Medical journals usually ask researchers to make their raw data available to others, to facilitate research, and keep fraud to a minimum. But many researchers are ignoring the request, says Pharmalot's Ed Silverman. According to a recent study in PLoS One, there is a wide variation in data sharing policies, and researchers tend to interpret those policies in their own ways, Silverman says. The study authors reviewed hundreds of papers from 50 high impact medical journals, and found that 22 of the 50 journals required sharing of raw data as a condition of publication. In addition, another 22 of the journals encourage data sharing, without making it a condition of publication. When the researchers examined specific papers in those journals, they found that 149 papers weren't subject to any data-sharing policy, and 208 "did not fully adhere to data availability instructions," Silverman says. The remaining 143 papers the researchers studied "did follow data availability instructions [and] did so by publicly depositing only the specific data type as required, making a statement of willingness to share, or actually sharing all the primary data," he adds. The researchers also found that scientists are reluctant to voluntarily share their data, and suggested that more journals adopt clear and precise data-sharing procedures and make sure the policies are consistently followed.