A Data Driven Approach to Understanding Social Movements

A Data Driven Approach to Understanding Social Movements

By Leo Villalobos and Ryan J. Ward

The utility of social movements as a means to achieve societal, cultural, or political change has been well established throughout history. In particular, adversaries of the United States (U.S.) have demonstrated their ability to leverage social movements to shape the operational environment in favor of their objectives. However, U.S. military doctrine addresses social movements exclusively as resistance movements as part of unconventional warfare.[i] This view, however, neglects the utility of social movements to disrupt or degrade adversarial efforts in strategic competition through non-violent means. Special Operations Forces (SOF) must explore the utility of social movements in the context of Irregular Warfare (IW) to remain competitive in a complex and dynamic global environment.

Social movement scholars assert that movements that primarily employ non-violent tactics are more likely to be viewed as legitimate.[ii] When legitimacy is established, social movements are more likely to garner increased support from the local populace, as well as from members within the target government.[iii] Non-violent resistance opens avenues for resistance operations to disrupt and degrade strategic competitors while delegitimizing the target government. Previous research indicates that non-violent resistance might be the most effective means to achieving a social movement’s political objectives.[iv] The data presented in this article corroborates this finding and expands on it by showing that while multiple variables affect the probability of success, none have a more significant effect than the tactics (violent vs. non-violent) employed. These findings provide important insights for understanding social movements as a component of irregular warfare.

In order to develop concepts for the application of social movements in IW, it is necessary to assess the validity of non-violence as the preferred method of resistance. Much has been written on this topic, but there is relatively limited research using this specific data-driven approach.[v] To fill this gap, we use statistical models to identify factors that significantly affect movement success.

HYPOTHESES

Social movement theory provides a wide range of explanations for why social movements materialize. While some argue that adequate resources are required to shape movement grievances into collective action, others make compelling arguments that the political environment must also be conducive for a movement to materialize.[vi] We sought to expand beyond the motivations and grievances that spark a movement and instead focus on analyzing factors that influence a movement’s outcome once it is already underway. Specifically, “what factors influence the overall likelihood of a social movement’s success?”

Political process theory suggests that the political environment presents activists with opportunities to mobilize to achieve their objectives.[vii] We are not testing this assertion but rather theorizing that those political opportunities would most likely be found in democratic states. This theory led to the following hypotheses:

H1: Social movements are more likely to experience success in governments with higher levels of democracy than in autocratic governments.

H2: Social movements that primarily employ non-violent tactics are more likely to experience success than violent movements.

            Datasets

To test our hypotheses, we used three distinct datasets: the Non-violent and Violent Campaigns and Outcomes (NAVCO), the Center for Systemic Peace’s Integrated Network for Societal Conflict Research (INSCR), and the World Bank’s World Development Indicators (WDI).

The NAVCO dataset sponsored by Harvard University provides data on 622 maximalist campaigns ranging from 1900 to 2019.[viii] It represents the most comprehensive of its kind and covers the longest available timespan. It provides binary coding for the type of movement (“violent or non-violent”), movement outcome (“success, failure, limited, ongoing”), and movement objective (“regime change, secession, or self-determination”). The dataset’s creators defined maximalist campaigns as attempting to achieve regime change, secession, or self-determination.[ix] However, this definition is potentially problematic since non-violent resistance movements can have a range of goals that fall short of total regime change or independence. Grievances that do not seek to obtain maximalist objectives can be addressed through increased inclusion, greater civil liberties, or enhanced economic rights, to name just a few that this dataset excludes. On the one hand, this dataset is limited in that it excludes other movement objective outcomes, but on the other, it is streamlined and standardized for maximum integrity and usability. If anything, including only maximalist campaigns further underlines the argument on the utility of non-violent resistance to effect significant political change and concessions.

We made two modifications to the original dataset. First, to analyze the findings in more specific regions and make the data more relatable to a military audience, we added a continent classification and a combatant command (COCOM) classification, divided into AFRICOM, CENTCOM, EUCOM, INDOPACOM, NORTHCOM, and SOUTHCOM. These classifications are graphically depicted in Figure 1.

Figure 1.         Geographic Combatant Commands[x]

The division into a COCOM classification allows the analysis of holistic trends and patterns for social movements worldwide. Additionally, it looks at the trends and patterns in specific regions of interest. A copy of the updated tables with the COCOM variable inserted is available upon request.

We incorporated the INSCR’s annual rating of democracy to autocracy levels of all countries with total populations over 500,000 from 1800 to 2018.[xi] The specific subset of data included for this research is the Polity2 variable, which allows for the use of the polity data with time-series-based analysis. The dataset converts the Polity data to a conventional polity score ranging from -10 to +10, indicating a full autocracy to full democracy range and all ratings in between. For the purposes of this study, these scores were further subcategorized into autocracies (-10 to -6), anocracies (-5 to 5), and democracies (6 to 10).[xii]

The WDI includes multiple financial and population metrics for countries around the world.[xiii] This dataset was added to account for population and gross domestic product per capita (GDPPC). Though this dataset adds to the study, it also presented limitations because it only encompasses 1945 to 2015. The effect of this limitation is likely minimal because most of the movements tracked in the NAVCO dataset align with the years represented in WDI data. Despite the limitations, the datasets allowed the researchers to account for and measure the effect of population and relative wealth as control variables for the hypothesis.

            Overall Movement Statistics (Preexisting Data)

Initially, the research team used the NAVCO and INSCR datasets to compare the total number of movements by COCOM. These comparisons are displayed in Figure 2.

Total movements by COCOM (top left), total failed movements by COCOM (top right), total successful movements by COCOM (bottom left), movements by regime type (bottom right).

Figure 2.          COCOM Preliminary Analysis

The y-axis shows the COCOM where each movement occurred, and the x-axis depicts the total number of movements for each respective category. In the aggregate, the top three COCOMs for total movements are AFRICOM, EUCOM, and INDOPACOM. However, when the number of successful movements are analyzed, SOUTHCOM overtakes EUCOM. The difference between these two theaters was particularly interesting to the researchers because historically, SOUTHCOM nations are considered less affluent and stable than EUCOM nations. Additionally, if higher levels of democracy increase the probability of social movement success, then one should expect higher success rates in EUCOM than in SOUTHCOM. This initial assessment fueled further interest in identifying the factors that influence success. Finally, as we compare the type of government that spawns the most movements, it is observed across the board that anocracies are far more likely to face violent and non-violent resistance.

            COCOM Breakdown

Further breaking down the movements visually by region, outcome, movement objectives, and type of movement yields the following results (see Figure 3).

Movement outcomes (top left), movement objectives (top right), effect of violence on success (bottom left), effect of violence on failure (bottom right).

Figure 3.          COCOM Breakdown

Figure 3 (top left) demonstrates that, in the aggregate, failures far outpace successes in all COCOMs except SOUTHCOM. Figure 3 (top right) reveals that the overwhelming majority of all movements in the dataset were seeking regime change. When analyzing data relating to the type of movement and movement outcome, it becomes apparent that although failure far outpaces success, when a movement is successful, it is often characterized as a non-violent movement. At this point, however, it is not yet clear whether success is more likely in democratic or autocratic forms of government when measured against the type of movement.

Finally, we visualized the number of social movements across time before conducting statistical analysis using the third dataset. Figure 4 plots the beginning year of each movement on the x-axis and the number of incidents on the y-axis.

Figure 4.         Social Movements from 1909-2019

The chart shows spikes in the number of movements immediately following periods of significant unrest. This effect can be seen in the late 1940s through the 1960s as decolonization took place. Another spike occurred following the collapse of the Soviet Union in 1989. Finally, a spike again occurred during a period of unrest in the Middle East (2001 to 2019) that saw the rise of Arab Spring movements and the U.S. Global War on Terror. This visualization lends some credence to the idea that regional instability and transitional governments may also contribute to the rise of social movements.

Given this initial assessment, we determined that it was vital to incorporate several variables to determine their effect on the likelihood of success. As suggested by our hypotheses, the variables that we were most concerned with are the effect that polity levels and non-violence have on the probability of a movement succeeding.

We used the following unit of analysis and variables to test the hypothesis. The unit of analysis is movement-year. The dependent variable is a dichotomous indicator of movement success. The independent variables potentially influencing this outcome include the number of years a movement takes, the violent or non-violent nature of the movement, and the polity level of the country where the movement occurred. Additional control variables that the team factored into the model include the goal of the movement, GDPPC, and population size at the time of the movement.

RESULTS

We estimated a logit model regression analysis with the dependent, independent, and control variables factored in. The dependent variable was the probability of success for a movement, measured against multiple other independent (Polity Levels, Non-Violence, Ongoing Years) and control (Population, GDPPC, Movement Objective) variables. The team ran eight separate logit models, gradually factoring in more control variables with each model. The final two models also incorporate a quadratic effect of polity and an interaction effect of polity and non-violence. Table 2 reports the results:

Table 1.           Regression Table Results