Quantitative Methods for Historians

A Guide to Research, Data, and Statistics

Konrad H. Jarausch, Kenneth A. Hardy

• Editorial: The University of North Carolina Press
• Año: 2016
• ISBN:
  9781469621470
• Categoría: History
• Categoría: Historiography

The pioneering texts in quantitative history were written over two decades ago, but as a command of methodological context, computer experience, and statistical literacy have become increasingly important to the study of history, the need for an introductory text addressing these matters has increased. Quantitative Methods for Historians is a theoretical and practical guide for the application of quantitative analysis in historical research. It is designed for students of history and related disciplines who are curious about the possibilities of quantification and want to learn more about its recent development. Integrating the use of the statistical packages SAS and SPSS with the quantitative method, the authors discuss techniques for defining a problem, proceed to the building of a data set and the use of statistical methods, and conclude with the interpretation of results. The data set section concentrates on the basics of formalized research, discussing the coding process and the more complicated problems of data transformation and linkage. The statistical parts systematically build upon traditional fundamentals and introduce new analytical techniques for qualitative variables. Intended as a working introduction to quantitative methods, this guide also provides additional information on advanced statistical techniques and discusses questions of historical computing, reflecting critically on the proper role of quantitative methods.

Chapter 1:

The Scope of Quantitative History

Even after one generation of use, quantitative methods are still controversial in historical research and writing. Scholars continue to react more emotionally than rationally to the possibilities and problems of quantification in history. Because of math phobia or hostility against technology, humanists tend to charge the computer with reductionism and oversimplification. Fascinated with technical gadgets or scientific advances, social scientists make counterclaims of innovative research and accuse traditionalists of obfuscation. Some of this ritualized debate is based on ignorance and distortion, since detractors rarely have any firsthand experience in quantitative work while statistical zealots are often impatient with older methods which they have left behind. Much controversy is also due to contradictory ideographic or nomothetic conceptions of the purpose and method of historical research. Gradually this recognition began to lead, as in the case of the civilized exchanges between Robert Fogel and G. R. Elton (1983), to a “more ecumenical view.” But just when it seemed that quantitative techniques had been accepted in the profession, a fresh wave of postmodernist, feminist, and populist criticism started to call their legitimacy once again into question. In order to dispel old misunderstandings and to counter fresh misconceptions, it is necessary to take a closer look at the actual role of quantification in history: What is quantitative history, which are its topical varieties, and how can one learn more about it?

DEFINITION AND DEBATE

Unfortunately, there is no clear and simple definition of “quantitative history.” Instead, three overlapping uses recur in the literature. First, quantification refers to works with tables and graphs that measure and count evidence. This is simply a more rigorous and formalized description of the past which has been practiced since the ancient Greeks. It is only a small step from saying that someone was “tall,” “many” soldiers were killed in battle, or people experienced a “severe” famine to expressing these attributes in numbers to make them comparable over time and between cases. Second, beyond describing events, figures offer more analytical possibilities than regular words due to their formal properties. The large variety of statistical procedures makes it possible to evaluate causative connections through testing hypotheses in a more stringent way. The rhetoric of statistical analysis can be a great help in clarifying and falsifying interpretations. Finally, most quantitative historians also use the computer for handling their data and describing or analyzing it statistically. To be sure, small data sets can be investigated with a scratch pad or pocket calculator. Moreover, there is a rapidly growing field of formalized historical research such as textual analysis or data basing which does not involve statistics. But the former does not exclude the computer, while the latter often also involves word counting and eventually shades over into statistical analysis.

As a response to the deficits of philological techniques for analyzing mass data, quantitative methods impressively extend historical tools in certain areas of research. To begin with, the computer makes it possible to handle much greater masses of evidence than by laboriously sorting and resorting a card file filled with confusing and overlapping references. With a self-indexed data base, a scholar can make the computer quickly find all the relevant text passages for a specific purpose. Moreover, quantification permits greater precision of statements. When confronted with a manuscript census list, the historian no longer has to scan a portion of thousands of entries (which might be unrepresentative) in order to get a qualitative impression (or “feel”) of the evidence. Instead, one can draw a statistical sample and with great confidence describe the general properties of an enormous number of individual cases quite accurately. Finally, the greatest advantage of quantitative analysis is the increased clarity and complexity of explanations which it makes possible. Far from oversimplifying the past, quantification allows the systematic consideration of a larger number of causative factors by defining and testing their relationship statistically. In a cross-disciplinary context, Eric Monkkonen argues that “quantitative history gives depth and subtlety to positivist social science while giving public policy implications to historical research” (1984). The data handling, precise description, and interpretative complexity of quantitative history can therefore remedy a double deficit, namely, the lack of time perspective in social science and lack of generality in history.

The practice of quantitative methods has, nevertheless, shown certain shortcomings which have resulted in a number of often justified criticisms. The initial overselling by cliometric zealots who wanted to reduce the entire analytical universe to a series of equations has created a backlash which stresses that quantitative methods have failed to live up to expectations. Indeed, there has been no total historiographical revolution and proponents have had to show greater modesty in their claims. For many readers quantitative research also remains boring and inaccessible. Tedious tables and formulas are certainly not as entertaining as juicy gossip about royal mistresses. But the arcana of classical philological criticism are not self-evident either. Improved training might demystify statistics, elegant style can make even numbers come to life, and welldesigned graphs are often more intuitively intelligible than a convoluted narrative. Another widespread charge is that quantitative results are trivial, since they prove only what has been known already. While there is an element of truth in the contention that quantitative conclusions are rarely so surprising that they have not been anticipated in some older account, the elimination of other, equally plausible explanations through a confirmed hypothesis represents significant scholarly progress. Objections also rest on different sets of priorities: For a scholar interested in the motivation behind President Lincoln’s Gettysburg address, quantitative methods appear irrelevant, whereas for a colleague fascinated by the causes of the Great Depression, they are imperative. Such real or imagined defects have prompted, in Lawrence Stone’s phrase, a programmatic “revival of narrative” seeking to recreate popular mentalities (1979). In spite of their stridency, most criticisms of numerical antiquarianism or statistical dehumanization are groundless, since they invalidate the misuse of quantification rather than its proper application as an additional tool in historical research.

VARIETIES OF QUANTITATIVE HISTORY

Arguments about the merits of quantitative methods are likely to continue until their contribution to historical understanding has become so substantial as to preclude further debate. In several important research fields quantitative methods have already profoundly reoriented the entire agenda (Possekel 1990). For example, the first area to be transformed by quantification was economic history. In the 1950s and 1960s the study of economic development moved from the collection of descriptive time series to neoclassical economic analysis, called econometrics. This “new economic history” quickly became so rigorously statistical and vigorously theoretical that it has moved largely out of history departments into economics departments at the universities. Less interested in particular businesses or the economic life of one town, cliometricians such as D. North (1981) turned to the general and the lawlike analysis of entire economies. Preoccupied with questions of economic growth, scholars analyzed the importance of railroads “counter-factually” by asking, What if they had not been built? (R. Fogel 1964) or investigated the profitability of antebellum slavery in the South (Fogel and Engerman 1974). Provoking much controversy, these innovative works changed the terms of the debate from a focus on individuals to factors, from questions of entrepreneurship to models of development, from an occasional illustrative table to long strings of complex equations. Propelled by the theoretical and anticontextual bent of modern economics, the transformation of economic history has been astonishingly complete, so that quantitative methods have come to dominate perhaps even excessively. But when somewhat rehistoricized and reculturalized, the stringency of the statistical and pronouncedly theoretical orientation of this cliometrics still holds promise for the explanation of a considerable part of economic development.

Another major research area radically transformed by quantitative methods has been historical demography. The compilation and investigation of vital statistics, initiated by mercantilist governments in the eighteenth century, has been greatly facilitated by the computer, since it involves large masses of data on “life events” such as births, deaths, and the like. The necessary older summary statistics have become easier to compute from aggregate figures or individual cases. Moreover, entire new categories of sources such as parish books or civil registers could now be transformed into machine-readable form in order to be processed statistically. Much effort went into the reconstruction of national demographic histories, such as those at the Princeton Population Research Center or the Cambridge Group for Population Research, in order to explain the “population transition” from preindustrial high mortality/nuptiality/birthrate to postindustrial stability on a much lower level (Laslett 1965; Wrigley and Schofield 1981). Inspired by the case study approach of the French journal Annales, another set of researchers concerned itself with “family reconstitution” in order to detect local and regional patterns of inheritance and reproductive activity. Somewhat at the borders of the historical discipline and sometimes housed in independent institutes, this quantitative demographic history developed a set of sophisticated measures and explanations of population development. But it also spurred the emergence of a nonquantitative, anthropologically oriented “family history,” which explores the affective, economic, and sexual dimensions of family life in the past and in turn challenges demographers to develop more complex accounts of motivation (Maynes 1981 and Hareven 1978). One of the more exciting subfields has been the quantitative investigation of human health and disease (e.g., Bourdelais and Raulot 1987).

Closer to traditional historical concerns was the emergence of a new political history on the basis of quantitative methods. Encouraged by the triumph of behavioralism in political science a generation earlier, during the 1960s some American political historians began to consider whether similar concepts and survey methods might not also be applied to the study of the evolution of the political system in the past. Eventually three distinctive subfields developed out of this impetus. First, many voting or electoral studies (like the work of Lee Benson [1964]) analyzed the rich evidence of U.S. election data through the last couple of hundred years. Second, there was also much research on parliamentary behavior of representatives on the basis of “roll-call” analysis of legislative voting (by scholars such as T. B. Alexander [1967]). Finally, a cluster of analyses of the collective biographies of political elites (legislators, senators, etc.) sought to link socioeconomic background variables with political decisions on the state as well as national level (W. O. Aydelotte [1977] for Great Britain). The lively debates of the new political historians brought forth such concepts as the notion of “critical elections” in which party affiliation realigned or argued in “ethnocultural” terms of electoral loyalties. Many other political historians continued to write about power, patronage, and sectional conflict, rejecting the “new political history” as retrospective political science. But one of the pioneers of this recasting of political history, Allan Bogue, concludes persuasively: “Even if some of the criticisms are valid, much of the work of the social science historians will stand; and our conception of the political past of the United States will remain substantially altered by their efforts” (1983). This new political history has recently also found a home in Western European election, parliamentary, or elite studies, demonstrating that its methodological impulses are proving fruitful beyond the United States.

A final area of historical inquiry which has been greatly stimulated by the introduction of quantitative methods is the new social history. Propelled by radical democratic or Marxist concerns, this recent and most dynamic sector of historical inquiry shifted attention away from traditional elites to the often mute masses and addressed the power of social structures or customs rather than of political events. Obviously not all of the agenda, as defined by new journals such as P. Stearns’s Social History or H.-U. Wehler’s Geschichte und Gesellschaft, was amenable to quantification, especially when touching on collective beliefs or cultural customs. But many other concerns did and do involve quantification. For instance, the Tillys’ work on collective violence is based on extracting a strike statistic from newspaper accounts (before governments collected such figures) and on analyzing the causative impact of such factors as urbanization, industrialization, and the like (1975). Stephan Thernstrom’s (1964 and 1973) or Michael Katz’s (1975) exemplary studies of social mobility or of class structure in nineteenth-century cities such as Newburyport or Hamilton analyze manuscript census data quantitatively. Fritz Ringer’s (1979) or my own (1982) work on educational expansion and opportunity tests contemporary assertions of elitism against educational statistics in part derived from matriculation records. A complete listing of subtopics of expanding social historical fields, such as women or crime, would go on and on. In contrast to econometrics, the more open-ended subject matter and the less developed links to sociological theory have allowed quantifiers to dominate only a part of the field and forced them into a healthy dialogue with their qualitatively oriented colleagues. While they have also reshaped the research agenda through prosopography or case studies, quantitative social historians have been compelled to maintain greater methodological balance in their work.

Though quantitative methods have not nearly invaded all of historical writing, they have revitalized research significantly. To mention just a few areas, psychohistory, intellectual history, political biography, or diplomatic history have hardly been touched by quantification. But the impact of the quantitative revolution on those fields where statistics could be applied with profit has been profound. It has produced much innovative research which has answered many of the old questions and raised new ones. During the last three decades, quantification in history has also reached a certain maturity. Even skeptics admit that the civil war among historians has ended in the triumph of some of the revolutionaries within the guild. Tentative beginnings in the 1950s gave way to exciting departures during the 1960s when quantitative methods were programmatically hailed as saviors or denounced as destroyers of the profession. In the 1970s the quantitative movement succeeded in institutionalizing itself and in transforming the research agenda in a variety of fields through major substantive contributions. During the 1980s the quantitative wave seemed to have crested and receded somewhat, since trendsetting scholars turned to “softer” innovative methodologies and returned from analytical to narrative history. Due to the rise of deconstruction, the surge of feminism, and the arrival of everyday history, quantitative history is no longer the latest fad in the 1990s. Older buzzwords such as coefficients, models, and modernization have been replaced by new “in” terms like textuality, gendering, or experience and so on. But in contrast to debilitating self-doubts, research practice demonstrates that quantitative methods are here to stay in an increasing portion of historical works. Repeated surveys of the United States, Britain, and Germany (Sprague 1978 and Johnson 1988) reveal that quantification has been accepted by a major sector of the profession as a practical and worthwhile addition to historical methods to be applied as a matter of course. Hence it is necessary for productive historians to have some grasp of its procedures in order to understand its limitations and possibilities.

INFORMATION SOURCES AND LEARNING AIDS

Historians wishing to inform themselves about quantitative methods can turn to a growing number of institutionalized sources of information. Although they are a minority within the profession and may well remain so in the immediate future, quantifiers are now represented in most larger departments of history. Many colleges also offer some introductory course in quantitative methods for historians, while others force students to gather such skills in neighboring departments. Better training is provided by short courses at the Inter-University Consortium for Political and Social Research (Michigan), formerly at the Newberry Library and now at Northwestern (Chicago), or in other summer schools. Organizational progress has been even more impressive. While the original quantitative methods committee of the American Historical Association still provides some impulses, a separate vigorous scholarly organization did spring up during the 1970s, aptly named the Social Science History Association. The annual meetings of the SSHA in early November are the most sophisticated interdisciplinary forum for the discussion of advanced quantitative work anywhere. Historical data-base development is the central concern of the newly founded Association for History and Computing (AHC), based in Britain but rapidly expanding throughout Europe. Since the early 1980s an International Commission for the Application of Quantitative Methods in History (INTERQUANT) holds cross-national workshops and sponsors a program at every International Congress of Historical Sciences meeting. Much of the advance of quantitative scholarship has finally come from a spate of innovative journals such as the Journal of Interdisciplinary History (published at Tufts), Social Science History (sponsored by the SSHA), Historical Methods (having grown out of a technical newsletter), Historical Social Research (supported by the German QUANTUM group), and the latest entries such as Histoire et Mesure (produced in Paris) and History and Computing (published by Oxford University Press).

Interested students can also consult a considerable introductory literature. While they still contain some relevant information, the once exciting introductions by Edward Shorter (1971) or Charles Dollar and Richard Jensen (1971) now seem curiously out of date. Somewhat more useful is Roderick Floud’s text (first released in 1973 and updated in 1979), but its British economic history perspective emphasizes concerns different from the needs of most American researchers. A series of interesting anthologies of the early 1970s (Aydelotte, Bogue, and Fogel, The Dimensions of Quantitative Research in History; Lorwin and Price, The Dimensions of the Past; or Rowney and Graham, Quantitative History) contain some pioneering work that is still valuable. While the show debates between William Aydelotte and Jack Hexter (in Aydelotte 1971) or Robert Fogel and G. R. Elton (1983) produced the expected fireworks, several special issues of journals provide more up-to-date insight into the current discussion without and within quantitative history: Theodore K. Rabb’s essay collection on the new histories in the JIH of 1983, Peter Smith’s issues on quantification and epistemology in the HM of 1984–85, and Jarausch’s own survey of “quantitative history in international perspective” in the SSH of 1984. For the data-base ferment one might want to consult also Peter Denley and Deian Hopkin’s collection on History and Computing (1987) as well as Heinrich Best’s abstracts of the Cologne Computer Conference of September 1988. Though the dynamic development of quantitative methods prevents the emergence of an immutable “canon,” the continuing debate is producing enough practical experience with techniques and consensus on the interpretative implications of procedures to make another attempt at synthesis necessary and possible. The present introduction therefore intends to present the diverse possibilities of quantitative methods not in the breathless enthusiasm of initial discovery but in the somewhat chastened confidence of mature experience.

“Should I dare to quantify or not?” Unfortunately, there is no clearcut and generally applicable answer to this crucial question. The decision for or against quantitative methods depends above all upon the specific research problem confronted by the scholar. Since quantitative methods have no magic powers nor is the computer a futuristic toy, a rational choice presupposes a realistic knowledge of the advantages or disadvantages of quantification. Only when one understands the concrete workings as well as the intellectual implications of quantitative techniques can one decide whether the investment of time and energy into learning them is likely to yield enough interpretative insights to make it worthwhile. This text seeks to help h...