The UCDP dataset: now with geography!

The article by Sundberg and Melander was an interesting article, which for me brought up questions about spatial scales and situating the data within geography and GIS. One thing I noticed immediately about the article was the map, as this is often what non-geographers immediately think of GIS and geography. I was disappointed that the authors didn’t map the trends of organized violence (i.e., state-based, non-state and one-sided), because it would have been a very interesting visualization to see, for example, where state-based violence is occurring the most. They have represented it temporally in a line graph, but it would have added to the analysis to represent it geographically. Perhaps they didn’t include it because it would have just reiterated already known information? (For example, it’s perhaps already well-known which countries or cities in Africa experience the most state-based violence.)

For me, the article raised as many questions about spatial scales as it did about open data. The authors write that previous research has been focused mainly on violence at the country/year level, but they argue for more sub-national studies, saying that they might help shed more light on the underlying mechanisms of violence. I agree, and think that mapping examples of violence at the sub-national level would allow for more thorough examination of all the variables that contribute to violence, because these variables would certainly change from country to country.

Overall, I found the article very interesting, but a bit difficult to situate the topic in GIScience or even in geography. It seemed like the authors were incorporating the spatial data as simply another facet of their data, along with other factors like time and type of violence, rather than framing it as an investigation fundamentally based in geography. For the authors, GIS is a tool they use to georeferenced their data and make a nice-looking map. This is a fine approach – but it leaves me wondering how the article would be different if the approach was embedded in geography, rather than incorporating geography as one aspect of the data.



Comments are closed.