Posted 25 May, 2011 by Christopher Wilson

Visualizing Protests For Media-Bias And Sectarianism

Reuters ran a story last month alleging media bias in Al Jazeera’s coverage of the Arab Spring. This spawned a short flurry of online commentary, some posts more vitriolic than others. It also raised the awkward issue of how sectarianism impacts the regional spread and response to unrest, which is worth more considered attention than I have seen it given in mainstream reporting.

The thrust of the Reuters piece was that Al Jazeera, much applauded for their critical role in disseminating information on protests in Tunisia and Egypt (praise rightly deserved), had turned a blind eye to Bahraini protests out of deference to Qatari royal interests.

Despite a wealth of material, there were no stirring montages featuring comments by protesters or scenes of violence against activists in Bahrain. Al Jazeera has produced such segments to accompany Egyptian and Tunisian coverage.

The threat posed by Bahrain’s protests was closer to home. Their success would have set a precedent for broader public participation in a region ruled by Sunni dynasties. More alarming for those dynasties, it would have given more power to Bahrain’s majority Shi’ites, distrusted by Sunni rulers who fear the influence of regional Shi’ite power Iran.

This is a serious charge, especially as Al Jazeera continues to function as a beacon for free expression and independent journalism in the region (and the world, as long as we are comparing relative merits). For less informed news consumers it is an intuitively compelling explanation for the dramatic variance in Gulf reactions to protests in Egypt and Bahrain. As the GCC sent troops into its small island neighbor, and Iraq’s accusations of secular bias derailed the Arab League Summit, it was not difficult to imagine that Al Jazeera might in fact be bowing to Royal pressure.

Indeed, one must assume that some degree of ideological bias influences coverage by any given media organization. This might be due to lack of interest and priorities, as was arguably the case of Western media indifference to protests in Gabon, or  protests around the world that have not been included in the “Arab Spring” discourse.  In a spectacle as highly politicized as the protests and revolutions that came tumbling out of the Maghreb countries in early 2011, one would expect regional, cultural, sectarian and ideological biases to play out in news coverage.

To take a look at whether this was the case, we started collecting data on media coverage in late April. Unable to map radio and television coverage on the fly (as important as they are) we charted the number of online articles within specific sub-domains for a series of media organizations using Google search queries. There were some interesting results (AJ coverage was indeed low for Bahraini protests, which received just one online story from Fox News), but the process was interrupted by a TDS interview sprint at the end of April, before we could collect a full set of numbers . When we resumed data collection last week, differences were immediately obvious, showing almost none of the simple biases we expected along a Gulf / Western split.

In order to put the media coverage numbers into context, we also gathered a significant amount of data to indicate the severity of unrest in each country, including the largest number of protesters reported, deaths and arrests. Both of these data sets were then visualized by Andy Kirk at, producing the two-part graphic below.

Clicking on the image will take you to html version that allows you to click through all media on a single map (beware, takes a minute to load each media into the cache the first time, we will look into speeding it up). A jpg version that includes country populations and lays out a map for each media can be downloaded or viewed at closr. We also welcome input and comments on the raw data, our methodology, and plans for future visualizations. Also check out Andy’s blogpost on the visualization process, where he discusses some of the challenges involved in mashing this kind of highly heterogeneous and dynamic data.

As noted, the resulting maps do not immediately support the biases we sought to explore. Here are a few immediate observations:

  • Though Al Jazeera has a slightly lower relative coverage of Bahrain than Egypt, Yemen or Syria (83: 204/118/99), this does not indicate a blanket lack of coverage. Of course a more nuanced investigation of content (or even moving beyond online article counts to broadcast media) would likely present a more complete picture. Of course, as mentioned above, data from last week is quite different from data at the end of April, in which the AJ score was significantly lower compared to other countries. It would be interesting to see if this is due to changes in AJ’s coverage respective to other countries, and to what degree that is dependent on events in Bahrain.  We will take a stab at segmenting this into timeframes to answer that question, and will come back with an update soon.
  • The one expected deviation noticable here is that Press TV, widely recognized as a propaganda machine for Iranian authorities, has dramatic increases in coverage for Bahrain, Saudia Arabia and Yemen.
  • Otherwise, perhaps the most immediately compelling variance is the vast predominance of articles by western media. This is undoubtedly a function of  language bias in the data set (only English represented here, Arabic forthcoming), as is belied by the preliminary data collected on Arab language media coverage ( feel free to have a look and try your own hand at vizualisation as we move ahead).
  • Within the Arab language media organisations, Al Jazeerra is a clear leader on content, as is BBC for the western media. Western media appear moreover to follow the same general trend across countries, with the exception of Reuters, which seems to generate a lot more coverage of the overall less covered countries, such as Jordan, Kuwait, Morocco and Oman.

Read about the visualization and data in greater detail here. We will be revisiting this with new data (including the Arabic language coverage and time sequencing), and coming back with more in-depth analysis. In the meantime, feel free to make observations or suggestions in the comments below or via email, or try your own hand at visualisation approaches.

We would love feedback and alternative suggestions for how to tackle this complex data set.

Related articles