To deal with ever-larger datasets, media scholars are increasingly using computational analytic methods. This article focuses on how the traditional (manual) approach to conducting a content analysis-a primary method in the study of media messages-is being reconfigured, assesses what is gained and lost in turning to computational solutions, and builds on a "hybrid" approach to content analysis. We argue that computational methods are most fruitful when variables are readily identifiable in texts and when source material is easily parsed. Manual methods, though, are most appropriate for complex variables and when source material is not well digitized. These modes can be effectively combined throughout the process of content analysis to facilitate expansive and powerful analyses that are reliable and meaningful.Keywords: content analysis; computational social science; computational content analysis; hybrid content analysis; digital research methods; media analysis; algorithms T he abundance of digitized data has become a defining feature of modern society, and particularly of communication that is expressed through digital, social, and mobile platforms: 308 The ANNALS OF The AMeRICAN ACADeMY communication and media research, especially, the possibilities are great: as computational tools and processes have become easier to employ (e.g., via opensource software), as large-scale datasets of digital media content have become more readily attainable (e.g., by scraping tweets or websites), and even as print media content, such as old newspapers and books, have become increasingly available in digital form, new types of large-scale, algorithm-driven analyses of media content have become possible, enabling scholars to address novel questions. To cite just one example, Colleoni, Rozza, and Arvidsson (2014) used more than a billion data points to reveal key differences in the structures of political homophily among Democrats and Republicans on social networks such as Twitter. Such "naturally occurring" data such as public tweets, and the growth in computing capacity that facilitates the collection and analysis of such voluminous data, marks a key turn toward computational social science (Shah, Cappella, and Neuman, this volume).As a distinct approach to social inquiry, computational social science is characterized by research that (1) uses large, complex datasets; (2) often involves social and digital media sources; (3) employs algorithmic or computational solutions to generate patterns and inferences from data; and (4) is applicable to social theory in a wide variety of domains (Shah, Cappella, and Neuman, this volume). examples of such research may be found across a range of disciplines, including a growing number and variety at the intersection of the social sciences and the digital humanities (Bruns 2013). Much of this work involves the quantitative analysis of textual content. Yet unlike traditional content analyses-which rely predominantly on human judgments-these studies are largely driven by algorithms and framework...