Keywords
Climate Change, SDG #16, Indices, Weighting, Conflict, Uganda
Start Date
15-9-2020 12:00 PM
End Date
15-9-2020 12:20 PM
Abstract
Resource and income insecurity are linked to the occurrence of violent conflict, and are expected to worsen under projected climate change, signifying the need for climate-conflict assessment tools that are informed by expert knowledge of expected assessment input weights, but which can be assessed empirically—a characteristic more associated with traditional statistical models. This capstone demonstrates a potential climate-sensitive conflict risk assessment for Uganda, using an index with expert-informed expected weights as a statistical predictor of the probability of conflict occurring. Assessment results found a climate impact on conflict probability of 0.37% increase for every 1% increase in climate anomaly, spatially variable key drivers of conflict risk, and the location of subnational conflict risk hotspots—information which could be used in policy planning. However, based on an analysis of the model output, the expert-designed index does not perform well in predicting conflict occurrence when compared to a strictly statistical model using the same inputs. This suggests that solely theoretically-chosen weights limit the performance of the index in logit models, and that a different means of incorporating expected weights would improve the hybridization of theoretical indices and statistical models.
Climate Change as a Threat Multiplier: Assessing Conflict Risk in Uganda
Resource and income insecurity are linked to the occurrence of violent conflict, and are expected to worsen under projected climate change, signifying the need for climate-conflict assessment tools that are informed by expert knowledge of expected assessment input weights, but which can be assessed empirically—a characteristic more associated with traditional statistical models. This capstone demonstrates a potential climate-sensitive conflict risk assessment for Uganda, using an index with expert-informed expected weights as a statistical predictor of the probability of conflict occurring. Assessment results found a climate impact on conflict probability of 0.37% increase for every 1% increase in climate anomaly, spatially variable key drivers of conflict risk, and the location of subnational conflict risk hotspots—information which could be used in policy planning. However, based on an analysis of the model output, the expert-designed index does not perform well in predicting conflict occurrence when compared to a strictly statistical model using the same inputs. This suggests that solely theoretically-chosen weights limit the performance of the index in logit models, and that a different means of incorporating expected weights would improve the hybridization of theoretical indices and statistical models.
Stream and Session
false