Patrick Dunleavy @Write4Research and London School of Economics With inputs from: Jane Tinkler (Nine Dots Prize) , Sierra Williams (PeerJ), David Ross (Sage Open Access) , Kieran Booluck (LSE Impact blog) , Beth Clark (LSE Library, Head of the Digital Scholarship and Innovation Group) , and Nancy Graham, (LSE Library, Research Support team head) Executive Summary Many current academic citation and referencing practices are out of date and dysfunctional, especially in leading only to closed-access and paywall sources, or in providing only details of ‘legacy’ print formats. The central principles of this Digital Style Guide are that 1 All citations/ references should lead wherever possible to a digital text, database, or other information source. 2. The primary source cited should wherever possible be open access — so that it is immediately available in one click to readers. Closed access and paywall sources should always be cited as well, but now in a secondary role. 3. Where an open access digital source is not available, referencing only closed access and paywall sources will have to suffice, as now. However, authors should try to provide clickable links to additional information useful to readers. We suggest economical ways of doing this. 4. Adapting to this new orientation requires some detailed but consequential changes in the items included in all citations, and in how they are triggered in texts and other outputs. We set out a modernized approach to citing, adapted to 21st century academic and scientific practice in an increasingly globalized academia. TOPICS COVERED A. Major issues and suggested innovations B. Smaller suggested changes in citations procedures C. Detailed Guide: How to cite different kinds of core academic sources A. Major Issues and Suggested Innovations 1. Seeking to provide digital open access citations raises a number of fundamental issues which need to be tackled in a concerted way. The rationale for open access and open science 2. The suggestions in this style guide are designed to radically and quickly improve the communication between university researchers and very large groups of readers worldwide who do not have comprehensive access to paywalled sources (such as most academic journals and books). An expensive facility like Science Direct, which allows those with very high quality access arrangements to the right journals to immediately consult any referenced sources (wherever it is located), shows the enormous potential here. 3. The main sets of people who would be helped by the far wider open access approach set out here are skilled knowledge workers: - Researchers, teachers and graduate students in smaller or less elite universities. Worldwide we estimate that there are around 60 million people working or studying in higher education. Staff and students at around 2,500 elite or better-funded universities have excellent digital access to journal resources via their libraries. But many other institutions have only limited or very partial access to journals, and often now relatively few books beyond online textbooks. - Teachers and students in near-university colleges and in further education systems, who number many millions but often have little access to any sources beyond textbooks. - Executives and professional in private sector corporations — ranging from multi-national through medium sized to small firms. Business people generally have no regular or easy access to academic journals and few research books — outside small specialist staffs in major corporation R & D labs. - Civil servants and government or public sector officials have little access to journals or academic books, and restricted budgets to acquire any more. Again there are some exceptions for small staffs in specialist R & D labs, and partial exceptions for public health care system doctors and some health professionals. - Staff in charities, non-governmental organizations, think tanks, and many professional organizations may have small libraries of key books but generally cannot afford journal subscriptions. - Most professionals in private practice and independent scientific research and advisory roles. - Consultants, media intermediaries, journalists and bloggers. 4. All these groups now play essential and enduring roles in sustaining contemporary complex civilizations, and in co-producing and disseminating knowledge that is vital for academic and scientific development. Collectively some experts have suggested that these kinds of knowledge workers number at least 120 million people worldwide. The goals of ‘open science’ and the open access movement generally recognize that university researchers must work co-operatively with as many others as possible to maximize the value of scientific and academic work. Increasingly too these goals have been strongly supported by government funding agencies across the UK and the European Union, Australia and other countries; and by almost all the world’s largest philanthropic foundations. For instance, every journal article that is to count for universities in attracting research support from the UK government in its 2021 audit of research (the ‘REF’) must be available in an open access form. Paywall-only articles are disqualified from the exercise and will attract no grant funding (although there are a few exceptions). 5. In addition, systematically fostering open access can help a far wider range of people with the interests and skills relevant to scientific and academic work to stay in touch with the disciplines from which they have graduated, and to find out about new areas of knowledge advance. Key groups here are: - independent researchers and authors working outside of a university affiliation; - PhDers and others studying at advanced level but not full time, or not at elite institutions; - school and college teachers; and - people who might have liked to pursue further study or an academic career but could not for various logistical or financial reasons; - interested alumni from your own university, especially those recently graduated, and others with particular aptitude or interest in an area of work; and - ‘lay’ people who have many good reasons to learn about and critically assess specialist knowledge in particular areas — such as patients with particular diseases or conditions, and their families and carers; or members of the public concerned about pollution problems, environmental conservation, historical heritage preservation or cultural development and creative arts. Huge numbers of people in all these categories constitute the ‘informed public’ whose support is crucial for the continuing development of scientific research and academic work across many different fields. Prioritizing open access sources 6. We are still transitioning to fully open access publication, and probably will be for the next fifteen to twenty years. In this interim period, different types of sources have to be handled in somewhat different ways. (i) The simplest case is where the full, final text is gold open access (OAg) , free of charge to all readers, with no paywall. For OAg texts published in as articles in a journal and as books, this solution generates the smallest and most efficient references, with no need for duplication. Include the digital object identifier (DOI) number in a web-clickable format (i.e. http://doi.org/ followed by the DOI number). If an OAg source has no DOI, there should be a permanent URL, which can be used instead. (ii) Where a paywall article or book has a ‘green’ open access (OA) version , the author has deposited a final manuscript version or even a proof read/edited version of the paper in a university or institutional repository, from which it can be downloaded free of charge. Here cite the green OA source first. Give the Repository Name and a permanent URL. The best institutional repositories will have their own distinctive shortened URL links (e.g. a Stanford University one looks like this: http://stanford.io/2fLBDow ); or they will issue DOI numbers for all their documents; or preferably do both. In addition , the paywall version of the text is then always cited second (see examples below). The formatting here is broadly as now, except for some small changes in conventions suggested below. (iii) For texts without a formal green OA version , the paywall or closed access version is cited first, again broadly as now. However, wherever authors know of a version of the text available on one or more widely used commercial sources — especially ResearchGate, Academia.edu or Mendeley — links to them should be cited second. These sites do not meet the ‘purist’ definition of green open access. Nor are their URL links guaranteed to be as permanent as institutional repositories. However, they are extensively used already by scientists and academics, and their value is widely appreciated by researchers and students. These sites do also provide ‘enduring URLs’, which empirically are much less likely to go out of date and be subject to ‘link rot’, and thus should stay valuable for readers over many years. (iv) Where a paywall text Z is digitally inaccessible but does have an earlier or alternate version Y , authors should cite Y as a separate entry, and link Z to Y. Check to see if you can find a related version of the text — such as a pre-print, conference paper or working paper. Text Y may differ from the final ‘version of record’ in Z. Yet Y helps most readers by bearing a ‘close family resemblance’ to the inaccessible text, and in many cases it may be near-identical. At the end of the citation for text Z, just put: ‘See also text Y’ . (v) Finally, if none of these strategies is possible , the closed or paywall source is cited alone, again broadly as now. Page numbers and precise quotation 7. Precision in quotation is an important scholarly virtue, enabling readers to check that a view attributed to another author has not been misquoted, manipulated or taken out of context. In the legacy print model page numbrs were the essential vehicle for achieving precision . However: - Page numbers are already completely different in ‘early online’ versions of journal articles that are later grouped into volumes and issues for a print version — often now with a lag of up to 18 months. In a digital world these early online versions may in fact remain fully available, creating long-running and not time-limited ‘version control’ problems. (And of course, early online versions are often lodged near-permanently in Mendeley and other peer-based repositories from which they are widely available still). Some top journals have responded by shifting to ‘continuous online’ publishing, sensibly abandoning altogether the ‘print legacy’ concept of issues (with pagination defined across multiple papers). - In digital texts page breaks can only be held the same as in print versions by using restrictive formats, like PDFs, which have many disadvantages still. Most HTML format articles are continuous and do not indicate where print page breaks occur. 8. As soon as we assign priority to open access and digital sources these already serious obsolescence problems become critical. Many important digital sources (like blogs) have no page numbers or page breaks, but can still be lengthy and substantial contributions to scholarly debate. And with other digital texts (such as books on Kindle or in ebook versions) readers themselves control the font size, and so change how pages appear and where page breaks occur. 9. We suggest that page numbers should be scrapped completely. Instead precise quotation should be realized in the following way. If author A wants to specifically cite a particular piece of author B’s work, then A must quote a minimum five or six word phrase in their main text that is distinctive to the passage being cited. A reader following the digital trail then clicks on A’s URL link to B’s text, crosses to it and from the top of B’s text uses Ctrl+F plus the quote text to find the exact passage involved. 10. In practice this means that we must completely abandon the familiar but vague style of reference where an author summarizes and cites a text, without specifically quoting from it. Without page numbers we can only cite like this: ‘Shifting functions from the government sector to off-budget bodies has been a core element in new public management (NPM) approaches (Dunleavy and Margetts, 2013)’. But we achieve precision by instead including a findable and distinctive short quote: ‘Margetts and Dunleavy (2013) consider “ externalizing services from the government sector to off-budget bodies” as a core element in new public management (NPM) approaches’. 11. Note that fixed numbering systems for exhibits — such as Figure, Table, and Box numbers — remain perfectly viable in the digital open access citation system. So too do section or paragraph numbers, often used in ‘grey literature’ reports; and numbering systems from older main texts, such as the books, chapters and verses of the Bible. Enhancing the reproducibility and replication of research 12. In some STEM disciplines — especially medicine, psychology and bio-sciences — some recent research has suggested that the results from many peer reviewed articles published in top science journals cannot be reproduced by the same investigators using the same techniques, nor replicated by other teams. Such concerns are corrosive of the public trust in science. Other studies in the social sciences have found that research teams can rather easily make mistakes in creating, cleaning or analysing datasets. 13. In response, increasing numbers of journals in quantitative disciplines have begun to require that researchers whose articles are accepted also make available a ‘replication archive’. This should contain their key data, along with sufficient information on variables and methods to allow other users to check their methods, processes and results in detail, and if appropriate to seek to replicate them. Such enhanced replicability requirements are harder to develop in the qualitative social sciences, such as studies relying on non-attributable interviews. The issue also remains very controversial in some humanities subjects, especially perhaps for historical or literary work based on un-digitized and hard-to-access paper archives. 14. Fostering greater open access can help to improve replicability in two main ways. First, a shift to more of a ‘show and tell’ ethos might (for example) see historians depositing with their articles or books their own numerous photographs of key passages from otherwise undigitized documents. Second, the computer scientist Eric Raymond summed up the ethos of the ‘open source’ movement in software development by saying: ‘Given enough eyeballs, all bugs are shallow’. By helping more readers to immediately find more source materials in their full form, open access powerfully counteracts the risks of academics misinterpreting other authors’ work; of researchers misreading or misquoting sources such as datasets, data tables or archive entries; or of technical mistakes in analysis going unnoticed in algorithms, statistics, spreadsheets, software or other models. 15. For all qualitative work in the social sciences, the Princeton political scientist Andrew Moravscik suggested two key rules for ‘ active citation’ in 2009: ‘Rigor : Any critical and contested substantive empirical point in a scholarly case study should be backed by a precise and annotated citation to one or more presumptively primary sources. Transparency : Citations must contain a hypertext link to a reproduction or transcript of some part of the source’. Moravcsik argues that these steps can boost the reliability of qualitative research by enhancing the quality of scholarship; encouraging wider criticism; creating richer secondary analysis and meta- analysis; and fostering greater inter-disciplinary engagement. A later 2014 piece develops his theme that ‘Transparency is the cornerstone of social science ’ — especially complete clarity in what the data actually show; how analyses have been conducted; and ‘production transparency’ covering how sources were selected. Our suggestions below are strikingly consistent with this approach, and could help create some of the building blocks that Moravcsik advocates. 16. In some preliminary groundbreaking work on these ideas, it became clear that the current partial development of digitization is seen by some humanities scholars as a basis for negating any move to greater open access sourcing of historical data and documentation by researchers themselves. On this view, academics have no role to play in making sources more open via their research: we should instead wait for some future, official ‘big bang’ digitization effort to get funded. To the contrary, we believe with Moravcsik that the everyday work of scholarship can rapidly have transformative effects on the digital availability of documents and information. To take a small example from the UK government. Here documents often become inaccessible in practical terms after any change of government (when ministers and policies shift). Whitehall departments send almost all their previous materials to the National Archives, which then takes time to get them back up online, and then only in part. Yet if academics have stored their own downloaded copies of reports etc on Mendeley or other peer-store sites, these materials remain immediately available for use by scholarly communities. What we are NOT saying 17. Earlier versions of many of the ideas here have been imperfectly expressed in blogposts, which have in turn generated very substantial volumes of Twitter commentary. Many Twitter participants have apparently responded to previous tweets they saw, but without reading the original blogposts, therefore missing some context for the ideas discussed. And scholars in some disciplines and specialisms — especially historians and archivists — have clearly reacted very adversely to some of the issues and sensitivities raised for them here. 18. So here is a list of completely false propositions that were generated in the Twitterstorm, and which we are NOT saying or advocating. B. Smaller suggested changes in citation procedures For all items Give Surname, Full First Name for all authors, and optionally one more initial as well. Rationale — simply giving one initial and surname is radically insufficient to create distinctive identities in a globalized academic world. Use semi-colons to separate out author name in long lists. Rationale — existing listing practices (using just commas and stops) assume that (Western) first names and surnames are clearly distinguishable. This is not the case with authors from many different countries and cultures. When OA and published texts are both available, always assume that the final, formally published text is the author’s preferred version. Rationale — this will generally be the case, and in STEM sciences especially the journal paper is also the ‘version of record’. The only exception occurs if you know that the author has explicitly signalled something contrary — e.g. by putting up an OA version that is newer, revised or updated compared to the paywall version (like a ‘Director’s Cut’ of a film). Use shortened URLs to replace any lengthy URL, especially perhaps non-permanent URLs. Rationale — For aesthetic reasons it is best to avoid long URLs, which are also liable to transcription errors, especially if there are gaps or underlines in addresses. Many long addresses (e.g. for blogposts and magazines) also repeat the title of pieces. A key disadvantage though is that if shortened URLs are wrongly transcribed, they become irrecoverable — so their resilience is low. In addition, there are different types of shortened URLs. Short links from university repositories are clearly best for their own contents. Some academic purists criticize commer