Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 246

246 QUANTIFICATION AND EXPERIMENT 14

Quantitative History

Margo Anderson

WHAT IS QUANTITATIVE HISTORY? of multiple events or phenomena. Such a standpoint creates a different set of issues for Quantitative history is the term for an array analysis. A classic historical analysis, for of skills and techniques used to apply the example, may treat a presidential election as methods of statistical data analysis to the a single event. Quantitative historians con- study of history. Sometimes also called clio- sider a particular presidential election as one metrics by economic historians, the term was element in the universe of all presidential popularized in the 1950s and 1960s as social, elections and are interested in patterns which political and economic historians called for characterize the universe or several units the development of a ‘social science history’, within it. The life-course patterns of one adopted methods from the social sciences, household or family may be conceived as and applied them to historical problems. one element in the aggregate patterns of fam- These historians also called for social scien- ily history for a nation, region, social class or tists to historicize their research and con- ethnic group. Repeated phenomena from the sciously examine the temporal nature of the past that leave written records, which read social phenomena they explored. For both one at a time would be insignificant, are par- types of questions, historians found that they ticularly useful if they can be aggregated, needed to develop new technical skills and organized, converted to a electronic database data sources. That effort led to an array of and analyzed for statistical patterns. Thus activities to promote quantitative history. records such as census schedules, vote tal- Classical historical research methodology lies, vital (e.g., birth, death and marriage) relies upon textual records, archival research records; or the ledgers of business sales, ship and the narrative as a form of historical writ- crossings, or slave sales; or crime reports ing. The historian describes and explains par- permit the historian to retrieve the pattern of ticular phenomena and events, be they large social, political, and economic activity in the epic analyses of the rise and fall of empires past and reveal the aggregate context and and nations, or the intimate biographical structures of history. detail of an individual life. Quantitative The standpoint of quantitative history also history is animated by similar goals but takes required a new set of skills and techniques as its subject the aggregate historical patterns for historians. Most importantly, they had to Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 247

QUANTITATIVE HISTORY 247

incorporate the concept of the data set and growth and expansion of the United States data matrix into their practice. Floud (1972: had long required American historians to 17) defined the data set as ‘a coherent selec- consider quantitative issues in their study of tion of data from the whole range of historical the growth of the American economy, popu- data available to the historian, and it is lation and mass democracy. Thus, for exam- selected because it relates closely to the ques- ple, Frederick Jackson Turner’s classic 1893 tions that the historian wishes to consider.’ essay on ‘The Significance of the Frontier in The myriad instances of a phenomenon— American History’ was largely based on a for example, all United States presidential reading and interpretation of the results of elections—form the cases of the data set. The the 1890 population census. pieces of information collected about the But true ‘data analysis’ in the current sense cases—for example, the candidates running, had to await the growth of the social and sta- the year of the election or the vote totals— tistical sciences in the first half of the twenti- become the variable characteristics of the eth century, and the diffusion to universities data set, that is, the varying characteristics of in the 1950s of the capacity for machine tab- any particular case. The historian arranges ulation of numerical records, and then of the data in tabular form, that is, in a matrix of mainframe computing in the 1960s. One can rows and columns, ‘consisting of a number see the emerging field exemplified in semi- of rows, which will normally represent cases, nal studies in the late 1950s and early 1960s. and a number of columns, which will nor- In 1959, for example, Merle Curti and his mally represent variables’ (Floud, 1972: 18). colleagues at the University of Wisconsin The creation of quantitative data sets thus published The Making of an American required the historian to carefully compile Community: A Case Study of Democracy in a consistent information about the phenome- Frontier County. Curti et al. (1959) explored non to be investigated, and prepare the data Turner’s thesis with an in depth look in tabular form. Historians then were pre- at the mid-nineteenth century history of pared to apply the techniques of statistical Trempeleau County, Wisconsin, including its data analysis to the data set to answer the records of newspapers, diaries, private research question posed. papers and county histories. But they also In short, to make effective use of quantita- added data analysis of the employment pat- tive evidence and statistical techniques for terns derived from the individual-level fed- historical analysis, practitioners had to inte- eral census manuscripts for the censuses grate the rapidly developing skills of the from 1850 through 1880. social sciences, including sampling, statisti- Similarly, the ‘new’ economic historians cal data analysis and data archiving into their of the 1950s challenged the conventional historical work. That task led to the develop- wisdom of the day on several key issues in ment of new training programs in quantita- economic history. One debate centered tive methods for historians, to the creation of on the ‘necessity’ of the US Civil War. new academic journals and textbooks, and to Historians at the time argued that the war had the creation of data archives to support the been ‘unnecessary’ since the institution of research. race-based slavery would collapse under the weight of its unprofitability. In contrast, eco- nomic historians employed economic theory EARLY EFFORTS and data on output of southern agriculture to argue that the southern agricultural economy Historians had made use of quantitative evi- could have survived profitably into the twen- dence prior to the 1950s, particularly in the tieth century using slave labor (Conrad and fields of economic and social history. The Meyer, 1958). Robert Fogel challenged the Annales school in France pointed the way in conventional wisdom on the centrality of the pre-World War II period. The rapid railroads for the industrial development of Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 248

248 QUANTIFICATION AND EXPERIMENT

the United States. Making use of economic analysis as part of its summer program in theory, carefully compiled data series, and quantitative methods. The course continues the logic of the counterfactual, Fogel to be offered each summer. At the Newberry argued that canals would have also suc- Library in Chicago, from 1971 to 1982 ceeded as a transportation system underpin- Richard Jensen spearheaded a summer pro- ning nineteenth-century American industrial gram in quantitative methods for historians. development (1964). By the early 1980s, about 40 percent of ‘New political historians’ such as Lee history graduate programs offered training in Benson, Allan Bogue, Richard P. quantitative history as part of the graduate McCormick, and political scientists with his- curriculum (Bogue, 1983: 220ff.).3 torical interests, such as Warren Miller and Additional institutional infrastructure of Walter Dean Burnham, translated the emerg- quantitative history can also be dated to the ing techniques of political scientists analyz- 1960s. New journals, textbooks, and edited ing contemporary election results and voter collections also promoted the growth of surveys to historical questions, and opened quantitative history. The Historical Methods up dramatic new insights into American Newsletter, for example, began publishing in political history.1 The new political historians 1967, and was renamed Historical Methods identified the parameters of party systems, in 1978. The Journal of Interdisciplinary developed the theory of the critical election, History began publication in 1970. The and argued that underlying structures of elec- Social Science History Association (SSHA) toral politics were accessible through histor- was founded in 1974 and the first issue of its ical analysis of voter turnout and election journal, Social Science History, appeared in results. In 1964 in England, demographers 1976. SSHA became the professional venue and historians founded the Cambridge Group for bringing together historians who con- for the History of Population and Social sciously adopted theory and methodology Structure and began a forty-year project to from the social sciences and social scientists retrieve, assemble and reconstruct 400 years doing historical work. The cross-fertilization of the family history of Britain.2 has continued, and, as noted below, many of The new possibilities of quantitative the innovations in quantitative history have history fit well with other trends within the been developed by scholars with formal discipline of history, particularly with the training in the social sciences and appoint- growth of social history and calls for what ments in departments of economics, demog- Jesse Lemisch (1967) called ‘history from raphy, sociology, anthropology, geography the bottom up’—that is, for historians to and political science. treat the lives of ordinary people, to comple- Textbooks in quantitative history began to ment the study of elites. By the mid-1960s, appear in the early 1970s, and many have the interest in the new techniques led the been published since.4 Numerous edited American Historical Association to recog- volumes introduced the new field and nize that ‘quantification in history’ would techniques to professional and student audi- require new skills and institutions within the ences.5 Finally, researchers created data historical profession. The AHA created a archives. In the United States, the Inter- Quantitative Data Committee to consider the university Consortium for Political Research issues. Summer institutes and classes in (ICPR) was founded in 1962 primarily by quantitative methods for historians were held political scientists. Renamed the Inter- in 1965, 1967 and 1973 at the University of university Consortium for Political and Michigan, Cornell University and Harvard Social Research (ICPSR) in 1975, the University respectively. In 1968, the Inter- Consortium has also pioneered in the university Consortium for Political Research creation and preservation of historical data at the University of Michigan began offering collections. The United States National a four-week course in quantitative historical Archives and Records Administration (NARA) Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 249

QUANTITATIVE HISTORY 249

created an electronic records preservation violence in the past. Historians of the family program in the early 1970s for federal gov- have examined patterns of inheritance and the ernment data that was ‘born digital’ inter-generational transfer of wealth. The (Ambacher 2003; Adams, 1995, forthcom- emerging work of ‘anthropometric’ history— ing; Fishbein, 1973). Similar work began in the study of living standards and well-being Britain with the founding of the UK Data in the past using measures of height, weight, Archive in 1967.6 stature and disease in the past—has cast an Thus by 1980, historians had take major even wider net, aiming to evaluate compara- steps to establish the institutional structures tive living standards over centuries and ulti- necessary to integrate quantitative history mately millennia.7 into larger historical practice. That infra- Making such studies possible was an structure has, if you will, both matured and explosive growth in the data sets informing faced challenges in the generation of work quantitative history. Quantitative history, like since, and in many ways quantitative history other branches of the social sciences, is still a work in progress. Nevertheless, it is requires what was once called ‘machine- possible to identify the types of questions readable’ (and are now known as ‘elec- quantitative history was intended to and has tronic’) data for analysis. Though there are been able to address; the major types of data some examples of large-scale data analysis sets that have developed and the key charac- undertaken by manual systems of tabulation teristics of historical data sets; and the most and statistical analysis, most notably the commonly used techniques within the field. nineteenth-century tabulations of census or That background in turn provides the frame- vital registration records, social science data work for a review of a number of method- in the modern sense required the develop- ological issues historians uniquely face, for a ment of machine tabulation devices, counter review of the achievements of quantitative sorters, and other mechanized calculators. history, and for a discussion of emerging The first system was the Hollerith system of issues. punch-card tabulation used for the 1890 American population census; the social and statistical sciences grew with the new QUESTIONS, DATA AND ISSUES IN machinery. By the 1940s, social scientists CREATING HISTORICAL DATA SETS had developed rules and procedures for col- lecting quantitative data to make best use of Quantitative history has been most successful machine tabulation and analysis. These con- in addressing big questions about long-term ventions included the fixed format data historical patterns of change. Practitioners matrix, the classification of variables into have achieved important results by assem- nominal, ordinal, interval and ratio variables, bling substantial amounts of numeric or the organization of questionnaires and survey countable information, and organizing it into forms to facilitate conversion to punch-cards tabular data matrices for statistical analysis. for analysis, and coding systems such as the The first generation of studies focused espe- Likert scale. Quantitative historians inherited cially on the history of the family and social these practices and adapted this existing structure, trends in economic growth and technology and set of conventions to their change, patterns of electoral behavior and historical project. The soon recognized that voter participation, or the record of inter- they had to solve major new methodological generational social mobility and living stan- and logistical problems before the potential dards. More recently, the examples have for quantitative history could be achieved. proliferated. Historians of crime and the The first problem derives from the larger criminal justice system, for example, have evidentiary issue faced by all historians, retrieved court and newspaper records to namely, that historical analysis must rely on examine the long-term patterns of crime and the extant record of the past. Historians are at Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 250

250 QUANTIFICATION AND EXPERIMENT

the mercy of their subjects’ penchant and Methods, in particular, became the venue for capacity for preservation. And before 1890, identifying, debating and proposing method- that is, for most of the historical record of ological solutions to these issues. human history, no preserved data were A related issue is the set of rules for ‘machine-readable’. Thus all potential histor- extracting the information from a text-based ical data had to be created from surviving, evidentiary source to create a data set. usually text-based, records and converted to Historical archives frequently contain text- machine-readable or electronic format. Even based records that lend themselves to data set records collected in the twentieth century and construction, but require considerable con- informed by the conventions of the emerging ceptual work before they can be manipulated social sciences frequently no longer exist in statistically. Historians have made use of machine-readable format. Thus, the United sales invoices, wills, parish registers and case States Census Bureau, for example, pre- files of charity or social welfare agencies, for served the original paper census question- example, and have had to create the cases naires from the eighteenth century forward. and variables from the extant texts. But census officials did not retain the punch- Historians have had to solve these method- cards they used to tabulate the censuses from ological questions as they select the evidence 1890 to 1960. These cards were destroyed to be analyzed and create the code-book for once the results of the census appeared in the data set. Whether one is analyzing exist- published form. Thus historians interested in ing tabular data from the past—for example, reanalyzing the microdata from past cen- the records of imports and exports of a nation suses faced creating, or recreating, the over a period of years, or the published machine-readable records. results of a census—or whether one is creat- Quantitative historians faced additional ing a data set from text-based sources, the major methodological problems resulting historian needs to define the case or unit of from the recalcitrance of the existing archival analysis, define the characteristics or vari- historical records. All historians face the ables to be selected to characterize the cases problems of missing data, and the difficulties within the data set, and define the coding of interpreting illegible, damaged, incom- system used to organize the source informa- plete or destroyed records. For quantitative tion for the data set. Several examples of the historians, though, aiming to translate the issues involved best illustrate the work of archival record to a data matrix for statistical quantitative historians. analysis, these questions of data quality are particularly difficult. Cases and variables for a data matrix require precise conceptual and COMPILATION AND ANALYSIS OF operational definitions, as do the allowable PREVIOUSLY PUBLISHED DATA entries for particular cell values within the matrix, since the goal of statistical analysis is The most accessible sources for quantitative to assess extent, central tendency and disper- historians were data that were already sion of any particular characteristic. What published in tabular format. The first genera- does one do if the records for a year or period tion of quantitative historians in particular of years are missing? How does one handle compiled data sets from existing, usually illegible entries in the records of a company’s aggregated, published data sources—for finances? How does one know if the probate example, tabulated census results, election records found in a county archive are com- results, government reports of tax collec- plete? Historians have had to confront the tions, imports and exports, and data from requirements for case and variable definition, trade publications. Assembled into time classification and coding in building a data series, such data permitted researchers to set. The solutions to these problems emerged undertake basic analyses of historical trends with the overall field. The journal, Historical and use regression models to correlate the Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 251

QUANTITATIVE HISTORY 251

determinants of change. For example, Walter volume through collaboration and by build- Dean Burnham’s 950+-page study of ing historical public-use microdata samples, count-level presidential election results, pub- or PUMS files. Starting with the 1900 lished in 1955, included a compilation of census, historians proposed to create histori- results from state archives and newspaper cal PUMS files that would be similar to the sources, and a discussion of the methodolog- contemporary PUMS files that the Census ical issues he faced in compiling the data. Bureau has created since 1970. In the late Combined with denominator data from 1980s, researchers at the University of census results that allowed the researchers to Minnesota, initially led by Russell Menard, measure turnout, the new data set permitted Steven Ruggles and Robert McCaa, began Burnham and his colleagues to begin the systematic retrieval of the historical census analysis of historical election analysis data from the United States, and more (Burnham, 1955). In similar ways, economic recently from other nations. The Integrated historians made particularly good use the Public Use Microdata Sample (IPUMS) data compiled in statistical abstracts, such as Project and the International IPUMS project the Statistical Abstract of the United States, have created microdata samples for the published annually since 1878. United States from all the censuses from 1850 to 2000, and are now collecting such data for many nations of the world. The data are easily downloadable from the web. The CONVERTING TABULAR DATA IN researchers have also built the code-books, MANUSCRIPT FORM TO ELECTRONIC technical support materials, and research FORMAT bibliography necessary for the user to under- stand the context of the questions and A second source of quantitative data were responses to the census.10 archived tabular records in text-based for- mat, probably best illustrated by individual- level census manuscript schedules. See Creating Tabular Data from Text 8 Figure 14.1, a facsimile of the 1950 US Based Records Census population schedule. For the United States, such original census The most time-consuming type of data set responses are available for all the federal creation is the conversion of text-based censuses except 1890, and are available for records to matrix format. For existing tabular public use through 1930.9 The schedule is data, whether in manuscript or published already in a matrix format, with rows of form, the basic framework of the matrix is cases and columns of variables. The original given in the original source. For text-based difficulty with using these records is their records with no tabular structure, it is up to volume. With one record per person for the the researcher to create the code-book, and censuses of 1850 and later, data set creation thus all the variable definitions and coding for a large portion of the population was rules. Figure 14.3,11 an illustration of a record beyond the capacities of an individual of a slave sale in antebellum America, illus- researcher. The first generation of quantita- trates the issues.12 tive historians resolved this problem by sam- There are thousands of such records in pling, and usually by organizing a research newspapers, private collections and archives, project of a particular locale. The historical and, if marshaled for analysis, provide social mobility studies were designed as detailed, if somewhat gruesome, evidence of community studies to solve the problem of this chapter in American economic history. the volume of data. Robert Fogel and Stanley Engerman com- Later generations of quantitative histori- piled such records for their study, Time on the ans have by and large solved the problem of Cross (1974) from the New Orleans Slave Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 252

252 QUANTIFICATION AND EXPERIMENT

Figure 14.1 Facsimile of 1950 Census Schedule for Orange City, Iowa

Market. ICPSR Study 7423 contains the data of analysis (the slave), sampling (2.5 percent and code-book for the New Orleans Slave or 5 percent, depending on the year of sale), Sale Sample.13 number of variables (46), and codes. Each For their sample, Fogel and Engerman con- decision extracted a piece of information verted the text-based records into cases and from the original text-based records, and had variables and codes, making decisions on unit implications for ultimate analysis. The final Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 253

QUANTITATIVE HISTORY 253

Figure 14.2 Facsimile of 1950 Census Schedule for Orange City, Iowa

data set contained 5009 records, and included Fogel and Engerman used the data to analyze information on the characteristics of the slave the inter-state slave trade, and to address (e.g., age, sex, occupation, color), the terms questions about the economic viability of the of the sale (e.g., the date, price, whether paid slave economy (Fogel and Engerman, 1974). in cash, the number of slaves sold together), The work of building the corpus of and information on the buyer and seller. machine-readable databases began in the Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 254

254 QUANTIFICATION AND EXPERIMENT

Figure 14.3 Slave Bill of Sale, Davidson County, Tennessee, 1833

1960s, and continues both with small compi- to a manageable level. Just as one does not lations and large collaborative data projects. need to survey the entire electorate to In addition to the IPUMS project mentioned develop quite precise estimates of the ulti- above, one can find large-scale historical data mate election results, so historians studying compilations of cost-of-living studies, elec- family structure or economic activity or con- tion results, crime data, and the records of the sumer behavior have not had to record all heights and weights of people in the past. The such behavior for study. As noted above, the creation and retrieval of historical data has process of creating historical data sets is suf- also led to revision and improvement of data ficiently time-consuming to strongly recom- series compiled in earlier years and to the mend sampling strategies designed to reduce analysis of the history of data development. the volume of coding and data entry to the Most recently, for example, economic histori- minimum necessary for robust analysis. Thus ans have produced a new ‘millennial’ edition the original users and secondary users of the of the Historical Statistics of the United archived historical data sets need to attend to States (Carter et al., 2006), which promises to sampling strategy and introduce appropriate provide opportunities for even more quantita- sample weights and measures of error into tive historical analysis. the analysis. A more difficult issue is the one facing the historian who cannot be sure that she knows ANALYZING HISTORICAL DATA SETS what the universe of cases actually is. Do the extant newspaper reports of lynchings, for Sampling and the Universe of Cases example, encompass all lynchings (Griffin et al., 1997; Tolnay and Beck, 1995)? Are the As have data analysts in the other social records of wills filed with a particular county sciences, historians have made use of the complete, or might some have been destroyed theory of probability sampling to reduce the or lost over the centuries? These dilemmas volume of information for a particular study have their analogues in non-quantitative Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 255

QUANTITATIVE HISTORY 255

research. But as with code-book creation, the with surveys of individual voters. Indeed the research must provide best estimates of National Election Survey, conducted since answers to such questions before analysis, 1948, has itself become an historical source and a substantial methodological literature of changing electoral behavior. But histori- has emerged to address the issues, often with ans cannot go back and survey voters from specific reference to the kind of data set being the election of 1860, and thus must make use compiled. of the aggregate election results and the eco- logical characteristics of the voting units— Techniques of Analysis e.g., precincts, districts or counties—that provided the vote. Ecological inference suf- Statistical analysis of historical data has fers from the threat of the ecological fallacy, ranged from elementary data analysis of the that is, the danger of wrongly inferring indi- patterns of central tendency and dispersion of vidual level behavior from the patterns of the phenomena under study to elaborate aggregates. Practitioners of quantitative explanatory models of events and behavior. history have taken up new methods devel- Much historical quantitative analysis has oped by political scientists and have devoted been descriptive, simply excavating and doc- good effort to minimizing, if not completely umenting patterns of change and activity in solving, this dilemma. With historically quantitative form that cannot be revealed by minded political scientists, they have pro- traditional historical analysis. Thus much duced a methodological literature and new work—important work—is simple counting techniques that have produced rigorous of a phenomenon, and describing trends over results.14 time. The second contribution is serious atten- Somewhat more elaborate analysis tion to the development of statistical tech- involves determining the correlates of the niques to conceptualize and model time and phenomenon under study, or building a temporal explanations. The methodological model to explicate more complex patterns in bread and butter for all historians is ‘thinking the data. Here the standard bivariate and in time’ (Neustadt and May, 1986), and that multivariate techniques of statistics provide standpoint has prompted historians and his- the tools necessary for the analysis. torically attuned social scientists to think Quantitative historians have borrowed about how to develop techniques of statistical heavily from sociology, political science, analysis suitable for the goals of historical demography and economics, and made use analysis. of the classic linear regression model and its Historians think about questions of what is variants as the workhorse technique for more an event, how is it bounded and measured; complex analysis. Statistical packages, such what is a turning point; what is a transition; as SPSS, SAS, STATA and the like underpin what is a conjuncture or a rupture; and how the analysis of quantitative historical work, is a period of time organized and bounded. as they do for the social sciences. Economists and other social and biological There is some evidence that quantitative scientists have developed techniques to mea- history has begun to have an impact on the sure time series and temporal and cyclical larger methodological practice of the social events, for example, life cycles. The entry of sciences, as quantitative historians have quantitative historians into these discourses brought their methodological expertise to the has been a useful clarification of the method- social sciences. Two brief examples should ological issues involved. For example, the illustrate that impact. phrase, ‘longitudinal analysis’ that social sci- The first is development of the field of entists use does not necessarily privilege time ecological regression, particularly for analy- as a central concern for analysis. Historians sis of electoral patterns. Political scientists and social scientists who make temporal can supplement analysis of election results analysis such a central concern have thus Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 256

256 QUANTIFICATION AND EXPERIMENT

argued for the need to add methods that will include history in the main governmental address ‘thinking in time’ to the standard foundation for funding academic research. repertoire of statistical techniques. Such Accordingly, quantitative history projects in techniques as sequence analysis, event the United States have had major difficulty in history analysis and the methodological dis- competing with both large-scale humanities cussions surrounding autocorrelation in time grant projects, such as compilations of archival series analysis have usefully been enriched papers, and with large-scale long-term by the growth of the field of quantitative research projects in the social sciences such as history.15 the National Election Study or the Panel Study of Income Dynamics. Allan Bogue (1983) iden- tified the chronic problems of funding faced by THE COSTS OF DOING quantitative historians in the late 1970s. They QUANTITATIVE HISTORY remain unsolved as the concrete example which follows illustrates. The cost of scholarly work in quantitative Robert Fogel, by any measure, represents history, like the cost of all scholarly work, one of the most successful and innovative can be measured in terms of both time and quantitative history scholars in the field, yet money required for the scholarship to flour- even he has faced major funding obstacles. ish. The largest change in the working envi- Fogel was awarded the Nobel Prize in ronment since the 1960s is that computing Economics in 1993, and in his autobiograph- costs, which were quite expensive in the ical statement prepared for the award, he early years of the field, have dropped as the described his career and acknowledged the larger information revolution has developed. problems of funding he faced, particularly, as To my knowledge, there is no extant schol- he put it, for the ‘current research projects on arly analysis of the costs of quantitative which I reported in the Prize Lecture’. The history versus traditional history, though I Center for Population Economics at the suspect that the underlying funding situation University of Chicago and the Walgreen for quantitative historians has had an effect Chair provided funding when federal grants on the progress of the field. would not. ‘The data on health conditions’, In the early years of the development of he wrote: quantitative history, in the United States the Social Science Research Council, the comes from a project called ‘Early Indicators of Later Work Levels, Disease, & Death’ which is trac- American Historical Association, the National ing nearly 40,000 Union Army men from the cra- Endowment for the Humanities, and the dle to the grave. It takes over 15,000 variables to National Science Foundation, as well as describe the life-cycle history of one of these men. research universities around the country, all These life-cycle histories are created by linking provided sponsorship of the field by funding about a score of data sets. It took more than half a decade of work to investigate the potential of grants for data development, conference spon- these data sets, work out procedures for data sorship and the institutional work required to retrieval and file management, and to establish the promote the field. This early institutional sup- feasibility of the enterprise in our own minds. port was aimed at jump-starting the field, not at providing sustained long-term support. The site committee of the National Institutes of Health which reviewed the original project pro- Related to this, the National Endowment for posal in 1986 agreed that such a project could in the Humanities, the main federally sponsored principle make a significant contribution to an grant agency for historians, has a much lower understanding of the process of aging, but they funding level than the federal funding agencies were skeptical about the quality of some of the that support related social science research— data, about whether the software and program- ming procedures we had developed by that time for example, the National Science Foundation were adequate for the management of such a or the National Institutes of Health. The United large data set, and about whether the project States, unlike European nations, does not could be completed within the proposed budget. Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 257

QUANTITATIVE HISTORY 257

To resolve these doubts it was necessary to draw a (Bridenbaugh,1962), memorably labeling it a six percent subsample which linked together all ‘bitch goddess’ (Bogue, 1983). Even during of the separate sources and which demonstrated the effectiveness of the software by analyzing the the period of the rapid growth of quantitative information in the subsample. It took an additional history in the 1960s and 1970s, ‘traditional’ four years to complete the second phase of the historians expressed doubts about the new justification of the project. Thus nearly a decade of methods, challenging them as reductionist, preliminary research, much of it funded by brittle and not pertinent to the main goal of the Walgreen and the CPE, was required before the project was accepted by the peer reviewers of NIH historical narrative. Critics were extremely and NSF.16 (Fogel, 1993) dubious of the ‘scientific’ claims of quantita- tive historians, and resisted the challenge of Despite such barriers, quantitative histori- the quantifiers that traditional historical writ- ans have been able to take advantage of the ing was not theoretically rigorous or concep- technological developments in computing tually consistent. and data management to make major In the 1980s, some of the original propo- advances in the ease of analysis, in terms of nents of the field also renounced their earlier both time and money. For example, historians enthusiasm and suggested that quantitative of the 1960s through the 1980s who wished to methods had not fulfilled their promise. Most have access to the archived data sets at notable among these critics were Lee Benson ICPSR had to order tapes and paper code- and Lawrence Stone, early enthusiasts who books which were delivered by mail. The tape had changed their minds (Benson, 1984; was then mounted on a mainframe computer, Stone, 1977, 1979). Such recantations gave to be accessed in a statistical package run in a support to the anti-quantifiers at a time when mainframe environment (with computer major new methodological challenges were usage often charged by the university in the facing historians, most notably from the same way that phones or paper were postmodernists and what came to be called charged). By the early 1990s, users could ‘the cultural turn’. Through this welter of access files using FTP (file transfer protocol), debate, quantitative practitioners continued and micro-computers on university desktops their efforts, somewhat chastened by their were providing direct access to statistical fall from the heights of fashion of earlier packages, even if those programs were some- years, but grounded sufficiently institution- times still lodged on a mainframe. By the ally and intellectually to continue to work.17 mid-1990s, desktop computing had replaced Through some twenty years of debate, nei- mainframe computing for most applications, ther side of the traditional/quantitative divide and by the early 2000s, ICPSR initiated ‘won’ their arguments. Rather, by the 1990s, ICPSR Direct, the application that permitted the debate cooled into something of an an authorized user to download data files and uneasy truce, with practitioners acknowledg- PDF code-books directly to a desktop. ing some of the points of their opponents, but agreeing to disagree on the larger validity of their enterprise.18 In practical terms, quantita- CRITIQUES OF QUANTITATIVE tive techniques did not become a routine part HISTORY of history graduate student training as they did in the social sciences, but have remained From the outset of the development of the a specialty of some historians in some grad- field of quantitative history, powerful critics uate training programs, considered more akin have challenged practitioners on their work, to language requirements for reading histori- and even challenged the usefulness of the field cal literature and texts of a non-English- itself. In the early 1960s, Carl Bridenbaugh speaking society than to a methodological devoted a portion of his 1962 American necessity for all practicing historians. This Historical Association Presidential Address compartmentalization of the skills of quan- to a condemnation of quantitative history tification for historians has in turn affected Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 258

258 QUANTIFICATION AND EXPERIMENT

the practice of quantitative historians within When quantitative history as a field was in its the larger history profession. most rapid initial development, most tradi- History as a field has maintained its roots tional historians labored much as their as a ‘humanities’ discipline and quantitative nineteenth-century predecessors had with historians’ connections to the social sciences pen, pencil, typewriter and note-card as tech- seem to many to be a betrayal of the histori- nological support. Bibliographic work cal project. The methodological ‘training entailed using library card catalogs or reading gap’ has meant that when quantitative histo- large indexed tomes of articles, books, com- rians research and write for other historians, pilations, and the like. ‘Data management’ as opposed to other social scientists, they meant developing a file of index cards, not an cannot expect their readers to appreciate or electronic spread sheet or database. even understand the technical issues Secretaries typed manuscripts for publication, involved in their work. The history profes- and though some large research institutions sion has maintained its commitment to had introduced line editors for manuscript accessible writing as well, and thus when production by the 1970s, these were writing for the broader audience of histori- machines for staff, not faculty or students. By ans, quantitative historians have had to avoid the 1980s, the situation changed. Desktop technical jargon—for example, by avoiding computers proliferated and for most histori- the use of variable names in the explication ans, word processing opened up the possibil- of a model—and be mindful to explicate ity of the electronic future. By the 1990s, their arguments clearly. email replaced typed letters. After 1995, the The critiques have also encouraged quantita- content on the internet exploded, and first tive historians to attend to the limitations in sta- bibliographical work, and then much actual tistical methodology for analyzing historical archival work, shifted to a computerized for- processes, as discussed above. Much of this mat. In short, non-quantitative historians had new work on statistical techniques for analyz- come to operate in a technological environ- ing temporal processes is still in development ment that was very similar to their quantify- and has yet to provide enough empirical work ing peers. Most recently, cheap computing to demonstrate the robust nature of the new has made multimedia evidence—visual and techniques, and hence convince non-quantita- oral, video and audio—accessible to the prac- tive historians, as well as the larger social ticing historian. One can see these develop- science community, of the need to integrate ments in particularly acute form in the explicitly temporal analysis into basic methods. developing field of historical geographic But the promise is there, and as noted below, information systems, or historical GIS. GIS, there are encouraging signs on the horizon. was until quite recently, a very expensive technology, and thus adding historical maps to geographic databases has only just begun. THE FUTURE OF THE FIELD As with the digitizing projects of the 1960s and 1970s, the payoff for the large initial The intellectual achievements of quantitative costs of first translating maps to a new history in conjunction with the larger infor- medium to become ‘data’, and then the devel- mation technology revolution makes the opment of new theory, software programs and prognosis for the future of the field better methods to make the best use of these new today than it has been for many years.19 data, are just beginning (Knowles, 2006). Almost a half-century on, one can look back More broadly, the effect of these technolog- at steady development, though not always ical changes has been to produce a conver- in a satisfyingly linear pattern.20 Perhaps gence of work of what one might call the most interesting recent development is ‘technologically enabled’ history. Traditional the impact of the information technology historians and humanists in general—for revolution on the larger practice of historians. example, in the work of Franco Moretti Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 259

QUANTITATIVE HISTORY 259

(2005)—also now work with electronic data- Abbott, Andrew and Tsay, A. (2000) ‘Sequence analysis bases, learn new computer programs to ana- and optimal matching methods in sociology’, lyze the rapidly proliferating data, and explore Sociological Methods and Research, 29: 3–33. new forms of presentation of the results of Adams, Margaret (1995) ‘Punch card records: Precursors of electronic records’, American Archivist, their analysis. Quantitative historians had to 58 (Spring): 182–201. learn the skills necessary to prepare and pre- ——— (forthcoming) ‘Analyzing archives and finding sent statistical results in print. Historians more facts: Use and users of digital data records’, Archival generally are using visual images, audio and Science. video in their presentations, not as ‘illustra- Alter, George (1988) Family and the Female Life tion’ to enhance or supplement an analysis but Course. Madison: University of Wisconsin Press. as core evidence for analysis.21 Alter, George and Gutmann, Myron (1999) ‘Casting Richard Steckel (2005) recently proposed an spells: Database concepts for event-history analysis’, agenda for what he called ‘Big Social Science Historical Methods, 32 (4): 165–76. History’, which would extend the capacities of Ambacher, Bruce I. (ed.) (2003) Thirty Years of quantitative history and translate some of its Electronic Records. Lanhan, MD: The Scarecrow Press. methods of work to non-quantitative projects.22 Annual Report of the American Historical Association Andrew Abbott (2005) has also proposed (1893). such possibilities. As with the first generation Aydelotte,William, Bogue,Allan and Fogel, Robert (eds) of quantitative history, these large agendas (1972) The Dimensions of Quantitative Research in will require collaborative efforts to manage History. Princeton: Princeton University Press. the enormously expanding data infrastructure Benson, Lee (1957) ‘Research problems in American and the myriad computer technologies political historiography’, in Mirra Komarovsky (ed.), required to make best use of the expanding Common Frontiers of the Social Sciences. Glencoe, corpus of digitized historical evidence, and to IL: Free Press, pp. 113–83, 418–21. develop appropriate theoretical approaches to ——— (1961) The Concept of Jacksonian Democracy: such historical work. New York as a Test Case Princeton: Princeton University Press. ——— (1984) ‘The mistransference fallacy in explana- tions of human behavior’, Historical Methods, 17 (3): ACKNOWLEDGEMENTS 118–31. Bogue, Allan G. (1983) Clio and the Bitch Goddess: * This essay has been improved considerably Quantification in American Political History. Beverly by comments from colleagues, particularly Hills: Sage Publications. ——— (1986) ‘Systematic revisionism and a genera- Peggy Adams, Erik Austin, Morgan Kousser, tion of ferment in American history’, Journal of Jim Oberly, Lex Renda, Jack Reynolds, Contemporary History, 21 (2): 135–62. Steve Ruggles, Carole Shammas and Dan ——— (1990) ‘The quest for numeracy: Data and Scott Smith, The overall interpretation and methods in American political history’, Journal of remaining errors are all mine. Interdisciplinary History, 21 (1): 89–116. Bourke, Paul, DeBats, Donald and Phelan, Thomas (2001) ‘Comparing individual-level voting returns with aggregates: A historical appraisal of the King REFERENCES AND SELECT solution’, Historical Methods, 34 (3): 127–34. BIBLIOGRAPHY Bridenbaugh, Carl (1962) ‘The Great Mutation’, AHA Presidential Address. Available at: http://www.histo- Abbott, Andrew (2001) Time Matters: On History and rians.org/info/AHA_History/cbridenbaugh.htm. Method. Chicago: University of Chicago Press. (Accessed on May 15 2006.) ——— (2005) ‘Looking backward and looking for- Burnham, Walter Dean (1955) Presidential Ballots, ward, social science history at 2000’ in Harvey Graff, 1836–1892. Baltimore: Johns Hopkins University Leslie Page Moch and Philip McMichael with Julia Press. Woesthoff (eds), Looking Backward and Looking ——— (1970) Critical Elections and the Forward: Perpspectives on Social Science History. Mainsprings of American Politics. New York: Madison: University of Wisconsin Press, pp. 69–72. W.W. Norton & Co. Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 260

260 QUANTIFICATION AND EXPERIMENT

Burton, Orville Vernon (2002) Computing in the Social ——— (2003) The Slavery Debates: A Retrospective. Sciences and Humanities. Urbana: University of Baton Rouge: Louisiana State University Press. Illinois Press. ——— (2004) The Escape from Hunger and Cameron, Sonja and Richardson, Sara (2005) Using Premature Death, 1700–2100: Europe, America, Computers in History. New York: Palgrave and the Third World. New York: Cambridge Macmillan. University Press. Carter, Susan, Gartner, Scott, Haines, Michael R., Olmstead, Fogel, Robert and Costa, Dora L. (1997) ‘A theory of Alan, Sutch, Richard and Wright, Gavin (2006) Historical technophysio-evolution, with some implications for Statistics of the United States: Millennial Edition. New forecasting population, health care costs, and pen- York: Cambridge University Press. sion costs’, Demography, 34 (1): 49–66. Chambers,William Nisbet and Burnham,Walter Dean (eds) Fogel, Robert William and Elton, G.R. (1983) Which (1967) The American Party Systems: Stages of Political Road to the Past? New Haven: Yale University Press. Development. New York: Oxford University Press. Fogel, Robert and Engerman, Stanley (1974) Time on Chopwhittle, Darcy and Taleglad, Lars Mooson (2001) the Cross, Vols. 1 & 2. Boston: Little Brown. ‘Review of Philinda Blank’s When the Cows Come Goldin, Claudia (1997) ‘Exploring the “Present Through Home: Barn Architecture and Changes in Bovine the Past”: Career and family across the last century’, Public Space,’ Social Science History, 25 (5): The American Economic Review, 87 (2): 396–9. 609–14. (See note 19.) Graff, Harvey, Moch, Leslie Page and McMichael, Philip Conrad, Alfred and Meyer, John (1958) ‘The economics with Woesthoff, Julia (eds) (2005) Looking Backward of slavery in the ante-bellum South’, Journal of and Looking Forward: Perpspectives on Social Science Political Economy, 66 (2): 95–130. History. Madison: University of Wisconsin Press. Curti, Merle, Curti, Margaret Wooster, Daniel, Robert, Greif, Avner (1997) ‘Cliometrics after 40 Years’, The Livermore Jr., Shaw, and Van Hise, Joseph (1959) The American Economic Review, 87 (2): 400–3. Making of an American Community: A Case Study of Griffin, Larry (1993) ‘Narrative, event-structure analysis, Democracy in a Frontier County. Stanford: Stanford and causal interpretation in historical sociology’, University Press. American Journal of Sociology 98: 1094–133. Darcy, R. and Rohrs, Richard C. (1995) A Guide to Griffin, Larry and Isaac, Larry W. (1992) ‘Recursive Quantitative History. Westport, CT: Praeger. regression and the historical use of “time” in time- Dollar, Charles M. and Jensen, Richard J. (1971) series analysis of historical process’, Historical Historian’s Guide to Statistics: Quantitative Analysis Methods, 25: 166–79. and Historical Research. New York: Holt, Rinehart Griffin, Larry J., Clark, P. and Sandberg, J. (1997) and Winston, Inc. ‘Narrative and event: Historical sociology and lynch- Feinstein, Charles and Thomas, Mark (2002) Making ing’, in W. Fitzhugh Brundage (ed.), Under Sentence History Count: A Primer in Quantitative Methods for of Death: Lynching in the New South. Chapel Hill: Historians. New York: Cambridge University Press. University of North Carolina Press, pp. 24–47. Fishbein, Meyer H. (ed.) (1973) The National Archives Gutmann, Myron and Alter, George (1993) ‘Family and Statistical Research. Athens, OH: Ohio University reconstitution as event history analysis’, in David Press. Sven Reher and Roger S. Schofield (eds), Old and Fitch, Nancy (1986) ‘Statistical fantasies and historical New Methods in Historical Demography. Oxford: facts: History in crisis and its methodological impli- Oxford University Press, pp. 159–77. cations’, Historical Methods, 18 (4): 239–54. Hacker, David and Fitch, Catherine (eds) (2003a) Floud, Roderick (1972) An Introduction to Quantitative ‘Building Historical Data Infrastructure: New Projects Methods for Historians. Princeton: Princeton of the Minnesota Population Center’, Historical University Press. Methods 36 (1). Floud, Roderick, Wachter, Kenneth and Gregory, ——— (eds) (2003b) ‘Building Historical Data Annabel (eds) (1990) Height, Health and History: Infrastructure: New Projects of the Minnesota Nutritional Status in the United Kingdom, Population Center’, Historical Methods 36 (2). 1750–1980. New York: Cambridge University Press. Harvey, Charles and Press, Jon (1996) Databases in Fogel, Robert W. (1964) Railroads and American Historical Research: Theory, Methods and Economic Growth: Essays in Econometric History. Applications. London: Macmillan Press Ltd. Baltimore: Johns Hopkins University Press. Haskins, Loren and Jeffrey, Kirk (1990) Understanding ——— (1993) Nobel Prize in Economics Autobiographical Quantitative History. New York: McGraw-Hill. Statement. Available at: http://www.nobelprize.org/ Heckman, James J. (1997) ‘The value of quantitative economics/laureates/1993/fogel-autobio.html. evidence on the effect of the past on the present’, (Accessed on May 15, 2006.) The American Economic Review, 87 (2): 404–8. Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 261

QUANTITATIVE HISTORY 261

Hudson, Pat (2000) History by Numbers: An Meyer, John R. (1997) ‘Notes on cliometrics’ fortieth’, Introduction to Quantitative Approaches. New York: The American Economic Review, 87 (2): 409–11. Oxford University Press. Monkkonen, Eric (2001) Murder in New York City. Isaac, Larry W. and Griffin, Larry (1989) ‘Ahistoricism Berkeley: University of California. in time-series analyses of historical process: Neustadt, Richard E. and May, Ernest R. (1986) Critique, redirection, and illustrations from US Thinking in Time: The Uses of History for Decision labor history’, American Sociological Review, Makers. New York: Free Press. 54: 873–90. Moretti, Franco (2005) Graphs, Maps, Trees: Abstract Jarausch, Konrad and Hardy, Kenneth (1991) Models for a Literary History. New York: Verso. Quantitative Methods for Historians: A Guide to North, Douglass C. (1997) ‘Cliometrics – 40 years later’, Research, Data, and Statistics. Chapel Hill: University The American Economic Review, 87 (2): 412–14. of North Carolina Press. Palmquist, Bradley (2001) ‘Unlocking the aggregate King, Gary (1997) A Solution to the Ecological Inference data past – Which keys fit?’ Historical Methods, 34 Problem: Reconstructing Individual Behavior from (4): 159–69. Aggregate Data. Princeton, NJ.: Princeton University Redding, Kent and James, David R. (2001) ‘Estimating Press. levels and modeling determinants of Black and Kousser, J. Morgan (1973) ‘Ecological regression and White voter turnout in the South, 1880 to 1912’, the analysis of past politics’, The Journal of Historical Methods 34 (4): 141–58. Interdisciplinary History, 4 (2): 237–62. Reiff, Janice (1991) Structuring the Past: The Use of ———(1974) The Shaping of Southern Politics: Computers in History. Washington, DC: American Suffrage Restriction and the Establishment of the Historical Association. One-party South, 1880–1910. New Haven: Yale Reynolds, John F. (1998) ‘Do historians count anymore’, University Press. Historical Methods, 31 (4): 141–8. ——— (1984) ‘The revivalism of narrative: A response Rowney, Don Karl and Graham, James Q. Jr. to recent criticisms of quantitative history’, Social (eds) (1969) Quantitative History: Selected Readings Science History, 8 (2): 133–49. in the Quantitative Analysis of Historical Data. ——— (1986) ‘Must historians regress?’ Historical Homewood, Ill.: Dorsey Press. Methods 19 (2): 62–81. Shreibman, Susan, Siemens, Ray and Unsworth, John ———(1989) ‘The state of social science history in the (2004) A Companion to Digital Humanities. Malden, late 1980s’, Historical Methods, 22: 13–20. MA: Blackwell Publishing. ———(2001a) ‘Evaluating ecological inference: An Shammas, Carole (1990) The Pre-industrial Consumer introduction’, Historical Methods, 34 (3): 100. in England and America. Oxford: Clarendon Press. ——— (2001b) ‘Ecological inference from Goodman to Shammas, Carole, Salmon, Marylynn and Dahlin, Michel King’, Historical Methods, 34 (3): 101–26. (1987) Inheritance in America from Colonial Times Knowles,Anne Kelly (2006) ‘GIS and history’, Paper pre- to the Present. New Brunswick, NJ: Rutgers sented at the Annual Meeting of the American University Press. Historical Association, Philadelphia. Shorter, Edward (1971) The Historian and the Lemisch, Jesse (1967) ‘The American Revolution seen Computer: A Practical Guide. Englewood Cliffs, NJ: from the bottom up’, in Barton Bernstein (ed.), Prentiss Hall. Towards a New Past: Dissenting Essays in American Silbey, Joel, Bogue, Allan and Flanigan, William (eds) History. New York: Vintage Books, pp. 3–45. The History of American Electoral Behavior. Lewis, Jeffrey B. (2001) ‘Understanding King’s ecologi- Princeton: Princeton University Press. cal inference model: A method-of-moments Smith, Dan Scott (1984) ‘A mean and random past: The approach’, Historical Methods 34 (4): 170–88. implications of variance for history’, Historical Lorwin, Val and Price, Jacob (1972) The Dimensions of Methods, 17: 14– 48. the Past: Materials, Problems, and Opportunities for ——— (1992) ‘Context, time, history’, in Peter Karsten Quantitative Work in History. New Haven: Yale and John Modell (eds), Theory, Method, and Practice University Press. in Social and Cultural History. New York: New York McCormick, Richard P. (1966) The Second American University Press, pp. 13–32. Party System; Party Formation in the Jacksonian Era. Steckel, Richard (2005) ‘Presidential Address’, Annual Chapel Hill: University of North Carolina Press. Meeting of the Social Science History Association, McDonald, Terrence (1986) The Parameters of Urban Portland, Oregon. Fiscal Policy: Socioeconomic Change and Political Steckel, Richard H. and Floud, Roderick (eds) (1997) Culture in San Francisco, 1860–1906. Berkeley: Health and Welfare During Industrialization. University of California Press. Chicago, IL: University of Chicago Press. Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 262

262 QUANTIFICATION AND EXPERIMENT

Steckel, Richard and Rose, Jerome (eds) (2002) The 11 Nattional Archives and Records Administration, Backbone of History: Health and Nutrition in the ‘Inside the National Archives – Southeast Region, Western Hemisphere. New York: Cambridge 1825-1863 Slave Sale Documents’. Available at: University Press. http://www.archives.gov/southeast/exhibit/2.php. Transcription of Slave Sale Document in Figure 14.3 Stone, Lawrence (1977) ‘History and the social sciences Know all men by these presents, That I, Albert G. in the twentieth century’, in C. Delzell (ed.), The Ewing, of the county of Davidson and state of Future of History. Nashville: Vanderbilt University Tennessee have this day for and in consideration of Press. pp. 3–42. five hundred dollars, to me in hand paid by Joseph ——— (1979) ‘The revival of narrative: Reflections on Woods and John Stacker, Trustees for Samuel a new old history’, Past and Present, 89: 3–24. Vanleer, his wife and chldren, under the will of Swierenga, Robert (ed.) (1970) Quantification in Bernard Vanleer, now recorded in the office of the American History: Theory and Research. New York: Davidson county court, state of Tennessee, bargained Atheneum. and sold unto said Trustees, a certain negro boy Tolnay, Stewart E. and Beck, E.M. (1995) A Festival of name George aged about seventeen years; which said slave I warrant to be sound and healthy; and I Violence: An Analysis of the Lynching of African- also will warrant the right and title of said slave, unto Americans in the American South, 1882–1930. said Trustees, their heirs, executors, &c. &c. and that Urbana: University of Illinois Press. said negro boy George is a slave for life. Turner, Frederick Jackson (1893) ‘The Significance of the Witness my hand and seal, this Sixth day of Frontier in American History’, Annual Report of the November 1833. American Historical Association: 199–227. A.G. Ewing Whaples, Robert (1991) ‘A quantitative history of the Journal of Economic History and the cliometric revo- Frederick Bradford Orville Erving lution’, Journal of Economic History, 51 (2): 289–301. Nov. 6. 1833.

12 For other examples of slave sale documents, see the Slave Documents Collection from the Enoch NOTES Pratt Free Library, Baltimore, Maryland, available at http://www.pratt.lib.md.us/exhibits/slavery/ 1 See, for example, Benson (1957; 1961); 13 The data set and code-book are available at: Burnham (1970); Chambers and Burnham (1967); http://www.icpsr.umich.edu/cocoon/ICPSR- Richard P. McCormick (1966). STUDY/07423.xml. 2 For information on the Cambridge Group, see 14 See, for example, Kousser (1973, 1974). For their website, http://www-hpss.geog.cam.ac.uk. recent methodological developments in the field and 3 See also Kousser (1989) and Reynolds (1998). their impact in history, see King (1997), and the arti- 4 See for example, Darcy and Rohrs (1995); Dollar cles in the Summer and Fall 2001 (34 (3 & 4)) issues and Jensen (1971); Feinstein and Thomas (2002); of Historical Methods on the time period by: Kousser Floud (1972); Haskins and Jeffrey (1990); Hudson (2001a, 2001b); Bourke et al. (2001); Redding and (2000); Jarausch and Hardy (1991); Shorter (1971). James (2001); Palmquist (2001); and Lewis (2001). 5 See, for example, Aydelotte et al. (1972); Lorwin 15 See Abbott (2001); Abbott and Tsay (2000); and Price (1972); Rowney and Graham (1969); Silbey Alter and Gutmann (1999); Alter (1988), Gutmann et al. (1978); Swierenga (1970). and Alter (1993); Griffin (1993); Griffin and Isaac 6 ICPSR, founded in 1962 as ICPR, changed its (1992); Isaac and Griffin (1989); Reher and Schofield name to the Inter-university Consortium for Political (1993). On time series, see also McDonald (1986). and Social Research (ICPSR) in 1975. On quantification and historical explanation, see 7 See, for example, Floud et al. (1990); Monkkonen Smith (1984; 1992). (2001); Shammas et al. (1987); Shammas (1990); 16 For the results of this research, see Fogel Steckel and Floud (1997); Steckel and Rose (2002). (2004); Fogel and Costa (1997). 8 The schedule in Table 14.1 is available on the 17 For discussion of Benson’s change in position IPUMS website at http://www.ipums.umn.edu/usa/ and critiques of the change, see Bogue (1986; 1990) voliii/form1950.html. and Kousser (1986). See also Fogel and Elton (1983); 9 The United States maintains census schedules as Kousser (1984); Fitch (1984); and Fogel (2003). confidential records for 72 years. The 1890 Census 18 For a hilarious parody of the issues involved, manuscript schedules were destroyed by fire in 1921. see the Winter 2001 issue of Social Science History. 10 See the special issues of Historical Methods Outgoing editors Paula Baker and Elizabeth Faue (Hacker and Fitch 2003a; 2003b) on ‘Building published reviews by Darcy Chopwhittle and Lars Historical Data Infrastructure: New Projects of the Mooson Taleglad of Philinda Blank’s (2001) When the Minnesota Population Center’ and the website of Cows Come Home: Barn Architecture and Changes IPUMS at www.ipums.org for details. in Bovine Public Space (2001). Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 263

QUANTITATIVE HISTORY 263

(The reviewed book does not exist, though per- ‘Cliometrics After 40 Years’. Papers in this section haps it might. Many people contributed to the include Goldin (1997); Greif (1997); Heckman review; Paula Baker and Elizabeth Faue take respon- (1997); Meyer (1997); and North (1997). See also sibility for it.) Whaples (1991). 19 On anthropometric history, for example, see 21 See, for example, Burton (2002); Cameron and the Summer 2004 Special Issue of Social Science Richardson (2005); Harvey and Press (1996); Reiff History, Volume 28, no. 2, guest edited by John (1991); Shreibman et al. (2004). Komlos and Jorg Baten. For the impact of the IPUMS 22 Steckel listed the large data projects social project, see the bibliography of work listed on the science historians have produced in the last genera- IPUMS website, http://www.ipums.org. For recent tion and then added his own wish list: including an evaluations of ‘social science history’ as a field, see inventory all archeological sites; an inventory all arti- Graff et al. (2005). facts at these sites; a database on natural disasters 20 For retrospectives on quantitative history, see and human history; and an international catalogue Reynolds (1998). For retrospective analysis of ‘clio- of films and photos. He called for extending the dig- metrics’, see the special section of the ‘Papers and itization of all extant manuscript censuses in the Proceedings of the Hundred and Fourth Annual past; a digitized and annotated collection of diaries; Meeting of the American Economic Association’ in voting records at the precinct level; and probate The American Economic Review (1997), 87 (2), on records. Outwaite-3587-14.qxd 6/12/2007 5:32 PM Page 264