We've been doing a lot of talking about Data Management Plans at MPOW, and I recently came across the following article from PLOS One: A funder-imposed data publication requirement seldom inspired data sharing.
In it, the authors take a look at 315 research projects that had been funded by the Exxon Valdez Oil Spill Trustee Council (EVOSTC) between 1989 - 2010 to see how many data sets could be obtained. This is an interesting funding agency, as it funded research from many different disciplines, including the Social Sciences. As you might guess, the number of retrieved data sets is pretty darn low (26%), but the single biggest reason was that the authors were unable to reach the original researcher. I guess that shouldn't surprise me as much as it did. The other thing I found surprising is that they found slightly lower retrievability for research that was conducted after the EVOSTC enacted a formal data policy in 1995. I guess that's the sound bite for the article title!
But the other thing we've been talking about at MPOW is what kind of reward might entice researchers to post their data, and this article addressed that as well:
"Inherent in all of the above hurdles to data recovery is the absence of a reward structure for sharing data. There is no reward for the time investment required to learn how to curate data or construct data packages (including well documented and formatted error-free data, with project and file level metadata) in cases where original data curation was insufficient. The focus in science has traditionally been on production and citation of publications, and as such, a process for identification and citation of manuscripts has been well developed. Similar attribution and recognition of data would incentivize the archiving and sharing of data in the scientific community. One way publications are tracked and cited is through the use of digital object identifiers (DOIs), a tool that is also increasingly being used to attribute data. DOIs are particularly important for data identification because unlike manuscripts, data can be regularly updated or exist in multiple formats or subsets, so identifying specific versions of the data via a DOI is key to data proper data attribution and use. Although the use of data DOIs is a relatively new practice, they should facilitate more routine data citation as compared to traditional methods. Incorporating data citations into a scientists’ overall research output alongside journal publications should further incentivize data sharing."
So they're making some claims that they can't back up with this research, but which certainly makes sense. The one thing that might get researchers to post their data sets is better recognition in the tenure and promotion process of the work that's involved in making this information available online.
As a snarky aside, I wonder if the EVOSTC website has seen a redesign since it was first launched? :-0