Meta-Research: How problematic citing practices distort science

Abstract

Citing practices constitute a core element in scientific research and communication in which they serve several important functions. They both comprise the principal unit of science’s social reward system and they establish epistemic genealogy, showing the foundations on which claims are built. Distorted or problematic citing behaviour can hence have major harmful consequences for the well-functioning of the scientific enterprise. In this article we employ a case study approach to show how, even in the field of research integrity, or more broadly, meta-research, researchers commonly adhere to suboptimal citing practices. Our findings highlight, among other shortcomings, an apparent lack of critical engagement with the cited literature, leading to incorrect reproductions of claims and overgeneralisation of supposed research findings. We link such problematic citing practices to the state of crisis that science is currently claimed to face, discussing how these practices might be both a symptom and a cause of these crises.

Non‐author entities accountable for retractions: A diachronic and cross‐disciplinary exploration of reasons for retraction

“…As disclosed in the retraction notices, use of third parties' retracted research or publications led to the retraction of news articles, letters, commentaries, reviews, and meta‐analyses but did not result in the retraction of other researchers' original research articles. However, given the trends of citations to retracted publications reported in numerous previous studies (e.g., Al‐Ghareeb et al, 2018; Bolland et al, 2021; Dinh et al, 2019; Fanelli et al, 2021; Horbach et al, 2021; Moylan & Kowalczuk, 2016; Rubbo et al, 2019; Suelzer et al, 2019), it would be very likely that findings of some original research articles based on third parties' retracted publications may have been adversely affected and thus should have been retracted. Therefore, third parties' liability for retractable publications is likely to be much greater than that indicated by the frequencies presented in Table 2.…”

Relatedly: Scaffolding Literature Reviews with Existing Related Work Sections

“…Another future direction for further improving Relatedly is to analyze the importance of each reference in the context of the paragraph they were mentioned. For example, NLP techniques such as [70] could be used to estimate the level of influence of each references in a paragraph, so that Relatedly could mitigate the effects of bulk and passing citations [26,27,37].…”

Automated citation recommendation tools encourage questionable citations

“…In the worst case, critical engagement could even harm career opportunitiesbe it through substantive (e.g., being proven wrong) or social ramifications (e.g., lawsuits and bullying). The 'publish or perish'-culture even incentivizes exerting less diligence in digesting the literature (e.g., Edwards & Roy, 2017), as closely reading and evaluating cited literature is not rewarded and its neglect not likely to be punished (Horbach, Aagaard, & Schneider, 2021).…”