Performance Modeling: A Data Scientific Operationalization of Tilly's Theory

Abstract

This article aims to revitalize and extend Charles Tilly’s theory of contentious performances by presenting a new and fruitful method for discovering, measuring, and modeling them. The article begins with a restatement of Tilly’s theory that performances are flexible, semi-scripted sets of coherent action occurring within protest events (i.e., contentious gatherings), then describes how Tilly attempted to measure performances. Next, it describes a novel method for discovering performances that efficiently merges hand-coding and automated text analysis techniques. The merits of the approach are demonstrated across three studies of news data reporting on the Occupy movement of 2011. The first shows that the method is able to easily identify performances common to the social movements repertoire, such as marches and demonstrations. The second shows that the method can identify performances common to the Occupy movement but rarely included in event catalogues, like setting up encampments and targeting banks. The third study discovers performances never before measured in quantitative studies: performances that only occur through the interaction of protesters and police. These ‘dances,’ namely curfew disputes and traffic battles, highlight the need for more research understanding the relationships between intra- and trans-event conflict for social movement campaign outcomes. The article discusses how its novel approach is both more efficient to put into practice and much more adequate to Tilly’s theory of performances than traditional hand-coding approaches. The article then previews future work, discussing how the approach may be used to model the incidence of various performances as a function of political opportunity structures, and the temporal location of an event within a longer movement campaign. The article concludes with an invitation to use and improve upon the performance modeling approach.