in the long view, “we should have confidence in the fact that eventually science is self-correcting.” It may take a long time, during which multiple and competing analyses coexist. “The anxiety that we feel about many different people looking at the data and coming up with different interpretations is somewhat misplaced. I do not think we should be afraid of that future.”
Other workshop participants worried that researchers will not have the time to rebut all misconceived analyses. Incorrect information can have extremely harmful results, observed Robert Califf, Duke University Medical Center. “There will be consequences of people being killed by poor use of data because, if it hits the news, a lot of people will stop taking their medications … based on what is in the news,” he said. “You can kill people with bad science very quickly. It is a problem that we are going to have to grapple with.”
Kelly Edwards, acting associate dean at the graduate school and associate professor in the Department of Bioethics and Humanities at the University of Washington, described some of the cultural barriers to greater data sharing—along with several ways of overcoming those barriers.
These arrangements may be necessary, but they are insufficient, said Edwards. Trust also needs to be relational, with contracts serving as a way to punctuate what has already been agreed to rather than the sum total of how a relationship will work. Different elements enter into relationship trust. In some cases, people share core values and interests or are committed to a common cause. Someone may have another person’s best interests in mind, as in the doctor–patient relationship. Relational trust can be built on transparent and consistent (or logical) rules, and trust can