In partnership with the Institute of Development Studies (IDS), ACLED is pleased to present the following policy briefing, working paper, and blog post as part of our joint project to produce a robust evidence base on the opportunities and limitations of social media data on violence monitoring to inform emergency and crisis response, in the context of monitoring in Kenya. The project is made possible by the Economic & Social Research Council (ESRC).
Policy Briefing: Assessing the Strength of Different Violence Monitoring Systems in Crises
Violence monitoring systems can play a vital role in tracking, managing, and responding to violence. Such systems typically rely on one or a combination of strategies for data collection, including old and new media monitoring. In spite of the widespread use of violence monitoring systems there is limited information on their comparative opportunities and limitations.
Drawing on research conducted during the 2017 Kenya elections, this briefing explains why policymakers and practitioners should continue to invest in combined approaches to violence monitoring that make use of both old and new media to play to their relative strengths while remaining aware of limitations and biases in both.
Authored by Caitriona Dowd, Patricia Justino, Gauthier Marchais, and ACLED Research Director Roudabeh Kishi.
Access the briefing here; for a PDF, click here.
Working Paper: Comparing ‘New’ and ‘Old’ Media for Violence Monitoring and Crisis Response in Kenya
Social media and digital technologies are changing the way information about political violence is collected, disseminated, analysed and understood. Effective early warning and crisis response increasingly depends on the availability of timely, reliable reports of violence, and a growing body of research on violence relies on the availability of reliable violent event data to understand patterns, dynamics and trajectories of violence. While biases in traditional media – newspapers and print media – have been analysed and documented in the literature, there is relatively little information about biases in relation to new and emerging sources of data.
This paper seeks to determine the comparative opportunities and limitations of ‘new’ and ‘old’ data sources for early warning, crisis response, and violence research. We compare the information set produced through social media violence reporting with conventional violence reporting around the August and October 2017 Kenyan elections. Specifically, we leverage data from a sample of social media reports of violence through public posts to Twitter. These reports are compared with events coded from media and published sources coded by ACLED along three dimensions: (1) the geography of violence reporting; (2) the temporality of reporting; and (3) the targeting of reporting.
Authored by Caitriona Dowd, Patricia Justino, Gauthier Marchais, and ACLED Research Director Roudabeh Kishi.
Access the paper here; for a PDF, click here.
Blog Post: Using digital media for real-time violence monitoring and response
Systems for reporting violence can play a vital role in reducing, preventing and responding to unrest. Policymakers and first responders depend on the availability of timely, reliable data, in order to react quickly, deploy resources effectively and prevent further escalation. Conflict researchers also depend on accurate records of violence in order to better analyse, understand and ultimately contribute to reducing violence.
A recent workshop hosted by the Institute of Development Studies brought together expert researchers, practitioners and policymakers from the UK and Kenya to reflect on the potential opportunities, and practical limitations, of using digital technology to promote peace, monitor violence, and support early warning and crisis response.
The workshop, held in London, took place in the aftermath of Kenya’s historic 2017 elections when in an unprecedented move Kenya’s Supreme Court annulled the August 2017 elections, leading to a controversial re-run in October. Throughout this period, digital peace activists and crisis response teams concerned with the possibility of violence established social media and digital platforms to help citizens report unrest and facilitate timely, preventive action. Given Kenya’s long history of electoral violence, the stakes were extremely high. Discussions at the workshop reflected on the success, efficacy and future of these and similar initiatives.
Keyboard warriors: Violence reporting and response
Data for violence reporting systems are usually collected from ‘old’ media, such as newspapers or radio reports, or ‘new’ social media, digital platforms, and other crowdsourcing systems, which have proliferated in Kenya since the height of election-related violence in 2007/8.
While each reporting system can provide critical information, there is limited robust research on their comparative reliability and comprehensiveness. An ESRC-funded project on New and Emerging Forms of Data for Crisis Response, led by IDS with the Armed Conflict Location & Event Data Project, has supported research into the comparative differences between ‘old’ media reports of violence, and those collected through ‘new’ crowdsourced social media and digital platforms in Kenya.
Research highlights significant media biases in selection and coverage of conventional media reporting on violence, while systematic issues of access, as well as distortion and biases in social media can determine whose voices are amplified by digital technology. Together, these findings suggest that neither traditional media nor digital platforms equally reflect the experiences of the most vulnerable, who are often key populations in early warning and crisis response.
In the course of the workshop, three key themes emerged in discussions: the geography of violence reporting; the temporality of coverage; and the changing nature of the new and old media landscape.
Who did what, where? The geography of risk reporting
Workshop discussions reflected how different data sources depict different ‘conflict landscapes’. Research in this project compared reports of violence from traditional media sources, like newspapers and radio, with accounts on Twitter. The findings from this comparative analysis suggest that Twitter tends to capture more violence in relatively wealthier areas, with higher population density and higher levels of economic development. By contrast, traditional media sources appear to have wider geographic coverage and capture more information about events that occur in more rural and less economically developed areas. This reflects characteristics of typical Twitter users, how Twitter is used, and the disparity of access to communications infrastructure across a country like Kenya. These findings are in line with earlier, preliminary analyses from the same project.
These are significant findings because internet-based technologies are sometimes claimed – uncritically – to be essential for democratising the media landscape. While they may serve this function in some regards, when it comes to reporting violence, it appears that a platform like Twitter may instead reproduce biases in the representation of relatively wealthier, urban populations, and may be less effective than traditional media in reflecting the experience and security threats faced by poorer, rural populations. Relying solely on social media platforms such as Twitter can therefore make certain types of violent events less visible in the public sphere. These findings are limited by the fact that they apply only to data collected through Twitter, and not to increasingly popular, closed platforms like WhatsApp or Telegram, through which users share information among a typically smaller number of contacts. Exploring how closed and open social media platforms differ in their use and content with regards to violence reporting is an important avenue for future research, and for building on the results of projects such as Social Media and Security in Africa.
Just in time: The temporality of violence reporting
Workshop participants also discussed the timeliness of reporting across different systems over time, that the levels of precision of such temporal information, and how reports reflect priorities for system users. The project found that in general there was not a significant difference in how quickly different platforms reported on violence.
Within this general comparison, however, there were some important differences. Twitter sources were the fastest source of information in the period immediately surrounding critical junctures – like the elections themselves and the earlier party primaries. However, outside a relatively narrow window during which the election focused the attention of Twitter users, newspapers and other traditional media sources tended to be the fastest and most comprehensive source of information.
The relevance of these differences depends on what the information will be used for. Discussions at the workshop reflected on the differences between practitioner and researcher end users, and the common trade-off in real-time systems between speed of information and the reliability of that information. For policymakers interested in violence prevention through early warning and crisis de-escalation, the timeliness of real-time information may be the most important consideration. Researchers who seek to contribute to violence prevention through analysis of detailed datasets may be more concerned with the accuracy and reliability of information, and less so with its real-time availability.
Truth is the first casualty: #FakeNews and the changing media landscape
Participants reflected on the changing media landscape – both new and old – in Kenya and elsewhere, and how this might affect future crisis response. IDS partners, the Centre for Human Rights and Policy Studies (CHRIPS), based in Nairobi, Kenya, discussed the circulation of fake news reports and images during Kenya’s elections, and led a reflection on the changing nature of Kenya’s media landscape. Significant developments have included recent government attempts to legislate against inaccurate reporting on social media, and the risks this may pose to freedom of speech; as well as the increasing shift of Kenya social media users to closed and encrypted messaging platforms like WhatsApp and Telegram. While it is not clear whether these developments impacted on the outcome of the elections, they may significantly impact on the use of social media for crisis reporting.
Discussions also highlighted the growing interdependence of ‘new’ and ‘old’ media – with traditional journalists heavily represented in crisis reporting on Twitter, and social media in part feeding into the issue attention cycle of more conventional media outlets. Ultimately, discussions pointed to the importance of crisis responders and policymakers seeing traditional and new media sources as complementary, rather than separate and in competition, and the need to diversify and triangulate sources of reporting to capture the most accurate and actionable information. This is particularly significant in a global environment in which traditional media sources are under attack or being delegitimised, while the reliability, credibility and representativeness of many social media reports remain a barrier for crisis responders.
The workshop provided a valuable opportunity to present and critique preliminary findings. As the project comes to a close, the research will be shared more broadly in a forthcoming working paper and policy briefing, which will highlight several recommendations for policy makers and researchers looking to maximise new media platforms for crisis response.
Authored by Patricia Justino, Gauthier Marchais, and ACLED Research Director Roudabeh Kishi.