How Events Enter (or not) Data Sets: The Pitfalls and Guidelines of Using Newspapers in the Study of Conflict

While conflict event datasets are increasingly used in contemporary conflict research, important concerns persist regarding the quality of the collected data. Such concerns are not necessarily new. Yet, because the methodological debate and evidence on potential errors remains scattered across different sub-disciplines of social sciences, there is little consensus concerning proper reporting practices in codebooks, how best to deal with the different types of errors, and which types of errors should be prioritised. In this paper, we introduce a new analytical framework -i.e. the Total Event Error (TEE) framework- which aims to elucidate the methodological challenges and errors that may affect whether and how events are entered into conflict event datasets, drawing on different fields of study. Potential errors are diverse and may range from errors arising from the rationale of the media source (e.g. selection of certain types of events into the news) to errors occurring during the data collection process or the analysis phase. Based on the TEE framework, we propose a set of strategies to mitigate errors associated with the construction and use of conflict event datasets. We also identify a number of important avenues for future research concerning the methodology of creating conflict event datasets.