Quantitative Methods in the Humanities, by Claire Lemercier and Claire Zalc
Total Page:16
File Type:pdf, Size:1020Kb
Quantitative Methods in the Humanities, by Claire Lemercier and Claire Zalc. Translated by Arthur Goldhammer. University of Virginia Press, 2019. Pp. 188. $19.50 paper. ISBN 9780813942698. By Ryan Light University of Oregon [email protected] Methodology, Comparative and Historical Sociology, Culture Word Count: 1,068 The digitization of everything poses serious questions for social scientific research from the practical to the ethical: How should we understand these digital data? What principles should govern our treatment of the people who create these data even when their identity is unclear or anonymous? Of course, not quite everything is digital, especially historical information, and that itself poses an important question for researchers: What remains hidden within our analog archives or what is altogether lost? The methods required to address these questions are proliferating alongside the data, with sophisticated tools drawn from the interdisciplinary fields of computational social science and the digital humanities. However, these fields are often preoccupied with technological development at the expense of domain understanding. How do domain experts develop the technical toolkit to incorporate digital data into their analyses? Claire Lemercier and Claire Zalc’s Quantitative Methods in the Humanities: An Introduction, translated by Arthur Goldhammer, provides a thoughtful and elegantly written overview of the methodological landscape that will prove useful to those wanting to expand their analytic toolkit. The elephant in the room: Why should a social scientist read a book on “the humanities?” First, the title is somewhat misleading: The text focuses on historical and qualitative data that are often the subject of the humanities, but are also foundational to interpretive approaches in the social sciences as well. In other words, the title undersells the potential audience of this volume that is largely bounded by data and not by discipline. Second, the authors draw from a deep understanding of the sociological literature on quantitative historical methods from Abbott’s work on sequence analysis to Ragin’s work on QCA to Tilly’s various formal approaches among others. The book provides an outline of methodological options for those interested in digital and/or quantifiable data – especially digitized or potentially quantifiable historical information – whether in the humanities and social sciences. Moreover, Lemercier and Zalc focus on best practices for these specific techniques with attention to conducting future studies using these methods, but also, as they write, “hope that our readers will become critical consumers of quantitative research who neither fetishize numbers nor fear them” (pg. 1). The introductory chapters provide an overview of the history of quantitative approaches to history, while also outlining the strengths and weaknesses of using statistics to understand it. The authors strike a balance between enthusiasm for what these approaches can accomplish, while also cautioning against blind faith in enumerating the past. Rather, quantitative methods complement interpretive modes of understanding working in conversation with microhistories focusing on individuals and individual agency. The methods, Lemercier and Zalc argue, are best suited for a broader, macroscopic view of historical events and social worlds: “Quantitative data make it possible to refine accepted interpretations, to ask new questions, and to proceed to more individualized analyses after identifying broad patterns” (pg. 25). The authors rightly emphasize that collecting quantifiable data is often difficult – more difficult than simply “scraping” websites – and that this effort needs to be balanced with the potential reward. Moreover, while most information has the potential to be quantified in one way or another, the focus should be on research questions or the goals of the research project. The authors provide concise introductions to complicated ideas like various sampling strategies. Here, they also offer practical data collection advice as “The 10 Commandments of Inputting Data,” such as treating missing data as data – “There should be no empty cells in your file” (pg. 59) – saving your data frequently, and coding dates in three columns. The middle chapters of the book consist of a methods buffet with a focus on the following approaches: standard regression, factor analysis, social network analysis, sequence analysis, event history analysis, data visualization, and several forms of text analysis. The goal of these chapters is not to teach the reader how to conduct a regression, network, or sequence analysis in a specific computer program – you will need additional texts for these tasks. Instead, these chapters provide an overview of the goals of these techniques, the type of data they for which they are best suited, their strengths and limitations, and examples of their use in the social scientific literature. The visualization chapter, as a case in point, offers numerous practical guidelines for building useful visualizations. Lemercier and Zalc, for example, highlight how visualization is “always abstraction” built from a series of decisions that the analyst makes and the goal is to “remain faithful” to the underlying data, while also generating a legible analytic device (pgs. 126-129). They illustrate the pitfalls of illegible visualizations through a series of network graphs moving from a “spaghetti monster” – a densely connected graph that looks like a “hairball” of lines and points – to more useful iterations that select on particular types of ties. While a basic lesson, it is a pragmatic one and one that is likely to stick with prospective network researchers. The authors speed through these techniques at a brisk clip which keeps the reader engaged, yet also allows for occasional ambiguities that would likely be addressed in a more extended treatment. For example, I would need a lengthier discussion to be convinced that the primary purpose of network analysis is to “give a brief description of a network too complex to be described in a few sentences or readily drawn.” I am also not convinced that it is possible to adequately introduce topic models in five paragraphs. However, these are ultimately minor quibbles as the tradeoff is worth it. Lemercier and Zalc successfully provide an engaging and concise introduction to these sophisticated approaches. Some of the most difficult aspects of the research process are learning which approaches are best suited to answer what questions given particular data and research objectives. In other words, it can be difficult to know where to begin. This book provides one solution by offering an overview of the main techniques used in the quantitative analysis of qualitative and/or historical data. Lemercier and Zalc keep best practices at the fore focusing on how particular analytic approaches are better suited for particular research questions, warning readers to avoid fetishizing quantification, and highlighting how productive these tools can be. The results of these techniques, the authors emphasize in relation to text analysis, “do not replace, but support and complement historical interpretation” (pg. 154). Future research, perhaps conducted by people who read this important introduction, will likely further develop the connection between quantitative and interpretive methods. .