Access to Mobile Phones Increases Protests Against Repressive Regimes

I recently shared a draft of my first dissertation chapter which consists of a comprehensive literature review on the impact of Information and Communication Technologies (ICTs) on Democracy, Activism and Dictatorship. Thanks very much to everyone who provided feedback, I really appreciate it. I will try to incorporate as much of the feedback as possible in the final version and will also update that chapter in the coming months given the developments in Tunisia and Egypt.

The second chapter of my dissertation comprises a large-N econometric study on the impact of ICT access on anti-government protests in countries under repressive rule between 1990 and 2007. A 32-page draft of this chapter is available here as a PDF. I use negative binomial regression analysis to test whether the diffusion of ICTs is a statistically significant predictor of protest events and if so, whether that relationship is positive or negative. The dependent variable, protests, is the number of protests per country-year. The ICT variables used in the model are: Internet users, mobile phone subscribers and number of telephone landlines per country-year. The control variables, identified in the literature review are percentage change in GDP, unemployment rate, the degree of autocracy per country-year, internal war and elections.

A total of 38 countries were included in the study: Algeria, Armenia, Azerbaijan, Bahrain, Belarus, Burkina Faso, Burma, China, Cote d’Ivoire, Cuba, DRC, Egypt, Gabon, Guinea, India, Iran, Iraq, Jordan, Kazakhstan, Kenya, Malaysia, Morocco, Pakistan, Philippines, Russia, Saudi Arabia, Singapore, Sudan, Syria, Tajikistan, Thailand, Tunisia, Turkey, Ukraine, United Arab Emirates, Uzbekistan, Venezuela and Zimbabwe. I clustered these countries into 4 groups, those with relatively (1) high and (2) low levels of ICT access; and those with (3) high and (4) low levels of protests per country-year. The purpose of stratifying the data is to capture underlying effects that may be lost by aggregating all the data. So I ran a total of 5 regressions, one on each of those four country groups and one on all the countries combined.

All five negative binomial regression models on the entire 18-year time panel for the study data were significant. Of note, however, is the non-significance of the Internet variable in all models analyzed. Mobile phones were only significant in the regression models for the “Low Protest” and “High Mobile Phone Use” clusters. However, the relationship was negative in the former case and positive in the latter. In other words, an increase in mobile phone users in countries with relatively high ICTs access, is associated with an increase in the number of protests against repressive regimes. This may imply that social unrest is facilitated by the use of mobile communication in countries with widespread access to mobile phones, keeping other factors constant.

These findings require some important qualifications. First, as discussed in the data section, the protest data may suffer from media bias. Second, the protest data does not provide any information on the actual magnitude of the protests. Third, economic data on countries under repressive rule need to be treated with suspicion since some of this data is self-reported. For example, authoritarian regimes are unlikely to report the true magnitude of unemployment in their country. ICT data is also self-reported. Fourth, the data is aggregated to the country-year level, which means potentially important sub-national and sub-annual variations are lost. Fifth and finally, the regression results may be capturing other dynamics that are not immediately apparent given the limits of quantitative analysis.

Qualitative comparative analysis is therefore needed to test and potentially validate the results derived from this quantitative study. Indeed, “perhaps the best reason to proceed in a qualitative and comparative way is that the categories of ‘democracy’ and ‘technology diffusion’ are themselves aggregates and proxies for other measurable phenomena” (Howard 2011). Unpacking and then tracing the underlying causal connections between ICT use and protests requires qualitative methodologies such process-tracing and semi-structured interviews. The conceptual framework developed in Chapter 2 serves as an ideal framework to inform both the process-tracing and interviews. The next chapter of my dissertation will thus introduce two qualitative case studies to critically assess the impact of ICTs on state-society relations in countries under repressive rule. In the meantime, I very much welcome feedback on this second chapter from iRevolution readers.

13 responses to “Access to Mobile Phones Increases Protests Against Repressive Regimes

  1. Pingback: Tweets that mention Access to Mobile Phones Increases Protests Against Repressive Regimes | iRevolution --

  2. That’s great stuff.

    Have you considered using other independent variables such as newspaper circulation, radio and TV penetration?

  3. Hey Patrick – great work.

    Wonder if you can control for state failure… Michael Best has found a relationship, as reported here: and I similarly did some rough work with the Foreign Policy State Failure Index that connected ICT development and state failure, inversely.

  4. Certainly state failure is more a scale than a binary. FP’s underlying data goes back 5 years: but not sure about the Brookings data that Best used. I have the FP data cleaned if you want it.

  5. Pingback: Choosing a mobile development platform « MIT GLOBAL CHALLENGE

  6. Pingback: Access to Mobile Phones Increases Protests Against Repressive Regimes | iRevolution | De camino a la abulia

  7. Pingback: Online privacy special series from MediaShift. A cyberculture roundup. | Erkan's Field Diary

  8. Pingback: Recent Linkage 6 « Signifying Media

  9. Hey Patrick,

    Thanks for sharing. An ambitious and interesting read. Some comments and a couple questions.

    Do you distinguish between independent variables and control variables in order to specifically highlight the ICT data? If so, that’s fine, although they are all technically explanatory/independent variables.

    Also, would be interesting to see how the model holds as you ‘relax’ the qualifications of the countries. I.e. instead of limiting to those 38, broadening the pool further.

    What was your aim in including both GNI and GDP as explanatory variables? As they effectively measure the same thing (plus or minus foreign income), do you think there might be some multicollinearity problems?

    To echo Graham from the comments on your lit review chapter, when you get a chance to pull together the bibliography, I would also be interested in checking out some of the authors you cited…

    Best of luck moving forward!

    A fellow humanitarian GIS geek and budding quantitative researcher,

  10. i live in zimbabwe and will send you an article i am preparing on some sentiments you are suggesting

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