Comparing ‘New’ and ‘Old’ Media for Violence Monitoring and Crisis Response: Evidence from Kenya

Abstract

This paper assesses the comparative opportunities and limitations of ‘new’ and ‘old’ data sources for early warning, crisis response and violence research by comparing reports of political violence, and both violent and peaceful demonstrations, produced through social media and traditional media during the Kenyan elections in August and October 2017. We leverage data from a sample of social media reports of violence through public posts to Twitter and compare these with events coded from media and published sources by the Armed Conflict Location & Event Data Project (ACLED) along two dimensions: 1) geography of violence; and 2) temporality of reporting. We find that the profile of violence recorded varies significantly by source. Records from Twitter are more geographically concentrated, particularly in the capital city and wealthier areas. They are timelier in the immediate period surrounding elections. Records from ACLED have a wider geographic reach, and are relatively more numerous than Twitter in rural and less wealthy areas. They are timelier and more consistent in the run-up to and following elections. While neither source can reveal the ‘true’ violence that occurred, the findings point to the value of drawing on a constellation of various source types given their complementary advantages.

Food and water insecurity as causes of social unrest: Evidence from geolocated Twitter data

“…There is precedence for using Twitter data in these settings. Dowd et al (2018), for instance, compare Twitter data to ‘old’ types of data to evaluate their effectiveness in social conflict prediction. Likewise, Anderson & Huntington (2017) conduct a quantitative content analysis of 4,094 tweets concerning the 2013 Colorado floods to examine how sarcasm and incivility is adopted in Twitter discussions of climate change.…”

“…Accordingly, and considering the volume and variability of Twitter data, we chose to center our analysis on cities and the type of social conflict that occurs predominantly in these settings, namely social unrest (Hendrix & Haggard, 2015; Bellemare, 2015). This focus minimizes several biases specific to Twitter data, such as the tendency to vastly overrepresent urban areas, especially in eastern Africa (Dowd et al, 2018). Next, we focused on a specific country that offers an effective test case for the scenario outlined above.…”

“…Kenya also offers an advantage with respect to our ability to effectively operationalize our anticipated Twitter data. Dowd et al (2018: 13), for instance, write that Kenya has ‘one of the highest levels of telecommunications and internet infrastructure development on the African continent. In 2017, there were 39.1 million registered mobile phone users in Kenya, and an estimated 40.5 million Internet users, out of an estimated population of 44 million.’ In terms of social media, usage is high with an estimated ‘2.2 million monthly active Twitter users’ (Dowd et al, 2018: 13).…”

“…Second, we rely on daily geolocated Twitter data to identify the causal effects of each type of staple insecurity on social unrest in real time and place , using urban centers in Kenya as the focus of analysis. Communication technologies in Kenyan cities are well developed, and millions of residents have access to Twitter (Dowd et al, 2018; Sumbeiywo, 2018), ensuring these data capture perceived food and water insecurity levels and their local prevalence in a given day.…”

An exploration of the association between fuel subsidies and fuel riots

“…Our data could therefore suffer from different types of biases (e.g. towards larger, more important events) that are very common when performing global-level research such as in our case and difficult to avoid (Dowd, Justino, Kishi, & Marchais, 2020). The full database is included as an Appendix with references to the articles.…”