Sharing Research Data (1985) / Chapter Skim
Currently Skimming:
Sharing Research Data in the Social Sciences
Sharing Research Data in the Social Sciences
Pages 39-88
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 39... ...
Traugott are at the Inter-university Consortium for Political and Social Research, Center for Political Studies, Institute for Social Research, University of Michigan. An earlier draft of this paper was discussed at length by Stephen Fienberg, Clifford Hildreth, Margaret Martin, Miron Straf, Joe Cecil, and Terry Hedrick.
|
From page 40... ...
It is only in very recent years, for example, that concerted efforts to develop bibliographic control over computer-readable data collections have begun, and there is as yet no centralized reference service for computer-readable social science data. Failure to move more rapidly toward acceptance and implementation of the principle of open access to basic data is sometimes asserted to be a reflection of the supposed transitional nature of the social sciences—from essentially literary values, with their emphasis upon private and unique individual creativity, to the scientific values of public and cooperative pursuit of cumulative knowledge.
|
From page 41... ...
The second section provides an indication of the magnitude of data sharing that now occurs. The third section considers technical obstacles to generalized access to basic data in usable form and suggests means by which some of these obstacles might be overcome.
|
From page 42... ...
At the same time, however, researchers faced high costs for data collection arid for processing data to computer-readable form, uneven access to computational facilities and capabilities among social scientists, and the possibility and value of multiple uses of data collections. Hence the early literature emphasized need for mechanisms that would facilitate generalized access to data and to computational capabilities required for their use.
|
From page 43... ...
These facilities were viewed, in some cases, as functioning analogously to the laboratories and the research installations of the physical sciences. They would provide mechanisms to implement the obligations of original data collectors to share their data with other researchers.
|
From page 44... ...
However, there are also occasional reports of fraudulent research, some of them with continuing and even dire consequences. For these reasons the opportunity for verification using original data is often seen as a vital element of the research process and as dictating generalized access to data.
|
From page 45... ...
Secondary Analysis The value of data collections for extended, or secondary, analysis is, of course, frequently discussed. The research potential of a welldesigned data collection is rarely exhausted by the original data collector, and data collections usually have value beyond those for which they were originally designed.
|
From page 46... ...
By using original data in this fashion, students are able to more directly experience the research process and come to better understand the empirical bases and the contingent nature of research findings. In a more general sense, instructional use of empirical data improves social scientific and numeric literacy and enhances students' critical capacity to evaluate the results of applications of social science methods, whether reported in scholarly publications or in the mass media.
|
From page 47... ...
There is widespread agreement that the former category of data should be shared and made generally available in a timely fashion, although there is less agreement as to what constitutes "timely." Sharing smaller data collections, particularly those created at individual expense, is often seen as less important, and obligations to provide access to such data are considered less pressing. These distinctions seem to be based on the presumed lesser value of smaller data collections for the purposes of secondary analysis, the sources of financial support for data collection, and the greater ease and lower cost at which smaller data collections can be duplicated.
|
From page 48... ...
Such data collection efforts usually involve large investments of time and energy, and to duplicate them is obviously wasteful. Of greater importance, data collections of this sort often draw on multiple sources, some of which may not be easily accessible, and often use complex derived measures and aggregations.
|
From page 49... ...
Indeed, one indication of the importance of data sharing is the development in the United States and other nations during the past two decades of numerous organizations that serve as mechanisms to provide general access to the basic data of social science research. These facilities include national indeed, international "social science data archives" in the academic sector, venous private organizations that provide access to data, as well as organizations that maintain and disseminate data collected by government agencies.
|
From page 50... ...
The size of ICPSR data holdings is a concrete indication of the willingness of researchers to share data. The data holdings include virtually all forms of social science data and span much of the spectrum of social science research.
|
From page 51... ...
Data Collections as National Resources A further indication of the incidence of data sharing is of a different order. In recent years research funding agencies have supported several major data collection efforts that are explicitly designed to serve the research interests of extended communities of scholars rather than those of individual researchers or research groups.
|
From page 52... ...
Information about the use of the third and fourth data collections noted above is more limited. ICPSR, however, has furnished well over 1,000 copies of specific files from the General Social Survey series to various institutions, and the Roper Center for Public Opinion Research, which also distributes the data, has supplied additional copies.
|
From page 53... ...
Nothing approaching complete information is available, and it is certain that these illustrations provide only a very partial indication of the incidence of data sharing and of multiple applications of shared data collections. Taken in total they strongly suggest, however, that data sharing has become an important mechanism to support research and teaching in the social sciences.
|
From page 54... ...
By prerecording this kind of descriptive information in computer-readable form in a database dictionary, the actual retrieval and analysis of data is greatly simplified. Indeed, the development of database dictionaries, begun in the late 1960s, stands as an important innovation in facilitating ready access to and use of large and complicated data collections.
|
From page 55... ...
and Roistacher et al. (19801; the problems arise because the original data collectors and processors are not aware of the existence of the standards or they are simply not followed.
|
From page 56... ...
Data Access We argue here that technical obstacles to data sharing are largely related to the practices of original data collectors and processors rather than to He peculianties of computers and data processing equipment. We have referred, however, to data sharing that involves actual transferral of copies of data collections, whether directly from one researcher or installation to another or through an
|
From page 57... ...
Secondary analysis often requires that researchers combine data from diverse data collections to create a new data collection designed for new research goals. The ready availability of data collections means that researchers can carry out exploratory analyses to design new data collection efforts, to assess the efficacy of particular measures and questions, and to perform preliminary tests of hypotheses.
|
From page 58... ...
Sharing data may be desirable, it may contribute to the development of knowledge, and it may facilitate the research of others, but it has no place on the curriculum vita. In fact, data sharing may hurt: premature release of data may allow another to publish it first, and any sharing deprives the original investigator, and perhaps students and colleagues, of long-term opportunities to mine data collections.
|
From page 59... ...
In these situations, more adequate research support would speed both processes.4 It is also sometimes argued that funding is adequate to support data collection and analysis but insufficient to support the documentation, cleaning, and processing of data to forms adequate for use by secondary analysts. As suggested above, however, adherence to basic standards of data preparation from the beginning of data collection would probably reduce rather than increase costs and would produce data collections adequate for secondary analysis.
|
From page 60... ...
appropriate steps to accord recognition to original data collectors and to ensure that they obtain the benefits of initial publication. Misuse of Scientific Data Another area of value conflict involves the possible misuse of scientific data.
|
From page 61... ...
It is unlikely that social scientists can achieve open and general access to such data. But if a data-sharing no was more fully institutionalized within the social sciences, such firms might be encouraged to provide at least limited access to their data, perhaps in the form of `'public-use tapes," for social science research.
|
From page 62... ...
Another approach involves provision of custom data reductions and analyses: for example, some organizations maintain confidential data collections and provide, to user specifications, subsets of data, summary measures, or analytic results that do not allow identification of indi
|
From page 63... ...
Since the original data collectors bear the costs of maintaining data collections, they suffer at least the distractions involved in honoring requests for data. If requests for data are numerous, those distractions may become intolerable and, for that reason, the data may become unavailable or may not be preserved for extended periods.
|
From page 64... ...
These include nationally oriented social science data archives in the academic community, which function in more or less general-purpose fashion in that they are oriented toward several or all social science disciplines. A number of agencies of the federal government also have data centers that maintain, manage, and disseminate data produced by those agencies.
|
From page 65... ...
Consequently, they are well integrated into the research community. They also relieve onginal data collectors of the burdens of maintaining and supplying data to oth
|
From page 66... ...
Intermediary facilities also have disadvantages, some of which were alluded to above: the overhead expenses required to maintain them; their distance from the original data-collection process; and their intermediary nature itself, sometimes interpreted as posing bamers between original data collectors and others with whom data might be shared. But at this point the advantages for data sharing of intermediary facilities seem to greatly outweigh their disadvantages.
|
From page 67... ...
The range of data resources and the conditions under which they are available are highly varied, but at least two facilities- one on the sun and one in medicine appear markedly similar to the social science data archives described above. For physicists and astronomers interested in data on the sun, there are a variety of data collections available from the World Data Center A for Solar-Terrestrial Physics in Boulder, Colorado.
|
From page 68... ...
The Laboratory Animal Data Bank (LADB) is a second example of datashanog facilities of this sort.
|
From page 69... ...
Very little sharing of the data on which research reports are based seems to occur, and data sharing is not widely advocated as a desirable or necessary practice. While nearly all biomedical researchers would agree in principle to make basic data available to other researchers, the practice is seemingly rarely followed.
|
From page 70... ...
They also suggest both similarities and differences between the social and other sciences. In numerous scientific areas there appears to be widespread interest in the development of data centers to collect, maintain, and provide access to basic data, and a number of such centers seem similar, on superficial examination, in many essential functions to the dat~archives and facilities of the social sciences.
|
From page 71... ...
In the social sciences emphasis is placed on sharing data to allow their use for secondary analysis in other words, for new research applications. In the other sciences it appears Mat heavier, although not exclusive, emphasis is placed on amassing data collections to serve as base-line data against which researchers can compare data that they have collected through their own experiments and observations.
|
From page 72... ...
Citation practices could be improved to provide better recognition of original data collectors. Secondary analysts could be expected to provide complete citations of the data collections used and to acknowledge the original data collectors; journal editors might require such citations as a condition of publication.
|
From page 73... ...
The advantages of data-shanng organizations are several: they remove the burdens of supplying data from original data collectors; they maintain data collections and so the cumulative and combinatorial values of data are more likely to be realized; and they cross disciplinary boundaries so that interdisciplinary use of data is facilitated. Data that threaten the privacy of individuals and the rights of organizations pose special problems.
|
From page 74... ...
In the case of data collections supported by government funding agencies, sponger action is possible. Item 754 of the National Science Foundation's Grant Policy Manual "Rights in Data Banks and Software," is a significant step toward stating a general norm of data shanng.
|
From page 75... ...
There is also no indication of expectation that data would be conserved for any extended period to allow realization of the cumulative value of data collections. Obviously we cannot expect a policy Hat specifies precise procedures for all occasions.
|
From page 76... ...
9. A number of the social science data archives mentioned above and in the appendix perform related functions beyond those of storing, processing, and disseminating machine-readable data.
|
From page 77... ...
Madison, Wisconsin 53706 Data Archives Library, Institute for Social Science Research, 1101 Gayley Center, 405 Hilcard Avenue, University of California, Los Angeles, California 90024 Data Bank, Institute for Behavioral Research, York University, 4700 Keele Street, Downsview. Ontano, Canada Data Library, 6356 Agricultural Road, Room 206, University Campus, University of British Columbia, Vancouver, British Columbia, Canada V6T 1W5 Data Library, Survey Research Center, University of California, Berkeley, California 94720 Data Resources, Inc., 29 Hartwell Avenue, Lexington, Massachusetts 02173 Data User Service Division, Bureau of the Census, U.S.
|
From page 78... ...
Box 1113, Palo Alto, California 94302 Public Opinion Survey Unit, University of Missouri, Columbia, Missouri 65201 Roper Public Opinion Research Center, Box U-164R, University of Connecticut, Stores, Connecticut 06268 Social Data Exchange Association, 229 Waterman Street, Providence, Rhode Island 02906 Social Science Computer Research Institute, 621 Mervis Hall, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 Social Science Data Archive, Laboratory for Political Research, 321A Schaeffer Hall, University of Iowa, Iowa City, Iowa 52240 Social Science Data Archive, Survey Research Laboratory, 414 David Kinley Hall, Urbana, Illinois 61810 Social Science Data Archive, Box 596, University of Notre Dame, Notre Dame, Indiana 46556 Social Science Data Archives, Department of Sociology and Anthropology, Carleton University, Colonel By Drive, Ottawa, Ontario, Canada K1S SB6 Social Science Data Center, University of Connecticut, Stom, Connecticut 06268 Social Science Data Center, University of Pennsylvania, 353 McNeil Building, CR, 3718 Locust Walls, Philadelphia, Pennsylvania 19104 Social Science Data Library, Manning Hall 026A, University of Norm Carolina, Chapel Hill, Norm Carolina 27514 Social Science User Service, Princeton University Computer Center, 87 Prospect Avenue, Princeton, New Jersey 08540 Social Security Administration, Office of Research and Statistics, Room 1120,, Universal North Building, 1875 Connecticut Avenue, N.W., Washington, D.C.
|
From page 79... ...
1976 The historian and social science data archives in the United States. American Behavioral Scientist 19:419 442.
|
From page 80... ...
Bryant, F.B, and Wortman, P.M. 1978 Secondary analysis: the case for data archives.
|
From page 81... ...
1971 The relation of social science data archives to libraries and wider information networks.
|
From page 82... ...
Social Science Information 8:153-158. 1969b The future development of social science data archives in Latin America.
|
From page 83... ...
Hofferbert, R.I., and Clubb, J.M., eds. 1976 Social science data archives: applications and potential.
|
From page 84... ...
1966 A social science data archive for Asia, Africa and Latin America.
|
From page 85... ...
International Social Science Journal 16:49 62. 1965 Second conference on data archives in the social sciences, Paris, 2~30 September 1964.
|
From page 86... ...
1963 Conference on data archives in the social sciences. Social Science information 2:109-114.
|
From page 87... ...
1967 Social Implications of Social Science Data Archives. Technical Memorandum 379/000/00.
|
From page 88... ...
Voss, P.R. 1977 Population data in social science data archives: the survey holdings of the Roper Public Opinion Research Center 14:141-144.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.