A Tale of Two Datasets: Historical Research and Food Riots in Early Modern France

Speaker: Cedric Chambru, ENS Lyon
Series: Core Seminar in Economic and Social History

On May 2nd 1775, helpless police watched rioters gather on the market of Gournay-en-Bray in Normandy to lower the price of grain. Reading accounts from the provincial administration, Jean Nicolas noted that the composition of the crowd was unknown. Working on similar documents, Cynthia Bouton identified female protestors. Were mistakes made? This type of discrepancy raises questions about the reliability of quantitative studies to investigate such questions as the agency of women in social conflict during the early modern period. More generally, such discrepancies seem to validate recent concerns about the reproducibility of research across all fields of social sciences and humanities. In the 1980s, scholars in Europe and the United States launched projects to study the incidence and character of collective violence in the early-modern and revolutionary eras. Some of these scholars specifically targeted food riots for analysis because they seemed to signal the clash of popular politics with shifts in economic and social policy on the eve of the age of revolution. In this paper, we propose to assess concerns related to reproducibility and explore the mistakes made in two unique and independent large-scale research projects on riots in early modern France. Our objective is to systematically study the extent to which these two projects, which relied on similar historical sources, resulted in the creation of comparable datasets. By comparing the original records established by the two researchers, we analyse the role played by errors of palaeography, categorisation, coding, or interpretation in the analysis of the same event. We further include one additional regional sample created to expand the initial work of Jean Nicolas in order to discuss how more thorough investigations in the archives might help us revise existing conclusions about the historical trends in food riots. Overall, we highlight why and how inadvertent errors of sampling by historians could threaten the reliability of historical research and the findings of studies using quantitative historical data.