International Relations Nonviolent Resistance Datasets
Charles Butcher
  • LAST REVIEWED: 03 June 2019
  • LAST MODIFIED: 26 July 2017
  • DOI: 10.1093/obo/9780199743292-0216


The waves of protest that swept across the Middle East during the so-called Arab Spring, and its widely varying outcomes, once again focused attention on the dynamics of nonviolent forms of contention such as demonstrations, strikes, and boycotts. Recent research suggests that nonviolent tactics are an increasingly common way in which people resist authoritarian regimes. The quantitative study of nonviolent resistance involves using a relatively large number of observations and, often, statistical modelling, to explore questions such as why some countries see major nonviolent protests while others experience civil war, why organizations choose violent and nonviolent tactics, whether nonviolent tactics are an effective form of resistance, and why some resistance campaigns succeed in achieving their aims while others fail or transition into civil wars. Depending on the research question, quantitative data can be useful for exploring the dynamics of nonviolent resistance across a large number of cases and wide expanses of time or for understanding in detail how nonviolent campaigns unfold over time in particular cases. Recent years have seen a growing availability of quantitative data on aspects of nonviolent conflict, especially with a “boom” in the availably of events data based on automated coding of newswire sources (see GDELT: Global Data on Events, Location, and Tone [Leetaru and Schrodt 2013, cited under Events Data]; ICEWS (Integrated Crisis Early Warning System coded event data [see Boschee, et al. 2015, cited under Events Data]). Important developments have also taken place with the release of cross-national data on nonviolent (and violent) “campaigns” (i.e., Nonviolent and Violent Campaigns and Outcomes Dataset [see Chenoweth 2011, cited under Campaigns]) and covering organizations and their choices of violent or nonviolent tactics in resistance (i.e., Minorities at Risk Organizational Behavior Data [see Asal, et al. 2008, cited under Organizations]). As just one example, these data are useful for researchers aiming to model how variations in tactical choices, mobilization capacity, and organizational profiles (as just a few examples) influence the short- and long-term outcomes of nonviolent resistance movements with statistical techniques, either within individual countries or across multiple countries or campaigns. This article provides an annotated bibliography of currently available data sets on nonviolent conflict. It is ordered by unit of analysis, beginning with data sets that take the “campaign” as the unit of analysis, then moving to events-based data, organizational-level data, and individual-level data. The focus of this entry is on cross-national data sets, but the article concludes with the most important data sets that concentrate on single-case studies. For each of these data sets information is provided on the unit of analysis, spatial and temporal scope, and data collection methods as well as an outline of the variables included that relate to nonviolent contention. Where possible, links are provided that provide access to the data.


Important developments in the study of nonviolent conflict have taken place with the release of cross-national data sets of nonviolent “campaigns.” The release of the Nonviolent and Violent Campaigns and Outcomes data (NAVCO), versions 1.1, 2.0 has probably had the greatest impact on the field (Chenoweth 2011, Chenoweth and Lewis 2013). NAVCO 1.1 and 2.0 take the “campaign” as the unit of analysis and NAVCO 2.0 disaggregates these campaigns into campaign-years. Erica Chenoweth (defines campaigns as a “series of observable, continuous, purposive mass tactics or events in pursuit of a political objective” (Chenoweth 2011, p. 3). This includes violent and nonviolent campaigns and the data are restricted to campaigns with the “maximalist goals” of regime change, independence, resistance to foreign occupation, and “in some cases,” “major social change’ (for example, in apartheid South Africa). A key difference between most campaign level data sets and events data sets is the use of secondary accounts, expert opinions, and even crowd-sourcing as source material. This stands in contrast with event data sets that rely heavily on media reporting. The NAVCO data were collected with what the authors describe as a “consensus” method. They first gathered information from secondary sources and assembled a list from existing sources. Then this preliminary list was circulated among a dozen experts on armed and unarmed insurrections to probe whether the cases were correctly coded as violent or nonviolent, successful or unsuccessful, and to suggest additions or omissions from the case list. Thus, the final data set reflects a combination of secondary material and expert consensus on the universe of cases, their primary method of contention, and their outcomes. Although not “ready-made” for quantitative analysis the Global Nonviolent Action Database out of Swarthmore University (Lakey 2016) is also based primarily on secondary accounts and includes information on tactics, outcomes, and participants, along with case narratives for more than 1,000 campaigns with diverse goals from moderate reformist to maximalist campaigns for regime change or democratization. A similar data collection method characterizes the Global Digital Activism Dataset (Joyce, et al. 2013) and while Sutton, et al. 2014 concentrates on repression episodes, the authors also rely on newswire and secondary accounts. Each of these data sets is discussed briefly below.

  • Chenoweth, Erica. Nonviolent and Violent Campaigns and Outcomes Dataset. Vol. 1.1. University of Denver, 2011.

    Key data set that has underpinned the expansion of quantitative studies of nonviolent resistance and foundation for the findings in Why Civil Resistance Works by Erica Chenoweth and Maria Stephan. NAVCO covers 323 cases of violent and nonviolent campaigns for regime change or territorial independence from 1900 to 2006, and records whether a campaign was partially or fully successful, the size of the movement and the severity of repression, among others. Latest version is 1.1 2011, available online.

  • Chenoweth, E., and O. A. Lewis. “Unpacking Nonviolent Campaigns: Introducing the NAVCO 2.0 Dataset.” Journal of Peace Research 50.3 (2013): 415–423.

    DOI: 10.1177/0022343312471551

    NAVCO 2.0 is very similar to NAVCO 1.1 but includes yearly variation within campaigns. NAVCO 2.0 contains all variables from NAVCO 1.1 and adds new variables, such as the extent to which the campaign constructs alternative institutions, the sources of foreign support, and the withdrawal of foreign support, in addition to measures of the number of new organizations that join and the organizational structure of the campaign. The latest version is 2.0 2013, available online.

  • Joyce, M., R. António, and P. N. Howard. Global Digital Activism Data Set. ICPSR34625-v2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2014-06-12, 2013.

    DOI: 10.3886/ICPSR34625.v2

    Covers digital activism campaigns by nongovernment organizations across 151 countries from 1982 to 2012 with 1,180 cases. Campaigns were sourced through crowdsourcing and secondary and primary sources (including citizen journalists and other nontraditional media). Variables include the duration of digital campaigns, the type of digital platform utilized, the purpose of digital activities (i.e., mobilization, framing), organizational structure, and targets. The latest version is 1.0, available online.

  • Lakey, G. “Global Nonviolent Action Database.” Swarthmore, PA: Swarthmore College, 2016.

    Not structured for quantitative analysis but contains information on more than 1,000 nonviolent campaigns for democracy, human rights, and economic conditions, among others, and has constituted an important source for other data collection projects. Information is collected largely from primary media sources and secondary academic accounts. GNVAD contains information, for each campaign, on six measures of success and failure, levels of mobilization, participating social groups and organizations, and external and internal allies as well as case narratives for each campaign.

  • Sutton, Jonathan, Charles R. Butcher, and Isak Svensson. “Explaining Political Jiu-Jitsu Institution-Building and the Outcomes of Regime Violence against Unarmed Protests.” Journal of Peace Research 51.5 (2014): 559–573.

    DOI: 10.1177/0022343314531004

    The unit of analysis is the “repression-episode” against unarmed protestors based on the Uppsala Conflict Data Program’s one-sided violence data set. Records data on whether mobilization increased or decreased after the episode, whether international sanctions were imposed, and whether the movement was linked to a broader NAVCO campaign. The latest version is 1, available online for purchase or by subscription.

back to top

Users without a subscription are not able to see the full content on this page. Please subscribe or login.

How to Subscribe

Oxford Bibliographies Online is available by subscription and perpetual access to institutions. For more information or to contact an Oxford Sales Representative click here.