Conflict due to climate change: a univariate causal analysis
In this series of blog posts, we will try to investigate the role of climate change on the conflict in Ethiopia using various quantitative methods. Our research approach will be from simple to complex where we first implement a descriptive and correlation analysis between climate change as approximated by changes in drought index (Evaporative Stress Index) and conflict as the occurrence of armed conflict events over a period of time. In future posts, we will relax this assumption to see the effect of various factors such as demographic and economic variables on conflict and by implementing advanced mathematical and machine learning algorithms.
We use the conflict data from the ACLED database in csv file format and drought risk index data from climserve in satellite imagery format. A range of data processing techniques were used to make the data appropriate for the analysis which includes web parsing, transformation, merging and encoding using python libraries. The cleaned data set covers the period from early 2011 to late 2020 on a monthly basis and at the district level with a total number of 49,265 rows and 9 columns.
Descriptive and correlation analysis paves the way to finding causality which is a difficult but important problem to solve in the field of data science. Here, we use a simple trend analysis and Pearson(pairwise) correlation test to detect the presence of statistically significant relationships.
The following line chart briefly shows the overall relationship of district-level drought risk index and mean frequency of conflict per month.
From the above visual, we can expect a positive relationship between drought and conflict frequency meaning both factors manifest a similar trend of increasing, decreasing or no change over time.
The following charts depict a disaggregation of conflict types and drought over time. Now, we can see that visual inspection becomes increasingly difficult for some of the conflict types to decide due to the magnitude of the relationship between variables. As such, we need a scientific way of detecting this relationship where statistical tests become useful to implement.
As mentioned earlier, the Pearson correlation test was used to test the direction and influence of the relationship between drought risk index and conflict records. Accordingly, we find that conflict is significantly correlated with the presence of drought. Specifically, there is statistical evidence that drought is related to the occurrence of protests, riots and violence but not with battles.
Correlation table: drought index and conflict
In this short blog post, we have seen the relevance of statistical analysis in determining relationships. Most importantly, we have seen that climate has a clear role to play in the pattern of conflict in Ethiopia. In future posts, we will attempt to investigate key questions such as whether there is a cause and effect relationship between the two variables and if we can predict future conflict events using past drought records.
1.According to ACLED, Battle is a violent interaction between two politically organized armed groups at a particular time and location. Explosions/Remote violence: one-sided violent events in which the tool for engaging in conflict creates asymmetry by taking away the ability of the target to respond.
Violence against civilians: violent events where an organised armed group deliberately inflicts violence upon unarmed non-combatants. Riots: violent events where demonstrators or mobs engage in disruptive acts. Protests: a public demonstration in which the participants do not engage in violence, though violence may be used against them. Strategic developments: contextually important information regarding the activities of violent groups that is not itself recorded as political violence, yet may trigger future events or contribute to political dynamics within and across states.
By Yared Hurisa, Applied Scientists and AI for Peace Adviser