A central attribute of any academic article is clear evidence that the article’s authorsFootnote ¹ have read and critically evaluated the extant scholarship. To demonstrate their critical appreciation of research on their topic or their acknowledgement of their debt to other researchers, authors must cite the sources which they have read, evaluated, and used in their work. This brief essay considers what is needed in citing sources. It does not consider the varieties of form of citation – Oxbridge versus Harvard versus Vancouver or psychologists. Indeed, to do so would be to sink into an immensely detailed, Kafkaesque labyrinth. Rather, I would like to explore what should be cited, how much citation is enough/too much/too little, and what are some of the doubtful or even fraudulent citation practices to be avoided at all costs. This is because, despite some clear and accepted but tacit ‘rules’, citation of sources is a contested item with some vexed possibilities.

What to cite requires scholarly skills, and some ethical and moral decision-making. As honest scholars, we must acknowledge our sources of information. However, generations of students (in humanities and social sciences) have asked their tutors, ‘How much is enough?’. And – it seems – some journal authors should have asked the same question.

Citation is core to academic and scholarly progress. Authors need to show they have a strong critical understanding of broad theoretical and / or empirical / policy literature, precisely so that they can show how their research findings are contributing to the scholarship of the topic. Moreover, researchers do not act in a vacuum. Their ideas / perceptions / theories and methods have been learned from other researchers, and authors need to show from where the foundation ideas of their research have come.

Some authors are much too frugal in acknowledging their sources of ideas or their appreciation of the research area. One article recently submitted to this journal had only five citations for an (admittedly under-length) 4000-word article. A manuscript with no or very few citations is not an academic article but rather an essay – that is a reflection, discussion, or commentary on a particular topic. Essays are written by scholars, journalists, authors, or analysts to present an often valuable argument or perspective, but they are not scholarly pieces of research. Rather, an essay offers readers the authors’ views, understandings and arguments, but without substantiated evidence of where the authors have gained their ideas, or how they think their ideas will contribute to research and scholarship. As such, the essay does not need to show how it is adding to wider aspects of the field of research. That is the task of the academic article or academic book, where the authors demonstrate their evident contributions to the extant research, while also acknowledging their scholarly debts.

On the other hand, some authors have far too many citations. Another potentially interesting article submitted recently contained dozens of citations on each page. The article comprised an occasional sentence followed by a desperately long, boring list of citations (and no apparent appreciation of the substance of anything cited). To us sour and cynical editors, such long undifferentiated lists suggest that the authors have downloaded a list of titles and may not have even read critically some of the cited articles. While this is pretty easy to do these days, it is not good scholarship. (The extent to which AI should be used for developing literature reviews is still under discussion – and being carefully avoided here).

Certainly, it is easy enough to obtain a long list of articles on a topic. Just ask your favourite AI friend, such as ChatGP or Copilot, to make a list of important and relevant articles using chosen search terms. Your BFF AI will then seemingly do your literature review in under a minute! Another recent (and rejected) article submitted to this journal offered two pages of citation over the course of a 4-page ‘literature review’ – which of course was not a literature review at all but a lightly linked list of possible sources. (See Kelly Reference Kelly2025). Perhaps these authors were trying to show they had read **everything** on their topic. Instead, however, they showed there was a lot of material on the topic but they offered nothing about its quality, theory, criticality, diversity direction, or emphasis. In terms of citing sources in literature reviews and the like, the point to remember is that authors need to demonstrate that they have read critically and understood the debates, so they can acknowledge their scholarly debts and demonstrate their own contributions to scholarship or policy.

In terms of acknowledging scholarly debt, authors should always cite sources that inform assumptions in an article. This is especially true of general statements that underpin the authors’ assumption. A recent article claimed that ’the gig economy is a labour market model based on the completion of short-term tasks known as gigs’ (with no further citation for ten lines). Authors of another article began their discussion, ‘Since Global Value Chains (GVCs) are the best drivers of economic growth, and depend on the link with trade patterns, our research will …. .’. Again, no citation for a couple of paragraphs. Yet within both these assertions are some evidently contestable elements. Authors must – at least – identify authentic and recognised sources, not only for quotes and paraphrased material but also for significant ideas on which assumptions, theories or choices of methodology are made. To add an in-text citation (for this journal) is a small demand, but it substantiates the provenance of authors’ ideas – surely the foundation of good research?.

So what is sufficient but not excessive citation? There is no set rule. Knowing what levels of citation are apt is a learned skill requiring novice researchers to reflect on what is ethical, appropriate and useful to cite. Outside of a literature review, and as a rough guide only, there may be 2–3 citations in each paragraph when introducing an article or its theoretical or methodological frameworks and perhaps 1–2 citations per paragraph in other sections, sometimes fewer, such as when, for example, discussing results. The ethically sound author learns to acknowledge (or even boast about) their debts for their underpinning intellectual ideas or even the basic sources of data, i.e., it is good to show your reader that you have read the classics as well as recent great research. Indeed, do not hesitate to highlight your excellent research skills – and your wise and critical appreciation of your sources. These are good practices that, in time, become automatic. On the other hand, to not acknowledge your sources – the great, the good and the doubtful – is to plagiarise. To plagiarise is to defraud, to claim, (overtly or tacitly) that the ideas, concepts, perspectives, heuristics, or frames of reference as one’s own. If you do not cite your sources, then your reader will logically assume they are your own ideas / methods theories …. . So – paradoxically and contrary to concerns of over-citation noted above – it is better to acknowledge one’s scholarly debts as clearly as possible. In so doing, you ensure the trustworthiness of your science and the validity and substance of your research.

Of course, not all citations are good citations. Practice in recent years has shown the growth in new problems with citations. Firstly, there have been what kindly folk call ‘ghost citations’. These are not quite honest, in some form or other. One form of ghost citation is almost traditional. It has been around for decades – and still gives editors (and supervisors) much annoyance. In these (version A) ghost citations, the authors will claim the (undoubtedly excellent) scholarship of, say, Geoff Harcourt (economics) or Richard Hyman (industrial relations). To make the claim that Harcourt or Hyman has said XYZ, gives considerable substance to the authors’ assertions or assumptions. These two scholars are very highly respected in their fields, and their scholarship is similarly very sound. However, some authors who acknowledge their debt to the ideas of great scholars do not cite any actual work of Harcourt or Hyman. Rather, they cite a textbook or a recent (and possibly obscure) journal article which, in turn, had claimed to cite Harcourt or Hyman. However, unless they checked the verity of the citation – by checking Hyman’s or Harcourt’s actual cited publication – how can they be sure of the integrity of their claim? That was the ‘old’ ghost citation – the author makes claims for legitimacy based on great and grand scholars whose work they have seemingly not read!

Perhaps even more morally bankrupt has been the growing use of false citations, or references to non-existent publications (Biagoli and Lippman Reference Biagioli and Lippman2020; Chambers and Mol Reference Chambers and Mol2022; Naddaf and Quill Reference Naddaf and Quill2026; Reznik and Hosseini Reference Resnik and Hosseini2026). Some call them ‘ghost citations’; others define them as hallucinated, fabricated, or zombie citations or ‘Frankencitations’. Many of us would just call them fraud.

These are technology-enabled, made-up citations, seemingly created by large language models (LLMs) (Warner Reference Warner2026). The growth of these invented citations (of non-existent publications) has become an evident problem, particularly for the physical / engineering / medical sciences. This is because contemporary fierce competition among scholars, especially for laboratory space and costly equipment and materials and / or the potential for commercialisation and commercial gains, has put immense pressures on researchers (Bauchner and Rivara Reference Bauchner and Rivara2026) to increase output, leading to, sometimes unethical, shortcuts.

Moreover, while the problem is rather less in the social sciences and humanities, Resnik and Hosseini (Reference Resnik and Hosseini2026), drawing on the outstanding work of Retraction Watch (n.d.), have noted that it is still evident. ‘In scholarly publications, there have been some noteworthy examples of hallucinated citations. For instance, in a paper published in the journal Academic Ethics, 19 out of 29 citations were hallucinated.’ As Resnik and Hosseini (Reference Resnik and Hosseini2026) continue, ‘A hallucinated citation may support flawed hypotheses or theories, lead researchers down blind alleys, waste scarce resources and undermine the overall integrity, reliability, and trustworthiness of research’.

And that of course is a serious challenge to the core attributes of much public scholarship – its trustworthiness, founded in its evident academic integrity (Gray Reference Gray2026; Bruton et al Reference Bruton, Macchione, Brown and Hosseini2025). Indeed, these are the basis of all good scholarship, whether for a book, a scholarly journal, or a conference presentation. If we cannot trust others’ scholarship, if we are not building on the research of all of our ancestors, then our work is merely a flutter of self-indulgence.

We, who are scholars – inside or outside the academy – research because we like to investigate, question, explore, and build on the research that has gone before. In so doing, we have built our skills in research, analysing, entangling and disentangling ideas, and communicating those ideas to students in classrooms and in publications. Most of us are seeking to build on the ideals of integrity, ethicality and expanding knowledge to make the world a better place. The use of fabricated / autonomous / manufactured citations is contrary not only to those fine ideals but also to the functionality and effectiveness of scholarship.

In these respects, the ELRR is wholly committed to upholding academic integrity by working to ensure that fraudulent citations of any kind will have no place in this journal.