Looking Past the Spark to Find the Fuel of the Arab Spring Fire
Abstract
The field of statistical conflict prediction addresses region-wide analysis in eras of stable conflict and peace. This study improves upon those prediction ratesin times of volatile conflict and peace seen during the Arab Spring of 2011 to 2015. During this time, higher rates of conflict transition in certain Middle Eastern and North African countries occurred than normally observed in previous studies. Due to the fact that previous prediction models decrease inaccuracy during times of volatile conflict transition and since the proper strategy for handling the Arab Spring has been highly debated, this study considers alterations to previous studies to understand the effects of the Arab Spring on conflict prediction over a five-year period. This study identifies which countries were affected by the Arab Spring, and then apply logistic regression to predict a country's tendency to suffer from high-intensity, violent conflict. A large number of open-source variables are incorporated by implementing an imputation methodology useful to conflict prediction studies in the future. The imputed variables are implemented in four model building techniques: Purposeful Selection of Covariates, Logical Selection of Covariates, Principal Component Regression, and Representative Principal Component Regression resulting in accuracies exceeding 90 . Analysis of the models produced by the four techniques supports hypotheses which propose political opportunity and quality of life factors as causations for increased instability following the Arab Spring.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 22, 2018
- Accession Number
- AD1056301
Entities
People
- Luke M. Brantley
Organizations
- Air Force Institute of Technology