The authors have a conflict of interest or not, right? Mistaken. Digital Science research has discovered a network entangled by missing statements, errors and subterfuge, which highlights the need for a more careful evaluation of published research.
In this year World Conference on Integrity of ResearchA team of digital science researchers led by Pitha Sarkar presented a poster with findings of their deep immersion about the statements of conflict of interests (IOC). Entitled Conflict of interest: a data -based approach for the categorization of COI statementsThe initial objective was to analyze COI’s statements in order to create a binary model that determines whether or not a statement of conflict is present in an article.
However, not everything was as it seemed. While some articles had no IOC and others had one present, those present covered several different areas, which led the team to think that Cois could represent a spectrum instead of binary options.
Gold standard
The conflict of interest is a crucial aspect of academic integrity. Properly declaring an IOC statement is essential for other researchers to evaluate any potential bias in academic articles. However, those same researchers often find COI statements that are inappropriate or misleading in some way, even if they are present.
The digital science team: all working on the integrity of research with Dimensions – Soon he realized that the data could be further exploited to better explore the wealth inherent in the statements of Coi nuanced. After greater investigation and analysis, it was clear that COI’s statements could be classified into six different types:
- None declared
- Membership or employment
- Funds received
- Shareholder, interested or property
- Personal relationship
- Donation
This analysis involved writing out hundreds of COI statements with natural language processing tools (NLP). The objective was to create a gold standard that could be used to classify all other COI statements, however, despite the team’s diligence, a significant challenge persisted in the form of the ‘data asymmetry’, which can be defined as An imbalance in data distribution within a data set that can affect data processing and analysis.
Fatal failure
An irresistible conclusion for data asymmetry was simple: that the authors did not sincerely report their conflicts of interest. But could this really be true?
The standard gold approach came from manuals and experts notes to develop an automatic annotation process. However, despite the ability of the algorithm to automatically score 33,812 articles in just 15 minutes, the asymmetry that had initially identified persisted, which leads to the theory of false reports for the authors (see Figure 1 of the COI poster) .
To reaffirm this hypothesis, when the Retraction watches database It was analyzed, the worrying trend, including the discrepancy between the category of informed COI and the reason for the retraction, became even more apparent (see Figure 2 of the COI poster).
In addition, when the team continued with the investigation, they discovered that there were 24,289 documents overlapping in GBQ dimensions and retraction clock, and among those documents, 393 were retracted due to the conflict of interest. Of those 393 documents, 134 had a statement of IOC, however, 119 declared that there was no conflict to declare.
Conclusion
A sub -registration and erroneous information report of conflict of interests or types can undermine the integrity of academic work. Other problems of integrity of research around paper factories, plagiarism and predatory magazines have already damaged the confidence that the public has with published research, so other problems with the IOC can only worsen the situation. With the evidence of these findings, it is clear that all parties interested in the research publication process must adopt standard practices to inform critical trusted markers such as COI to defend transparency and honesty in academic efforts.
To end a positive note, this research poster received second place at the World Research Conference of 2024, which shows that team’s research has already attracted considerable attention among those who seek to safeguard the integrity of research and the trust in science.
You can find the poster in Figshare: https://doi.org/10.6084/m9.figshare.25901707.v2
Partial data and code for this project are Also available in Figshare.
For more information on the subject of research integrity, see the details of the Catalyst Digital Science subsidy prize for 2024, which focuses on digital solutions around this issue.
About the author
Simon LinacreContent Manager, Brand & Press | Digital science
Simon has 20 years of experience in academic communications. He has given conferences and published on the issues of the Bibliometry, the ethics of the publication and the impact of the research, and has recently written a book on predatory publications. Simon is an ALPSP tutor and has also served as COPE trustee.
#Shining #light #conflict #statements
