Social Web: An Empirical Analysis of Networked Political Protests

I kicked off the first panel presentation of the conference on the social web and networked political protests by introducing the preliminary quantitative findings of my dissertation research. The question I ask in the first part of my dissertation is whether the rapid diffusion of information communication technology has had any statistically significant impact on anti-government protests in countries under repressive rule? Or do authoritarian regimes maintain the upper hand? My  research was funded by Harvard University’s Berkman Center’s and my presentation is available on Slideshare.


While there are many qualitative case studies out there and numerous anecdotes, the question that motivates my interest in this research is whether all these instances of digital activism actually add up to anything.  For example, DigiActive systematically documents instances of digital activism around the world, sharing best practices and lessons learned. But few studies (if any) seek to quantify the impact of the information revolution on state-society relations in repressive contexts.

Berkman Center fellow Victoria Stodden and I recently carried out a large-N quantitative study on the impact of ICT diffusion on World Bank measures of democracy and governance. While we drew on data from 180 countries, we also took a subset of these countries, namely autocracies, and tested whether an increase in Interent access and mobile phones had any impact on one indicator in particularly, political stability. We found that both variables, Internet and mobile phones, were statistically significant, and negative, with the mobile phones coefficent being larger the Internet coefficient. This would suggest that mobile phones have more of a disruptive impact on repressive regimes.

One question that naturally follows is what form that immediate instability takes? Does the rapid diffusion of communication tools facilitate the organization, mobilization and coordination of anti-government protests, which then contributes to political instability? I decided to find out whether new communication technologies do lead to more frequent protests. Please see the Slideshare presentation for specifics on the regression analysis (thanks to Dr. Stodden and Dr. Woodard for their assistance on the quantitative analysis). As my research is still ongoing and my findings preliminary, I hesitate to make any definite conclusions. With this caveate in mind, here are the tentative results.


Somewhat surprisingly, the results suggest that an increase in the number of Internet users in countries under repressive rule leads to a decrease in the number of protests. Perhaps even more surprising is the fact that mobile phones turned out to have no statistically significant impact on the frequency of protests. The reason I find these findings surprising is that Internet access in repressive environments is extremely limited compared to mobile phones.


The above results are equally surprising. The blue curve is a time series of observed political protests. The green curve is my complete model which includes ICT independent variables as well as my political and economic control variables. The red curve represents the model without the ICT variables. The large difference between the two adjusted R squared figures is rather striking, not to mention that an R squared of 0.613 seems rather high.

In sum, I’m rather skeptical about these results and specifically asked my fellow participants for their feedback. One colleague rightly mentioned that frequency of protests does not provide information on the magnitude of protests—an issue I noted in my dissertation proposal. The same colleague recommended that I include literacy rates as a control variable while Professor Dieter Rucht expressed his concern about the quality of the protest dataset I am drawing on.

The dataset was developed using automated natural language parsing of Reuters news wires. Although Professor Rucht mistakenly assumed that I am using the KEDS dataset, his remarks on the nature of media reporting still hold. There is a considerable amount of literature out there on media bias, which I have reviewed for my dissertation research, and which Professor Rucht echoed. In particular, he is concerned that as Reuters opens new offices in a particular country, this might lead to “over reporting” of protest events compared to other countries. Interestingly enough, however, the frequency of protests in the 22 countries I analyzed goes down with time.

In any case, these kinds of concerns are precisely why my dissertation takes a nested analysis approach, or a mixed method approach. The first part of my research seeks to carry out a large-N quantitative study while the second part entails field based research to carry out qualitative comparative case studies on the impact of ICTs in repressive environments. Whether qualitative findings support my quantitative ones remains to be seen. In the meantime, I plan on running additional regressions and models to double-check my findings.

Patrick Philippe Meier

5 responses to “Social Web: An Empirical Analysis of Networked Political Protests

  1. Pingback: » Towards Networked Protest Politics - Tag eins » Politik in der digitalen Gesellschaft

  2. Pingback: Politischer Aktivismus im Internet ein Hindernis? « Aaron Bruckmillers Weblog

  3. Pingback: Digital Media & Repressive Regimes: Reshaping Public Spheres « iRevolution

  4. Pingback: The Alternative’s alternative, Evgeny Morozov – Politics Unlimited | UK politics news

  5. Pingback: P2P Foundation » Blog Archive » Are current networked protests disaggregating the disaggregators?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s