Political Violence in Turkey

Abstract

This thesis aims to create a novel dataset on political violence to estimate the accurate number of political violence incidents and fatalities in Turkey: Political Violence in Turkey Event Dataset (POLVITED). I argue that the reporting bias and differences in the scopes and definition of event datasets impact the quantitative study of conflict and violence. To eliminate the reporting bias and create a comprehensive dataset on political violence, I apply Matching Event Data by Location, Time and Type (MELTT) method to the Global Terrorism Database(GTD) and Uppsala Conflict Data Programme (UCDP-GED). The finding suggests that UCDP-GED were able to record 65% of the total number of political violence incidents in Turkey, and GTD was able to record 45% of them while 10% is coded in both datasets. POLVITED enables the researchers to capture the accurate picture of political violence incidents in Turkey by eliminating the double recording and alleviating the reporting bias. Lastly, the comparison of POLVITED and official reports demonstrates that, although the reporting bias is inevitable, integration alleviates the problems associated with it by increasing the coverage of the dataset.