Redundant publication and salami slicing: the significance of splitting data

Abstract

When researchers are under pressure to constantly increase the number of publications to their name, the temptation to split one set of results, or data set, into many articles is understandable. Though alluring, this is generally considered unethical practice. Whilst not such a serious problem as fraud or plagiarism, such over-publishing wastes the time of editors and reviewers (refereeing multiple articles, etc.), and may mislead the readers as well as waste their time. Producing many articles from a moderately sized research project might give it undue significance – something which could initially appear beneficial to the research team responsible. But splitting the data into segments may also affect the statistical significance of each part and possibly undermine the findings themselves, thus changing an important result into several moderately interesting results. Such splitting of results to produce multiple papers is called redundant publication or salami slicing, and is sometimes derisively described as the least publishable unit. Redundant publication is not a new problem, as evidenced in a 1995 editorial from the New England Journal of Medicine: 'Nobody is well served by the practice of reporting the same study in two journals, publishing a review of the same subject nearly simultaneously in two journals, or splitting a study into two or more parts and submitting each to separate journals.'1 The editorial also acknowledged justifiable reasons for redundant publication; for example, wishing to reach different audiences. Other justifications may include a desire for a longer discussion on one aspect of the results than is allowed in the standard article word count. It is very difficult to estimate the prevalence of redundant publication (or redundancy), as the basic plagiarism-checking tools will often not find it. But the CV of any researcher will identify if they have followed the route of least publishable unit, or collated their research into larger, more impactful articles. Whilst the practice of generating inflated numbers of articles is generally frowned upon, it changes from being an irritation to unethical behaviour when later papers ignore earlier ones, and do not acknowledge that they stem from the same research. Such declaration is needed for the same reasons that Conflict of Interest statements are required – to provide transparency. The Committee on Publication Ethics (COPE)2 places emphasis on citation in their definition of redundant publication: 'When the same (or substantially overlapping) data is presented in more than one publication without adequate cross-referencing/justification, particularly when this is done in such a way that reviewers/readers are unlikely to realise that most or all the findings have been published before.' Authors must decide how best to write up their research, where to submit it, and which audiences they want to reach. They could decide to split their data into several articles rather than one3 and their decision may be valid and reasonable. However, they should have a good reason for their decision and be prepared to declare it. If they are not prepared to do this, then unfortunately it will be assumed that there was no good reason.4 Also, if they do not cite the other works based on the same research, this implies a certain amount of duplicity and even a wish to mislead the readers. Authors need to carefully consider their publication strategy and consider both long- and short-term repercussions. Redundant publication or salami slicing might appear immediately rewarding, but in the longer term may undermine a reputation – and without due acknowledgement can directly harm it.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 23 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.