Navigation – Plan du site

AccueilNuméros45DossierRediscovering the Early History o...

Dossier

Rediscovering the Early History of Social Scientific Data Centers in the US

Redécouvrir les débuts de l’histoire des centres de données sur les sciences sociales aux États-Unis
Emily Hauptmann
p. 43-67

Résumés

Les centres de recueil de données en sciences sociales sont depuis longtemps essentiels à la réalisation de la recherche dans le domaine aux États-Unis. Pourtant, il y a encore eu peu d’enquêtes historiques sur leur origine et le début de leur trajectoire institutionnelle. On s’appuie ici sur une série de travaux publiés par des chercheurs en sciences sociales qui ont participé à l’essor des centres de données, sur les archives institutionnelles non publiées des deux premiers centres de données en sciences sociales fondés dans l’après-guerre, le Roper Center for Public Opinion Research et l’Inter-University Consortium of Political and Social Research (ICPSR), ainsi que sur des évocations plus récentes de cette période. Ces diverses sources permettent de retracer les aspirations intellectuelles et professionnelles qui ont conduit à la création de tels centres, mais aussi les obstacles majeurs qu’ils ont dû affronter. L’article s’achève par une réflexion sur ce en quoi la connaissance et la recherche actuelles en sciences sociales sont tributaires du genre d’hégémonie acquise par les centres de données de l’époque.

Haut de page

Texte intégral

Introduction

  • 1 I use the term “data center” for institutions that acquired data sets from multiple sources, integr (...)

1Social scientific data centers, now so firmly established in the US, grew out of small repositories struggling to be noticed and needed. By the mid-1960s, however, many new repositories for social scientific data were springing up in a climate of great anticipation and excitement. In what seems to foreshadow the recent academic buzz about “big data,” people who devoted their professional lives to building what became the first data centers predicted they would fundamentally change how social science was done.1 And they were right.

  • 2 I came across a similar analogy drawn by De Chadarevian and Porter (2018, 549) who compare investig (...)
  • 3 Converse, 1964, 274. Changing “the social organization of research” was a frequently mentioned aim (...)

2For as important to contemporary social science as data centers have turned about to be, however, there has been little historical investigation of their emergence. Though some accounts of the increased emphasis on quantitative proficiency in the social sciences during the Cold War mention them, for most social scientists today, data centers, like the power plants and servers that make home and workplace computing possible, are as essential as their institutional histories are invisible.2 By investigating the climate in which the first social scientific data centers emerged in the US, this paper aims to make this history visible again. In doing so, it also illuminates how these institutions began to change “the social organization of the research community.”3

  • 4 I focus throughout this paper on political science, the discipline whose history I have studied the (...)
  • 5 On private philanthropic foundation funding for the behavioral sciences, see Hauptmann (2022) and S (...)

3As one of the most durable but understudied social scientific institutions to emerge during the early postwar period, I believe the data center was crucial to establishing and then solidifying the primacy of quantitative analytic methods in political science and many other social sciences.4 Though accounts of how collecting and analyzing quantified data became dominant in postwar political science, sociology, anthropology, economics, and psychology differ in timing, degree, etc., some common themes run through them. First, most of the awards made by private philanthropic foundations and later by the National Science Foundation (NSF) supported social scientists who sought to collect and analyze quantified data. This was true of the Ford Foundation’s short-lived but munificent Behavioral Sciences Program in the 1950s, as well as smaller but crucial Carnegie Corporation and Rockefeller Foundation grants. Then, from the early 1960s on, the longer-term programs of the NSF took the lead in building institutions devoted to the collection and analysis of quantified social scientific data.5 Second, new organizations designed to produce large-scale externally funded research during and after World War II formed at many universities. Like their colleagues in the “hard sciences,” social scientists began doing full-time, capital-intensive team research in these new institutes or centers loosely modeled on the newly prestigious university weapons labs. Third, some of the first university-based social science researchers learned about collecting, storing, and analyzing quantified data working for federal agencies or private corporations. Their work there also taught them that such research depended on a sizeable staffs to provide technical support and computing services. The overall shape of the first social scientific data centers grew out of the confluence of these factors.

  • 6 Bouk, 2018; Igo, 2018.
  • 7 Lepore (2020) documents the Simulmatics Corporation’s attempts to use computer-aided analyses of su (...)

4Booms are accompanied and sustained by widespread enthusiasm. There were plenty of people who became enthusiastic about creating social scientific data centers in the postwar US, including a multi-disciplinary contingent of anthropologists, demographers, economists, political scientists, sociologists, social psychologists, and scholars of communications. Intense as their data hunger was, however, academics were far from the only group to feel it. Nor were they the first. A number of private corporations and government agencies had been collecting would-be data, making it machine-readable, and analyzing it for decades.6 Data hunger inside and beyond the academy became even keener in the 1950s and 1960s.7

  • 8 Geiger, 1993.

5Much of my recent work explores the relation between the broader intellectual, social, political, and economic features of the mid-twentieth century US and the academic knowledge cultures and institutions that emerged and flourished then. Set within this frame, the enthusiasm for social scientific data centers reads as one specific manifestation of the massive expansion of the research missions of US universities in the postwar period. As the research missions of universities expanded, the meaning and practice of research also changed to favor teams of researchers supported by extramural funds who worked at sites outside the older departments and colleges.8 This paper focuses on how these reconfigurations of research were enabled by the postwar rise of social scientific data centers. Once established, data centers exercised an enduring influence on the shape of social scientific research by means of the multiple educational and technical services they began to offer and the networks of researchers they built and sustained.

  • 9 The Roper Center began operations in 1957, the Inter-University Consortium for Political Research ( (...)

6To recover the history of the first social scientific data centers, I rely on a range of published works by social scientists who participated in the data center boom, unpublished institutional records of the two earliest social scientific data centers founded in the postwar period, the Roper Center for Public Opinion Research and the Inter-University Consortium of Political and Social Research (ICPSR), as well as more recent historical accounts of this period.9 These sources all speak to the intellectual and professional aspirations that drove the creation of social scientific data centers as well as the most fundamental challenges they faced.

A Pre-History of the Social Scientific Data Center Boom

  • 10 Lemov, 2015.
  • 11 Bouk, 2018; Igo, 2018, 221-263.
  • 12 This was true of Rensis Likert and Angus Campbell of Michigan’s Institute for Social Research. Like (...)

7As much as what counts as “data” seems self-evident to distinct groups of professionals today, in the mid-twentieth century, questions about what to collect were hard to separate from questions about how such collections should be stored, shared and promoted as sources for research and analysis. Though there were some indefatigable researchers who built and organized their own collections, like the anthropologist Bert Kaplan who created a “database of dreams” stored on Microcards,10 most took the more practical approach of adapting or extending already extant collections of corporate and government data. Before social scientists had either the means or the interest to do so, federal agencies like the Bureau of Labor Statistics, Census Bureau, Internal Revenue Service, and Social Security Administration had been amassing and analyzing huge quantities of data on US citizens for decades. Private corporations, including insurance companies, financial institutions, and market research firms had been doing so as well.11 Some social scientists who would later be considered academic pioneers of data collection and analysis first honed these skills working for government agencies or private corporations.12

  • 13 The kinds of materials made into “data” differed by discipline. While surveys and polls were centra (...)
  • 14 Converse, 1987, 180-182.
  • 15 Converse (1987) discusses what she terms survey researchers’ “migration to the universities” in Par (...)
  • 16 In a letter Baxter sent to a number of potential donors to the Roper Center in 1959, he added, “Thi (...)
  • 17 Hauptmann (2016, 2022) discusses 1950s private philanthropic support that made these surveys possib (...)

8Surveys and polls were just one particular type of data produced, stored, and analyzed by private firms and the US government.13 Though relatively new, private polling firms like Crossley, Gallup, Harris, and Roper were increasingly familiar and prominent in the mid-century US. During World War II, the US Department of Agriculture and the War Department had also set up small survey research units of their own. But by 1946, these government-sponsored research centers had wound down and closed up.14 After World War II, public opinion surveys continued to be done either by private firms or a handful of groups occupying what was initially a marginal niche on college and university campuses.15 That survey data became as important as it did to postwar US social science, was, in part, a fortuitous result of how this data became available to academics. The Roper Center grew out of Elmo Roper’s 1947 donation of his firm’s surveys to Williams College, where Roper’s war-time OSS colleague, James Baxter, was president.16 ICPR grew out of the voter surveys that Michigan’s Survey Research Center (SRC) had done since 1948. Both of these would-be data centers began with significant bodies of data that were not widely available to others.17 That is, data derived from surveys and polls became important to US social science not because of any pre-existing, widespread conviction about its merits but because of the resources available to those who collected and analyzed it.

  • 18 Hastings, 1957; Hastings, 1963, 592.
  • 19 The Roper Center, now housed at Cornell University, provides an extensive account of its history on (...)

9Should the survey and polling data collected and produced by these private firms and academic centers be stored and maintained? Should it be made available and distributed to others? And could social scientists not currently involved in these centers be persuaded to use and add to them? These were live questions for the supporters of the first academic social science data repositories. The Roper Center’s close connection to private polling firms made it possible for it to store survey data produced not only by the many offices of the Roper firm but also to receive data produced by Gallup and many other private firms.18 ICPR grew out of the University of Michigan’s Survey Research Center (SRC), which did contract work for government agencies and private corporations and received significant research grants from private philanthropies.19 Both centers are still alive and well today.

  • 20 For instance, Converse (1964) uses “data archive,” “data library,” and “data bank,” without drawing (...)

10Social science data centers like these have become so integral to the contemporary infrastructure of research that understanding the imaginative leaps their earliest supporters had to make and the obstacles they faced requires patient historical excavation. One part of this imaginative work was a struggle to name what they were trying to create. Groping for the right model, they asked: do we want to build something like a library? a bank? a treasury? an archive? Or does it need to be something that isn’t fully like any of these? The earliest plans for what became social scientific data centers in the 1950s and early 1960s were shot through with trying on and setting aside each of these models in rapid succession.20

  • 21 Warren Miller would later say that it was this Lucci and Rokkan report that inspired him to build I (...)
  • 22 See the discussion of the “Farmington plan” of inter-library lending in Converse (1963a, 8-9).

11One of the earliest proposals to build such an institution, written under the auspices of a Ford Foundation grant to Columbia University’s School of Library Services, oscillated between calling what was being proposed a “library” and an “archive.” The first of the two main authors of the proposal, Columbia’s York Lucci, opted for the library model, calling what he envisioned a library of survey research data, given the library-like challenges he imagined these institutions would face.21 Not only would they have to decide who could use their materials and how, but they would also have to work out an analogue to an interlibrary loan system for dealing with one another.22 And they would have to develop policies for adding to and managing their collections. As I discuss below, in the rush to build viable data centers that began in the 1960s, their architects juggled making cooperative plans while also trying to promote the distinctive virtues of their collection and sharing practices.

  • 23 Lucci and Rockan, 1957, 118.
  • 24 Loughnane and Aspray, 2018, 220-221, 228-229.
  • 25 Converse, 1962, 4.
  • 26 Internal references suggest this undated piece was written in the 1970s.

12The “library” model, however, did not seem apt to everyone. For instance, unlike his co-author Lucci, Stein Rokkan favored the term “archive” for what he had in mind. Like archives, he argued these entities would also have to acquire, maintain, and conserve material that was not available elsewhere—and, ideally, provide “services” to the researchers who used them.23 “Archive” was probably even more widely used than “library” during this early planning period. Ralph Bisco, ICPR’s technical services director, used it well into the late 1960s; it is still in use today.24 Still others ventured that these institutions would most resemble banks in that they should try to increase the value of the data deposited in them by loaning it out for secondary analyses and then capitalizing on the new questions and conclusions it yielded in others’ hands. Along these lines, when ICPR was just beginning its operations, Philip Converse imagined it becoming an “information treasury” that would grow as research results “feed back into25” it. Somewhat later, ICPR’s Richard Hofferbert (n.d.) expressed the hope that continued sharing and dissemination of data would lead to “a significant multiplication of return on… investments” made both by those who created it and also by those who archived and distributed it.26

  • 27 Lucci and Rokkan, 1957, 48-82.
  • 28 Lucci and Rockan, 1957, 51-54. Well into the 1960s, making arguments for the merits of secondary an (...)

13The tensions between these different models notwithstanding, the success of any one of them hinged on persuading social scientists that research based on secondary analyses of data was not only legitimate but potentially important.27 Doing that persuading was another element of the imaginative work the architects of the social science data center undertook. Secondary analyses, or using existing bodies of data to answer new questions, have become so well-established in contemporary social scientific practice that mid-twentieth-century arguments for them seem puzzling, like pushing vigorously at an open door. Yet in 1957, Lucci recognized that the success of his proposal for a “library” of survey data hinged on dispelling what were then widespread doubts about the scholarly legitimacy of such analyses. To counter such doubts, Lucci argued that secondary analyses of survey data could spark new lines of research, train students in data analysis, and make social scientific knowledge more cumulative.28 Still, he acknowledged, there was no way to be sure social scientists would indeed make use of such resources. They would have to be built, made accessible, and publicized first.

  • 29 Truman, 1991, 143-145. When Elmo Roper’s 1943 donation of a number of studies done by his firm to W (...)
  • 30 Political scientist V.O. Key wrote that if “younger men in political science” would draw on Roper’s (...)
  • 31 Baxter, 1960. Baxter’s full pitch read as follows: “As an historian I know something about how valu (...)
  • 32 Glenn, 1973 cited in Loughnane and Aspray, 2018, 216-217.

14These problems were keenly felt by people connected to what became the Roper Center. For instance, the political scientist David Truman, the first curator of what was still the Roper Collection, noted that few social scientists at Williams or anywhere else showed much interest in this collection of polls in the late 1940s.29 Once the Roper Center began operations in 1957, it sought to get social scientists’ attention by publicizing endorsements from well-known social scientists vouching for the potential importance of research using Roper’s holdings.30 Convincing social scientists to use Roper’s resources, however, proved a persistent problem. When Williams’ president, James Baxter, made pitches for financial support to foundations and corporations in 1960, he stressed how little the Roper materials had been used: “only the surface of this treasure house has been scratched.”31 Even in the early 1970s, not much had changed. According to the sociologist Norval Glenn, only a handful of sociologists were undertaking secondary analyses of the Roper Center’s data; most continued to rely on “original data gathered at great expense”32 instead.

  • 33 Hauptmann 2020, 110. “Willy Loman” is the harried travelling salesman of Arthur Miller’s 1949 play, (...)
  • 34 Scheuch, 2003, 393.

15Those involved in creating ICPR were also keenly aware of this problem, especially Michigan’s Warren Miller, who traveled to campuses all over the US, laying the groundwork for “the consortium” of colleges and universities that would become the basis of ICPR. Had secondary analyses of survey data already been well-established research practices, Miller would not have had to drum up demand for the data and data services ICPR proposed to offer. However, since this was not the case, Miller later said he became the “Willy Loman of political science,” a traveling salesman trying to persuade his fellow social scientists to pay for something they didn’t already think they needed.33 Miller’s recollections were confirmed by Erwin Scheuch, the developer of one of the first data archives in Germany, who said in retrospect that getting mid-twentieth-century social scientists to take secondary analyses seriously was partially a matter of “public relations.”34 Regardless of whether repositories holding social scientific data should be most like libraries, banks, treasuries, or archives, those who advocated for them understood that their long-term survival depended on making potential patrons aware of their existence and convincing them that using their holdings could open up new research possibilities.

The Social Science Data Center Boom Begins

  • 35 Larsen, 1992.
  • 36 Solovey, 2020.
  • 37 Solovey, relying on figures from Larsen (1992) concludes that NSF never allocated more than 6% of i (...)

16The Roper Center and ICPR’s ambitions to become widely known and widely used data centers required significant resources. Each had received grants or gifts that had gotten them started; these, however, fell far short of what they needed to keep them going. Nor were the fees each charged its members enough to sustain them. Crucially, multiple 1960s grants from the NSF allowed both to continue and expand their operations. Even though, as both Otto Larsen35 and Mark Solovey36 have shown, the NSF made only a small fraction of its grants to the social sciences, its patronage was nevertheless vital.37 Without NSF support, social science data centers might not have survived into the 1970s; if they had, they certainly would have been much weaker. Funding data archives and data centers was one of the first major commitments the NSF made that affected the shape of US social science.

  • 38 Converse, 1963c.
  • 39 The US Bureau of Labor Statistics and the Bureau of the Census (1966, 99-100).
  • 40 Bisco, 1966.
  • 41 Current values calculated using the US Bureau of Labor Statistics Inflation Calculator (https://www (...)
  • 42 Bisco, 1966, 97, 102, 104-106.

17In 1963, Philip Converse of Michigan’s Survey Research Center and ICPR reported that an NSF official said he was “getting inundated by specialized repository requests.”38 Though it is not possible to know how many such requests it received and declined, NSF did make a considerable number of grants to social science data repositories in the 1960s. When ICPR’s Bisco published an annotated list of data archives of interest to political scientists in 1966, all the archives he listed in the US (except the two US government agencies39) were connected to colleges or universities—and all were new.40 For many of these new institutions, NSF support was crucial. The Roper Center received just under $260,000 from the NSF from 1962 to 1967; ICPR received just under $1 million from 1963 to 1969; Roper’s awards are equivalent to roughly $2.4 million in 2024; ICPR’s to $8.3 million.41 Besides Roper and ICPR, NSF supported other data archives as well. It also underwrote the creation of the Council on Social Science Data Archives (CSSDA) to coordinate the activities of all these new institutions.42

  • 43 Aronova, 2017, 307-308, 313.
  • 44 For a list of all thirteen of the disciplines involved in the International Geophysical Year, see A (...)
  • 45 Aronova, 2017, 318-319. “Deluge” in the sub-heading on p. 316; “accumulations of data” in the quota (...)
  • 46 Shapley, quoted in Aronova, 2017, 308.

18Exciting and ambitious as these developments no doubt seemed to social scientists, they were small compared to the creation of massive World Data Centers during the International Geophysical Year (1957-1958). This international data collection, sharing, and dissemination project involved thousands of scientists working in all branches of geophysics and the creation of at least fifteen data centers.43 Storing and disseminating the “deluge” of data from meteorologists, seismologists, oceanographers, and others that began flowing into these World Data Centers, however, continued to pose challenges well into the 1960s.44 Scientists closest to this effort in the US asked for a concerted effort on the part of “all major federal scientific agencies” to figure out how to make these vast and increasing “accumulations of data” most usable.45 There were no blueprints for how to manage this “big data collecting binge.”46

  • 47 Kraus, 2011, 4-8; Miller, 1971, 54-67.

19Nor was the data archive creation boom confined to the academic world. In 1965, a prestigious committee of academics and government officials recommended establishing a National Data Center that would pool, standardize, and then distribute data collected by different federal government agencies.47 Fast-paced and broad in scope, the expanding data center boom was the steady backbeat of scientific research in the 1960s.

  • 48 Garfinkel, 2000, 17-18.
  • 49 Garfinkel goes on to explain that Herman Hollerith, the nineteenth century inventor of the punched (...)
  • 50 Aronova notes that most of the physical scientists who contributed data to the World Data Centers n (...)
  • 51 Lucci, 1957, 14-17. Lucci itemizes the neglect and destruction of data stored on punched cards in a (...)

20Advances in computing technology no doubt fueled this boom. The US Census Bureau had begun to experiment with storing and analyzing data on punched cards in the late nineteenth century,48 but in the academic world, this approach was still a relative novelty in the 1940s and 50s.49 Even those academics who stored their data on punched cards did not necessarily keep them or assume they would be of any value to others. The cards were bulky but also fragile, prone to being torn by the machines that read them—machines to which many social scientists did not have reliable access in the first place.50 And, as discussed above, for at least the first fifteen years after World War II, few social scientists accorded secondary analyses of data the same level of scholarly legitimacy as studies based on original data collections. For any and all of these reasons, a significant quantity of data stored on punched cards was discarded or destroyed up through the mid-twentieth century.51

  • 52 Grant Proposal to NSF, 1968, 94-95.

21When magnetic tape began to replace the bulkier, heavier stacks of punched cards in the 1960s, storing, analyzing, and distributing machine-readable data became easier. As social scientists’ grant proposals during this decade make clear, the latest large, expensive computers were ever more essential parts of the infrastructure of research.52 Both these developments made sharing data easier—no longer did “sharing” require mailing someone a large, heavy stack of punched cards or expecting them to travel to an institution with a computer capable of reading them. Technological advances like these undeniably made the data center boom possible and accelerated it once it began.

Competition and Coordination during the Boom

  • 53 Some subject-specific or regional archives include Purdue Opinion Panel’s collection of its surveys (...)
  • 54 The grants Roper received from NSF discussed above overwhelmingly supported “Acquisition of Foreign (...)

22In contrast to the Roper Center and ICPR’s aims to become full-fledged data centers, many other subject-specific or regionally-focused data archives that sprouted up at universities across the US deliberately chose narrower remits. Their creators identified and then occupied particular data-collecting and distributing niches, launching projects that may well have gone undone had they not taken it upon themselves to do them.53 Roper and ICPR’s ambitions, however, were broader. For one, each sought to acquire data from a wide range of researchers and firms. Each also offered institutions and individuals access to the data they stored as well as to a range of “data services” for an annual membership fee. Finally, each (at least for a time) offered specialized data analysis training to members.54 On the one hand, people at both Roper and ICPR energetically affirmed high-minded values like cooperation, sharing, and free exchange. On the flipside of these affirmations, however, they also criticized those amongst them who were not playing by the emerging rules—and worried that they themselves might be so criticized.

  • 55 Lucci and Rokkan, 1957, 27-28, 46.

23In the proposal for a library of survey research data discussed above, Lucci argued that survey researchers were clearly ready for such an institution by noting that nearly all social science research centers claimed they made their data freely available to any researcher who asked for it. He then pointedly noted that one institution—Michigan’s Survey Research Center (SRC)—said it did not. To explain its policy of less than free exchange, the representative of the SRC Lucci interviewed said the Center preferred to have more control over how its data were used than simply making it available to anyone who asked.55 That SRC, one of the biggest academic survey research centers, was wary of granting others free access to its data posed a significant obstacle to the library Lucci envisioned.

  • 56 Miller often referred to ICPR as “the consortium.”
  • 57 Miller Papers, 1959, 2; Hauptmann, 2020, 107-108.

24The correspondence of key SRC figures offers ample evidence of this wariness, though it was sometimes accompanied by the worry that SRC might be seen as uncooperative or miserly by others. In one notable case, Michigan’s Miller faced a dilemma when a prominent colleague at another institution requested a substantial chunk of SRC data. On the one hand, Miller didn’t want to give this colleague all he’d asked for since “our apparent willingness to ‘give away’ data destroys a minor part of the prize which we were thinking of offering for participation in the consortium.”56 But on the other hand, given that his plans for ICPR hinged upon people and institutions outside Michigan joining it, Miller saw that it would be wise to “enhanc[e] the popular image of us as a generous and cooperative institution,” particularly since he had heard some grumbles from political scientists outside Michigan’s SRC about its “non-participation in the Roper Collection at Williams.”57 Even before ICPR was officially launched, its founder-to-be was already thinking about how to attract potential members but also how to avoid being perceived as uncooperative.

  • 58 Hastings, 1957, 1.

25Given that the Roper Public Opinion Center had already begun its operations in 1957, Miller had to consider how to position the institution he wanted to build in relation to it. Even if there might be “room” for multiple social scientific data centers, their early leaders still had strong incentives to pitch them as unique or preeminent to possible funders. For instance, when Philip Hastings, the first head of the Roper Center, wrote to Elmo Roper to ask for long-term financial support for the Center, he stressed that based on the Center’s acquisitions of polls from multiple commercial firms in the US and abroad, “we have in the making what can become not just a highly important, but in fact, the opinion research center.”58 Roper’s Statement of Purpose, published around the same time, included testimonials from well-known social scientists that also vouched for its great potential as a resource for researchers and teachers. The space Roper and then ICPR were seeking to occupy was too new, their institutional fortunes too uncertain to allow either to view the other’s development with equanimity.

  • 59 Hastings, 1964.
  • 60 Miller, 1964b, 1-2.
  • 61 Miller, 1964a, 3-4.
  • 62 Wahlke, 1964a, 2, 3a.
  • 63 Lane, 1964, 1.

26Such concerns persisted well into the 1960s. Just a few years after ICPR’s 1962 founding, Miller and the board members argued amongst themselves about how to position the data center they were trying to build in relation to Roper, especially given Roper’s recent creation of an International Survey Library Association (ISLA).59 As ICPR’s first director, Miller worried that Roper/ISLA would deny ICPR access to the data it held and impose excessive fees on individual members of ICPR who sought it. To avoid a stalemate unfavorable to ICPR, Miller argued for formally conceding the Roper Center’s preeminent position as “the central storage facility for survey materials collected in America as well as in other countries”60 while also continuing informal negotiations to persuade Roper to lower barriers to its data for ICPR and its members.61 However, others who served on ICPR’s governing council or as official representatives of member institutions lobbied for a more confrontational stance. For instance, John Wahlke of SUNY-Buffalo, who served on ICPR’s governing council, pushed back strongly against what he saw as the overly conciliatory tone of Miller’s proposals. ICPR’s council, Wahlke wrote, thought Miller’s principles conceded far too many “sovereign rights for data-supplying archives” and complained that Miller “mislabeled” Roper’s mainly commercial survey holdings a “general” archive, compared with the “far broader range of data” Wahlke projected ICPR would ultimately acquire.62 Similarly, Robert Lane of Yale advised that ICPR should just start accumulating as much especially international “poll material” as possible to break Roper’s “relatively monopolistic position.” If there were multiple centers that had significant amounts of US and international poll data, Lane argued, Roper would be forced “to cooperate to stay in business.” Lane argued that pursuing such an aggressive acquisitions strategy was ICPR’s best bet for improving its position in relation to Roper: “Agree to nothing that would limit the Consortium acquisitions… and quietly move in the direction of making Michigan a general repository for a wider range of public affairs survey materials”.63

  • 64 Lowi, 1964, 1-2; emphasis in original.

27Wahlke and Lane’s attempts to persuade Miller to confront or try to undercut the Roper Center, however, were mild compared to the bellicose tack taken by Theodore Lowi. In a letter to Miller and Wahlke, Lowi used a nuclear arms race simile to argue that though ICPR’s position was strong at the moment, “like the American Atomic monopoly, such advantages wither if not used at the proper time—which is before there are other club members.” Though Roper “is weak at the moment,” Lowi warned, it probably wouldn’t stay so for long. The time was right, therefore, for ICPR to move in the direction of centralization: “If we aim now toward the destruction of ‘state sovereignty’ [for data archives] we’ll have done ourselves a big favor.” Such “total fusion” is not only “desirable,” Lowi argues, but “is also the only justifiable position, because foundation and government moneys are making all these efforts possible.”64

  • 65 Bisco, 1966, 104-106; Bisco 1967, 64-66; Farber and Beck, 1971, 238-239.
  • 66 Farber and Beck, 1971, 239. For a discussion of a more robust and longer-lasting set of institution (...)

28Such competitive strategizing, however, was not the only order of the day. In 1962, the same year ICPR was launched, it also oversaw the creation of another body, the Council of Social Science Data Archives (CSSDA). This council aimed to set general policies for all member archives and to manage relations between them;65 any early competition between ICPR and Roper was mediated by the policies set by this new but short-lived organization. By the end of 1969, the CSSDA was no more.66

  • 67 Converse, 1962b. Among those included in the earliest CSSDA discussions were representatives from U (...)
  • 68 Miller 1964a. According to Bisco (1966, 96) computing facilities at MIT’s Center for International (...)
  • 69 Converse, 1963b.
  • 70 Bisco, 1966, 105; Farber and Beck, 1971, 238.

29Still, during its short organizational life, the CSSDA was able to do two crucial things: first, it worked quickly to develop a cooperative federation among the newer data archives with ICPR at its center.67 Then, several years later, with the strength of these numbers, the group presented its ideas for “inter-archive” cooperation to Roper in 1964. It was around this time that a representative for the Roper Center, MIT’s Ithiel de Sola Pool, first joined the governing board of the Council.68 Though the Roper Center seems to have had some representation on the Council from 1964 through 1969, in its first few years, CSSDA focused on pooling the efforts of non-Roper data repositories to make a collective pitch for NSF support. Michigan’s Converse reported to the group in 1963 that he thought the time was right to make a proposal to the NSF for several reasons. For one, Converse stressed that Henry Riecken, Director of the Social Science Program at NSF, seemed “downright enthusiastic” about data repositories. Converse also reported that Riecken suggested an application from CSSDA would not undercut individual entities’ chances of securing NSF support.69 When CSSDA began its official life as an organization in 1965 (with a constitution, by-laws and an office in New York), it did so with the NSF support Converse had predicted it was well-positioned to secure.70

  • 71 This group was called the Committee of Eight. See Converse (1964, 273, 275) for a list of its membe (...)
  • 72 Converse, 1964, 276, 279.
  • 73 Idem, 278, 280.
  • 74 Idem, 284.
  • 75 Idem, 280.

30Representing a larger group of people involved in these efforts, Converse wrote in Public Opinion Quarterly that building a network of repositories seemed the most prudent course at this early stage of the development of social science data centers.71 While individual repositories should still retain some autonomy, belonging to a coordinated network would make any future moves toward centralization easier.72 Those who participated in this network were learning how “wasteful” a wholly decentralized array of repositories would be compared to “the development of appropriate social organization to harness the new potentials for the common weal.”73 Converse imagined that the “common weal” of future social scientists would be served by a network of linked computers that allowed data to be accessed and transmitted to anyone anywhere in the US. This vision of a fast and frictionless future should start with a commitment to “standardization” of coding and formatting data.74 All this, Converse conceded, “comes with a substantial price tag, a fair portion of which must be borne by collective subsidy.”75 Imagined along these lines, the future of social science would be more organized, more centralized, and more expensive.

  • 76 Solovey, 2013, 166-169; Solovey, 2020.
  • 77 Riecken, 1986, 211.
  • 78 See Converse (1963b) for “enthusiastic.” Riecken (1986) does not specifically discuss NSF support f (...)
  • 79 Solovey, 2013, 157-163.
  • 80 Solovey, 2020, 196.

31Beginning in the 1960s, NSF grants fueled and supported this vision, investing resources in numerous social science data repositories and the CSSDA. Overall, however, the social sciences were still new, small, and vulnerable in the 1960s NSF.76 Especially if one keeps the overall weakness of the social sciences at NSF in mind, Henry Riecken’s retrospective assessment offers some insight into how supporting data centers may have fit into NSF policy. Riecken, who served as the NSF’s Social Science Director from 1958 to 1965, retrospectively characterized how he and his predecessors advocated for the social sciences as a “strategy of protective coloration, of allying one’s cause with [the] stronger… physical and biological sciences”.77 Given the prominence of the late 1950s World Data Centers devoted to geophysical data discussed above, Riecken’s “enthusiastic” attitude towards social scientific data centers may well have been a specific expression of this “strategy of protective coloration.”78 Historian of science Mark Solovey confirms Riecken’s assessment, noting that social science projects most easily allied to the sciences at the core of the NSF were more likely to be promoted by social science directors and approved by the NSF’s board.79 This continued to be true of NSF policies in the 1970s, when “the agency’s natural-science oriented leadership viewed [large database] projects favorably because they seemed, in a sense, familiar.”80 The NSF’s enthusiasm for data repositories was clearly evident to social scientists in the 1960s, even if they didn’t fully understand what motivated it.

The Rise and Fall of the Proposed National Data Center

  • 81 Kraus, 2011, 6-9.
  • 82 As Kraus’s (2011) detailed account of the effort to create a National Data Center makes clear, the (...)
  • 83 Idem, 9-10.
  • 84 Idem, 12.

32Another uncertainty hanging over all academic data centers during the 1960s was whether the federal government would build a data center of its own—and if it did so, how its operation might impinge upon their plans. Momentum for a National or Federal Data Center had been building as early as the mid-1950s, reaching its peak a decade later. A committee of the Social Science Research Council, led by Yale economist Richard Ruggles, made one of the first public cases for a National Data Center in 1965. Ruggles’ case was supported by the leaders of several federal agencies, especially by Raymond Bowman of the Bureau of the Budget.81 For social scientists seeking access to federal agency data, the appeal was clear.82 Not only would the proposed National Data Center handle all requests for federal data; it would also standardize data from different agencies, making comparing and combining them feasible in the foreseeable future.83 For federal agencies, the National Data Center promised efficiencies of several kinds: savings of time and money once each agency was no longer required to collect and maintain its own data and quicker, more comprehensive internal policy analyses using standardized federal data.84

  • 85 Idem, 13-22. For other recent historical treatments of this episode, see Bouk (2018), Igo (2018, 22 (...)
  • 86 Kraus, 2011, 26-30.

33The advocates of the proposed National Data Center, however, said little about what it offered the general public, apart from a tepid afterthought that some of its benefits to social scientists and federal policy-makers might somehow eventually benefit everyone else. As it turned out, the envisioned centralization of federal data, so benign, sensible, and efficient a proposal to its defenders, seemed deeply threatening to its numerous public and Congressional critics. Opponents of the proposal made passionate cases against it, arguing this centralized store of federal data could be used for blackmail, surveillance, and political persecution. These arguments frightened many and caught the proposal’s supporters off-guard.85 Though supporters attempted to boost the public image of the proposed data center and quell concerns about the potential dangers it posed, salvaging even a more modest version of this proposal proved impossible. By 1970, efforts to create a National Data Center ground to a halt.86

  • 87 Bisco, 1967, 68-70.
  • 88 Bisco, 1967, 70.
  • 89 It was not until the late 1960s that computing facilities on a number of university campuses began (...)

34What consequences did the development of this project, as well as its eventual collapse, have for Roper and ICPR? The proposed National Data Center was unlikely to make these redundant for several reasons. First, its proposed mission focused on managing data generated by federal agencies rather than the survey and polling data at the heart of ICPR and Roper’s holdings. Second, the National Data Center was subject to a far higher level of public scrutiny than university-based data centers. This explains why people connected with social science data centers watched the unveiling and eventual collapse of the proposed National Data Center with more interest than alarm. For example, in early 1967, ICPR’s Bisco commented favorably on what the planned National Data Center might do to improve the preservation and usability of federal government data. Though he implied that the federal government was lagging a bit behind university-based data centers in managing, standardizing, and providing access to the data its many agencies generated, Bisco did not regard the National Data Center as encroaching upon non-governmental data centers’ acquisitions or services.87 Instead, Bisco seemed only mildly worried that the “adverse publicity” the proposal had received from “sensational and distorted” media coverage might undermine existing data centers.88 On balance, Bisco hoped the National Data Center proposal would survive its very public trials, but in the mid-1960s, ICPR, Roper, and other university-based data centers faced no comparable challenges.89 Instead, the collapse of this proposal may have cleared the way for academic data centers to acquire broader swaths of federal government data.

Intellectual Aspirations that Fueled the Boom

35Up until this point, I have focused primarily on the financial and organizational challenges the architects of the first social scientific data centers faced. I do not want to suggest, however, that they were driven by professional or institution-building motives alone. Had those who participated in it not also had powerful intellectual aspirations, the whole data center boom would not have been as intense as it was. What, then, were some of those intellectual aspirations?

  • 90 Cited in Scheuch, 2003, 387-388.
  • 91 Lemov, 2015, 212.

36For one, those who devoted their professional lives to building data centers believed they could generate far deeper and broader knowledge about society and politics than what early twentieth-century social science had been able to muster—provided these data centers were well-stocked, expertly staffed, and widely used. For instance, the developer of Yale’s Human Relations Area Files projected in 1949 that analyses of the collection’s vast array of data about interpersonal relations in hundreds of cultures across the world might “prove” that the family—perhaps even the nuclear family—was a universal feature of human culture.90 Similarly, in the late 1950s, the anthropologist who promoted a Microcard storage system for his and other anthropologists’ research records believed that his “database of dreams” was uniquely poised to offer a “global understanding of man.”91 Grandiose and misguided as these aspirations may seem, it is hard to imagine people dedicating themselves to the daily drudgery that it took to build and promote the infrastructure of social scientific data centers without the impetus of such heady visions.

  • 92 Allen, 1967b, 5. The WPA was the Works Progress Administration, a sweeping New Deal employment prog (...)

37When people connected to Roper and ICPR expressed their intellectual hopes for what the institutions they were building might make possible, they too argued that the depth and breadth of computer-aided analyses of vast quantities of data promised to reveal general, fundamental patterns beyond the reach of what human analysts could achieve on their own. The realization of these hopes, however, required huge prior investments of time and labor. Even when survey data was already in machine-readable form, data center staff still had to check it for errors, “clean” it, and put it into formats to make it compatible with other data sets. Making the older paper records of electoral results and Congressional roll call votes machine-readable required even more work. For example, some collections ICPR acquired were truly gargantuan, like the eight-ton Congressional Vote Analysis collection received from Columbia University in 1967 on a long-term loan. This collection, originally done “under WPA auspices” in the 1930s, recorded and analyzed all Congressional votes from the time of the late eighteenth-century Continental Congress up through the late 1930s.92

  • 93 NSF Grant Proposal, 1964, 2-5; Austin, 2011. Austin (2011) offers an excellent account of ICPR’s ea (...)
  • 94 In an undated glossary, ICPSR’s Carolyn Geda defined data as “The quantities, characters or symbols (...)
  • 95 Allen 1967a, 3.
  • 96 Grant Proposal, 1964, 2.

38ICPR also took part in an ambitious project to collect huge quantities of older US election results in order to standardize them and make them available for secondary analyses. This project, sponsored initially by the SSRC’s Committee on Political Behavior and the American Historical Association, sought to collect information about US elections as far back as 1824. This meant first locating the most complete and reliable county-level election results across the US (a herculean task overseen by Walter Dean Burnham over several years in the early 1960s); then, at ICPR, the work of “cleaning” or standardizing these results to make them machine-readable began. NSF sponsored the ICPR phase of this project, beginning in 1963 with a series of substantial grants.93 The work of transforming these records into machine-readable data took years.94 In 1967, Howard Allen, ICPR’s Director of Data Recovery, spelled out the difficulties of working with so large and unruly a set of political records; turning 150 years of electoral records into data meant dealing with changing county boundaries and the names of over 800 political parties.95 In 1968, ICPR applied for an extension of the NSF grant to support this enormous task. Before they were fully aware of these challenges, ICPR’s architects optimistically claimed that turning these older results into machine-readable data “represents an increment that is exponentially more useful and valuable”96 once they were combined with the electoral data ICPR already had. The hope was that computer-aided “longitudinal” analyses drawing on this huge mass of data might yield general insights into long-term patterns of political behavior—analyses not even the largest, most well-informed team of researchers could perform on their own. To many social scientists today, such a claim about the value of longitudinal analyses seems self-evident. In the 1960s, however, the promise of research grounded in longitudinal analyses was still a case that had to be made.

  • 97 Hastings,1959.
  • 98 Hastings, 1963, 592, 597.
  • 99 Rohde, 2011; Lepore, 2020, 205-232.

39The people who built ICPR hoped that casting a wider net deeper into the US’s political past would yield ever more fundamental insights into American political behavior. Roper’s aims were more focused on building a large, cross-cultural collection of survey materials—and what drawing on such materials might allow social scientists to do. Just a few years after the Roper Center began operations, its director, Philip K. Hastings, was pitching the benefits of accessing the Center’s collections to US foreign policy agencies.97 By 1963, Hastings reported that one-third of the Center’s holdings were from non-US sources “with thirty-nine countries represented” and cited a number of “cross-national” studies that had made use of these materials.98 MIT’s Pool, who represented Roper at the meetings of the CSSDA discussed above, took ambitions regarding what social scientists could do with survey material from outside the US several steps further. In the late 1960s, Pool went to Vietnam as a representative of the Simulmatics Corporation, having won a Department of Defense contract to assess the psychological states of Vietnamese villagers by means of surveys and tests. Though the Department of Defense later withdrew its support for a project it deemed a failure on multiple levels, Pool’s ambitions for what data-driven social science meant and could do were among the most grandiose of the decade.99

  • 100 Rohde, 2017.

40Anticipations of the new vistas of social and political knowledge computer-aided analyses might open up began to give way to far bleaker views. As Joy Rohde100 argues, the development of computer systems to predict international crises that began in the late 1960s was driven by a profound “epistemic insecurity.” The flip side of a sunny, confident optimism about all the computer-aided human mind can know is a crushing awareness of the slowness, partiality, and incorrigible flaws of human cognition.

Conclusion

  • 101 Ribes and Jackson, 2013, 164. The “petacenter” refers to twenty-first century “computing centers th (...)

Data demand and build the human, organizational, and infrastructural worlds around them—enforcing a burden of care and work that disappears beneath (but ultimately constitutes) the futuristic possibilities of the peta center.101

  • 102 Aronova, 2017, 321.
  • 103 Bisco, 1964, 19. Along similar lines, Solovey (2020, 282-285) analyzes the narrowing consequences o (...)

41Though some of the early “the sky’s the limit” optimism about computer-aided social science has surely faded, the conviction that the most reliable social science must be grounded in the bedrock of data cleaned, managed, and expanded by data centers remains. Despite having lost their futuristic gleam decades ago, the sheer inertial bulk of social scientific academic data centers and the services they provide continue to dominate contemporary social science. These institutions, however, support only particular kinds of intellectual ambitions; as a consequence, their dominance has narrowed the epistemic range of social science in the US. Accessing and analyzing huge masses of cleaned social and political data may appeal to those seeking previously undiscovered correlations within it or general, global explanations of social and political patterns. By contrast, data centers offer little to nourish the intellectual ambitions of those interested in the atypical, local, or structural aspects of social and political life. As Aronova notes in her discussion of the various geophysical disciplines, those scientists most devoted to explaining “local particularities” of ice and glacier formations saw “little value” in working with other scientists’ raw data or in creating standardized machine-readable abbreviations of their own work.102 Similarly, many social scientists committed to interpretive analyses would not sign on to Bisco’s view that “the most creative parts of empirical research involve operationalizing concepts and developing testable theory” by consulting data collected and processed by a data center.103

42Investigating how social scientific data centers came to be, therefore, also entails investigating the ascendance of the research practices their advocates favored. As I have shown, in order to get these data centers built in the first place, these advocates had to do some rhetorical heavy lifting to convince their colleagues that social science grounded in secondary and longitudinal analyses of large data sets was legitimate and worth doing. That game was largely won in the 1970s with the advent of “big social science” at the NSF—grants averaging half a million dollars for long-term data collection and curation projects.104 As part of this development, when the NSF announced its long-term support for Michigan’s election studies in 1977, it effectively nationalized it as well; Michigan’s election studies became the American National Election Studies (ANES).105 By the end of the twentieth century, analyses tied to “big social science” had already become so much a part of the mainstream of social science that it was their challengers, not their defenders, who struggled to make themselves heard. Data centers have been crucial to reproducing this new “social organization of research” with secondary and longitudinal analyses of large curated, cumulative bodies of data at its core. The institutional success of the social scientific data center has been so complete that, like its history, the epistemic victory it brought about has become all but invisible. Yet the primacy of quantitative, data-driven social science in the twenty-first century US is surely one of the most important consequences of the rise of the social scientific data center. Rediscovering this institution’s history begins to make these consequences visible too.

Haut de page

Bibliographie

Archival Sources

University of Michigan Records, Bentley Historical Library, Ann Arbor Michigan

ICPSR Records

Allen, H., 1967a, “Memo from Director of Data Recovery, Subject: ‘Availability of historical election data’, February 24”, Folder: Historical Archive, 1967, Box 14.

Allen, H., 1967b, “Memo from Director of Data Recovery, Subject: ‘Progress Report on Historical Data Projects May 30”, Folder: Historical Archive, 1967, Box 14.

Bisco, R. L., 1964, “Information Services for Political Science: Progress and Prospects”, Paper delivered at 1964 meeting of the American Political Science Association, Chicago, Folder: Development Activities and Plans, 1963-1965, Box 14.

Converse, P. E., 1962a, “Letter to Karl Deutsch, Member of Committee of Eight, October 26”, Folder: Committee of Eight, Correspondence and Minutes, 1962, Box 1.

Converse, P. E., 1962b, “Memo to Committee of 8, ‘Background, Issues & Tentative Agenda for Meeting, December 1-2’, November 8”, Folder: Committee of Eight, Correspondence and Minutes, 1962. Box 1.

Converse, P. E., 1963a, “Toward the Development of an Inter-University Data Archive System for Comparative Research”, Proposal labeled “First Draft, January, 1963”, hand-dated “1.31.63.”, Folder: Organization and Administration, Committee of Eight, Proposals, 1963, Box 1.

Converse, P. E., 1963b, “Letter to Charles Glock, Survey Research Center, U.C. Berkeley, May 2”, Folder: Organization and Administration, Committee of Eight, Correspondence and Minutes, 1963, Box 1.

Converse, P. E., 1963c, “Memo to Committee of 8, June 28”, Folder: Organization and Administration, Committee of Eight, Correspondence and Minutes, 1963, Box 1.

Geda, C., N.d., “Basic Glossary of Data-Processing Terms”, Box 14.

Grant Proposal, 1964, “A Proposal for a Study of the National Electorate, 1964”, No Author, May 13, Box 16.

Grant Proposal, 1967, “To Ford Foundation for Funds to Support an Expansion of Archival Resources”, No Author, Box 16.

Grant Proposal, 1968, “Proposal to NSF for Funds for Supplementary Support of a Specialized Facility for Social Science Research”, No Author, Box 16.

Hofferbert, R., 1975, “Memo to Jerome M. Clubb, ‘Supervision of Data Acquisition’”, Folder: Correspondence and Memoranda folder, 1964, 1966, 1974-75, Box 14.

Hofferbert, R., N.d., “Archiving of Machine-Readable Policy Data: Activities of the ICPR”, Folder: Resource Development and Services, Archival Activities, Reports and Reviews, 1962-1978, Box 15.

Lane, R., 1964, “Letter to Miller, November 7”, Folder: Correspondence To/From ICPSR Executive Directors, 1964-2001, 1 of 3, Box 3.

Lowi, T., 1964, “Memo to Miller and John Wahlke: ‘Anti- Inter-University Cooperation’, November 2”, Folder: Correspondence To/From ICPSR Executive Directors, 1964-2001, 1 of 3, Box 3.

Miller, W. E., 1962a, “Memo to ICPR Council, ‘Estimated cost of establishment of data repository’, October 18”, Folder: Resource Development and Services, Archival Activities: Formation of a Data Repository, 1962, Box 14.

Miller, W. E., 1962b, Memo to ICPR Council, “Conversations with Peter Rossi and Jack Feldman of NORC, October 4, 1962” n.d. Folder: Resource Development and Services, Archival Activities: Formation of a Data Repository, 1962, Box 14.

Miller, W. E., 1964a, Memo to Official Representatives, ICPR, re. “Inter-archival cooperation,” September 1964. Folder: Correspondence To/From ICPSR Executive Directors, 1964-2001, 1 of 3, Box 3.

Miller, W. E., 1964b, “Proposed Principles for Inter-Archive Cooperation,” n.d.; attached to Miller, 1964a. Folder: Correspondence To/From ICPSR Executive Directors, 1964-2001, 1 of 3, Box 3.

Miller, W. E., 1997, “Oral history interview. Oral history interview. (Erik Austin, interviewer). July 20”, VHS tape and typed transcript, Institute for Social Research (ISR) Oral History project.

NSF Grant Proposal (no author), 1964, “A Proposal for Funds to Support the Addition of Data to the Inter-University Consortium for Political Research Data Repository”, February, Box 16.

Wahlke, J., 1964, “Council, ICPR, to Official Representative, ICPR, Subject: ‘Proposed Principles for Inter-Archival Cooperation’”, Draft, n.d., Folder: Correspondence To/From ICPSR Executive Directors, 1964-2001, 1 of 3, Box 3.

W. E. Miller’s Personal Papers

Miller, W. E., 1966. “Proposal Rating Sheet, Proposal S 70163-P (An African Social Science Data Archive)”, Folder: Correspondence, Topical—Proposal Reviews, NSF: Miscellaneous, 1962-67. Box 3.

Special Collections, Williams College, Williamstown, Massachusetts

Office of the President Records (President James P. Baxter)

Baxter, J. P., 1959, “Baxter to Joel I. Brooke, 30 Rockefeller Plaza, March 12”, Box 69 (Roper Center Correspondence).

Baxter, J. P., 1960, “Baxter to Henry Heald, Ford Foundation, February 22”, Box 69 (Roper Center Correspondence).

Hastings, P. K., 1957, “Hastings to Elmo Roper, July 11”, Box 69. (Roper Center Correspondence).

Statement of Purpose, N.d., “Roper Public Opinion Research Center at Williams College”, Box 69 (Roper Center Correspondence).

Roper Public Opinion Research Center (Subject File)

“Williams Newsletter”, 1962, Fall.

Published and Online Works

Aronova, E., 2017, “Geophysical Datascapes of the Cold War: Politics and Practices of the World Data Centers in the 1950s and 1960s”, Osiris, 32, p. 307-327.

Austin, E., 2011, “ICPSR: The Founding and Early Years”, ICPSR website, online: https://page.hn/28vphd

Bisco, R. L., 1966, “Social Science Data Archives: A Review of Developments”, The American Political Science Review, 60 (1), p. 93-109.

Bisco, R. L., 1967, “Social Science Data Archives: Progress and Prospects”, Social Science Information, 6 (1), p. 39-74.

Bouk, D., 2018, “The National Data Center and the Rise of the Data Double”, Historical Studies in the Natural Sciences, 48 (5), p. 627-636.

Converse, J. M., 1987, Survey Research in the United States: Roots and Emergence, 1890-1960, Berkeley, The University of California Press.

Converse, P. E., 1964, “A Network of Data Archives for the Behavioral Sciences”, Public Opinion Quarterly 28 (2), p. 273-286.

De Chadarevian, S., Porter, T. M., 2018, “Introduction: Scrutinizing the Data World” Historical Studies in the Natural Sciences, 48 (5), p. 549-556.

Farber, K. A., Beck, C., 1971, “Council of Social Science Data Archives (CSSDA)”, Encyclopedia of Library and Information Science.

Frantilla, A., 1998, Social Science in the Public Interest: A Fiftieth-Year History of the Institute for Social Research, Ann Arbor, Bentley Historical Library.

Garfinkel, S., 2000, Database Nation. The Death of Privacy in the 21st Century, Sebastopol, O’Reilly and Associates.

Geiger, R. L., 1993, Research and Relevant Knowledge. American Research Universities Since World War II, Oxford, Oxford University Press.

Glaser, B. G., 1963, “Retreading Research Materials. The Use of Secondary Analysis by the Independent Researcher”, American Behavioral Scientist, 6 (10), p. 11-14.

Hastings, P. K., 1963, “The Roper Center: An International Archive of Sample Survey Data”, Public Opinion Quarterly, 27 (4), p. 590-598.

Hastings, P. K., 1964, International Survey Library Association of the Roper Center, Public Opinion Quarterly, 28 (2), p. 331-333.

Hauptmann, E., 2016, “‘Propagandists for the Behavioral Sciences’: The Overlooked Partnership between the Carnegie Corporation and the SSRC in the mid-20th century”, The Journal of the History of the Behavioral Sciences, 52 (2), p. 167-187.

Hauptmann, E., 2020, “Why They Shared. Recovering Early Arguments for Sharing Social Scientific Data”, Science in Context, 33 (2), p. 101-119.

Hauptmann, E., 2022, Foundations and American Political Science: The Transformation of a Discipline, 1945-1970, Lawrence, The University Press of Kansas.

Igo, S. E., 2007, The Averaged American: Surveys, Citizens, and the Making of a Mass Public, Cambridge, Harvard University Press.

Igo, S. E., 2018, The Known Citizen: A History of Privacy in Modern America, Cambridge, Harvard University Press.

Kraus, R. S., 2013, “Statistical Déjà Vu: The National Data Center Proposal of 1965 and its Descendants. Paper presented at the Joint Statistical Meetings”, Journal of Privacy and Confidentiality, 5 (1).

Larsen, O. N., 1992, Milestones and Millstones. Social Science at the National Science Foundation, 1945-1991, New Brunswick, Transaction Publishers.

Lemov, R., 2015, Database of Dreams: The Lost Quest to Catalog Humanity, New Haven, Yale University Press.

Lepore, J., 2020, If Then. How the Simulmatics Corporation Invented the Future, New York, W.W. Norton Publishing.

Loughnane, C., Aspray, W., 2018, “Rethinking the Call for a US National Data Center in the 1960s: Privacy, Social Science Research, and Data Fragmentation Viewed from the Perspective of Contemporary Archival Theory”, Information & Culture, 53 (2), p. 203-242.

Lucci, Y., Rokkan, S., and Meyerhoff, E., 1957, A Library Center of Survey Research Data: A Report of an Inquiry and a Proposal, New York, Columbia University School of Library Service.

Miller, A. R., 1971, The Assault on Privacy: Computers, Data Banks, and Dossiers, Ann Arbor, The University of Michigan Press.

Moustafa, T., 2024, “Political Science as a Dependent Variable: The National Science Foundation and the Shaping of a Discipline”, Perspectives on Politics, online: https://page.hn/2aesfh.

Ribes, D., Jackson, S. J., 2013, “Data Bite Man”, p. 147-166, in Gitelman, L. (ed.), “Raw Data” Is an Oxymoron, Cambridge, MIT Press, online: 10.7551/mitpress/9302.001.0001.

Riecken, H. W., 1986, “Underdogging: The Early Career of the Social Sciences in the NSF”, p. 209-226, in Klausner, S. Z., and Lidz, V. M. (eds), The Nationalization of the Social Sciences, Philadelphia, The University of Pennsylvania Press, online: 10.9783/9781512803013-011.

Rohde, J., 2011, “The Last Stand of Psychocultural Cold Warriors: Military Contract Research in Vietnam”, Journal of the History of the Behavioral Sciences, 47 (3), p. 232-250.

Rohde, J., 2017, “Pax Technologica: Computers, International Affairs, and Human Reason in the Cold War”, Isis, 108 (4), p. 792-813.

Rokkan, S., 1976, “Data Services in Western Europe. Reflections on Variations in the Conditions of Academic Institution-Building”, American Behavioral Scientist, 19 (4), p. 443-454.

Scheuch, E. K., 2003, “History and Visions in the Development of Data Services for the Social Sciences”, International Social Science Journal, 177, p. 385-399.

Solovey, M., 2013, Shaky Foundations. The Politics-Patronage-Social Science Nexus in Cold War America. New Brunswick, Rutgers University Press.

Solovey, M., 2020, Social Science for What? Battles over Public Funding for the “Other Sciences” at the National Science Foundation, Cambridge, MIT Press.

Truman, D. B., 1991, “Oral history interview”, p. 135-151, in Baer, M. A., Jewell, M. E. and Sigelman L. (eds), Political Science in America: Oral Histories of a Discipline, Lexington, The University of Kentucky Press.

Haut de page

Notes

1 I use the term “data center” for institutions that acquired data sets from multiple sources, integrated these into databases, and provided a variety of educational and technical services to users. In the period on which I focus here, many data repositories or data archives did not do all of these things. My focus here is on two that did—the Roper Center for Public Opinion Research, the Inter-University Consortium for Political Research (ICPR). Proposals for the National Data Center also emphasized that it would do all these things.

2 I came across a similar analogy drawn by De Chadarevian and Porter (2018, 549) who compare investigating the history of data to investigating “the domain of infrastructure that was to be kept under the street with wires, gas lines, and sewer pipes.”

3 Converse, 1964, 274. Changing “the social organization of research” was a frequently mentioned aim of ICPR’s. It is also mentioned on page 82 of a 1967 ICPR grant proposal to the Ford Foundation. Full citation in the Bibliography section under ICPSR Records.

4 I focus throughout this paper on political science, the discipline whose history I have studied the most closely, though I try to be explicit about when and how my analyses apply to other social sciences.

5 On private philanthropic foundation funding for the behavioral sciences, see Hauptmann (2022) and Solovey (2013). Private philanthropies, like the Rockefeller Foundation and the Carnegie Corporation, began making major grants to the academic social sciences as early as the 1920s. The NSF, however, was a postwar creation. And though it began making grants in 1950, it did not have a program for funding the social sciences until the late 1950s. For a recent account of the history of the NSF in the US that emphasizes the consistently meager funding allotted to the social sciences coupled with its favoring “scientistic” approaches to social science, see Larsen (1992) and Solovey (2020). For a detailed examination of NSF’s Political Science program’s overwhelming support for studies employing quantitative methods or formal models, see Moustafa (2024).

6 Bouk, 2018; Igo, 2018.

7 Lepore (2020) documents the Simulmatics Corporation’s attempts to use computer-aided analyses of survey data to advise political campaigns, predict urban unrest and plan psychological warfare operations in Vietnam. Several social scientists who worked with Simulmatics—Ithiel de Sola Pool and James Coleman—were also important participants in 1960s efforts to build and expand academic data repositories.

8 Geiger, 1993.

9 The Roper Center began operations in 1957, the Inter-University Consortium for Political Research (ICPR) in 1962. In 1975, the latter changed its name to the Inter-University Consortium for Political and Social Research (ICPSR). Since I am mostly concerned with its early years in the bulk of the paper, I call ICPR by its original name.

10 Lemov, 2015.

11 Bouk, 2018; Igo, 2018, 221-263.

12 This was true of Rensis Likert and Angus Campbell of Michigan’s Institute for Social Research. Likert, best known for the “Likert scale” for measuring attitudes, worked in the insurance industry in the 1930s. Both he and Campbell, the senior author of The American Voter, worked for the Division of Program Surveys under the Department of Agriculture during World War II (Frantilla, 1998, 16-18).

13 The kinds of materials made into “data” differed by discipline. While surveys and polls were central to social psychologists, sociologists, and political scientists, they were less so for economists—and much less so for anthropologists. For instance, as I discuss below, economists were especially interested in data produced by federal agencies that were not derived from surveys. And Yale anthropologists created a wide-ranging filing system that listed crucial cultural attributes mentioned in a variety of written materials. The creators of this project, the Human Relations Area Files (HRAF), began distributing microfiche versions of these files to subscribers in the late 1940s (Lemov, 2015, 137-138; Scheuch, 2003, 387).

14 Converse, 1987, 180-182.

15 Converse (1987) discusses what she terms survey researchers’ “migration to the universities” in Part Three of her history of survey research in the US.

16 In a letter Baxter sent to a number of potential donors to the Roper Center in 1959, he added, “This project for the Research Center lies very close to my heart, because of my interest in the field which dates from the days when Elmo and I were working for Bill Donovan in the OSS.” (Baxter, 1959) A Fall 1962 Williams Newsletter noted that Elmo Roper had met Baxter on an OSS mission in London regarding “the later landing in North Africa” (Williams Newsletter, 1962).

17 Hauptmann (2016, 2022) discusses 1950s private philanthropic support that made these surveys possible. As the unnamed authors of a 1967 ICPR grant proposal to the Ford Foundation noted, in the 1950s, Michigan’s Survey Research Center “had inadvertently acquired something approaching a monopoly of survey data pertaining to American national elections” (Grant Proposal, 1967, 79).

18 Hastings, 1957; Hastings, 1963, 592.

19 The Roper Center, now housed at Cornell University, provides an extensive account of its history on its website, as does ICPSR. The two entities on which I focus were not the only postwar data centers that tried to build a national, principally academic membership base. Chicago’s National Opinion Research Center (NORC) built such a base and made some moves towards acquiring data beyond that produced by its researchers. Peter Rossi, NORC’s director the 1960s, also tried (without success) to get federal support to build a National Behavioral Data Center in Chicago.

20 For instance, Converse (1964) uses “data archive,” “data library,” and “data bank,” without drawing any distinctions between them. “Data libraries” on page 275; “data bank” on page 276; “data archives” in title and throughout.

21 Warren Miller would later say that it was this Lucci and Rokkan report that inspired him to build ICPR—and that ICPR turned out to be more like a library than an archive (Miller 1997).

22 See the discussion of the “Farmington plan” of inter-library lending in Converse (1963a, 8-9).

23 Lucci and Rockan, 1957, 118.

24 Loughnane and Aspray, 2018, 220-221, 228-229.

25 Converse, 1962, 4.

26 Internal references suggest this undated piece was written in the 1970s.

27 Lucci and Rokkan, 1957, 48-82.

28 Lucci and Rockan, 1957, 51-54. Well into the 1960s, making arguments for the merits of secondary analyses continued to seem necessary. Glaser (1963) devotes an entire article to presenting such arguments.

29 Truman, 1991, 143-145. When Elmo Roper’s 1943 donation of a number of studies done by his firm to Williams College first became available to researchers in 1947, it was called “the Roper Collection.” The Roper Center was established a decade later.

30 Political scientist V.O. Key wrote that if “younger men in political science” would draw on Roper’s multiple surveys to test their “inferences,” their conclusions would be significantly stronger—so much so that “two and two make something more than four.” Economist James Tobin wrote, “empirical testing of hypotheses in social science will be much more powerful if it can mobilize the evidence of a large number of surveys (Roper Statement of Purpose, n.d., 6).” Internal references suggest this statement was drafted in the fall of 1957.

31 Baxter, 1960. Baxter’s full pitch read as follows: “As an historian I know something about how valuable a collection of this nature—the only one of its kind in the world—could be both to History and all the other Social Sciences. The range of materials as well as their relevance to a large number of problems of great current importance is impressive. Only the surface of this treasure house has been scratched.”

32 Glenn, 1973 cited in Loughnane and Aspray, 2018, 216-217.

33 Hauptmann 2020, 110. “Willy Loman” is the harried travelling salesman of Arthur Miller’s 1949 play, Death of a Salesman.

34 Scheuch, 2003, 393.

35 Larsen, 1992.

36 Solovey, 2020.

37 Solovey, relying on figures from Larsen (1992) concludes that NSF never allocated more than 6% of its budget to the social sciences (2020, 242-243). Often, it was less.

38 Converse, 1963c.

39 The US Bureau of Labor Statistics and the Bureau of the Census (1966, 99-100).

40 Bisco, 1966.

41 Current values calculated using the US Bureau of Labor Statistics Inflation Calculator (https://www.bls.gov/data/inflation_calculator.htm). Totals of original grants derived from https://www.nsf.gov/awardsearch/. The information this search tool provides for awards made prior to 1976, however, is sketchy; no information regarding the program making the award, the program officer, the principal investigator or the award abstract appears.

42 Bisco, 1966, 97, 102, 104-106.

43 Aronova, 2017, 307-308, 313.

44 For a list of all thirteen of the disciplines involved in the International Geophysical Year, see Aronova (2017, 309, note 8). Aronova notes that all of these disciplines had “direct military implications” (309).

45 Aronova, 2017, 318-319. “Deluge” in the sub-heading on p. 316; “accumulations of data” in the quotation from Marshall Jamison, p. 319

46 Shapley, quoted in Aronova, 2017, 308.

47 Kraus, 2011, 4-8; Miller, 1971, 54-67.

48 Garfinkel, 2000, 17-18.

49 Garfinkel goes on to explain that Herman Hollerith, the nineteenth century inventor of the punched card, also invented a tabulating machine that “read” them quickly. These inventions formed the basis for what would become IBM, the company that manufactured most computers in the mid-twentieth century. Lepore (2020, 68-73) provides an overview of the capabilities of IBM computers during this period.

50 Aronova notes that most of the physical scientists who contributed data to the World Data Centers never took the step of storing their data on punched cards for several reasons: the bulk and weight of the cards, the sprawling “variety of data formats” preferred by different fields as well as the marked differences in how valuable “raw data” (as opposed to reports) produced by other scientists seemed in particular fields. Instead, World Data Centers relied on “microphotography,” storing the data they held on microfilm or microcards (2017, 316-318; 321).

51 Lucci, 1957, 14-17. Lucci itemizes the neglect and destruction of data stored on punched cards in a variety of settings (14-17); he notes that in the US, the survey data collected during World War II by the Office of War Information gathered dust after that office closed. When NORC took steps to acquire it, so much was missing or incomplete that it was ultimately all destroyed in 1955 (17). The data that formed the basis of the four-volume American Soldier project was also nearly destroyed.

52 Grant Proposal to NSF, 1968, 94-95.

53 Some subject-specific or regional archives include Purdue Opinion Panel’s collection of its surveys of nation-wide samples of US high school students (Lucci, 1957, 30-31), the 1966 proposal to the NSF by two US sociologists to build an African Social Science Data Archive (Miller, 1966). Also, the data archives at MIT, University of Illinois, University of Iowa, and UCLA did not have national aims—each sought instead to provide services to people affiliated with these institutions or within their regions (Bisco, 1966, 102-104).

54 The grants Roper received from NSF discussed above overwhelmingly supported “Acquisition of Foreign Survey Data.” In the 1960s, ICPR charged its members $2,500 per year, Roper/ISLA, $1,000 (Bisco, 1966, 96; 99). ICPSR’s well-known training courses began in the 1960s; today, it offers a massive summer curriculum that enrolls thousands of students. The Roper Center also offered a “Social Science Research Training Institute” as late as 1973 (Roper Public Opinion Research Center subject file).

55 Lucci and Rokkan, 1957, 27-28, 46.

56 Miller often referred to ICPR as “the consortium.”

57 Miller Papers, 1959, 2; Hauptmann, 2020, 107-108.

58 Hastings, 1957, 1.

59 Hastings, 1964.

60 Miller, 1964b, 1-2.

61 Miller, 1964a, 3-4.

62 Wahlke, 1964a, 2, 3a.

63 Lane, 1964, 1.

64 Lowi, 1964, 1-2; emphasis in original.

65 Bisco, 1966, 104-106; Bisco 1967, 64-66; Farber and Beck, 1971, 238-239.

66 Farber and Beck, 1971, 239. For a discussion of a more robust and longer-lasting set of institutions to manage cooperation among principally European social science data archives, see Scheuch (2003, 386-392). Scheuch also comments here on what he saw as the challenges Roper’s ambitions and policies posed to inter-archive cooperation in Europe.

67 Converse, 1962b. Among those included in the earliest CSSDA discussions were representatives from UCLA, UC Berkeley, the University of Chicago, Yale, the University of North Carolina and the University of Cologne (Converse, 1962a; Bisco, 1966, 104-105).

68 Miller 1964a. According to Bisco (1966, 96) computing facilities at MIT’s Center for International Studies, with which Pool was affiliated, were able to process Roper data in ways that were not available at Roper’s facility in Williamstown. MIT’s computing in effect amplified what Roper could offer its members.

69 Converse, 1963b.

70 Bisco, 1966, 105; Farber and Beck, 1971, 238.

71 This group was called the Committee of Eight. See Converse (1964, 273, 275) for a list of its members and a discussion of its purpose.

72 Converse, 1964, 276, 279.

73 Idem, 278, 280.

74 Idem, 284.

75 Idem, 280.

76 Solovey, 2013, 166-169; Solovey, 2020.

77 Riecken, 1986, 211.

78 See Converse (1963b) for “enthusiastic.” Riecken (1986) does not specifically discuss NSF support for any social scientific data center; nor does he mention the World Data Centers. He does remark, however, that “improved methods for collecting and analyzing data” has underwritten “progress” not just in the social sciences but in “all sciences” (220-221).

79 Solovey, 2013, 157-163.

80 Solovey, 2020, 196.

81 Kraus, 2011, 6-9.

82 As Kraus’s (2011) detailed account of the effort to create a National Data Center makes clear, the social scientists involved in them were largely economists, like SSRC Committee chair Ruggles. This makes sense, since much of the data crucial to economists’ research was produced by federal agencies rather than private or academic survey organizations.

83 Idem, 9-10.

84 Idem, 12.

85 Idem, 13-22. For other recent historical treatments of this episode, see Bouk (2018), Igo (2018, 221-223), Loughnane and Aspray (2018), and Lepore (2020, 279-283).

86 Kraus, 2011, 26-30.

87 Bisco, 1967, 68-70.

88 Bisco, 1967, 70.

89 It was not until the late 1960s that computing facilities on a number of university campuses began to be attacked by anti-war protestors (Igo, 2018, 235).

90 Cited in Scheuch, 2003, 387-388.

91 Lemov, 2015, 212.

92 Allen, 1967b, 5. The WPA was the Works Progress Administration, a sweeping New Deal employment program begun in 1935.

93 NSF Grant Proposal, 1964, 2-5; Austin, 2011. Austin (2011) offers an excellent account of ICPR’s early history. Austin joined the staff of ICPR to assist specifically with its work on this massive quantity of election data.

94 In an undated glossary, ICPSR’s Carolyn Geda defined data as “The quantities, characters or symbols on which operations are performed by computers or other automatic equipment, and which may be stored or transmitted in the form of electrical signals, records on magnetic tape or punched cards, etc.” (5). Internal references suggest this glossary was produced in the mid-1970s.

95 Allen 1967a, 3.

96 Grant Proposal, 1964, 2.

97 Hastings,1959.

98 Hastings, 1963, 592, 597.

99 Rohde, 2011; Lepore, 2020, 205-232.

100 Rohde, 2017.

101 Ribes and Jackson, 2013, 164. The “petacenter” refers to twenty-first century “computing centers that have been established to store the tens of thousands of terabytes (a terabyte being a thousand gigabytes) of data flowing from dozens of meteorological satellites, hundreds of genomic sequencers, thousands of ecological field sites, and the millions of sensors at the Large Hadron Collider” (Ribes and Jackson, 2013, 151).

102 Aronova, 2017, 321.

103 Bisco, 1964, 19. Along similar lines, Solovey (2020, 282-285) analyzes the narrowing consequences of the NSF’s emphasis on “hard core” or “scientistic” approaches to social science; see especially his discussion of the “interpretivist challenge” to these approaches.

104 Larsen, 1992, 106-110.

105 https://electionstudies.org/about-us/history/

Haut de page

Pour citer cet article

Référence papier

Emily Hauptmann, « Rediscovering the Early History of Social Scientific Data Centers in the US »Revue d’histoire des sciences humaines, 45 | -0001, 43-67.

Référence électronique

Emily Hauptmann, « Rediscovering the Early History of Social Scientific Data Centers in the US »Revue d’histoire des sciences humaines [En ligne], 45 | 2024, mis en ligne le 07 février 2025, consulté le 06 février 2026. URL : http://journals.openedition.org/rhsh/9722 ; DOI : https://doi.org/10.4000/13auz

Haut de page

Auteur

Emily Hauptmann

Western Michigan University

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search