Solicited self‐referencing undermines the credibility of researchers and journals

Abstract

We wish to draw the attention of the readership of this journal and of the scientific community at large to what recently happened to a colleague of ours. The letter accompanying the editor’s request for major revision of a manuscript included the following surprising advice: “The Editors would also greatly appreciate you adding more than two but fewer than six references of articles published in [the Journal involved], above all articles published over the past 2 years.” A rapid survey among a few colleagues told us that this type of editorial policy is not as exceptional as one might believe. Another editor conveyed the same message to a prospective author: “We would like to emphasize that we attach great importance to cross referencing very recent material on the same topic in [this journal]. Therefore, it would be highly appreciated if you would check the last 2 years of [the same Journal] and add all material relevant to your article to the reference list”. These requests came from two journals with different impact factors (i.e. the index originally devised for the benefit of librarians to help them select scientific journals) [1]. The original goal of the impact factor as a tool for librarians has changed, and now it is widely used to rank the prestige of scientific journals among prospective contributors and ultimately to evaluate the scientific quality of contributors. It is likely that the forementioned editors’ requests for self-citation of articles published in the previous 2 years was meant to increase the number of citations and hence inflate the impact factors of their journals. While it is legitimate that a scientific journal is pleased to see its impact factor rising as a consequence of the increased recognition of its published articles, it is obviously unacceptable for the increase to be artificially triggered through the practise of soliciting self-citations. If this kind of request from editors spreads, prospective authors may soon feel obliged to refer to articles from the journal to which they are submitting their work, even when citations are unnecessary or irrelevant. With time this may become a custom and a way to capture the benevolence of editors, though not serving scientific merits and aims. Solicited self-citations not only bias a journal’s image, they also detract from a researcher’s authority and public image. Ultimately, solicited self-citations could well undermine the credibility of the peer review process, within the scientific community as well as in the public opinion, and hence of scientific information itself. How frequent is the habit of editors of soliciting self-citation? No information is available to our knowledge, and it could anyway only stem from people who, like us, inform the scientific community of these habits. The Institute for Scientific Information (ISI) calculates the impact factors of many scientific journals and publishes them yearly in the Journal of Citation Reports. [1] The ISI also calculates a ‘corrected’ impact factor after excluding the self-citations appearing in the journals examined (Table 1). What is the weight of self-citations on the impact factor of scientific journals? We took a look at the Journal of Citation Reports and found that in a sample of journals listed in the Hematology category ranked by their impact factor in 2010, the rate of self-citations ranged from 0 to 35% (Table 1). A poor correlation was observed between impact factors and percentages of self-citations (Pearson correlation coefficient, r = −0.30). Although in most instances self-citations do not change dramatically the impact factor, in some instances the latter is substantially inflated (Table 1). Admittedly in some instances a high rate of self-citations may be related, rather than to an editorial policy, to the fact that the corresponding journals cover a very narrow scientific field (such as, for instance, Haemophilia, see Table). With all this as background, we think that it might be useful if the International Committee of Medical Journal Editors (ICMJE) consider measures to counteract solicited cross-referencing. One obvious proposal seems to be to exclude the self-citations from the impact factor calculation. Perhaps this would prevent not only explicit requests for cross-referencing from editors but also conscious or unconscious spontaneous adoption of this habit by too zealous scientists. We thank L. Cortesi for the statistical analysis and J. Baggott for editing. Supported by internal institutional funds of the Mario Negri Institute, Milan. The authors state that they have no conflict of interest.