Building Historically-Oriented Datasets: A Practical Guide

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

Abstract

Political science has been experiencing an upsurge in the construction of original historically-oriented datasets. Whether broadly cross-national or richly subnational, these data collection efforts are increasingly recognized as invaluable to the field—providing resources of enormous long-term benefit and enabling richer empirical analysis of a broader range of research questions. Within methodological debates, much progress has been made in understanding the bias and limitations of such datasets and their use. At a pragmatic level, however, scant resources exist to help guide graduate students and other scholars learn how to collect, code, and document historically-oriented data. This article provides an initial practical guide based on our own experiences, failures, and hard-won lessons learned. It aims to make transparent the choices and trade-offs involved during the data generation process, while providing helpful tips and highlighting pitfalls to avoid. We begin by foregrounding the importance of scope, especially to early career researchers with limited resources, and ways to narrow the universe of cases to increase feasibility. We then discuss how to move from concepts to measures, properly document those choices in a codebook, and create an iterative process to flexibly but systematically refine coding criteria (e.g. a pilot project). We then consider issues of quality control, creating a manageable workflow process, and organizing documentation. Finally, we provide pointers on how to manage a team of research assistants and construct a reasonably accurate project budget, which is particularly important for grant applications.

Authors

• Erica Susanne De Bruin, Hamilton College

• Kristen A. Harkness, University of St. Andrews

• Jun Koga Sudduth, University of Strathclyde