There were some nice questions asked on Facebook in response to my post on the effectiveness of non-violent resistance. I’ll note that most of these are genericish “did the authors control for” like questions.
Did the authors control for magnitude of desired change or a priori likelihood? Nonviolent is likely to perform far better if it tends to be used on the easier cases.
From the paper:
6. Our use of “resistance” designates major nonstate rebellions, either armed or unarmed. Instead of using event count data, we identify campaigns—a series of repetitive, durable, organized, and observable events directed at a certain target to achieve a goal—as the main unit of analysis. We measure “effectiveness” by comparing stated group objectives to policy outcomes (e.g., states’ willingness to make concessions to opposition movements). This analytical distinction is imperfect, but others have used it with success. See Abrahms, “Why Terrorism Does Not Work.”
That, plus their case studies in the paper, give me some confidence that they aren’t looking at “easy vs hard” rather than “nonviolent vs. violent”. In the description of an earlier version of the dataset we get this nice discussion:
Especially among the nonviolent campaigns, there is real concern that the campaigns included are biased toward success, since it is the large, mature campaigns that are most commonly reported. Other would be nonviolent campaigns that are crushed in their infancy (and therefore fail) will not be included in this dataset. This is the major limitation in this study, and it is difficult to avoid. However, we did attempt to mitigate the effects of underreporting bias in several ways. First, we chose to compare nonviolent campaigns with their comparable counterparts in violent campaigns rather than to view nonviolent campaigns in isolation. We did this because there are many “non-starters” among violent campaigns as well as nonviolent ones, and the same underreporting bias exists within the study of violent insurgencies as with nonviolent insurrections. Therefore, we only investigate the outcomes of major nonviolent and violent campaigns — those that are already “mature” in terms of objectives and membership. We only included cases where the objective was at some point maximalist (i.e. regime change, secession, or self – determination) as opposed to limited (i.e. greater civil liberties or economic rights). Such limited goals (greater autonomy and significant institutio nal reform) are coded only when campaigns’ goals were less than maximalist for certain campaign years. Additionally, we only include nonviolent and violent campaigns where we were certain that more than 1,000 people were actively participating in the strug gle. For the nonviolent campaigns, we gleaned this information from the sources mentioned above. For violent campaigns, the information was implied from the 1,000 battle – death criteria for inclusion in the various datasets we accessed. Using this strict criteria or comparing major campaigns allows us to address questions of the relative effectiveness of resistance type among comparably developed mass movements. Critics may still be skeptical of the case selection, especially in the case of nonviolent conflict. Most concerning is the potential omission of failed nonviolent campaigns, which may not be captured in the dataset due to extreme repression or poor news sources. To address this concern, we made certain that the dataset reflects a consensus sample, which was circulated among the world’s leading authorities on nonviolent conflict to make sure we accounted for known failed campaigns. Unknown, failed, nonviolent campaigns are necessarily omitted from the dataset, just like unknown, failed, violent campaigns.
I think this raises my confidence that they are comparing likes to likes, but it highlights the threats to external validity. Without good data on early squashed campaigns, it’s a bit hard to say too much about what strategy people should consider when trying to get going. I suspect that growing a campaign has a suite of challenges all its own.
Several people suggest that the coding is wrong or distorted by rose colored glasses:
Not sure I buy the methodology here. Many of the peaceful examples they’re looking at had violent campaigns running in parallel or were themselves sometimes violent. The authors note this, but decided to call campaigns peaceful if they are historically viewed as being peaceful (using encyclopedias and biographies as evidence).
I think what they actually proved is that we’re more likely to view successful campaigns as less violent.
I’m not sure where the “many of the examples” is coming from. But let’s look at the actual methodology:
Labeling one campaign as “nonviolent” and another as “violent” is difficult. In many cases, both nonviolent and violent campaigns exist simultaneously among different competing groups. Alternatively, some groups use both nonviolent and violent methods of resistance over the course of their existence, as with the African National Congress in South Africa. Characterizing a campaign as nonviolent or violent simplifies a complex constellation of resistance methods.
To address these difficulties, we established some standards of inclusion for each of these categories. The list of nonviolent campaigns was initially gathered from an extensive review of the literature on nonviolent conflict and social movements. Then we corroborated these data using multiple sources, including encyclopedias, case studies, and a comprehensive bibliography on nonviolent civil resistance by April Carter, Howard Clark, and Michael Randle. Finally, the cases were circulated among experts in nonviolent conflict who were asked to assess whether the cases were appropriately characterized as major nonviolent conflicts, and also which notable conflicts had been omitted. Where the experts suggested additional cases, the same corroboration method was used. The resultant data set includes major resistance campaigns that are primarily or entirely nonviolent. Campaigns that committed a significant amount of violence are coded as violent. The data on violent campaigns are derived primarily from Kristian Gleditsch’s 2004 updates to the Correlates of War database on intrastate wars (COW), as well as from Kalev Sepp’s list of major counterinsurgency operations for information on conflicts after 2002.
This seems…reasonable. Saying that history is written by the victors and that people remember violent campaigns as non-violent is not very convincing. First, given how positively people seem to feel toward successful violence, why would they distort away from non-violence? Second, we’re not talking about victors per se, but scholars of nonviolent conflict. If we think these people are going to systematically get which conflicts are nonviolent wrong, well, that’s a pretty big indictment of the field. It’s, of course, possible, but I’d like to see at least a little more than bare assertion (esp when the fact of expert review is conveniently elided).
That being said (and being aware of some confirmation bias on my part) the basic concern (e.g., roughly, mixed or parallel campaigns might muddle causality or cause biased classification) is reasonable. I’m not too worried about it given their discussion above, but it’d be nice to get some direct reassurance on this.
Perhaps we can peek at the data to get a sense of how many coded ambiguous campaigns there are. Alas, the online browser is rubbish for this purpose. (No counts for searches and why people think horrible plotting on maps is a good thing eludes me.)
Fortunately, the data is available for download as an Excel spreadsheet of doom. (112k rows! Yay!) Unfortunately, afaict, it’s all event level data. I don’t see how that the campaigns were identified in the dataset (rather than by some external derivation). So, boo! I’m not able to immediately napkin this off. So, there’s an IOU!
I do think I was a little too bullish on the results. I still believe them, but I didn’t state the limitations. Yay for pushback!
While trying to figure out how to extract campaigns from the dataset (ongoing) I found some related stuff in the paper, Unpacking nonviolent campaigns: Introducing the NAVCO 2.0 dataset:
Scholars typically characterize campaigns as nonviolent or violent based on the primacy of resistance methods employed (Chenoweth & Stephan, 2011). To qualify as a nonviolent campaign, the campaign must have been prosecuted by unarmed civilians who did not directly threaten or harm the physical well-being of their opponent. Sharp (1973) has identified nearly 200 nonviolent resistance tactics, such as sit-ins, protests, boycotts, civil disobedience, mass noncooperation, and strikes. When a campaign relies almost uniformly on nonviolent methods such as these (as opposed to violent or armed tactics), we characterize the campaign as primarily nonviolent. The First Intifada, for instance, is often remembered to have been violent, due to youths’ rock-throwing and the bloody intra-Palestinian infighting that characterized the Intifada’s final two years (1992–94) and those afterward. However, the IDF’s figures on the First Intifada report that over 97% of Palestinian activities through 1992 were nonviolent or ‘unarmed’ (Pearlman, 2009). As such, it would be empirically accurate to call the 1988–92 phase of the First Intifada primarily nonviolent.
There’s definitely a possibility for recall distortion, but I’d guess ex ante that it was in favor of calling campaigns violent when mostly no. Small bits of violence get a lot of coverage (cf Seattle and black bloc generally) and are a standard way to disparage a campaign). Also:
NAVCO 2.0 also allows for mixed characterizations. It reports data on changes in primary resistance method, which codes whether nonviolent campaigns adopted violence in certain years (as the Defiance Campaign did in the 1960s in response to repression from the South African apartheid regime), and whether violent campaigns began to rely on mass civil resistance (as the Second Defiance Campaign later did in the early1990s). We also identified years where nonviolent campaigns coexisted with, tolerated, or adopted armed wings for self-defense or offensive purposes. These variables allow researchers to further investigate the role of strategic choice and identify periods in a campaign where the movement relied on both nonviolent and violent resistance, which occurs in about 30% of the campaign years.
I think if you are going to jump on them for the mixed bit, you have to delve a lot more. They seem to be doing a fair job of teasing it out.
Re: easy vs. hard, there’s this fascinating bit:
For nonviolent resistance, however, one sees a clear divergence in the determinants of campaign onset. Perhaps the most striking is that violent and nonviolent campaigns share only one determinant in common: population size.
Otherwise, Fearon & Laitin’s (2003) predictors are either completely reversed or insignificant. The only significant correlates of nonviolent campaigns are flatter terrain and older, more durable authoritarian regimes.
This suggests that nonviolent and violent campaigns emerge in very different types of countries. The substantive findings are counter-intuitive and paradoxical: violent campaigns seem to emerge where resistance is ‘easy’, whereas nonviolent resistance is emerging where resistance is supposedly ‘difficult’. These results reveal that nonviolent campaigns are emerging under quite unfavorable circumstances from a mobilization perspective, challenging many assumptions about the supposed ineffectiveness of nonviolent resistance against durable authoritarian regimes