Introduction. Oxford Journals adventures in open access CASE STUDY

Transcription

1 Oxford Journals adventures in open access Claire Bird 200 Claire Bird Learned Publishing, 21, doi: / X Claire Bird CASE STUDY Oxford Journals adventures in open access Claire BIRD Oxford University Press Claire Bird 2008 ABSTRACT. In 2004, Oxford Journals began experimenting with an author-side payment open access model for its flagship molecular biology journal, Nucleic Acids Research (NAR). Since then, around 70 of its approximately 200 journals haveadoptedanopenaccessmodelofsomekind, providing a unique perspective on the practicalities involved and the potential impact of open access on established academic journals. Under NAR s full open access model, submissions and author satisfaction remain encouragingly stable, and most NAR authors are paying the open access charges. NAR s income per article declined in 2005, but increased once more in The journal remains financially viable under the new model. Uptake of the optional Oxford Open model varies by discipline. It is highest in the life sciences: in the region of 17 25% for some molecular and computational biology journals. It is too soon to tell whether the Oxford Open model will have an impact on subscriptions, usage, and citations, but further research is under way. Introduction Open access to academic research is a topic guaranteed to spark a passionate response in some research circles, and incomprehension or indifference in others. A vast number of person-hours have been spent in debate about the principles and ethics of open access publishing, and its future viability. However, there is a considerable dearth of real data upon which the academic and publishing communities can base their decisions. As a university press, Oxford University Press has a strong imperative to explore any publishing model that might further the dissemination of research. In 2004 this led its journals division, Oxford Journals, to embark on a series of open access publishing experiments, with the ambitious aims of finding out what researchers really want, whether immediate open access increases dissemination (in terms of usage and citations), and, crucially, whether open access models can be sustained financially. Most informed participants understand that the costs of validating and disseminating research output must be covered somehow. Without the possibility of charging for access through subscriptions or licences, a common approach under an open access model is to cover publishing costs primarily through author-side payments. The journal business model is therefore turned on its head, from reader-side payment to author-side payment. It seems immediately clear that that this kind of model may only be viable in certain disciplines where authors have access to funds for publication. Gaining an understanding of the differences between disciplines was therefore another important aim of our experiments. This article provides an update on findings from Oxford Journals open access initiative so far.

2 Oxford Journals adventures in open access 201 Overview of Oxford Open models Oxford Journals experiments with open access are grouped together under the Oxford Open brand and can be divided broadly into two types: full open access models, where the entire journal is open access immediately upon publication; and optional open access models, where authors can decide whether or not to pay for immediate open access to their article. The latter gives rise to a hybrid journal made up of a variable combination of open-access and subscription-access content. In each case, we aim to obtain the revenue needed by each journal for open access publication through authorside charges. Nucleic Acids Research (NAR) isthelargest journal owned and published by Oxford Journals. It provides an example of a traditional subscription-based journal that has made a transition to full open access; significant funds now come from author-side charges. Since 2005, all new NAR content has been made freely available online immediately on publication. Subsequently the Oxford Open initiative was expanded to include an open access option; currently 68 further journals give their authors the option of paying to make their article open access immediately upon publication. Full open access: Nucleic Acids Research History of NAR s open access experiments NAR is a large and mature molecular biology journal, publishing its 36th volume in The journal publishes about 1,200 articles each year, and has an acceptance rate of around 40% (around 31% when the special issue and review articles are excluded). NAR operates in a highly competitive market, alongside many other high-profile publications. The journal is well respected, and is fortunate in having built a supportive and loyal author base. In 2004, in response to strong encouragement and support from NAR s authors and editors, NAR s two computational biology special issues (the annual Database and Web Server issues) were published under an open access model, partially funded by an authorside charge of 300/$500 per article. Based on this experience and positive feedback from the community, we then conducted a large-scale survey in 2004 of the journal s authors and readers. We asked them whether they would support full and optional models of open access for NAR (based on authorside charges of up to $1,500 per article), and then asked which model (full, optional, or subscription-based) they would prefer. NAR received 1,052 responses to this survey, representing a 14% response rate. 63% of respondents said that they would support a full open access experiment with NAR. Ina separate question, 54% said that they would prefer a full open access model, while 22% preferred an optional open access model. 19% said that they would prefer the journal to continue operating under its existing subscription model (the remaining 5% did not respond to the question). This response from the NAR community was an important driving force behind our subsequent decision to experiment with full open access. It is, however, also worth noting that additional factors were at play. At that time NAR was experiencing higher levels of subscription attrition than other comparable mature journals on the Oxford Journals list. This may have been influenced by NAR s move to a six-month free back-archive policy in November 2001; however, it is very likely that many factors were involved (e.g. the cancellation of duplicate print subscriptions). At the same time, NAR was growing significantly (in terms of the number of articles published), owing to an increase in submissions (Figure 1), and this growth necessitated above-inflation subscription price increases. As a university press, our focus is on fair pricing. Whilst we knew that the price per page for NAR would continue to be fair, we were concerned about the potential impact of continued price increases (in absolute and percentage terms) on NAR s subscription base. The combination of subscription attrition and the need for further subscription price increases was a concern, and reinforced our decision to explore alternative models. In January 2005, therefore, we launched a fully open access model for NAR, fundedby a combination of author-side charges, institutional memberships, and print subscriptions. the combination of subscription attrition and the need for further price increases reinforced our decision to explore alternative models

3 202 Claire Bird from the beginning we knew that the success of the journal relied on the continued support of its authors Figure 1. Number of NAR submissions received Table 1 provides an overview of the main elements of NAR s charging model under author-side payment open access in 2005 and in subsequent years. Under the institutional membership element of the model, corresponding authors based at institutions that have purchased institutional membership in a given calendar year are eligible for the reduced member-rate open access charges. Author behaviour From the beginning of the experiment we knew that the success of the journal relied more than ever before on the continued support of its authors. It is vital that the journal remains an attractive place to publish compared with its competitors. While other factors (including impact factor, speed, and quality of peer review and publication) are known to be of greater importance to the Table 1. NAR s full open access charging model Item Rate type Price Notes NAR institutional membership NAR print subscription NAR author-side open access charge 1,423/$2,459 1,500/$2,855/ 2,250 1,900/$3,515/ 2,850 Corresponding authors based at institutions with NAR membership are eligible for the member rates on author-side charges. 1,652/$2,855 1,652/$2,855/ 2,478 1,735/$3,210/ 2,603 In 2005, complimentary membership was included with a print subscription. This wasnotthecasein 2006 and Non-member 900/$1,500 1,000/$1,900 1,250/$2,370 Member 300/$ /$ /$1,185 Developing countries non-member 0/$0 to 300/$500 (depending on country) Developing countries member JISC Members: 0; non-members 600 0/$0 to 350/$665 (depending on country) 0/$0 0/$0 0/$0 0/$0 to 420/$800 (depending on country) N/A N/A Supported by a JISC grant to NAR in 2005 only

4 Oxford Journals adventures in open access 203 Table 2. Rates paid by NAR authors Period % requesting waiver % paying member rates % paying non-member rates 2005 ~8% (inc. 3% funded by JISC) est majority of authors, 1,2 the costs of publishing are undoubtedly a consideration for many, and thus the author-side charges required under an open access model might, at some level, become a barrier to submission. It is early days, but there is evidence to suggest that, by and large, this point has not yetbeenreached.asshowninfigure1,submissions have remained stable. (The steep increase experienced between 2002 and 2004 can be attributed to rapid growth of the two special computational biology issues mentioned above, as well as the implementation of online submission.) Table 2 shows the actual fees that NAR authors are paying to publish in the journal. The percentage of waivers granted has remained fairly constant at 6 7% as the open access charges have increased. Of those paying the requested charge, the percentageeligibleforreducedratesbyvirtueof institutional membership has varied. Most notably, in 2005, 64% of authors paid the member rates. This figure is higher than in subsequent years because complimentary institutional membership was included with a print subscription in A higher percentage of authors was entitled to member rates in 2006 than in 2007, because the complimentary institutional membership element of the 2005 print subscriptions was extended to January/February of In April 2006, we conducted another large-scale survey of the NAR community to try to understand how the journal s open access model was affecting authors and readers. 2 We received 1,144 complete responses (giving a 9% response rate). The responses were divided into different subsets depending on whether the respondent had recently published a paper in NAR. For example, 283 of the respondents stated that they had published one paper in NAR during 2005 (the first year of full open access) as the main/ corresponding author (we refer to this subset in our analysis of the results 2 as subset 2005: 1 paper: main author ). As part of the survey, we outlined the 2006 NAR open access charges ( 500/$950/ 750 for institutional members; 1,000/$1,900/ 1,500 for nonmembers) and asked respondents whether they would submit to NAR in 2006, if they had an appropriate paper. In the 2005: 1 paper: main author subset, after considering these charges, 49% said that they would publish an appropriate paper in NAR and wouldbeabletopaythefullopenaccess charge; 32% stated that they would publish a paper in NAR if their institution had membership of the journal; 11% stated that they would publish a paper in NAR but only if the open access charge could be waived. The results from this subset were not significantly different from all other respondents. In reality, as mentioned above, we are finding that only 6 7% of authors are requesting a waiver of the open access charge, and that the majority are paying the full open access charge. Author satisfaction Submission levels provide one indicator for author satisfaction with NAR. Wealsoconduct regular author and reader surveys, both through one-off, large-scale surveys as in April 2006, and on an ongoing basis. In the April 2006 survey, 92% of the 2005: 1 paper: main author subset gave the experience of publishing with NAR a satisfaction rating of 4 or 5 (where 5 equals very satisfied, and 1 equals not satisfied ). Six respondentsinthissubset(2%)selecteda rating of 1 ( not satisfied ). Following publication of his or her article, every NAR corresponding author is also invited to complete a survey on the experience of publishing in the journal. Responses received February July 2007, from 119 authors, show only 6 7% of authors are requesting a waiver of the open access charge

5 204 Claire Bird Figure 2. NAR income by type. for many authors, open access is just part of the picture when considering where to publish that NAR authors generally remain happy with the journal as a place to publish their work. 54% said that the experience was better than other publishers, 34% said that the experience was equal to other publishers, and 6% said the experience was worse than other publishers (the remaining 6% answered don t know or not applicable ). We also asked the 2006 survey respondents whether they would have published their paper in NAR if the journal had not offered open access. Of the 283 respondents in the 2005: 1 paper: main author subset, 79% said Yes, 14% said Don t know, and 7% said No. This illustrates that for many authors, open access is just part of the picture when they are considering where to publish their work, and is often not the deciding factor. It will be interesting to see whether this picture changes over time, perhaps through external pressures on authors from their funding bodies. The financial picture The overall financial aim for NAR s open access experiment was to maintain revenue levels. However we knew that it would be impossible to replace subscription revenue overnight with revenue from author-side charges, and so we aimed to make the transition to full open access via a mixed revenue model, funded by author-side charges, institutional memberships, and print subscriptions. Predictably, print subscriptions have declined under the open access model, by about 20 25% per annum, and so we have needed to increase the author-side charges steadily as shown in Table 1. Figure 2 shows how the composition of NAR s revenue has changed following the journal s move to full open access. In 2004, 83%ofthejournal srevenuecamefromsubscriptions, whereas in 2007, author-side charges made up 52% of the revenue, with print subscriptions dropping to 34%. Figure 3 illustrates how the revenue received per article has changed during NAR s transition was something of an anomaly, in that the journal received additional open access charge revenue from the two special issue experiments while still being sold under subscription as normal. The revenue per article fell in 2005 and 2006 (although there was a

6 Oxford Journals adventures in open access 205 Figure 3. NAR income per article. 6% increase in articles published in 2006), but increased in 2007, in part due to an increase in author-side charges. For NAR to remain viable we need to receive sufficient revenue to cover both direct costs (e.g. peer-review system, editorial honoraria, copyediting and typesetting) and indirect costs (e.g. staff salaries and overheads). In addition, we need to make a surplus, which, as a university press, we reinvest into further publishing developments, and directly into the academic community via contributions to our parent university. The surplus made by Oxford Journals over the last decade has been put to particularly important use in funding online development, including the introduction of online submission, rapid online publication models, linking functionality, and alerting services. These would not have been possible without thesurplusgenerated,andofcoursewewill need to continue to invest in developments inthefuture.atthesametimeasintroducing full open access for NAR, wehavebeen able to make some significant cost savings across the Journals division, primarily through new technology and outsourcing. Overall we are pleased to say that NAR remains financially viable at this time. Looking to the future, we must assume that print subscriptions will continue to fall, although they may of course plateau at some point. We will therefore need to increase author-side charges further indeed, we have increased author-side charges by 10% for Of course, the continued success of NAR depends on its authors being willing and able to pay these charges, and on funding bodies making the money available to them. We will continue to gather vital author feedback, and to do all that we can to make sure that NAR remains attractive to authors. Optional open access Overview of optional Oxford Open Following our initial experiments with NAR, and interest from other journal communities, in July 2005 we launched an optional open access model 3 for approximately 20 Oxford Journals-owned titles, across a range of disciplines. Since then, we have followed suit with further titles, including a number published on behalf of learned societies; at the time of writing, 68 journals are part of the initiative. Echoing the NAR institutional membership model, we decided to offer corresponding authors significant discounts on the charge payable for optional open access if they are based at institutions with an online subscription to the relevant journal. This adds value to the institutional subscription under the hybrid open access model. Current author-side charges for the majority of optional open access journals are 800/ $1,500/ 1,200 for those based at subscribing institutions and 1,500/$2,800/ 2,250 for others. Further discounts are available for authors based in developing countries. we need to receive sufficient revenue to cover both direct costs and indirect costs; in addition, we need to make a surplus

7 206 Claire Bird Table 3. Uptake of Oxford Open by subject area in 2007 Subject area No. of journals Articles published Open access articles Medicine 30 5, Life sciences 19 3, Social sciences and humanities Mathematics Total 65 10, %uptake uptake was highest in the life sciences Uptake Across the 65 journals offering the Oxford Open option in 2007, overall uptake was fairly low. Table 3 shows that it was highest inthelifesciences(averageuptakeapproximately 11%), unsurprising given that the open access movement has a higher profile here than in the other subject areas and that funding is usually greatest in the life sciences.averageuptakefortheparticipating medical and mathematics journals was 5% in 2007, followed by 2% for participating social science and humanities titles. Access to funding to cover open access charges is likely to be limited in the latter areas. However, these averages mask the fact that for a handful of journals in molecular and computational biology, uptake is considerably higher. For example, uptake for the Oxford Journals-owned title Human Molecular Genetics was approximately 17% in 2007, while for Bioinformatics it was around 24%. Like NAR, Bioinformatics is a well-established journal; it is publishing its 24th volume in 2008 and publishes around 600 articles each year. Computational biology is a booming area and open access is a particularly hot topic in this community. In a 2004 survey, we asked the journal s authors whether they would choose to pay in the region of 900/$1,500 for optional open access. Of the 901 respondents who answered this question, 44% said yes. 42% said no, because they did not have the necessary funds, while the remaining 14% said no, although they would have funds available for this. These results encouraged Oxford Journals to make Bioinformatics one of the first of our titles to introduce optional open access. The uptake of 24%, while higher than for other participating journals, is still lower than might have been expected from the survey response. The open access charging scheme for Bioinformatics is slightly more complicated than the standard optional Oxford Open model; there is a two-tiered charge depending on article type (Original Articles and shorter Application Notes). Overall, the majority of authors are based at institutions with online access to Bioinformatics, and therefore pay the discounted rates. For example, in 2006, 87% of authors choosing the open access option for Bioinformatics were eligible for the subscriber rates and therefore paid open access charges of between 400/$750 and 800/$1,500, depending on the article type. Effect on subscription prices When we launched the optional Oxford Open model we promised our library customers that the online subscription prices of participating journals would be adjusted in future years, taking into account the percentage of content published under the author-side payment open access model. Our standard policy for 2008 is to price the online-only (and print-only) subscriptions to our journals at 95% of the combined (print and online) price. In the case of journals with Oxford Open uptake, we have applied a further discount to the 2008 online-only price based on the amount of open access content published in This means that the average 2008 online-only price increase for the 28 Oxford journals with open access uptake in 2006 is just 1.7% much lower than the average increase of 6.9% across all our titles. Because pricing is dependent on

8 Oxford Journals adventures in open access 207 many factors, including variations in page extent and exchange rate adjustments, as well as open access uptake in previous years, open access adjustments do not always result in an actual price decrease from one year to the next; they may simply reduce the necessary price increase. However, when all factors are taken into account, eight Oxford Open titles are seeing an absolute reduction in price from 2007 to As yet, it is too early to say whether we will see greater subscription attrition for titles participating in Oxford Open due to the open access option, but we are of course watching this closely. Optional open access chargesarecurrentlythesameforthemajority of the participating journals. We are undertaking modelling to determine whether we need to introduce different rate bands based on each journal s own financial picture and the level of uptake that each title is experiencing. Usage and citations In order to determine whether open access can be shown to have an impact on online usage and citations, we are working with CIBER (University College London, Centre for Publishing) and LISU (Loughborough University), looking at trends for both NAR and optional Oxford Open journals. This is a complicated picture, with many factors at play, and it is still too soon to tell whether open access has a direct effect on the number of citations an article receives and on the time taken for an article to be cited. In their June 2006 report, 2 CIBER concluded that the significant increase in online usage experienced by NAR in recent years can largely be attributed to the opening up of the journal to search engines, although the move to full open access may have increased usage by a further 7 8%. The CIBER group is currently analysing usage data for open access and non-open access articles in optional Oxford Open journals with significant uptake, to see whether further conclusions can be drawn. Practicalities Along with valuable insights into the views and behaviours of our authors, and the potential impact of open access on journal finances, we have also learnt a great deal about the practicalities of implementing author-side payment open access models. The following is not intended to be an exhaustive list, but journals considering experimenting with author-side payment open access for the first time may find it helpful to consider these areas: Defining what is meant by open access: Itis widely accepted within the open access community that open access means more than simply free access. Peter Suber has written extensively on this in his SPARC Open Access Newsletter. 4 When we refer to open access content at Oxford Journals we mean that the content is not only freely available online, but also free to reuse for non-commercial purposes, without the need to seek permission. We make a clear distinction between free access content (for example, content previously published under subscription that we choosetomakefree12or24monthsafter publication) and open access content published under the open access author licence. However, other publishers may define open access differently; therefore, journals need to decide what they mean by open access and ensure that this is clear to authors and readers. The open access licence: Oxford Open content is published under the above terms with a Creative Commons Attribution Non-Commercial License ( creativecommons.org/licenses/by-nc/2.0/ uk/).itwasnotessentialthatwemovedto Creative Commons licensing for our open access content. However, the Creative Commons approach brings certain benefits; in particular, the terms of reuse are immediately available to the user and the machine-readable element (incorporated into the article tagging) helps search engines and other applications to identify aworkbyitslicenceterms.creativecommons is increasingly being seen as the standard licensing approach for open access content, and is recognised by the open access community. Timing of payment: From the beginning of our open access experiments we felt it was journals need to decide what they mean by open access and ensure that this is clear to authors and readers

9 208 Claire Bird models funded by author-side charges seem unlikely to be viable across all disciplines very important that the open access decision and payment process be kept separate from the editorial review process. Therefore, for optional Oxford Open journals, authors make their decision about open access after manuscripts have been accepted for publication. Likewise, for the NAR full open access model, open access payments and waivers are administered by Oxford Journals following acceptance. We believe that it is very important to retain editorial independence and for authors to be reassured that their open access decisions will not influence editorial decisions. The need for a considered waiver policy and mechanism: NAR s full open access model requires a clear mechanism for authors to apply for waivers on grounds of financial hardship. As the capacity to grant waivers is limited, it is also important to have a clear and consistent means of deciding who should be granted a full or partial waiver.thisisnotaneasytaskandisa matter to which we continue to give considerable thought. In doing so, we will bear in mind which funding bodies have stated that they will cover the costs of open access publication. In instances where waivers are granted, this is stated in the article s acknowledgements section. When an author changes his or her mind: We have also placed great emphasis on ensuring that open access options for authors are clear and widely promoted. However, authors may not always understand the charges involved, or may simply change their mind down the line after selecting open access on their author licence form. Under the optional model, our policy in these instances is to revert to the normal subscription model if we are able; for example, if the article has been published online ahead of print under our Advance Access model, but not yet in an issue, we can replace the article with a closed access version showing a standard (nonopen access) copyright line. However, if the article has already been published in an issue of the journal in its final definitive form, we do not reverse the open access status of the article. If this is the case we must then decide whether to pursue payment of the author-side charges, and this will depend on the individual circumstances of the case. To sum up, open access experimentation is providing us with hard data and practical experience of this emerging publication model, and we would encourage other journals and publishers to share findings from their own experiments. Models funded by author-side charges seem unlikely to be viable across all of the disciplines in which we publish, so we foresee a future of diverse models, ranging across full open access, optional open access, delayed free access (e.g. where journal content is made freely available 12 months after publication), subscription access, and every combination in between, depending on the journal. One size does not fit all we must continue experimenting and supporting a wide variety of publishing models, reflecting the different communities which we serve. Note This article is based on presentations given by the author at the following events: ALPSP Seminar: Experience of Open Access Looking at Data, 22 June 2007, London; and the STM Frankfurt Conference, 9 October References 1. Rowlands, I., Nicholas, D., and Huntingdon, P. Scholarly Communication in the Digital Environment: What Do Authors Want? CIBER, ucl.ac.uk/ciber/ciber-pa-report.pdf 2. Assessing the Impact of Open Access Preliminary Findings from Oxford Journals, June htm Claire Bird Senior Editor Oxford Journals Oxford University Press Great Clarendon Street Oxford OX2 6DP, UK Website: claire.bird@oxfordjournals.org