My inclination would be to say that data should nearly always be shared. If you share your data, this lets others check the conclusions you draw from the data, as well as verifying the accuracy of the data against other available sources. They might disprove your arguments, or lead you to improve your arguments, or, if they reproduce your results, they might help prove the validity of your arguments. But in either case, science progresses better, and the decisions made based on the science are more reliable, than if you keep the data secret.
I can see some possible exceptions, for instance where there are concerns about the privacy of research subjects, or where the data was gathered as part of a commercial endeavor that requires that the data be kept proprietary for the commercial project to be viable, or some such. Even there I stress that the exceptions are merely possible; perhaps on balance the data should be shared even then, and in any event, even if there’s a good reason for the data not to be shared, people should view the research skeptically because of the lack of sharing.
Saturday, November 28, 2009
EUGENE VOLOKH ON DATA SHARING AND CLIMATE RESEARCH, though this sentiment applies to all research. This is how peer review works. It is why we know that fusion power is not yet commercially viable, that perpetual motion machines don't and that certain medicines are dangerous or effective.
Posted by Curmudgeon at Saturday, November 28, 2009