movement appears to achieve a victory, when in fact the victory is insubstantial and allows the elite to improve their own situation. This criterion can also show if the pacifists are right when they say the government wants us to be violent, or if the opposite is true, that the elite want us to be nonviolent.

Finally, did a movement achieve any concrete gains that improve people’s lives, restore their dignity, or demonstrate that struggle is worth it and that the government is not omnipotent? From this criterion, we must exclude strictly formalistic gains, like pro-democracy movements that achieve free and fair elections, because this is a redundant victory that can only matter to those who have allowed themselves to believe that democratic government is somehow analogous to freedom or a better life. When the Soviet Bloc countries transitioned from dictatorship to democracy, citizens’ freedom of action did not at all increase, whereas their quality of life suffered dramatically. In other words, the achievement of democracy is solely a question of how power organizes itself, and not one that necessarily impacts how normal people live. If, however, successful resistance to a dictatorship means that people can take to the streets without fear of being arrested and tortured, then we can clearly count this as a concrete gain. Hopefully, the critical difference is obvious.[24]

In sum, the four basic criteria are:

1. whether a movement seized space for new social relations;
2. whether it spread an awareness of new ideas (and secondarily if this awareness was passive or whether it inspired others to fight);
3. whether it had elite support;
4. whether it achieved any concrete gains in improving people’s lives.

Because all of us are still at the mercy of an oppressive system, our focus must be on the strengthening of our struggles for freedom, dignity, and well-being. The above criteria measure the health of our struggles, and whether different methods avail us of what we need to have any chance of creating a new world.

Gelderloos then briefly covers more than three dozen large social movements and their characteristics, not as a comprehensive evaluation but a complication of the simplistic narrative pushed by nonviolence researchers. He concludes the chapter, in part:

Some of my characterizations could definitely be disputed: I do not claim to be an expert on the struggles presented above. However, after a fair evaluation based on the readily available information, what becomes indisputable is that since the end of the Cold War, nonviolent movements have had their greatest successes in effecting regime change, helping to inaugurate new governments that subsequently disappoint and even betray those movements. They have not succeeded in redistributing power in any meaningful way, or putting revolutionary social relations into practice, despite claiming victory numerous times. On the other hand, heterogeneous movements using conflictive methods and a diversity of tactics have been the most effective at seizing space and putting new social relations into practice.

[…]

In sum, a review of revolutions and social uprisings since the end of the Cold War demonstrates the following:

1. Movements that use a diversity of tactics are overwhelmingly more effective at seizing and defending space, and using that autonomy to put new social relations into practice, whether through practices of self-organization, collective self-defense, the reanimation of indigenous ways of life, or collectivization and communization (ending the alienation of capitalist property, which dictates that everything can be bought and sold, and putting our resources in common in a spirit of mutual aid rather than profit).[41]
2. Movements that use a diversity of tactics are more likely to spread, to inspire other people to take action, and they are much more likely than nonviolent movements to spread radical ideas and social critiques, whereas the majority of nonviolent movements are connected to populist complaints and watered-down slogans either lacking in social content or relying on the same social analysis disseminated by the mass media.
3. Nonviolent movements are exponentially more likely to receive substantial elite support. The primary case in which combative movements receive elite support is when they crop up in opposition to governments that are at odds with ruling states (as when NATO will support people rebelling against the Libyan government).
4. Excluding the achievement of free elections, which both combative and peaceful movements have proven effective at winning, movements that use a diversity of tactics have a better track record of achieving concrete gains.

The entire chapter — and Gelderloos’ entire book — is worth reading, especially during such incendiary times as these. Wash your hands, stay safe, and don’t play by our oppressors’ rules.