1Quantitative methods have been at the heart of recent historiographical debates in the history of economics (e.g., Edwards, Giraud, and Schinckus, 2018). The recently published Quantitative Methods in the Humanities, co-authored by Claire Lemercier and Claire Zalc, offers a valuable introduction to the topic. The book is addressed “to anyone who is grappling with historical sources” (1), i.e. to all social scientists interested in quantifying the past. The book aims to familiarize the reader with quantitative methods and their application to historical objects, and to offer a critical perspective on their strengths and weaknesses.
2The book is divided into three main parts: chapter 1 sketches the history of quantitative methods in history; chapters 2 and 3 discuss issues related to the selection of sources, and the building of data. The remaining chapters introduce a set of quantitative methods gathered around four themes: correlation and causality; network and trajectories; visualization of data; and textual analysis.
3Chapter 1 narrates a brief history of quantification in the field of history. It explains how quantifications became popular after World War II, in particular with the rise of the Ecole des Annales. At the time, historians used quantitative analysis for describing long term and structural changes. The chapter then shows how this first wave of quantification started to be challenged in the 1980s. Long term series came under increasing scrutiny, and they became increasingly criticized. First, such methods were viewed as anachronic since they employed present categories, for instance, wage-earner, for describing periods in which such categories did not exist (18). Second, microhistory became increasingly important in the 1990s. Microhistorians criticized quantification because it captures only broad aggregated facts, and completely abandons the description of individual and local characteristics (18). The chapter ends with a normative discussion of the possible renewal of quantitative methods (22-26). Lemercier and Zalc (23-26) assert that quantification is not only about “macro” phenomenon, but also serves the microhistorians’ historiography. For instance, the authors argue that quantitative tools, such as network analysis, give a broader perspective on individuals that proves valuable for identifying individual peculiarities (24). The chapter concludes with a nuanced opinion: quantification is a necessary but non-sufficient tool of investigation for historians. For the authors, quantification can be particularly useful to abandon bad habits among historians, such as the use of “examples (selected in a nonexplicit way from a not-so-clearly-defined population) as proof, and adverbs such as ‘often’ or ‘generally’ whithout the support of precise data” (25). Ironically, the main flaw of this chapter is precisely what the authors criticize. In particular, it proposes a historical analysis (that of the debates among historians since the post-war period) that is based on fuzzy quantitative arguments, like the extensive use of “many historians” (15, 18, 19, 20, 21), which are not supported by any data. Nevertheless, we can forgive this flaw in the chapter, which seeks to introduce the reader to the subject rather than to analyze rigorously the evolutions of historiography since the postwar period. However, it illustrates how difficult it is for historians to tell a story that does not use such fuzzy quantitative arguments.
4Chapter 2 deals with selecting sources and data samples. The authors first claim that any source can be quantified: statistics, individuals, texts, images, pieces of art, speeches, or correspondences. However, the fact that anything can be quantified does not imply that quantification should always be used. Quantification alone does not answer all questions historians may ask. What matters is understanding how quantitative methods work and when they should be employed. “[W]hat matters … is not their type or form, but their suitability for quantification appropriate to the goals of the research” (30). The chapter then discusses two popular sources used by historians. First “statistics from the past” (ie. any numbers collected by statisticians of the period examined). Second, the prosopography, i.e. “listing individuals sharing some attribute (such as Roman senators or university professors) and analyzing the data to find out what other characteristics member of the group shared” (33). Finally, the authors introduce the notion of sampling, ie. selecting a subset of the population to estimate the features of the whole population. The general message emerging from this chapter is that historians should “abandon any hope of completeness” for their sources. Rather than looking for the ideal source, historians should “learn how to combine different sources and play on their respective strengths, rather than to look for the ideal set of documents” (30). The chapter introduces the reader to some quantitative tools for such purpose (e.g. contingency tables and the chi2 test).
5Chapter 3 focuses on the transformation of sources into data. According to the authors, the conversion of sources into data is a two-step process. Firstly, there is the “input” stage, i.e., defining the “units” of interest (persons, books, etc.) and compiling information about them (age, name, etc.). This stage aims at collecting “as much information as possible in the language of the source” (53). For the authors, inputting is comparable to the fieldwork of sociologists and anthropologists, where historians are in “physical contact” with the source (55). The second step is the categorization of sources, i.e., putting the pieces of information into a finite number of categories. The authors borrow from Alain Desrosières (1991) the instrumental perspective of quantification. In particular, statistical categorizing is putting some singularities (e.g., individuals) into equivalence classes (e.g., unemployed) for a specific set of actions (e.g., “economic policy”). In this tradition of thinking, a category is not good or bad, realistic or unrealistic in absolute terms. The relevance of a category is always evaluated by its ability to describe and answer questions. Thus, for historians, the categorization process should be conceived “as a moment to reflect on the sources and the purpose of the research” (62).
6The four last chapters present more directly a series of methods, organized in four main topics. Chapter 4 discusses correlation and causality by comparing linear regression with factor analysis. Chapter 5 is dedicated to methods measuring the evolution of social interactions in space and time. The chapter focuses mostly on network and trajectories analysis. Chapter 6 analyses data visualization through two examples, network visualization, and the geographic information system. Lastly, chapter 7 introduces various kinds of textual analysis such as topic modeling.
7The discussions are rarely technical. The authors largely ignore mathematical and practical aspects (related to programming). Instead, the authors present how such methods might be used by giving numerous examples of applications (mostly from the field of history). For a hands-on understanding of the methods, the authors refer to a website (https://quanthum.hypotheses.org) which includes tutorials, programming languages, softwares, reading suggestions, etc. The main limitation of this guide is the neglect of technical and mathematical aspects of the different methods presented. It is clear that the ambition of this book is not to provide a mathematical introduction to quantitative analysis. The online resources provided by the authors are also extremely helpful to complement the book content. Yet, a minimum amount of formalism would have helped introduce certain methods to the beginner. For instance, it is not clear why the basic regression equation is included (76), while the mathematical bases of sequence analysis, event history analysis, network analysis, or textual analysis are not provided. If the reader is not aware of such methods, the authors’ discussion on these methods and their applications are harder to appreciate.
8Nevertheless, Lermercier and Zalc’s introduction is a must-read for historians of economics. The book offers, first of all, a great introduction to a list of quantitative methods that are increasingly used in our field. In particular, for the history of recent economics, given the rise in the number of publications (Claveau and Gingras, 2016), it is difficult to analyze without quantification the general trend of economics. Ambrosino and al. (2018) and Claveau and Herfeld (2018) advocate topic modeling and network analysis to understand the evolution of recent economics. Recent works use quantitative methods to build indicators on the recent state of the field. For instance, Truc et al. (2020) use citation analysis to construct a simple indicator for measuring the interdisciplinarity of a discipline. They show that economics is, along with business, the most insular (i.e. the less interdisciplinary) discipline in the social sciences. This book might be helpful for historians to understand such methods. Not only for incorporating them in their work but also to be able to criticize them. For instance, Truc et al. (2020) work on interdisciplinarity is a direct reply to recent claims from economists (Angrist et al., 2020) that economics became more and more open to other disciplines.
9The variety of methods presented in this book is also valuable. While some of them, like network analysis and topic modeling, have a high entry cost, Lemercier and Zalc show a wide range of simple tools that historians of economics could easily use. For instance, on several occasions (e.g., 155) the authors advocate that most research questions might be answered, or partly answered, with simple descriptive analysis, such as contingency tables. While we are used to statistical tools in the work of economists that we analyse, we rarely use them for our historical investigation. Fourcade and Khurana (2013, 122) give a recent example of the usefulness of such elementary analyses for the history of economics. To illustrate the major role played by business schools in economics in the last decades, they show that, in 2003-2004, the top 20 business schools in the United States delivered as many PhDs (549) in economics as the top 20 economics departments (637). In the same perspective, Claveau and Dion (2018) analyze the convergence between economics and central banks. Notably they show the increasing number of central banks economists publishing in leading monetary economics journals. In both instances, such quantitative analyses are not an end in itself, but intended to complete qualitative arguments.
10The third contribution of this book for historians of economics is that Lemercier and Zalc focus the discussion on practical tips and concrete examples for historians. While it is easy to find other introductions on quantitative methods, they rarely address historical analyses. This book emphasizes the numerous historiographical issues that historians may face with the use of these methods. I especially appreciate the original discussion of chapters 2 and 3 on how to select sources and build databases for quantification. If quantification has been the topic of recent discussion in the history of economics (Edwards, Giraud, and Schinckus, 2018; Cherrier and Svorenčík, 2018; Herfeld and Doehne, 2018; Ambrosino et al., 2018), they rarely discuss the process of selection of sources, their sorting, and their inputing into a database. For bibliometric sources, Cherrier & Svorenčík (2018, 373) raise some concerns on the subjective choice of a database (Web of Science, EconLit, Jstor). Moreover, building databases remains crucial for other increasingly popular methods used in the history of economics, like prosopography (Svorenčík, 2018). Lemercier and Zalc stress that this process is not only important for avoiding biases, but is also one of the most interesting ones. Selecting quantifiable sources and building a database are essential steps in designing a proper and innovative research question: “it makes it possible to discover peculiarities and expectations, which can lead to innovative questions” (52).
11Quantitative Methods in the Humanities is a very good guide for any historians of economics interested in quantifying the past. While the novice will find a nice introduction to quantification, this book also sums up for more advanced scholars, with special benefits to the historians of economics, a set of original methods and historiographical issues—on the selection of quantifiable sources, the anachronism of our data building, or the relationship between individuals and groups—debated by historians for decades.