Point of View: How open science helps researchers succeed
Abstract
Open access, open data, open source and other open scholarship practices are growing in popularity and necessity. However, widespread adoption of these practices has not yet been achieved. One reason is that researchers are uncertain about how sharing their work will affect their careers. We review literature demonstrating that open research is associated with increases in citations, media attention, potential collaborators, job opportunities and funding opportunities. These findings are evidence that open research practices bring significant benefits to researchers relative to more traditional closed practices.
https://doi.org/10.7554/eLife.16800.001Introduction
Recognition and adoption of open research practices is growing, including new policies that increase public access to the academic literature (open access; Björk et al., 2014; Swan et al., 2015) and encourage sharing of data (open data; Heimstädt et al., 2014; Michener, 2015; Stodden et al., 2013), and code (open source; Stodden et al., 2013; Shamir et al., 2013). Such policies are often motivated by ethical, moral or utilitarian arguments (Suber, 2012; Willinsky, 2006), such as the right of taxpayers to access literature arising from publicly-funded research (Suber, 2003), or the importance of public software and data deposition for reproducibility (Poline et al., 2012; Stodden, 2011; Ince et al., 2012). Meritorious as such arguments may be, however, they do not address the practical barriers involved in changing researchers’ behavior, such as the common perception that open practices could present a risk to career advancement. In the present article, we address such concerns and suggest that the benefits of open practices outweigh the potential costs.
We take a researcher-centric approach in outlining the benefits of open research practices. Researchers can use open practices to their advantage to gain more citations, media attention, potential collaborators, job opportunities and funding opportunities. We address common myths about open research, such as concerns about the rigor of peer review at open access journals, risks to funding and career advancement, and forfeiture of author rights. We recognize the current pressures on researchers, and offer advice on how to practice open science within the existing framework of academic evaluations and incentives. We discuss these issues with regard to four areas – publishing, funding, resource management and sharing, and career advancement – and conclude with a discussion of open questions.
Publishing
Open publications get more citations
There is evidence that publishing openly is associated with higher citation rates (Hitchcock, 2016). For example, Eysenbach reported that articles published in the Proceedings of the National Academy of Sciences (PNAS) under their open access (OA) option were twice as likely to be cited within 4–10 months and nearly three times as likely to be cited 10–16 months after publication than non-OA articles published in the same journal (Eysenbach, 2006). Hajjem and colleagues studied over 1.3 million articles published in 10 different disciplines over a 12-year period and found that OA articles had a 36–172% advantage in citations over non-OA articles (Hajjem et al., 2006). While some controlled studies have failed to find a difference in citations between OA and non-OA articles or attribute differences to factors other than access (Davis, 2011; Davis et al., 2008; Frandsen, 2009a; Gaulé and Maystre, 2011; Lansingh and Carter, 2009), a larger number of studies confirm the OA citation advantage. Of 70 studies registered as of June 2016 in the Scholarly Publishing and Academic Resources Coalition (SPARC) Europe database of citation studies, 46 (66%) found an OA citation advantage, 17 (24%) found no advantage, and 7 (10%) were inconclusive (SPARC Europe, 2016). Numerical estimates of the citation advantage in two reviews range from -5% to 600% (Swan, 2010) and 25% to 250% (Wagner, 2010). The size of the advantage observed is often dependent on discipline (Figure 1). Importantly, the OA citation advantage can be conferred regardless of whether articles are published in fully OA journals, subscription journals with OA options (hybrid journals), or self-archived in open repositories (Eysenbach, 2006; Hajjem et al., 2006; Gargouri et al., 2010; Research Information Network, 2014; Wang et al., 2015; Swan, 2010; Wagner, 2010). Moreover, at least in some cases, the advantage is not explained by selection bias (i.e., authors deliberately posting their better work to open platforms), as openly archived articles receive a citation advantage regardless of whether archiving is initiated by the author or mandated by an institution or funder (Gargouri et al., 2010; Xia and Nakanishi, 2012).
Open publications get more media coverage
One way for researchers to gain visibility is for their publications to be shared on social media and covered by mainstream media outlets. There is evidence that publishing articles openly can help researchers get noticed. A study of over 2,000 articles published in Nature Communications showed that those published openly received nearly double the number of unique tweeters and Mendeley readers as closed-access articles (Adie, 2014a). A similar study of over 1,700 Nature Communications articles found that OA articles receive 2.5–4.4 times the number of page views, and garnered more social media attention via Twitter and Facebook than non-OA articles (Wang et al., 2015). There is tentative evidence that news coverage confers a citation advantage. For example, a small quasi-experimental 1991 study found that articles covered by the New York Times received up to 73% more citations that those not covered (Phillips et al., 1991). A 2003 correlational study supported these results, reporting higher citation rates for articles covered by the media (Kiernan, 2003).
Prestige and journal impact factor
As Sydney Brenner wrote in 1995, ‘‘…what matters absolutely is the scientific content of a paper and…nothing will substitute for either knowing it or reading it’’ (Brenner, 1995). Unfortunately, academic institutions often rely on proxy metrics, like journal impact factor (IF), to quickly evaluate researchers’ work. The IF is a flawed metric that correlates poorly with the scientific quality of individual articles (Brembs et al., 2013; Neuberger and Counsell, 2002; PLOS Medicine Editors, 2006; Seglen, 1997). In fact, several of the present authors have signed the San Francisco Declaration on Research Assessment (SF-DORA) recommending IF not be used as a research evaluation metric (American Society for Cell Biology, 2013). However, until institutions cease using IF in evaluations, researchers will understandably be concerned about the IF of journals in which they publish. In author surveys, researchers repeatedly rank IF and associated journal reputation as among the most important factors they consider when deciding where to publish (Nature Publishing Group, 2015; Solomon, 2014). Researchers are also aware of the associated prestige that can accompany publication in high-IF journals such as Nature or Science. Thus, OA advocates should recognize and respect the pressures on researchers to select publishing outlets based, at least in part, on IF.
Fortunately, concerns about IF need not prevent researchers from publishing openly. For one thing, the IFs of indexed OA journals are steadily approaching those of subscription journals (Björk and Solomon, 2012). In the 2012 Journal Citation Report, over 1,000 (13%) of the journals listed with IFs were OA (Gunasekaran and Arunachalam, 2014). Of these OA journals, thirty-nine had IFs over 5.0 and nine had IFs over 10.0. Examples of OA journals in the biological and medical sciences with moderate to high 2015 IFs include PLOS Medicine (13.6), Nature Communications (11.3), and BioMed Central’s Genome Biology (11.3). The Cofactor Journal Selector Tool allows authors to search for OA journals with an IF (Cofactor Ltd, 2016). We reiterate that our goal in providing such information is not to support IF as a valid measure of scholarly impact, but to demonstrate that researchers do not have to choose between IF and OA when making publishing decisions.
In addition, many subscription-based high-IF journals offer authors the option to pay to make their articles openly accessible. While one can debate the long-term viability and merits of a model that allows publishers to effectively reap both reader-paid and author-paid charges (Björk, 2012), in the short term, researchers who wish to publish their articles openly in traditional journals can do so. Researchers can also publish in high-IF subscription journals and self-archive openly (see section "Publish where you want and archive openly"). We hope that in the next few years, use of IF as a metric will diminish or cease entirely, but in the meantime, researchers have options to publish openly while still meeting any IF-related evaluation and career advancement criteria.
Rigorous and transparent peer review
Unlike most subscription journals, several OA journals have open and transparent peer review processes. Journals such as PeerJ and Royal Society’s Open Science offer reviewers the opportunity to sign their reviews and offer authors the option to publish the full peer review history alongside their articles. In 2014, PeerJ reported that 40% of reviewers sign their reports and 80% of authors choose to make their review history public (PeerJ Staff, 2014). BioMed Central’s GigaScience, all the journals in BMC’s medical series, Copernicus journals, F1000Research, and MDPI’s Life require that reviewer reports be published, either as part of a prepublication review process, or subsequent to publication. Some studies suggest open peer review may produce reviews of higher quality, including better substantiated claims and more constructive criticisms, compared to closed review (Kowalczuk et al., 2013; Walsh et al., 2000). Additional studies have also argued that transparent peer review processes are linked to measures of quality (Wicherts, 2016). Other studies have reported no differences in the quality of open versus closed reviews (van Rooyen et al., 1999; van Rooyen et al., 2010). More research in this area is needed.
Unfortunately, the myth that OA journals have poor or non-existent peer review persists. This leads many to believe that OA journals are low quality and causes researchers to be concerned that publishing in these venues will be considered less prestigious in academic evaluations. To our knowledge, there has been no controlled study comparing peer review in OA versus subscription journals. Studies used by some to argue the weakness of peer review at OA journals, such as the John Bohannon ‘sting’ (Bohannon, 2013) in which a fake paper was accepted by several OA journals, have been widely criticized in the academic community for poor methodology, including not submitting to subscription journals for comparison (Joseph, 2013; Redhead, 2013). In fact, Bohannon admitted, ‘‘Some open-access journals that have been criticized for poor quality control provided the most rigorous peer review of all’’. He cites PLOS ONE as an example, saying it was the only journal to raise ethical concerns with his submitted work (Bohannon, 2013).
Subscription journals have not been immune to problems with peer review. In 2014, Springer and IEEE retracted over 100 published fake articles from several subscription journals (Van Noorden, 2014; Springer, 2014). Poor editorial practices at one SAGE journal opened the door to peer review fraud that eventually led 60 articles to be retracted (Bohannon, 2014; Journal of Vibration and Control, 2014). Similar issues in other subscription journals have been documented by Retraction Watch (Oransky and Marcus, 2016). Problems with peer review thus clearly exist, but are not exclusive to OA journals. Indeed, large-scale empirical analyses indicate that the reliability of the traditional peer review process itself leaves much to be desired. Bornmann and colleagues reviewed 48 studies of inter-reviewer agreement and found that the average level of agreement was low (mean ICC of .34 and Cohen’s kappa of .17) – well below what what would be considered adequate in psychometrics or other fields focused on quantitative assessment (Bornmann et al., 2010). Opening up peer review, including allowing for real-time discussions between authors and reviewers, could help address some of these issues.
Over time, we expect that transparency will help dispel the myth of poor peer review at OA journals, as researchers read reviews and confirm that the process is typically as rigorous as that of subscription journals. Authors can use open reviews to demonstrate to academic committees the rigorousness of the peer review process in venues where they publish, and highlight reviewer comments on the importance of their work. Researchers in their capacity as reviewers can also benefit from an open approach, as this allows them to get credit for this valuable service. Platforms like Publons let researchers create reviewer profiles to showcase their work (Publons, 2016).
Publish where you want and archive openly
Some researchers may not see publishing in OA journals as a viable option, and may wish instead to publish in specific subscription journals seen as prestigious in their field. Importantly, there are ways to openly share work while still publishing in subscription journals.
Preprints: Authors may provide open access to their papers by posting them as preprints prior to formal peer review and journal publication. Preprints servers are both free for authors to post and free for readers. Several archival preprint servers exist covering different subject areas (Table 1). (Note: The list in Table 1 is not all-inclusive; there are many other servers and institutional repositories that also accept preprints).
Many journals allow posting of preprints, including Science, Nature, and PNAS, as well as most OA journals. Journal preprint policies can be checked via Wikipedia (Wikipedia, 2016) and SHERPA/RoMEO (SHERPA/RoMEO, 2016). Of the over 2,000 publishers in the SHERPA/RoMEO database, 46% explicitly allow preprint posting. Preprints can be indexed in Google Scholar and cited in the literature, allowing authors to accrue citations while the paper is still in review. In one extreme case, one of the present authors (CTB) published a preprint that has received over 50 citations in three years (Brown et al., 2012), and was acknowledged in NIH grant reviews.
In some fields, preprints can establish scientific priority. In physics, astronomy, and mathematics, preprints have become an integral part of the research and publication workflow (Brown, 2001; Larivière et al., 2014; Gentil-Beccot et al., 2010). Physics articles posted as preprints prior to formal publication tend to receive more citations than those published only in traditional journals (Gentil-Beccot et al., 2010; Schwarz and Kennicutt Jr, 2004; Metcalfe, 2006). Unfortunately, because of the slow adoption of preprints in the biological and medical sciences, few if any studies have been conducted to examine citation advantage conferred by preprints in these fields. However, the growing number of submissions to the quantitative biology section of arXiv, as well as to dedicated biology preprint servers such as bioRxiv and PeerJ PrePrints, should make such studies feasible. Researchers have argued for increased use of preprints in biology (Desjardins-Proulx et al., 2013). The recent Accelerating Science and Publication in biology (ASAPbio) meeting demonstrates growing interest and support for life science preprints from researchers, funders, and publishers (Berg et al., 2016; ASAPbio, 2016).
Postprints: Authors can also archive articles on open platforms after publication in traditional journals (postprints). SHERPA/RoMEO allows authors to check policies from over 2,200 publishers, 72% of which allow authors to archive postprints, either in the form of the authors’ accepted manuscript post-peer review, or the publisher’s formatted version, depending on the policy (SHERPA/RoMEO, 2016). Of notable example is Science, which allows authors to immediately post the accepted version of their manuscript on their website, and post to larger repositories like PubMed Central six months after publication. The journal Nature likewise allows archiving of the accepted article in open repositories six months after publication.
If the journal in which authors publish does not formally support self-archiving, authors can submit an author addendum that allows them to retain rights to post a copy of their article in an open repository. The Scholarly Publishing and Academic Resources Coalition (SPARC) provides a template addendum, as well as information on author rights (SPARC, 2016). The Scholar’s Copyright Addendum Engine helps authors generate a customized addendum to send to publishers (Science Commons, 2016). Not all publishers will accept author addenda, but some are willing to negotiate the terms of their publishing agreements.
Retain author rights and control reuse with open licenses
To make their findings known to the world, scientists have historically forfeited ownership of the products of their intellectual labor by signing over their copyrights or granting exclusive reuse rights to publishers. In contrast, authors publishing in OA journals retain nearly all rights to their manuscripts and materials. OA articles are typically published under Creative Commons (CC) licenses, which function within the legal framework of copyright law (Creative Commons, 2016). Under these licenses, authors retain copyright, and simply grant specific (non-exclusive) reuse rights to publishers, as well as other users. Moreover, CC licenses require attribution, which allows authors to receive credit for their work and accumulate citations. Licensors can specify that attribution include not just the name of the author(s) but also a link back to the original work. Authors submitting work to an OA journal should review its submission rules to learn what license(s) the journal permits authors to select.
If terms of a CC license are violated by a user, the licensor can revoke the license and, if the revocation is not honored, take legal action to enforce their copyright. There are several legal precedents upholding CC licenses, including: (1) Adam Curry v. Audax Publishing (Court of Amsterdam, 2006; Garlick, 2006a); (2) Sociedad General de Autores y Editores (SGAE) v. Ricardo Andrés Utrera Fernández (Juzgado de Primera Instancia Número Seis de Badajoz, España, 2006; Garlick, 2006b); and (3) Gerlach v. Deutsche Volksunion (DVU) (Linksvayer, 2011). Through open licensing, researchers thus retain control over how their work is read, shared, and used by others.
An emerging and interesting development is the adoption of rights-retention open access policies (Harvard Open Access Project, 2016). To date, such policies have been adopted by at least 60 schools and institutions worldwide, including some in Canada, Iceland, Kenya, Saudi Arabia, and U.S. universities like Harvard (Harvard Library, Office for Scholary Communication, 2016) and MIT (MIT Libraries, Scholarly Publishing, 2016). These policies involve an agreement by the faculty to grant universities non-exclusive reuse rights on future published works. By putting such a policy in place prior to publication, faculty work can be openly archived without the need to negotiate with publishers to retain or recover rights; open is the default. We expect to see adoption of such policies grow in coming years.
Publish for low-cost or no-cost
Researchers often cite high costs, primarily in the form of article processing charges (APCs), as a barrier to publishing in OA journals. While some publishers – subscription as well as OA – do charge steep fees (Lawson, 2016; Wellcome Trust, 2016c), many others charge nothing at all. In a 2014 study of 1,357 OA journals, 71% did not request any APC (West et al., 2014). A study of over 10,300 OA journals from 2011 to 2015 likewise found 71% did not charge (Crawford, 2016). Eigenfactor.org maintains a list of hundreds of no-fee OA journals across fields (Eigenfactor Project, 2016). Researchers can also search for no-cost OA journals using the Cofactor Journal Selector tool (Cofactor Ltd, 2016). Notable examples of OA journals which do not currently charge authors to publish include eLife, Royal Society’s Open Science, and all journals published by consortiums like Open Library of Humanities and SCOAP3. The Scientific Electronic Library Online (SciELO) and the Network of Scientific Journals in Latin America, the Caribbean, Spain, and Portugal (Redalyc), each host over 1,000 journals that provide free publishing for authors.
Many other OA journals charge minimal fees, with the average APC around $665 USD (Crawford, 2016). At PeerJ, for example, a one-time membership fee of $199 USD allows an author to publish one article per year for life, subject to peer review. (Note: Since PeerJ requires the membership fee to be paid for each author up to 12 authors, the maximum cost of an article would be $2,388 USD. However, this is a one-time fee, after which subsequent articles for the same authors would be free.) Most Pensoft OA journals charge around €100–400 ($115–460 USD), while a select few are free. Ubiquity Press OA journals charge an average APC of £300 ($500 USD), with their open data and software metajournals charging £100 ($140 USD). Cogent’s OA journals all function on a flexible payment model, with authors paying only what they are able based on their financial resources. Importantly, most OA journals do not charge any additional fees for submission or color figures. These charges, as levied by many subscription publishers, can easily sum to hundreds or thousands of dollars (e.g. in Elsevier’s Neuron the first color figure is $1,000 USD, while each additional one is $275). Thus, publishing in OA journals need not be any more expensive than publishing in traditional journals, and in some cases, may cost less.
The majority of OA publishers charging higher publication fees (e.g., PLOS or Frontiers, which typically charge upwards of $1,000 USD per manuscript) offer fee waivers upon request for authors with financial constraints. Policies vary by publisher, but frequently include automatic full waivers for authors from low-income countries, and partial waivers for those in lower-middle-income countries. Researchers in any country can request a partial or full waiver if they cannot pay. Some publishers, such as BioMed Central, F1000, Hindawi, and PeerJ, have membership programs through which institutions pay part or all of the APC for affiliated authors. Some institutions also have discretionary funds for OA publication fees. Increasingly, funders are providing OA publishing funds, or allowing researchers to write these funds into their grants. PLOS maintains a searchable list of both institutions and funders that support OA publication costs (Public Library of Science, 2016). Finally, as discussed previously in the section "Publish where you want and archive openly", researchers can make their work openly available for free by self-archiving preprints or postprints.
Funding
Awards and special funding
For academics in many fields, securing funding is essential to career development and success of their research program. In the last three years, new fellowships and awards for open research have been created by multiple organizations (Table 2). While there is no guarantee that these particular funding mechanisms will be maintained, they are a reflection of the changing norms in science, and illustrate the increasing opportunities to gain recognition and resources by sharing one’s work openly.
Funder mandates on article and data sharing
Increasingly, funders are not only preferring but mandating open sharing of research. The United States National Institutes of Health (NIH) has been a leader in this respect. In 2008, the NIH implemented a public access policy, requiring that all articles arising from NIH-funded projects be deposited in the National Library of Medicine’s open repository, PubMed Central, within one year of publication (Rockey, 2012). NIH also requires that projects receiving $500K or more per year in direct costs include a data management plan that specifies how researchers will share their data (National Institutes of Health, 2003). NIH intends to extend its data sharing policy to a broader segment of its portfolio in the near future. Since 2011, the United States National Science Foundation (NSF) has also encouraged sharing data, software, and other research outputs (National Science Foundation, 2011). All NSF investigators are required to submit a plan, specifying data management and availability. In 2015, U.S. government agencies, including the NSF, Centers for Disease Control and Prevention (CDC), Department of Defense (DoD), National Aeronautics and Space Administration (NASA), and more announced plans to implement article and data sharing requirements in response to the White House Office of Science and Technology (OTSP) memo on public access (Holdren, 2013). A crowd-sourced effort has collected information on these agency policies and continues to be updated (Whitmire et al., 2015).
Several governmental agencies and charitable foundations around the world have implemented even stronger open access mandates. For example, the Wellcome Trust’s policy states that articles from funded projects must be made openly available within six months of publication, and where it provides publishing fee support, specifically requires publication under a Creative Commons Attribution (CC BY) license (Wellcome Trust, 2016b). The Netherlands Organization for Scientific Research (NWO) requires that all manuscripts reporting results produced using public funds must be made immediately available (NWO, 2016). Similar policies are in place at CERN (CERN, 2014), the United Nations Educational, Scientific and Cultural Organization (UNESCO, 2013), and the Bill & Melinda Gates Foundation (Bill & Melinda Gates Foundation, 2015) among others, and are increasingly covering data sharing. Funders recognize that certain types of data, such as clinical records, are sensitive and require special safeguards to permit sharing while protecting patient privacy. The Expert Advisory Group on Data Access (EAGDA) was recently established as a collaboration between the Wellcome Trust, Cancer Research UK, the Economic and Social Research Council, and the Medical Research Council to advise funders on best practices for creating data sharing policies for human research (Wellcome Trust, 2016a).
Researchers can check article and data sharing policies of funders in their country via SHERPA/JULIET (SHERPA/JULIET, 2016). BioSharing also maintains a searchable database of data management and sharing policies from both funders and publishers worldwide (Biosharing.org, 2016). Internationally, the number of open access policies has been steadily increasing over the last decade (Figure 2). Some funders, including the NIH and Wellcome Trust, have begun suspending or withholding funds if researchers do not meet their policy requirements (National Institutes of Health, 2012; Van Noorden, 2014; Wellcome Trust, 2012). Thus, researchers funded by a wide variety of sources will soon be not just encouraged but required to engage in open practices to receive and retain funding. Those already engaging in these practices will likely have a competitive advantage.
Resource management and sharing
In our researcher-centric approach, the rationale for data sharing based on funder mandates could be understood simply as ‘funders want you to share, so it is in your interest to do so’. That may be a compelling but dissatisfying reason to practice openly. Fortunately, there are other compelling reasons to share.
Documentation and reproducibility benefits
First, submitting data and research materials to an independent repository ensures preservation and accessibility of that content in the future - both for one’s own access and for others. This is a particular benefit for responding to requests for data or materials by others. Preparation of research materials for sharing during the active phase of the project is much easier than reconstructing work from years earlier. Second, researchers who plan to release their data, software, and materials are likely to engage in behaviors that are easy to skip in the short-term but have substantial benefits in the long-term, such as clear documentation of the key products of the research. Besides direct benefits for oneself in facilitating later reuse, such practices increase the reproducibility of published findings and the ease with which other researchers can use, extend, and cite that work (Gorgolewski and Poldrack, 2016). Finally, sharing data and materials signals that researchers value transparency and have confidence in their own research.
Gain more citations and visibility by sharing data
Data sharing also confers a citation advantage. Piwowar and Vision (2013) analyzed over 10,000 studies with gene expression microarray data published in 2001–2009, and found an overall 9% citation advantage for papers with shared data and advantages around 30% for older studies. Henneken and Accomazzi (2011) found a 20% citation advantage for astronomy articles that linked to open datasets. Dorch et al., 2015 found a 28–50% citation advantage for astrophysics articles, while Sears (2011) reported a 35% advantage for paleoceanography articles with publicly available data. Similar positive effects of data sharing have been described in the social sciences. Gleditsch et al., 2003 found that articles in the Journal of Peace Research offering data in any form – either through appendices, URLs, or contact addresses – were cited twice as frequently on average as articles with no data but otherwise equivalent author credentials and article variables. Studies with openly published code are also more likely to be cited than those that do not open their code (Vandewalle, 2012). In addition to more citations, Pienta et al., 2010 found that data sharing is associated with higher publication productivity. Across over 7,000 NSF and NIH awards, they reported that research projects with archived data produced a median of 10 publications, versus only 5 for projects without archived data.
Importantly, citation studies may underestimate the scientific contribution and resulting visibility associated with resource sharing, as many data sets and software packages are published as stand-alone outputs that are not associated with a paper but may be widely reused. Fortunately, new outlets for data and software papers allow researchers to describe new resources of interest without necessarily reporting novel findings (Chavan and Penev, 2011; Gorgolewski et al., 2013). There is also a growing awareness that data and software are independent, first class scholarly outputs, that need to be incorporated into the networked research ecosystem. Many open data and software repositories have mechanisms for assigning digital object identifiers (DOIs) to these products. The use of persistent, unique identifiers like DOIs has been recommended by the Joint Declaration of Data Citation Principles to facilitate data citation (Data Citation Synthesis Group, 2014). Researchers can register for a unique Open Researcher and Contributor ID (ORCID) (Haak et al., 2012) to track their research outputs, including datasets and software, and build a richer profile of their contributions. Together, these developments should support efforts to ‘‘make data count’’, further incentivize sharing, and ensure that data generators and software creators receive greater credit for their work (Kratz and Strasser, 2015).
In summary, data and software sharing benefits researchers both because it is consistent with emerging mandates, and because it signals credibility and engenders good research practices that can reduce errors and promote reuse, extension, and citation.
Career advancement
Find new projects and collaborators
Research collaborations are essential to advancing knowledge, but identifying and connecting with appropriate collaborators is not trivial. Open practices can make it easier for researchers to connect with one another by increasing the discoverability and visibility of one’s work, facilitating rapid access to novel data and software resources, and creating new opportunities to interact with and contribute to ongoing communal projects. For example, in 2011, one of the present authors (BAN) initiated a project to replicate a sample of studies to estimate the reproducibility of psychological science (Open Science Collaboration, 2012; Open Science Collaboration, 2014). Completing a meaningful number of replications in a single laboratory would have been difficult. Instead, the project idea was posted to a listserv as an open collaboration. Ultimately, more than 350 people contributed, with 270 earning co-authorship on the publication (Open Science Collaboration, 2015). Open collaboration enabled distribution of work and expertise among many researchers, and was essential for the project’s success. Other projects have used similar approaches to successfully carry out large-scale collaborative research (Klein et al., 2014).
Similar principles are the core of the thriving open -source scientific software ecosystem. In many scientific fields, widely used state-of-the-art data processing and analysis packages are hosted and developed openly, allowing virtually anyone to contribute. Perhaps the paradigmatic example is the scikit-learn Python package for machine learning (Pedregosa et al., 2011), which, in the space of just over five years, has attracted over 500 unique contributors, 20,000 individual code contributions, and 2,500 article citations. Producing a comparable package using a traditional closed-source approach would likely not be feasible, and would, at the very least, have required a budget of tens of millions of dollars. While scikit-learn is clearly an outlier, hundreds of other open-source scientific packages that support much more domain-specific needs depend in a similar fashion on unsolicited community contributions e.g., the NIPY group of projects in neuroimaging (Gorgolewski et al., 2016). Importantly, such contributions not only result in new functionality from which the broader scientific community can benefit, but also regularly provide their respective authors with greater community recognition, and lead to new project and employment opportunities.
Institutional support of open research practices
Institutions are increasingly recognizing the limitations of journal-level metrics and exploring the potential benefits of article-level and alternative metrics in evaluating the contributions of specific research outputs. In 2013, the American Society for Cell Biology, along with a group of diverse stakeholders in academia, released the San Francisco Declaration on Research Assessment (SF-DORA) (American Society for Cell Biology, 2013). The declaration recommends that institutions cease using all journal-level metrics, including journal impact factor (IF), to evaluate research for promotion and tenure decisions, and focus instead on research content. Additional recommendations include recognizing data and software as valuable research products. As of March 2016, over 12,000 individuals and more than 600 organizations have signed SF-DORA in support of the recommendations, including universities from all over the world. The 2015 Higher Education Funding Council for England (HEFCE) report for The Research Excellence Framework (REF) – UK’s system for assessing research quality in higher education institutions – also rejects the use of IF and other journal metrics to evaluate researchers for hiring and promotion, and recommends institutions explore a variety of quantitative and qualitative indicators of research impact and ways to recognize sharing of diverse research outputs (Wilsdon et al., 2015).
Several U.S. institutions have passed resolutions explicitly recognizing open practices in promotion and tenure evaluations, including Virginia Commonwealth University (Virginia Commonwealth University Faculty Senate, 2010) and Indiana University-Purdue University Indianapolis (Indiana University-Purdue University Indianapolis, 2016). In 2014, Harvard’s School of Engineering and Applied Sciences launched a pilot program to encourage faculty to archive their articles in the university’s open repository as part of the promotion and tenure process (Harvard Library, Office for Scholarly Communication, 2014). The University of Liège has gone a step further and requires publications to be included in the university’s open access repository to be considered for promotion (University of Liège, 2016). Explicit statements of the importance of open practices are even starting to appear in faculty job advertisements, such as one from LMU München asking prospective candidates to describe their open research activities (Schönbrodt, 2016).
Discussion
Open questions
The emerging field of metascience provides some evidence about the value of open practices, but it is far from complete. There are many initiatives aimed at increasing open practices, and not yet enough published evidence about their effectiveness. For example, journals can offer badges to acknowledge open practices such as open data, open materials, and preregistration (Open Research Badges, 2016). Initial evidence from a single adopting journal, Psychological Science, and a sample of comparison journals suggests that this simple incentive increases data sharing rates from less than 3% to more than 38% (Kidwell et al., 2016). More research is needed across disciplines to follow up on this encouraging evidence. UCLA’s Knowledge Infrastructures project is an ongoing study that, among other objectives, is learning about data sharing practices and factors that discourage or promote sharing across four collaborative scientific projects (Borgman et al., 2015; Darch et al., 2015).
Open research advocates often cite reproducibility as one of the benefits of data and code sharing (Gorgolewski and Poldrack, 2016). There is a logical argument that having access to the data, code, and materials makes it easier to reproduce the evidence that was derived from that research content. Data sharing correlates with fewer reporting errors, compared to papers with unavailable data (Wicherts, 2016), and could be due to diligent data management practices. However, there is not yet direct evidence that open practices per se are a net benefit to research progress. As a first step, the University of California at Riverside and the Center for Open Science have initiated an NSF-supported randomized trial to evaluate the impact of receiving training to use the Open Science Framework for managing, archiving, and sharing lab research materials and data. Labs across the university will be randomly assigned to receive the training, and outcomes of their research will be assessed across multiple years.
Preregistration of research designs and analysis plans is a proposed method to increase the credibility of reported research and a means to increase transparency of the research workflow. However, preregistration is rarely practiced outside of clinical trials where it is required by law in the U.S. and as a condition for publication in most journals that publish them. Research suggests that preregistration may counter some questionable practices, such as flexible definition of analytic models and outcome variables in order to find positive results (Kaplan and Irvin, 2015). Public registration also makes it possible to compare publications and registrations of the same study to identify cases in which outcomes were changed or unreported, as is the focus of the COMPare project based at the University of Oxford (COMPare, 2016). Similar efforts include the AllTrials project, run by an international team (AllTrials, 2016), and extending beyond just preregistration of planned studies to retroactive registration and transparent reporting for previously conducted clinical trials. Another example is the AsPredicted project, which is run by researchers at the University of Pennsylvania and University of California Berkeley, and offers preregistration services for any discipline (AsPredicted, 2016). To initiate similar research efforts in the basic and preclinical sciences, the Center for Open Science launched the Preregistration Challenge, offering one thousand $1,000 awards to researchers that publish the outcomes of preregistered research (Center for Open Science, 2016).
Openness as a continuum of practices
While there are clear definitions and best practices for open access (Chan et al., 2002), open data (Open Knowledge, 2005; Murray-Rust et al., 2010), and open source (Open Source Initiative, 2007), openness is not ‘all-or-nothing’. Not all researchers are comfortable with the same level of sharing, and there are a variety of ways to be open (see Box 1). Openness can be thus defined by a continuum of practices, starting perhaps at the most basic level with openly self-archiving postprints and reaching perhaps the highest level with openly sharing grant proposals, research protocols, and data in real time. Fully open research is a long-term goal to strive towards, not a switch we should expect to flip overnight.
Many of the discussions about openness center around the associated fears, and we need encouragement to explore the associated benefits as well. As researchers share their work and experience the benefits, they will likely become increasingly comfortable with sharing and willing to experiment with new open practices. Acknowledging and supporting incremental steps is a way to respect researchers’ present experience and comfort, and produce a gradual culture change from closed to open research. Training of researchers early in their careers is fundamental. Graduate programs can integrate open science and modern scientific computing practices into their existing curriculum. Methods courses could incorporate training on publishing practices such as proper citation, author rights, and open access publishing options. Institutions and funders could provide skills training on self-archiving articles, data, and software to meet mandate requirements. Importantly, we recommend integrating education and training with regular curricular and workshop activities so as not to increase the time burden on already-busy students and researchers.
Summary
The evidence that openly sharing articles, code, and data is beneficial for researchers is strong and building. Each year, more studies are published showing the open citation advantage; more funders announce policies encouraging, mandating, or specifically financing open research; and more employers are recognizing open practices in academic evaluations. In addition, a growing number of tools are making the process of sharing research outputs easier, faster, and more cost-effective. In his 2012 book Open Access, Peter Suber summed it up best: "[OA] increases a work’s visibility, retrievability, audience, usage, and citations, which all convert to career building. For publishing scholars, it would be a bargain even if it were costly, difficult, and time-consuming. But…it’s not costly, not difficult, and not time-consuming.’’ (Suber, 2012)
What can I do right now?
Engaging in open science need not require a long-term commitment or intensive effort. There are a number of practices and resolutions that researchers can adopt with very little effort that can help advance the overall open science cause while simultaneously benefiting the individual researcher.
Post free copies of previously published articles in a public repository. Over 70% of publishers allow researchers to post an author version of their manuscript online, typically 6-12 months after publication (see section "Publish where you want and archive openly").
Deposit preprints of all manuscripts in publicly accessible repositories as soon as possible – ideally prior to, and no later than, the initial journal submission (see section "Postprints").
Publish in open access venues whenever possible. As discussed in Prestige and journal impact factor, this need not mean forgoing traditional subscription-based journals, as many traditional journals offer the option to pay an additional charge to make one’s article openly accessible.
Publicly share data and materials via a trusted repository. Whenever it is feasible, the data, materials, and analysis code used to generate the findings reported in one’s manuscripts should be shared. Many journals already require authors to share data upon request as a condition of publication; pro-actively sharing data can be significantly more efficient, and offers a variety of other benefits (see section "Resource management and sharing").
Preregister studies. Publicly preregistering one’s experimental design and analysis plan in advance of data collection is an effective means of minimizing bias and enhancing credibility (see section "Open questions"). Since the preregistration document(s) can be written in a form similar to a Methods section, the additional effort required for preregistration is often minimal.
References
-
Do open-access articles have a greater research impact?College & Research Libraries 65:372–382.https://doi.org/10.5860/crl.65.5.372
-
Will open access get me cited? an analysis of the efficacy of open access publishing in political sciencePolitical Science & Politics 48:129–137.https://doi.org/10.1017/S1049096514001668
-
A catalogue of data preservation, management and sharing policies from international funding agencies, regulators and journalsA catalogue of data preservation, management and sharing policies from international funding agencies, regulators and journals, biosharing.org/policies.
-
The hybrid model for open access publication of scholarly articles: A failed experiment?Journal of the American Society for Information Science and Technology 63:1496–1504.https://doi.org/10.1002/asi.22709
-
Anatomy of green open accessJournal of the Association for Information Science and Technology 65:237–250.https://doi.org/10.1002/asi.22963
-
Knowledge infrastructures in science: data, diversity, and digital librariesInternational Journal on Digital Libraries 16:207–227.https://doi.org/10.1007/s00799-015-0157-z
-
Deep impact: unintended consequences of journal rankFrontiers in Human Neuroscience 7:291.https://doi.org/10.3389/fnhum.2013.00291
-
The E-volution of preprints in the scholarly communication of physicists and astronomersJournal of the American Society for Information Science and Technology 52:187–200.https://doi.org/10.1002/1097-4571(2000)9999:9999<::AID-ASI1586>3.0.CO;2-D
-
A reference-free algorithm for computational normalization of shotgun sequencing dataarXiv:1203.4802.
-
What lies beneath?: Knowledge infrastructures in the subseafloor biosphere and beyondInternational Journal on Digital Libraries 16:61–77.https://doi.org/10.1007/s00799-015-0137-3
-
The case for open preprints in biologyPLOS Biology 11:e1001563.https://doi.org/10.1371/journal.pbio.1001563
-
The open access advantage for American law reviewsEdison: Law + Technology 2015:1–22.
-
The integration of open access journals in the scholarly communication system: three science fieldsInformation Processing & Management 45:131–141.https://doi.org/10.1016/j.ipm.2008.06.001
-
Getting cited: Does open access help?Research Policy 40:1332–1338.https://doi.org/10.1016/j.respol.2011.05.025
-
Citing and reading behaviours in high-energy physicsScientometrics 84:345–355.https://doi.org/10.1007/s11192-009-0111-1
-
Posting your data: will you be scooped or will you be famous?International Studies Perspectives 4:89–97.https://doi.org/10.1111/1528-3577.04105
-
Making data sharing count: a publication-based solutionFrontiers in Neuroscience 7:9.https://doi.org/10.3389/fnins.2013.00009
-
The impact factors of open access and subscription journals across fieldsCurrent Science 107:380.
-
ORCID: a system to uniquely identify researchersLearned Publishing 25:259–264.https://doi.org/10.1087/20120404
-
Ten-year cross-disciplinary comparison of the growth of open access and how it increases research citation impactarXiv: cc/0606079.
-
Comparing the impact of open access (oa) vs. non-oa articles in the same journalsD-Lib Magazine 10:.
-
From toddler to teen: growth of an open data ecosystemeJournal of eDemocracy & Open Government 6:123–135.
-
Retraction noticeJournal of Vibration and Control 20:1601–1604.https://doi.org/10.1177/1077546314541924
-
Diffusion of news about researchScience Communication 25:3–13.https://doi.org/10.1177/1075547003255297
-
Investigating Variation in ReplicabilitySocial Psychology 45:142–152.https://doi.org/10.1027/1864-9335/a000178
-
The citation impact of open access agricultural researchOnline Information Review 34:772–785.https://doi.org/10.1108/14684521011084618
-
A comparison of the quality of reviewer reports from author-suggested reviewers and editor-suggested reviewers in journals operating on open or closed peer review modelsF1000 Posters 4:1252.
-
Scientific DataComment: Making data count, Scientific Data, 2, 10.1038/sdata.2015.39.
-
arXiv E-prints and the journal of record: An analysis of roles and relationshipsJournal of the Association for Information Science and Technology 65:1157–1169.https://doi.org/10.1002/asi.23044
-
Identifying the effect of open access on citations using a panel of science journalsEconomic Inquiry 52:1284–1300.https://doi.org/10.1111/ecin.12064
-
Does online availability increase citations? Theory and evidence from a panel of economics and business journalsReview of Economics and Statistics 97:144–165.https://doi.org/10.1162/REST_a_00437
-
Ecological data sharingEcological Informatics 29:33–44.https://doi.org/10.1016/j.ecoinf.2015.06.010
-
Impact factors: uses and abusesEuropean Journal of Gastroenterology & Hepatology 14:209–211.https://doi.org/10.1097/00042737-200203000-00001
-
The citation advantage of open-access articlesJournal of the American Society for Information Science and Technology 59:1963–1972.https://doi.org/10.1002/asi.20898
-
An Open, Large-Scale, Collaborative Effort to Estimate the Reproducibility of Psychological SciencePerspectives on Psychological Science 7:657–660.https://doi.org/10.1177/1745691612462588
-
BookThe reproducibility project: a model of large-scale collaboration for empirical research on reproducibilityIn: Stodden V, Leisch F, Peng RD, editors. Implementing Reproducible Research. CRC Press, Taylor & Francis Group. pp. 299–324.
-
Scikit-learn: machine learning in PythonJournal of Machine Learning Research 12:2825–2830.
-
Importance of the lay press in the transmission of medical knowledge to the scientific communityNew England Journal of Medicine 325:1180–1183.https://doi.org/10.1056/NEJM199110173251620
-
Data sharing in neuroimaging researchFrontiers in Neuroinformatics 6:9.https://doi.org/10.3389/fninf.2012.00009
-
Demographic and citation trends in astrophysical journal papers and preprintsArXiv: astro-ph/0411275.
-
Practices in source code sharing in astrophysicsAstronomy and Computing 1:54–58.https://doi.org/10.1016/j.ascom.2013.04.001
-
Code sharing is associated with research impact in image processingComputing in Science & Engineering 14:42–47.https://doi.org/10.1109/MCSE.2012.63
-
Open access citation advantage: An annotated bibliographyIssues in Science and Technology Librarianship 60:.https://doi.org/10.5062/F4Q81B0W
-
Open peer review: a randomised controlled trialThe British Journal of Psychiatry 176:47–51.https://doi.org/10.1192/bjp.176.1.47
-
Cost effectiveness of open access publicationsEconomic Inquiry 52:1315–1321.https://doi.org/10.1111/ecin.12117
-
BookThe Access Principle: The Case for Open Access to Research and ScholarshipMIT Press.
-
The Metric Tide: Report of the Independent Review of the Role of Metrics in Research Assessment and ManagementThe Metric Tide: Report of the Independent Review of the Role of Metrics in Research Assessment and Management, 10.13140/RG.2.1.4929.1363.
-
Do open access articles in economics have a citation advantage?Munich Personal RePEc Archive 56842:.
-
Self‐selection and the citation advantage of open access articlesOnline Information Review 36:40–51.https://doi.org/10.1108/14684521211206953
-
Analysis on open access citation advantage: an empirical study based on oxford open journals426–432.
-
The growing impact of open access distance education journals: a bibliometric analysisInternational Journal of E-Learning & Distance Education 24:.
Decision letter
-
Peter RodgersReviewing Editor; eLife, United Kingdom
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "The benefits of open research: How sharing can help researchers succeed" to eLife for consideration as a Feature Article. Your article has been reviewed by three peer reviewers and the eLife Features Editor (Peter Rodgers), and this decision letter has been compiled to help you prepare a revised submission.
All three reviewers have agreed to reveal their identity: Robert Kiley; Chris Gorgolewski; Vincent Lariviere.
General assessment:
This paper is a much-needed overview of benefits and practical advice about open research. Instead of preaching and using moral arguments, the authors focus on benefits to the researcher of doing research openly, and provide evidence that researchers who practice open research (making articles OA, sharing data, publishing code under an open licence etc.) enjoy significant benefits (more citations, more media coverage) compared with researchers who don't practice open research.
Essential revisions:
1) (Section 2.1) Although the authors do provide evidence that OA leads to more citations, I think they need to recognise that there is significant disagreement within the community about this. The paper does cite one study from Davis (Davis, 2011) but he has published quite a few papers on this topic: please consider citing another one of these studies.
2) (Figure 1) It is misleading to plot only the maximum citation advantage: please plot the median advantage and/or the range of advantages (or, alternatively, drop the figure).
3) (Section 2.3) I found this a little self-contradictory. At the start the authors argue (rightly) that IF are a flawed measure, but then go on to quote the IF of OA journals. I know they go to so say "we reiterate that IF are flawed[…]" but if they really believe this, then arguing that some OA journals have high IF doesn't make sense. Please consider deleting the passage "In the 2012 Journal Citation Report...choose between IF and OA)."
4) (Section 2.4; first paragraph) This is debatable and should be toned down a bit. Results obtained with F1000 tend to show the opposite (https://scholarlykitchen.sspnet.org/2013/03/27/how-rigorous-is-the-post-publication-review-process-at-f1000-research/).
5) (Section 2.5.1) I think this section should mention the recent ASAPbio meeting – and subsequent researcher survey – which seem to suggest that researchers in the life sciences have woken up to the potential of preprints (albeit 25 years after physicists reached the same conclusion!)
6) (Section 3.2) Right at the start of the article (Introduction, first paragraph) the authors get to the nub of the problem – namely that open practices could present a risk to career advancement. This in my opinion is the big issue. Until researchers are persuaded that making their outputs open is not going to adversely impact them (and ultimately come to believe that it will benefit them) then changing behaviour is always going to be difficult. As such I was surprised that section 3.2 seemed to suggest that Funder mandates are sufficient to bring about this change. Although I agree that some mandates are important, on their own they are not sufficient.
The article would be improved if it recognised that mandates are not enough and then set out a list of things funders could do to help move the needle in this space (e.g., maybe end-of-grant reports should actively recognise and reward behaviours like data sharing, undertaking peer review, publishing papers on preprint servers etc. – and be less fixated on counting journal article outputs.)
Further to this: the NIH requires grant applications to include a data-management plans: however, when the NIH is considering grant applications, it does not take into account if applicants have a history of sharing data, and it does not penalize applicants if data from previous grants have not been shared. Until data sharing becomes an important part of review procedure, change will remain slow.
7) (Section 2.7) To be transparent, please mention the higher APCs of PLOS journals, as well as of for-profit publishers like Elsevier, Wiley, Springer, etc. When taking about high APCs the authors may also wish to cite the data the Wellcome Trust (and others) have published on this topic (e.g. blog.wellcome.ac.uk/2016/03/23/wellcome-trust-and-coaf-open-access-spend-2014-15/).
8) (Section 4.2) Being devil’s advocate here: the authors report that research projects that share data produce twice as many publications as those that do not share data. Isn't the more likely explanation of this relation that successful projects that have published a lot can afford to share data because the risk and consequences of scooping are much lower/smaller?
9) (Table 1) Please add the following columns: "Ability to leave feedback", "Provides DOI", and "Indexed by Google Scholar".
10) (Figure 2) Please expand the caption for this figure to better explain the five different organisations shown in the figure, maybe by giving examples of each type of organization. Please also explain why there are categories for "research organisations", "funders" and "funders and research organisations". Also, please explain the category "multiple research organisations"
https://doi.org/10.7554/eLife.16800.007Author response
Essential revisions:
1) (Section 2.1) Although the authors do provide evidence that OA leads to more citations, I think they need to recognise that there is significant disagreement within the community about this. The paper does cite one study from Davis (Davis, 2011) but he has published quite a few papers on this topic: please consider citing another one of these studies.
To present this information in a more balanced way, we deleted the word ‘overwhelming’ from Section 2.1, first paragraph, and made other minor changes to wording of this section.
We added another paper from Davis (Davis, 2008):
P.M. Davis, B.V. Lewenstein, D.H. Simon, J.G. Booth, and M.J.L. Connolly. Open access publishing, article downloads, and citations: randomised controlled trial. BMJ, 337:a568, 2008.
We also added citations to 3 other studies which failed to find an OA citation advantage:
T.F. Frandsen. The effects of open access on un-published documents: A case study of economics working papers. Journal of Informetrics, 3(2):124–133, 2009.
P. Gaule and N. Maystre. Getting cited: does open access help? Research Policy, 40(10): 1332–1338, 2011.
V.C. Lansingh and M.J. Carter. Does open access in ophthalmology affect how articles are subsequently cited in research? Ophthalmology, 116(8):1425–1431, 2009.
We expanded Figure 1 to include some studies in which no OA citation advantage, or even a disadvantage, was found for certain disciplines (see below).
2) (Figure 1) It is misleading to plot only the maximum citation advantage: please plot the median advantage and/or the range of advantages (or, alternatively, drop the figure).
We revised Figure 1 to include mean citation advantages (medians are often not reported). These are now shown as a relative citation rate, instead of percentage. We also expanded the figure to include more disciplines and more studies, including some in which an OA citation advantage was not found.
3) (Section 2.3) I found this a little self-contradictory. At the start the authors argue (rightly) that IF are a flawed measure, but then go on to quote the IF of OA journals. I know they go to so say "we reiterate that IF are flawed[…]." but if they really believe this, then arguing that some OA journals have high IF doesn't make sense. Please consider deleting the passage "In the 2012 Journal Citation Report…choose between IF and OA)."
We appreciate the reviewer’s concern, but believe strongly that we have to discuss IF since it is often cited by researchers as a barrier to publishing openly. In numerous author surveys, researchers repeatedly rank impact factor and associated journal reputation as among the most important factors they consider when deciding where to publish. To support this, we added citations to the following author surveys:
Nature Publishing Group (2015): Author Insights 2015 survey. figshare. https://dx.doi.org/10.6084/m9.figshare.1425362.v7
Solomon DJ. (2014) A survey of authors publishing in four megajournals. PeerJ 2:e365 https://doi.org/10.7717/peerj.365
Given our researcher-centric approach, it is important to recognize concerns about IF as a practical, albeit regrettable, reality. We believe that ignoring this reality, and specifically removing the recommended passage with data on the IFs of OA journals, would weaken the paper. This information provides researchers with options that satisfy both their worry about publishing in high IF journals and their wish to do so openly. To clarify our goals with this section, we made several small changes in wording and sentence order, and added a closing statement, which reads:
“We hope that in the next few years, use of IF as a metric will diminish or cease entirely, but in the meantime, researchers have options to publish openly while still meeting any IF-related evaluation and career-advancement criteria.
4) (Section 2.4; first paragraph) This is debatable and should be toned down a bit. Results obtained with F1000 tend to show the opposite (https://scholarlykitchen.sspnet.org/2013/03/27/how-rigorous-is-the-post-publication-review-process-at-f1000-research/).
We reworded this section to read:
“Some studies suggest open peer review may produce reviews of higher quality, including better substantiated claims and more constructive criticisms, compared to closed review [Donovan, Watson and Osborne, 2015; McCabe and Snyder, 2015]. Additional studies have also argued that transparent peer review processes are linked to measures of quality. Other studies have reported no differences in the quality of open versus closed reviews [Wicherts, 2016; Rooyen et al., 1999]. More research in this area is needed.”
We have added the following references to controlled studies finding no difference in quality of open versus closed reviews:
S. Van Rooyen, F. Godlee, S. Evans, N. Black, and R. Smith. Effect of open peer review on quality of reviews and on reviewers’ recommendations: a randomised trial. BMJ, 318(7175):23–27, 1999.
S. van Rooyen, T. Delamothe, and S.J.W. Evans. Effect on peer review of telling reviewers that their signed reviews might be posted on the web: randomised controlled trial. BMJ, 341:c5729, 2010.
5) (Section 2.5.1) I think this section should mention the recent ASAPbio meeting – and subsequent researcher survey – which seem to suggest that researchers in the life sciences have woken up to the potential of preprints (albeit 25 years after physicists reached the same conclusion!)
We added mention of the ASAPbio meeting (end of Section 2.5.1) and cited the following references on the meeting and survey:
J.M. Berg, N. Bhalla, P.E. Bourne, M. Chalfie, D.G. Drubin, J.S. Fraser, C.W. Greider, M. Hen- dricks, C. Jones, R. Kiley, S. King, M.W. Kirschner, H.M. Krumholz, R. Lehman, M. Leptin, B. Pulverer, B. Rosenzweig, J.E. Spiro, M. Stebbins, C. Strasser, S. Swaminathan, P. Turner, R.D. Vale, K. VijayRaghavan, and C. Wolberger. Preprints for the life sciences. Science, 352(6288): 899–901, 2016.
ASAPbio. Opinions on preprints in biology. Accessed May, 2016 at http://asapbio.org/survey. Data available via figshare https://dx.doi.org/10.6084/m9.figshare.2247616.v1.
6) (Section 3.2) Right at the start of the article (Introduction, first paragraph) the authors get to the nub of the problem – namely that open practices could present a risk to career advancement. This in my opinion is the big issue. Until researchers are persuaded that making their outputs open is not going to adversely impact them (and ultimately come to believe that it will benefit them) then changing behaviour is always going to be difficult. As such I was surprised that section 3.2 seemed to suggest that Funder mandates are sufficient to bring about this change. Although I agree that some mandates are important, on their own they are not sufficient.
While we agree that mandates are unlikely to bring about the culture change we would like to see, there is evidence that mandates are effective in increasing rates of article and data sharing (see work from Harnad and colleagues, especially). More importantly, our goal with this section is not to argue that mandates are sufficient, but rather that “[researchers] already engaging in [open] practices will likely have a competitive advantage”.
The article would be improved if it recognised that mandates are not enough and then set out a list of things funders could do to help move the needle in this space (e.g., maybe end-of-grant reports should actively recognise and reward behaviours like data sharing, undertaking peer review, publishing papers on preprint servers etc. – and be less fixated on counting journal article outputs.)
We recognize in the subsequent section (section 4) that “[funder mandates] may be a compelling but dissatisfying reason to practice openly”. However, our primary target audience for this article is researchers, so we have focused on outlining the steps they can take and showing them “there are other compelling reasons to share”.
Further to this: the NIH requires grant applications to include a data-management plans: however, when the NIH is considering grant applications, it does not take into account if applicants have a history of sharing data, and it does not penalize applicants if data from previous grants have not been shared. Until data sharing becomes an important part of review procedure, change will remain slow.
We added mention of policy revisions implemented by NIH and Wellcome Trust, detailing how funds can be suspended or withheld if researchers do not comply with mandates (Section 3.2, last paragraph). We cited the following relevant references, one of which (van Noorden, 2014) discusses how both funders have already followed through on enforcement:
National Institutes of Health (NIH). Upcoming Changes to Public Access Policy Reporting Requirements and Related NIH Efforts to Enhance Compliance, 2012. Retrieved June, 2016 from http://grants.nih.gov/grants/guide/notice-files/NOT-OD-12-160.html. Last updated Feb., 2013.
Van Noorden, R. Funders punish open-access dodgers. Nature News, 2014. Retrieved June, 2016 from http://www.nature.com/news/funders-punish-open-access-dodgers-1.15007.
Wellcome Trust. Wellcome Trust strengthens its open access policy, 2012. Retrieved June, 2016 from https://wellcome.ac.uk/press-release/wellcome-trust-strengthens-its-open-access-policy.
7) (Section 2.7) To be transparent, please mention the higher APCs of PLOS journals, as well as of for-profit publishers like Elsevier, Wiley, Springer, etc. When taking about high APCs the authors may also wish to cite the data the Wellcome Trust (and others) have published on this topic (e.g. blog.wellcome.ac.uk/2016/03/23/wellcome-trust-and-coaf-open-access-spend-2014-15/).
We added mention of the higher APCs charged by some OA publishers, like PLOS and Frontiers (Section 2.7, last paragraph). We also felt the no-cost/low-cost examples here were numerous, so we have stricken two of them.
We added the suggested reference from Wellcome Trust, as well as one from Stuart Lawson documenting high APCs:
S. Lawson. APC data for 27 UK higher education institutions in 2015. figshare, 2016. Retrieved June, 2016 from https://dx.doi.org/10.6084/m9.figshare.1507481.v4.
Wellcome Trust. Wellcome Trust and COAF Open Access Spend, 2014-15,. Retrieved June, 2016 from https://blog.wellcome.ac.uk/2016/03/23/wellcome-trust-and-coaf-open-access-spend-2014-15/. Data available via figshare doi:10.6084/m9.figshare.3118936.v1.
We also added a citation to a new study of over 10, 300 OA journals, showing 71% do not charge an APC (Section 2.7, first paragraph), and that the average APC for OA journals is around $665 (Section 2.7, second paragraph):
W. Crawford. Gold Open Access Journals 2011-2015. Cites & Insights Books, 2016. Accessed June, 2016 via http://waltcrawford.name/goaj.html.
8) (Section 4.2) Being devil’s advocate here: the authors report that research projects that share data produce twice as many publications as those that do not share data. Isn't the more likely explanation of this relation that successful projects that have published a lot can afford to share data because the risk and consequences of scooping are much lower/smaller?
The authors of Pienta et al. admit that “It is unclear whether larger numbers of primary publications lead to data sharing or if sharing data leads to more primary publications”. However, the authors did control for factors such as Principal Investigator age, gender, career status, and funding history, as well as features of the grant such as duration as an indirect measure of grant size. None of these factors sufficiently explained the primary or secondary publication advantage conferred by data sharing.
9) (Table 1) Please add the following columns: "Ability to leave feedback", "Provides DOI", and "Indexed by Google Scholar".
We added the columns “Can leave feedback?” and “Third party persistent ID?”. The latter is broader and includes externally managed persistent identifiers such as DOIs, Handles, and ARKs. We added a footnote to the table saying that all the listed preprint servers and repositories are indexed by Google Scholar. After community feedback, we also added several relevant repositories.
10) (Figure 2) Please expand the caption for this figure to better explain the five different organisations shown in the figure, maybe by giving examples of each type of organization. Please also explain why there are categories for "research organisations", "funders" and "funders and research organisations". Also, please explain the category "multiple research organisations"
Based on the information provided by ROARMAP, we added the following explanation and examples to the figure caption:
“Data are broken down by policymaker type: funder (e.g. Wellcome Trust), joint funder and research organization (e.g. British Heart Foundation), multiple research organizations i.e. associations and consortia (e.g. Max Planck Society), research organization i.e. university or research institution (e.g. CERN), and subunit of research organization (e.g. Columbia University Libraries).”
https://doi.org/10.7554/eLife.16800.008Article and author information
Author details
Funding
National Institute on Aging (R24AG048124)
- Brian A Nosek
- Courtney K Soderberg
Laura and John Arnold Foundation
- Brian A Nosek
- Jeffrey R Spies
John Templeton Foundation (46545)
- Brian A Nosek
- Jeffrey R Spies
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
This paper arose from the ‘‘Open Source, Open Science’’ meeting held March 19-20th, 2015 at the Center for Open Science in collaboration with Mozilla Science Lab. This meeting was supported by the National Institute of Aging (R24AG048124), the Laura and John Arnold Foundation, and the John Templeton Foundation (46545). The authors thank all those who responded to our public calls for comment – especially Virginia Barbour, Peter Binfield, Nazeefa Fatima, Daniel S. Katz, Sven Kochmann, Ehud Lamm, Alexei Lutay, Ben Marwick, Daniel Mietchen, Ian Mulvany, Cameron Neylon, Charles Oppenheim, Pandelis Perakakis, Richard Smith-Unna, Peter Suber, and Anne-Katharina Weilenmann – whose feedback helped us improve this manuscript.
Publication history
- Received:
- Accepted:
- Accepted Manuscript published:
- Version of Record published:
Copyright
© 2016, McKiernan et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 36,167
- views
-
- 5,556
- downloads
-
- 473
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
Researchers can benefit from making their research findings freely available online.
-
- Biochemistry and Chemical Biology
The Parkinson’s disease (PD)-linked protein Leucine-Rich Repeat Kinase 2 (LRRK2) consists of seven domains, including a kinase and a Roc G domain. Despite the availability of several high-resolution structures, the dynamic regulation of its unique intramolecular domain stack is nevertheless still not well understood. By in-depth biochemical analysis, assessing the Michaelis–Menten kinetics of the Roc G domain, we have confirmed that LRRK2 has, similar to other Roco protein family members, a KM value of LRRK2 that lies within the range of the physiological GTP concentrations within the cell. Furthermore, the R1441G PD variant located within a mutational hotspot in the Roc domain showed an increased catalytic efficiency. In contrast, the most common PD variant G2019S, located in the kinase domain, showed an increased KM and reduced catalytic efficiency, suggesting a negative feedback mechanism from the kinase domain to the G domain. Autophosphorylation of the G1+2 residue (T1343) in the Roc P-loop motif is critical for this phosphoregulation of both the KM and the kcat values of the Roc-catalyzed GTP hydrolysis, most likely by changing the monomer–dimer equilibrium. The LRRK2 T1343A variant has a similar increased kinase activity in cells compared to G2019S and the double mutant T1343A/G2019S has no further increased activity, suggesting that T1343 is crucial for the negative feedback in the LRRK2 signaling cascade. Together, our data reveal a novel intramolecular feedback regulation of the LRRK2 Roc G domain by a LRRK2 kinase-dependent mechanism. Interestingly, PD mutants differently change the kinetics of the GTPase cycle, which might in part explain the difference in penetrance of these mutations in PD patients.