Research Infrastructure in the Behavioral and Social Sciences

If the history of science teaches us anything, it is that an infrastructure is an indispensable adjunct to the efforts of individual researchers. . . . In the social sciences, no less than the natural sciences, theoretical advances occur in conjunction with continuous innovations in collective resources for research (Prewitt, 1985:xv, xxi).

Much scientific research is done with tools and resources that are often referred to as research infrastructure. For instance, the Hubble telescope allows astronomers to study the structure of the universe and the origins of galaxies; extremely high-resolution microscopes allow biologists to study the interior of living cells; longitudinal sample surveys allow social scientists to study changes in the demographic characteristics of the population, changes in attitudes, and changes in behavior; functional magnetic resonance imaging allows behavioral and cognitive scientists to study how the brain responds to certain stimuli.

In 1998 the National Science Foundation's Directorate of Social, Behavioral, and Economic Sciences (NSF/SBE) reiterated its continued support for infrastructure investment and expressed an interest in increasing that investment. NSF/SBE asked a number of organizations, including the Commission on Behavioral and Social Sciences and Education (CBASSE) how much of its research funds should be devoted to infrastructure investments and how the its infrastructure investment process might be improved. In support of NSF/SBE's initiative, CBASSE conducted a workshop in November of 1997 to examine improvements that could be made in the funding of research infrastructure for the behavioral and social sciences. (The agenda and attendees are in Appendix A).

Over the years, CBASSE and its predecessor entities at the National Research Council (NRC) have conducted a number of studies related to collective resources for research infrastructure in the behavioral and social sciences. In 1969 and 1989, CBASSE reviews of social and behavioral science progress laid out research infrastructure priorities. In addition, CBASSE's Committee on National Statistics (CNSTAT) has published many reports on infrastructure issues relating to federal data, including questions of confidentiality, strengths of longitudinal data, and data priorities for specific issues such as aging populations. (A list of CBASSE and CNSTAT's recent, relevant publications is in Appendix B.) Several other NRC studies in other sciences have addressed research infrastructure issues (see, e.g., Institute of Medicine, 1990; National Research Council, 1991).

This short report is based on the papers and discussions from the research infrastructure workshop, CBASSE and CNSTAT reports about infrastructure issues, discussions within CBASSE and CNSTAT, and the research experience of the CBASSE



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 3
--> Research Infrastructure in the Behavioral and Social Sciences If the history of science teaches us anything, it is that an infrastructure is an indispensable adjunct to the efforts of individual researchers. . . . In the social sciences, no less than the natural sciences, theoretical advances occur in conjunction with continuous innovations in collective resources for research (Prewitt, 1985:xv, xxi). Much scientific research is done with tools and resources that are often referred to as research infrastructure. For instance, the Hubble telescope allows astronomers to study the structure of the universe and the origins of galaxies; extremely high-resolution microscopes allow biologists to study the interior of living cells; longitudinal sample surveys allow social scientists to study changes in the demographic characteristics of the population, changes in attitudes, and changes in behavior; functional magnetic resonance imaging allows behavioral and cognitive scientists to study how the brain responds to certain stimuli. In 1998 the National Science Foundation's Directorate of Social, Behavioral, and Economic Sciences (NSF/SBE) reiterated its continued support for infrastructure investment and expressed an interest in increasing that investment. NSF/SBE asked a number of organizations, including the Commission on Behavioral and Social Sciences and Education (CBASSE) how much of its research funds should be devoted to infrastructure investments and how the its infrastructure investment process might be improved. In support of NSF/SBE's initiative, CBASSE conducted a workshop in November of 1997 to examine improvements that could be made in the funding of research infrastructure for the behavioral and social sciences. (The agenda and attendees are in Appendix A). Over the years, CBASSE and its predecessor entities at the National Research Council (NRC) have conducted a number of studies related to collective resources for research infrastructure in the behavioral and social sciences. In 1969 and 1989, CBASSE reviews of social and behavioral science progress laid out research infrastructure priorities. In addition, CBASSE's Committee on National Statistics (CNSTAT) has published many reports on infrastructure issues relating to federal data, including questions of confidentiality, strengths of longitudinal data, and data priorities for specific issues such as aging populations. (A list of CBASSE and CNSTAT's recent, relevant publications is in Appendix B.) Several other NRC studies in other sciences have addressed research infrastructure issues (see, e.g., Institute of Medicine, 1990; National Research Council, 1991). This short report is based on the papers and discussions from the research infrastructure workshop, CBASSE and CNSTAT reports about infrastructure issues, discussions within CBASSE and CNSTAT, and the research experience of the CBASSE

OCR for page 3
--> members themselves. The topic of research infrastructure deserves a much more in-depth investigation than is possible here, given NSF/SBE's schedule to make changes in their investment strategy in research infrastructure this year. This report, therefore, is limited to a discussion of the processes to improve the allocation of scarce funds for research infrastructure in the behavioral and social sciences. This report: --   discusses briefly the definitions and dimensions of behavioral and social science infrastructure; --   summarizes the modest evidence on trends in the investments in infrastructure; --   recommends a selection process and criteria for deciding among infrastructure proposals; and --   suggests changes to facilitate the effective longer-term management of research infrastructure investments for the behavioral and social sciences. The Importance and Role of Infrastructure Background The strategic importance of research infrastructure to the long-term development of the behavioral and social sciences can be illustrated with examples of two of the most important forms of behavioral and social science infrastructure: sample surveys and functional magnetic resonance imaging (fMRI). Scientifically valuable sample surveys are those for which the sample population is of long-term interest and the survey questions have utility for testing theories and hypotheses. A good example of the importance of surveys can be seen in a recent finding from the National Long Term Care Surveys (NLTCS). The NLTCS showed that rates of chronic disability and institutionalization among older people in the United States are falling dramatically: a smaller proportion of older people are disabled, and disabilities among those having functional problems are less severe. In addition, reduction in the rate of disability is actually gaining momentum, even at very old ages (Manton et al., 1997). Longitudinal sample surveys are very important for observing behavior over time. For example, the Health and Retirement Survey (HRS) has shown that households headed by older persons are more likely to give than to receive money (transfers) across generations: 23 percent gave financial gifts of $500 or more to at least one child or grandchild in the preceding 12 months, but only 2 percent received financial assistance from children over a year's time (Population Reference Bureau, 1996). Both of these examples reveal important findings that would have been impossible to discover without the basic infrastructure of sample surveys.

OCR for page 3
--> The ability to “image” the structure and functioning of the brain is a fundamental advance that is being vigorously applied to the study of cognition. FMRI is revealing subtle and intriguing structural property changes and abnormalities that occur in disease and aging. Brain imaging techniques can be used to measure such functions as energy metabolism and activity, with studies done while subjects are actively involved in specific tasks or at rest. Such studies shed light on brain functioning during periods of severe dysfunction or relative remission, during treatment with medication, and so forth. Studies of cognitive functioning are revealing both specificity and generality of function: in some cases, specific brain sites are now identified with functions, while in others imaging is showing the distributed properties of functions. Definitions Although most researchers directly and indirectly depend on various kinds of infrastructure for their work, there is no formal or comprehensive definition of research infrastructure for the behavioral and social sciences (Johnson, 1997). The workshop participants spent considerable time debating the definition of infrastructure. NSF/SBE defines infrastructure in two main categories. One is multidisciplinary centers: They provide an opportunity for bringing together a critical mass of experts interested in common problems such as violence, environmental decision making, or cognitive science. A novel variation on the traditional center is the “virtual center” which is possible via the Internet (Bertenthal, 1998:3). Second is the much broader category of scientific instrumentation, with the subcategories of: (1) research instrumentation, tools that cost more than $10,000 and (2) research equipment, tools that cost under $10,000. There are at least five subcategories of research instrumentation (Bertenthal, 1998:46). (1)   platforms and observational systems (e.g., neural imaging equipment, observational coding systems); (2)   computational systems (e.g., supercomputers, mass storage devices, visualization systems); (3)   laboratory and analysis systems (electron microscopes, statistical software, image processing); (4)   communications and network systems (vBNS, Internet); and (5)   information systems and databases (digital libraries, large surveys). The workshop participants discussed other dimensions and alternatives to the NSF/SBE

OCR for page 3
--> classification of infrastructure that would include four major components: social, communicative, mechanical, and intellectual (see Johnson, 1997). Social infrastructure includes the resources needed to promote research collaboration within and among other fields. Various government agencies provide some of this infrastructure by promoting collaboration through joint requests for proposals, interdisciplinary meetings, and travel. Interdisciplinary centers for research are also a critical part of social infrastructure. Closely aligned with social infrastructure is communicative infrastructure, which includes the Internet and the many other forms of electronic communications. Workshop participants discussed the new availability of scientific reports and some academic journals in electronic formats. Dissemination of research through print journals as well as through professional meetings and specialized briefings is also part of this kind of communicative infrastructure. Mechanical and intellectual components of infrastructure come closest to the definitions that NSF/SBE uses. Mechanical infrastructure includes such equipment as magnetic resonance imagers and positron emission tomographers, as well as major computer equipment, including supercomputer systems (Edwards, 1997). It also includes many kinds of specialized equipment that different behavioral and social scientists use in their research, such as the equipment anthropologists use to date their field samples and the equipment that linguists use to record and group sounds. Much of the workshop discussion focused on intellectual infrastructure, which is similar to what NSF/SBE defines as instrumentation/information systems and databases. These databases are critical for a vast amount of the research in both the behavioral and social sciences. Another part of the intellectual infrastructure is the methodological developments essential to the sophisticated analysis of data collected. The development of methodological tools, such as statistical computer programs, facilitate data analysis (Eddy, 1986). “Methodology differs from other aspects of science in that if it remains separate from the rest of the science it serves, the science will atrophy or at least continue to be limited by methodological constraints that in fact have solutions” (Johnson, 1997:4). Other definitions of infrastructure come from the offices at the National Science Foundation and the National Institutes of Health that fund infrastructure activities across many sciences. The Office of Science and Technology Infrastructure (OSTI) and the National Center for Research Resources (NCRR) have similar definitions of different aspects of infrastructure, but they give different emphasis to different forms. (Appendix C summarizes their definitions of infrastructure). OSTI and NCRR generally have not funded infrastructure in the behavioral and social sciences in the past. Both existing and other possible definitions of research infrastructure are not mutually exclusive. For example, research centers and data archives may have tools and electronic

OCR for page 3
--> networking network infrastructure embedded in them. And, over time, infrastructure needs--and definitions--are likely to change with advances in the sciences, such as the development of new kinds of computers for geographic information systems and new magnetic resolution imagery for studying the brain. With regard to change, a major question arises in the case of databases because of their complexity and the need for historical consistency. “The costs of change would be disproportionately borne by the administrators, while the costs of maintaining the status quo are disbursed across the research community” (Smolensky, 1997:3). Issues of change and cost are also involved in archiving databases so that they are accessible to the research community. In 1962 the Inter-University Consortium for Political and Social Research (ICPSR) began archiving political and social digital data and making it available to member organizations. ICPSR and other archives face the issues of developing accessible documentation, ensuring the integrity of the data archived and the migration of the files from one electronic medium to another as technology changes (Vavra and Rockwell, 1997). International databases face additional problems of making the data comparable despite differences in national data collection methods and definitions. Another problem is that many datasets in other countries are not publicly available. “There is no international body, or union of national bodies, which supports such an infrastructure [of public use data tapes]” (Smeeding, 1997:13). The changes in the process of selecting infrastructure investments recommended in this paper would apply to any definition of infrastructure the NSF/SBE wants to use. If NSF/SBE accepts the recommendation to change the current selection process for infrastructure proposals, they will also want to reconsider their definition of what constitutes infrastructure proposals that would be covered by the new process. Trends in Infrastructure Investments There is an implacable law of the economics of knowledge, first stated as far as I know by the German physicist Max Planck; each incremental unit of new knowledge costs more than the last … for a constant increment of gain at the frontier we … find ourselves allocating more and more resources (Kennedy, 1997:11). Workshop participants spent some time discussing past trends in infrastructure investments. The Methods, Measurement, and Statistics program in NSF/SBE had twice the number of awards in 1996 as the Methods, Measurement, and Data Resources program did in 1980, but it did not have twice the funding. Although the two programs are not strictly comparable (see Levine, 1997), the rough comparison is suggestive because the needs and costs of behavioral and social science infrastructure have been increasing over time. Looking at the first 65 years of this century, a survey by Karl Deutsch and his colleagues found that only about 25 percent of the major scientific advances in the behavioral

OCR for page 3
--> and social sciences required intensive capital investments in the first three decades of the century; in the next 35 years, 62 percent of the major scientific advances rested on intensive capital investments (Deutsch et al., 1971). The perception of social science work as cheap--a notion that is widespread among laymen and some university administrators--seems based on the experiences before 1930, when only one-fourth of all major social science contributions required major amounts of capital. Since 1930 more than three-fifths of all contributions have required relatively large amounts of capital, particularly for survey research and large scale tabulations, and this proportion seems likely to increase in the future. If explicit quantitative results are desired, the requirement for capital support becomes still stronger. Low-budget research, the work of lone individuals, or work on nonquantitative topics may play a smaller and smaller role. The industrial revolution in the production of knowledge has not only reached a large part of the natural sciences but has reached the social sciences as well (P.457). The workshop participants agreed with Featherman (1997) that the Deutsch article needs to be updated to give a more current analysis of behavioral and social science trends in research infrastructure. There are only a few studies that have looked specifically at infrastructure issues in behavioral and social sciences. Almost 30 years ago, the Committee on Science and Public Policy (of the National Academy of Sciences) and the Social Science Research Council (1969) published a study of outlook and needs in the behavioral and social sciences. Much of the research infrastructure needs specified in the report focused on data issues. But the study also used a questionnaire of academic departments in universities to estimate the cost of research equipment per full-time-equivalent faculty member and by discipline within the behavioral and social sciences. (Psychology had the highest per faculty cost; political science and history had the lowest.) The report also estimated the space requirements of behavioral and social science academic departments and the cost of computer service per department. (Agricultural economics departments had the highest computer cost per department and anthropology had the lowest.) The report estimated that costs for research equipment would increase 56 percent during the next decade. In a study of research frontiers 10 years ago, experts in 30 different areas of behavioral and social science research described not only recent scientific advances, but also what would be needed to ensure future progress (Luce et al., 1989). The most frequently mentioned recommendations were for more infrastructure, such as computer resources, computerized data bases, longitudinal surveys, and interdisciplinary research centers. In no discipline was the existing infrastructure considered to be adequate for future scientific work. More recently, specific information on selected research infrastructure comes from three separate NSF surveys--on academic facilities, research instruments, and research