The free and open sharing of scientific information, data, and materials is a fundamental ideal underlying the social structure of science. The importance of data sharing is clearly illustrated by the inscription on the statue of Albert Einstein in front of the National Academy of Sciences: "The right to search for truth implies also a duty: one must not conceal any part of what one has recognized to be true." The case presented by Next Wave illustrates the powerful incentives that work to undermine the practice of sharing in science. In this case, protecting one's scientific priority (and that of a student) led to data withholding. However, in withholding data, researchers may unknowingly jeopardize several fundamental characteristics of the scientific enterprise.
This essay is divided into three sections. Initially, we discuss the importance of sharing to three key aspects of science. Then, we present the results of our research into the nature, extent, and consequences of data sharing and withholding in the genetic and the life sciences. Finally, we briefly explore the implications of our research for science policy and practice.
The Importance of Sharing in Science
At least three fundamental characteristics of science are based on data sharing. First, the prompt and widespread dissemination of data about ongoing research is one of the most efficient ways to alert scientists to the availability and comparative potential of new areas of research. Failure to disseminate data promptly and widely may result in fields of research that never reach their maximum potential, either because scientists repeat each others' work unnecessarily or because research opportunities go unexplored. These negative consequences of withholding on scientific progress are clearly illustrated by the Jones-Smith-Montgomery case study provided.
Second, data sharing is essential to maintaining the integrity of science. Publications, posters, abstracts, and--increasingly--presentations represent scientists' claims of truth to the scientific community. Often such claims are independently tested and ultimately verified or rejected by other scientists. It is this process of replication, along with peer review, that makes science self-correcting and ensures the accuracy of the scientific record. Data withholding potentially limits this fundamental quality control process when authors refuse to share unique resources (such as Montgomery's Abc1 antibody) necessary to confirm scientific claims.
Third, data sharing is vital to training the next generation of scientists. When their colleagues withhold data, young investigators' exposure to the most current ideas, materials, and information may suffer, and their skills may not be developed as fully as possible. Equally important, these young scientists may be more likely to withhold data themselves in the future and pass on a culture of secrecy to their future trainees.
Findings Related to Data withholding in Genetics
Until recently, little empirical evidence existed regarding the impact of data sharing and withholding on these and other aspects of science. To address this issue, we conducted a national study of scientific secrecy in academic genetics and the other life sciences. Our study consisted of a mailed survey of 2893 life science faculty in the 100 most research-intensive universities in the United States. The response rate was 64%, and the results were published  in the Journal of the American Medical Association in January 2002.
Our survey found that data withholding is fairly common in academic genetics. In the last 3 years almost half of academic geneticists (47%) had been denied access to published information, data, and materials by other academic scientists. However, only 12% of scientists reported engaging in such behavior.
Biomaterials were the resource most commonly withheld. Of people seeking access to biomaterials referred to in published work (such as the cell lines, tissues, antibodies, and reagents), 35% were denied access to these materials at least once in the last 3 years. Other forms of data that were less frequently denied were unpublished phenotypic information, information about lab procedures, and pertinent findings not included in a paper.
The motivations most frequently cited by investigators who withheld data were that sharing required too much effort (80%) and that scientists needed to protect the ability of a graduate student, postdoctoral fellow, or junior faculty member to publish (64%). About half (53%) denied requests for data in order to protect their own ability to publish in the future. Nearly half (45%) withheld data because of the financial costs of providing the requested information or materials.
Data withholding has a negative impact on each of the three fundamental aspects of science mentioned above. Almost three-quarters (73%) of geneticists reported that data withholding slowed the progress of research in their field of science, and 68% reported that data withholding had a negative effect on the progress of their own research. More than half of geneticists (56%) reported that data withholding reduced their ability to educate students and postdoctoral fellows, and 28% said they had been unable to replicate published research as a result of data withholding.
Implications for Policy and Practice
One critical question from a policy perspective is whether research progress is as rapid as it could be if data sharing were maximized. Our studies suggest that the answer is no. Almost three-quarters of geneticists in our research said withholding was slowing progress in their field. Some causes of data withholding, such as the struggle for priority, may be extremely difficult to change, but others identified by our research may not be. Scientists are most likely to be denied access to biomaterials, and the most commonly cited cause is the costs involved in the transaction. Recently, the federal government announced a draft statement that directly addresses this issue by offering explicit funding in research grants to support the costs of sharing biomaterials and other forms of data (see the National Institutes of Health's Draft Statement on Sharing Research Data ). Other organizations that support research should consider similar policies.
A second policy question concerns the impact of data withholding on trainees in science. As the case and our study demonstrate, students are likely to observe and experience the impact of the data sharing and withholding behaviors of faculty. Because many of these students lack experience in science, they may automatically model the data sharing and withholding practices of faculty without fully understanding the reasons and potential implications of such behaviors. Providing students with a forum to explore all aspects of this issue, whether in formal settings such as classes and seminars or in the laboratory, may be beneficial.
A scientific world completely free of data withholding is probably unachievable and, indeed, may be undesirable. Some types of data withholding are necessary to ensure that investigators get the credit that they deserve and that partly motivates their effort. However, research and policy development should optimize scientific progress by identifying and combating the remediable causes of data withholding (such as those due to a lack of funds). Furthermore, when faced with decisions about sharing, scientists should consider seriously the broader impact of their decisions on the fundamental characteristics of the scientific enterprise and, when at all possible, err on the side of openness. Doing so may increase the likelihood that our system of science will function to its maximum capacity in the years to come.
David Blumenthal, M.D., M.P.P. is director of the Institute for Health Policy and a physician at the Massachusetts General Hospital/Partners HealthCare System in Boston, Massachusetts. He is also a professor of medicine and a professor of health care policy at Harvard Medical School.
Eric Campbell is an instructor of health policy at the Institute for Health Policy at the Massachusetts General Hospital and Harvard Medical School.