Open access is one of the most exciting trends in technical science publishing. Open access journals make their technical articles free and open to the public.
Most technical science journals are not open access. As a general rule, they require a subscription or some other fee for access to the articles, unless the corresponding author of a given article forks over a lot of money upfront to reimburse the publisher.
The intent of open access is to disseminate scientific knowledge quickly, regardless of one's financial resources. The overall goals are to speed up the pace of scientific discovery, and introduce democracy and transparency into a science publishing system that is often closed and restrictive.
Open access journals explicitly require that the authors of the articles make their data open to other scientific investigators. Not doing so is a direct violation of the open access agreement, and fundamentally hinders the success of the open access model.
One would hope that authors of technical articles in open access science journals comply with this requirement. Unfortunately, Caroline Savage and Andrew Vickers (Memorial-Sloan Cancer Research Center, New York) report that they generally do not, at least for the two technical journals they analyzed.
An evaluation of two journals.
The scientists focused their study on two technical science journals, both published under the Public Library of Science umbrella: PLoS Medicine and PLoS Clinical Trials. They chose these two journals because they are open access journals, and their policies regarding data sharing (required) are quite explicit.
From these two journals, they selected ten articles (total). None of these articles provided original data in the "supporting information" supplement, which means that scientists interested in this data would be forced to contact the authors directly.
The scientists emailed a request for data to each of the ten corresponding authors of the articles in question. In all cases, the request for data was said to be motivated by a need for the original data and personal interest.
Furthermore, the emailed request noted that it was not intended to create undue excessive work for the corresponding author. The emailed request included a promise of coauthorship for any publications that came out of the scientists' use of the data.
If the answer to the request for data was no (or if no reply was given), one of the scientists pretended to be a member of the editorial board of the journal in question, reminding the corresponding author of his or her agreement to the explicit open access policies of the journal. A response was then awaited by the scientists.
Low compliance.
Two of the corresponding authors could not be contacted by email. Of the remaining eight, four explicitly refused to share the original data, and explicitly refused again upon the follow-up email, for a variety of reasons.
Of the remaining four, three did not respond to the first email. Of these three, two didn't respond to the second email either, and one explicitly refused on the second attempt.
Only one author responded to the original email request that he/she was willing to provide data, and requested a data analysis proposal. After receiving such a proposal, this author submitted the original data to the scientists conducting the compliance investigation.
The state of open access publishing.
There's no valid excuse for this lack of compliance with open access policies. If it's too much work for a scientist to compile the original data sets in a meaningful format, this suggests that the original data is unorganized, increasing the probability of flaws in the scientist's interpretation of the data.
On a more malicious note, this may also suggest that the original data never existed. This is clearly even worse than careless unprofessionalism, and would be deliberate scientific fraud.
These open access (non)compliance results suggest that open access is partially failing in its mission of placing specific scientific data into the public domain. I fully support open access in technical science publishing, and would like to see it continue and succeed.
Towards this goal, my personal recommendation would be for technical science journals to institute clear punishment protocols regarding noncompliance with open access agreements (e.g., banning noncompliant scientists from publishing in the journal again). The goal should be to encourage all contributors to the public access model to fully fulfill all of their agreements, and to stop certain scientists from exploiting the generosity of pioneering publications such as those under the Public Library of Science umbrella.
for more information:
Savage, C. J., & Vickers, A. J. (2009). Empirical Study of Data Sharing by Authors Publishing in PLoS Journals PLoS ONE, 4 (9) DOI: 10.1371/journal.pone.0007078