ICPSR: The Founding and Early Years
The following text was excerpted from a personal history written by Erik W. Austin, whose career at ICPSR lasted 41 years and included service as interim executive director. Austin says: "I have written these pages from my recall of events and developments described below. These recollections are my own, and do not reflect 'official' institutional history. Admittedly, some aspects of the Consortium's programs and activities will be slighted due to my tangential involvement in them. Contemporary survey data developments, the details of computational matters, and the Summer Program are three such areas that won't get as much ink as they undoubtedly deserve."
The Inter-university Consortium for Political Research (ICPR, as it was first known) was created in 1962 by political scientist Warren E. Miller as the data dissemination arm of the American National Election Studies (ANESi). It was shortly to become much more than that, but the initial genius of establishing an effective mechanism for sharing a seminal collection of scientific data cannot be overestimated. Theretofore, scientific data from all disciplines were jealously guarded by their developers/creators as private resources, no matter the source of funding that permitted their collection. The impulse at the University of Michigan's Institute for Social Research (ISR) to break with proprietary tradition was both strange and much welcomed in the social scientific community. In hindsight, we can think of this data sharing motive as a prerequisite of the "scientific ethic" of verification, replication, and validation. In the early 1960s, though, the concept of giving access to all interested scholars to one's basic (micro) data was so foreign as to be considered "revolutionary." Miller even likened it, in retrospect, to a violation of basic economic precepts: data were the scientist's capital, and "they weren't about to share their capital."ii
Erik Austin, author of this personal history of ICPSR, talks about the early years of the Consortium.
The history of the U. S. National Election Studies is chronicled elsewhere.iii From accidental and obscure beginnings in 1948iv, the study by 1960 had grown to become the premier assemblage of scientific data for an entire discipline (Political Science). Scholars clamored to get at these wonderful new data, and ISR's Survey Research Center (where the study was conducted) looked for a way to effectively provide access to these data. Recognizing that data preservation and dissemination were not cost-free, Warren Miller and colleagues at the University of Michigan and elsewhere conceived of a collegial ("consortial") mechanism for facilitating this type of data sharing. In the early 1960s, Miller assumed the role of Willy Lomanv and traveled from university to university in the U.S., seeking to convince each university's Political Science Department to contribute funds to support a permanent data dissemination mechanism in the form of an organization located at ISR. This was yet another radical idea—universities giving money to a "competitor" (and a powerful one at that) to support something in the "scientific commons." By the end of 1962, Miller had recruited twenty-onevi additional universities to contribute to the new organization, each paying $2,500 in annual dues/membership fees; by the end of the 1962-1963 academic year, the organization was up and running. [In August 1962, Warren Miller wrote this letter to ISR director Rensis Likert describing how ICPR would be organized. An early description of the Consortium written by Miller also appeared in the journal American Behavioral Scientist in 1963. —Ed.]
Mid- to Late 1960s
The initial focus of ICPR data archiving activities was on the ANES and other contemporary (i.e., 1950s and 1960s) collections of sample survey data, for the U.S. and even other nations. Within three years, however, Miller initiated the collection of another form of empirical data, that of "aggregate" data bearing on political and social phenomena in historical depth. The first of these data collection activities was spawned by the American Historical Association's Committee to Collect the Quantitative Data of American History. (The Committee's mission was soon enlarged to include all historical data.) The object of data collection was county-level election returns for the whole of the U.S. from 1824 onward. Miller raised funds from the National Science Foundation for this effort, which consisted of locating the most "authoritative" sets of election returns available, keying the data into digital form, and providing them to historians and historically minded political scientists. To oversee this "Data Recovery" activity, the AHA committee first recruited political scientist Walter Dean Burnham, then teaching at Washington University in St. Louisvii, and empanelled a committee of political historians to search for and evaluate extant (if competing) election returns. Burnham was succeeded by historian Howard W. Allen (Southern Illinois University), and then, in very early 1966, by another political historian, Jerome M. Clubb of Bowling Green State University. (Both Allen and Clubb took leaves of absence from their universities to come to ISR; Allen chose to return to his teaching position in Illinois, while Clubb remained at ISR after his initial leave period expired.viii) Feeling the need to add staff to this burgeoning enterprise, Clubb put out a call to the University of Michigan's History Department for a graduate student available to work on the project on a temporary, part-time basis. I answered that call.
The West Hospital building was home to ICPSR from 1962 to 1964
ICPR, as part of the Political Behavior Program of the Survey Research Center of ISR from 1962 until 1970, shared physical space with the rest of ISR. From 1962 to 1964, that space was located in West Hospital on the University's medical campus. Even though construction had begun in 1964 on a separate ISR structure (at 426 Thompson Street), the Institute was forced to vacate West Hospital to make way for a Medical School construction project. Temporary space for ISR for 1964-1965 was found in the former Argus Camera factory on 4th and William, in Ann Arbor's historic Old West Side. The original ISR building was completed in the fall of 1965, and ICPR staff (along with the rest of ISR) occupied the building in early 1966, ICPR setting up shop on the 3rd floor. Within three years of its completion, the structure at 426 Thompson was filled to overflowing as ISR continued to grow. One of the fastest-growing units was ICPR, and the Institute decided to relocate most of ICPR to rental space in downtown Ann Arborix. A new office building, the City Center Building on 5th and Huron (across from the old fire station), was chosen, and ICPR moved most of its staff there in 1968. Two floors were rented there, the third and the basement. ICPR's Survey Data Archive and Computer Support Groupx were placed on the third floor, and what became the Historical Archive was given offices in the basement, which it shared with ISR Publications' order fulfillment unit. ICPR would stay in the City Center Building for eight years.
Digitizing Historical Data
For nearly twenty years (1965-1985), the Historical Archive of ICPR/ICPSRxi engaged in a series of data conversion projects to digitize historical source materials and make them available to the scholarly community. Each of these projects was supported by grants from U.S. federal agencies or private foundations. Following the historical election returns collection, the next of these was the keying of basic population and demographic data from the published reports of the U.S. decennial Census from 1790 to 1970.xii Virtually all indicators from the published volumes of 1790-1840 were thus digitized. From the 1850 Census onward, the volume of data available exceeded the funds allotted for the conversion process, and selections had to be made. A committee of historians decided on which items would be digitized, and priority was given to ethno-religious indicators then in vogue in the quantitative political history field.xiii The project director (Jerome Clubb), however, being somewhat of an economic determinist, managed to see that economic indicators as well would be included in the data conversion process: such variables as agricultural and industrial production and employment, as well as government expenditures in a variety of arenas, were thus added to the database of historical census statistics at ICPR.
In 1968, ICPR obtained a grant from the Ford Foundation to digitize all of the Congressional roll call votes of the U.S. Congress from 1789 to the present. The chief source for this collection was a compilation of the roll calls and descriptions of the measures voted upon, assembled under the direction of political scientist Clifford Lord in the 1930s. Lord had overseen a project sponsored by the U.S. government's Works Progress Administration (WPA) during the Great Depression as a means of employing out-of-work academics and white collar workers. Paper tabulations of all the early roll call votes of Congress were stored in a thousand boxes in the Columbia University library. Transshipped to Ann Arbor, these materials became the source material for ICPR's exhaustive collection of Congressional voting records.xiv
Office suites in the new City Center office building were constructed specifically to meet ICPR's needs. In addition to staff offices, three specialized areas were created. One was a room that contained a dozen IBM keypunch machines, used by a staff of professional data entry personnel to key in the election, census, and roll call data. Another was a storage room containing reinforced shelving for the 1,000 boxes of the "Lord Collection" of roll call voting materials. The third special room housed the U.S. Census volumes, on wall-to-wall, floor-to-ceiling shelving; the room doubled as a conference room, with two long conference tables and a dozen chairs.xv The Historical Archive space in the basement of the City Center Building (CCB) was a busy but depressing dungeon, with no views of the outside world, save for the vibrations of the structure when fire engines rolled out of the fire station across the street. Perhaps because of its isolation, the basement of CCB became a very productive environment: nearly three dozen staff occupied the space, and helped prepare more than two million punch cards' worth of social and political data. Contacts with ICPR central (still housed at 426 Thompson Street) were few and far between, and interactions with the rest of ISR were virtually nonexistent.
Graduate student Joanna Tatomir loads cards into ICPSR's sole remaining punch-card machine in February 2011. She was inputting data from a study of retired people conducted in the 1950s that had not been used for publication at the time and had just been put away.
The "industry" of converting information residing on sheets of paper into digital "data" (stored on punched cards) had been developed at ISR before ICPR came into existence. In the 1950s and early 1960s, survey responses from paper questionnaires were hand-transcribed onto 80-column "codesheets" by staff trained to interpret the responses, convert them to numeric codes, and hand-write them into the correct positions on the codesheet. The codesheets were then turned over to professional data entry operators ("keypunch operators"), who converted them into "decks" of punched cards. The cards were then taken to a bulky piece of unit record equipment (called a "card reader") attached to a mainframe computer; the cards were fed into a hopper, the punches on them sensed electronically, and the information from the cards "written out" onto magnetic tape. Specialized computer software then "processed" the information on the magnetic tapes, moving the information from one tape to the next as each software program was executed. Initially the software programs were "run" by trained computer programmers, who customized the software at each run to accord with the specifics of the data being processed.
ISR had a six-person keypunch staff by the mid-1960s, sufficient for the first of ICPR's historical data conversion projects (the election returns collection). As the volume of ICPR data entry grew, and with the move of staff to the City Center Building, a specialized data entry staff was required just for ICPR keypunching. A time-saving method for preparing the information stored on sheets of paper was then developed, to avoid copying millions of numbers onto code sheets. This consisted of drawing lines between columns of numbers and writing notes on copies of the original data sheets, instructing the data entry operators where to put the data on the punch cards, and what to label them. Once "punched," the data (stored in long, narrow boxes of 2,000 punched cards per box) were hauled to ISR or the University's central computing facility for "reading in" and storage on magnetic tape. ICPR soon had to hire a full-time courier/messenger to carry these boxes of punched cards and the tapes they were to reside on to the central facilities. A series of hardy young men (and one woman!) were hired to ferry these boxes and cases of punched cards from the City Center Building to the data input stations at ISR (six city blocks away) or to one of the University's central computing facilities (one mile away, near the Exhibit Museum, or four miles away on the North Campus of the University). [The last of these couriers, whose duties went away with technological changes in computing procedures, hardware, and software, was David Kushner, who transitioned into a job in User Support at ICPSR.]
The work ways of the ICPR staff in the early 1970s were heavily dependent upon the IBM 029 keypunch machine. Not only was that piece of equipment crucial for large-scale data entry by professional keypunch operators, but it was used by most of the data processing staff as well to prepare instructions ("setups") for the computer programs they submitted to mainframe computers. The "keypunch room" at the City Center Building had been heavily soundproofed in an attempt to deaden the noise of a dozen machines clacking away. Entering that room when the 4-8 professional keypunch operators were all working was akin to visiting a factory with its assembly line running full-blast. Conversation was virtually impossible inside that room; several of us tried to time our visits to prepare necessary "setups" during the 15-minute periods each half day when the professionals were taking their required breaks. The professionals themselves relied upon the noise of the keypunch machine chisels perforating the cardboard punch cards to establish a "rhythm" for their work that resulted in average production capacity of up to 200 cards punched per hour. When I was first told of the reliance upon sound made by these professional data entry operators, I was incredulous. The keypunch supervisor who told me this then paused, and said "Erik, you know that all professional keypunch operators have experienced some hearing loss due to the noise of the machines. After twenty years in the profession, keypunch operators are almost completely deaf."
From 1968 to 1972, I was an Assistant Study Director (today's equivalent of a Graduate Student Research Assistant, but without the employer tuition obligation), first in the Survey Research Center and then (after 1970) in the Center for Political Studies. My functional title within ICPR in the mid-70s was "Assistant Archive Director for History." Michael Traugott was the Archive Director, and there were four other Assistant Directors: Robert R. Beattie (International Relations), Barbara Farah (Contemporary Politics), John Stuckey (Instructional Services), and Janet Vavra (Member Services). None of us (Traugott included) had received any formal training in what we were being asked to do: staff management (hiring, training, and occasionally firing employees), grant proposal writing, authoring technical documents on data processing and computational techniques, interacting with sponsor agency personnel, and representing our organization to various audiences and individuals. Our formal training had all been in the substance of the respective academic disciplines or sub-disciplines. We learned on the job by doing these tasks (some of us were better at such tasks than were others of us), and when we fouled up (through inexperience or bad judgment), we were told about it in no uncertain terms—usually by Jerry Clubb. In 1978, Mike Traugott became Director of Special Projects and Resource Development, and I was named Director of Archival Development.
iThe name of this, the longest-running continuous social science survey program in the world, was changed to the "National Election Study"—NES—in 1978.
iiInterview of Warren Miller by Erik Austin on August 4, 1982.
iiiSee, for example, Warren E. Miller, "An Organizational History of the Intellectual Origins of the American National Election Studies." European Journal of Political Research 25, no. 3 (1994), 247-65, and Philip E. Converse and Donald Kinder, "Voting and Electoral Behavior," in James S. House, et al. (eds.), A Telescope on Society: Survey Research and Social Science at the University of Michigan and Beyond (Ann Arbor, Michigan: University of Michigan Press, 2004), 70-97.
ivAlmost as an afterthought, ISR's Survey Research Center scientists added two questions to a national survey on foreign affairs that they were fielding in October of 1948. The two questions dealt with the upcoming U.S. presidential election the following month. The two questions were "Are you planning to vote in next month's presidential election?" and "What party are you likely to vote for, Democratic, Republican, or some other party?" With all the public opinion polls predicting a Dewey (Republican) landslide, the SRC scientists were mildly amazed that the results of their survey showed the presidential race "too close to call." When Truman actually won the closely contested election, the SRC scientists raised funds to conduct a follow-up survey with the same (October) respondents to determine more about the factors underlying voters' choices in that election. ISR sampling expert Leslie Kish frequently proclaimed that ISR was "lucky" to come so close in their survey to the actual election outcome, as the October survey (while nationally representative) consisted of only 662 respondents. Kish attributed ISR's superiority to its area probability sampling strategy, while nearly all public opinion polls at the time used a soon-to-be-discredited "quota sampling" scheme.
vLoman was the traveling salesman in Arthur Miller's play "Death of a Salesman." Warren called himself "Loman" in the oral history interview cited above.
viThe "original" ICPSR member institutions have been the subject of some dispute over the years. Faculty from more than two dozen universities "recalled" that theirs was one of the original members, thus adding up to more than the iconic "Michigan plus twenty-one" original members. Using membership records and contemporary testimony, I discovered that the "original" members of ICPSR were as follows:
viiBurnham brought intimate knowledge of U.S. elections to this project. His Presidential Ballots, 1836-1892 (Baltimore: Johns Hopkins Press, 1955) was a compilation of county-level election returns, and remains a much-cited classic even today. Burnham subsequently taught in the Government and Political Science departments at MIT and the University of Texas at Austin, from where he retired in 2003. Dean Burnham remained a staunch friend and supporter of ICPSR.
viiiClubb gave up a tenured position in the History Department at Bowling Green State University to remain at ICPR/ISR.
ixRemaining at ISR were the administrative offices of ICPR and the staff of the Summer Training Program.
xThe leader of ICPR's Computing group was Gregory Marks. He served in that capacity for over fifteen years, and led the Consortium through several technological "revolutions," including the transition from mainframe computing to mini-computers, and the introduction of personal computers (the first of which he assembled himself from a "Heath-Kit.") He was noted for his work developing the OSIRIS set of mainframe-based computer programs for data management and analysis; OSIRIS was exported to over 200 universities in its day.
xiWhen I joined ICPR, the group I worked in had no formal name; we were just the staff "working on those historical data." In 1968, the name "Data Recovery Group" was applied to us, and in 1970 we became the "Historical Archive."
xiiICPR was able to obtain most of the printed volumes from the censuses of 1900-1950 from the Census Bureau, at no cost to the project. For the nineteenth century source materials, we obtained blowbacks of census volumes that had been microfilmed. The quality of the reproductions, though, was so poor that most of the numbers could not be read. Instead we photocopied needed tabular materials from bound Census volumes from the University of Michigan Library collection of these volumes, which is quite complete.
xiiiThe best-known works from this school of historiography are Paul Kleppner, The Cross of Culture: A Social Analysis of Midwestern Politics, 1850-1900 (New York: Free Press, 1970), and Richard J. Jensen, The Winning of the Midwest: Social and Political Conflict, 1888-1896 (Chicago; University of Chicago Press, 1971).
xivAssigned to manage the historical roll call project was Santa A. Traugott, then a political science graduate student and research assistant for the American National Election Study. Santa went on to become an ICPSR Archival Assistant Director for Contemporary Politics, and then the Director of Studies of the National Election Study in CPS.
xvWhen ICPSR moved back to ISR, we were able to re-create this "library/conference room" facility. It endured in this form until the early 1990s, when the CPS leadership decided they wanted the room for their purposes. The chief of their purposes was to use the space as a conference room, with a decorative panel on one of the walls that had supported shelving for the Census volumes. When we pointed out that the room already WAS configured as a conference room, the reply was "yes, but it's all those books! They're ugly and old, some of them 100 years old!" We were forced to relinquish the room, and scatter the books around in various offices. We moved them twice thereafter, to Borders and then to Perry, but each time had to disperse the books instead of putting them in one central location.