Documents as Data: Challenges of Constructing a Peace Agreement Dataset

• In Event: Critical Challenges and Solutions towards (Re)Constructing Conflict Datasets

Abstract

While collecting publicly available documents appears a straightforward task, it nevertheless requires much planning and serious consideration of matters of selection and validity, reliability, and bias - and satisfying these requirements becomes more taxing and more important when the documents are coming from politically sensitive situations such as peace talks, and the data collection’s global scope requires translation in terms of both language and understanding of conceptual differences and equivalents across varied contexts. In this paper, the experience of constructing the PA-X Peace Agreements Database and Dataset serves to illustrate the common problems and quandaries of constructing a dataset that builds on a collection of documents with global scope, from the selection of documents that fit the inclusion criteria, dealing with the problems of translation and data coding, and the research ethics concerns that appear in the process of turning documents into data. The process through which the PA-X Dataset is created is also described in more detail, with a discussion of technical concerns, and those related to the creation of a data resource that caters to multiple audiences, and aims to be methodologically versatile, sustainable and effective, offering advice and lessons learned for both smaller- and larger-scale data collection efforts.

Author

• Sanja Badanjak, University of Edinburgh