Bar-Ilan, J., & Halevi, G. (2018). Temporal characteristics of retracted articles. Scientometrics, 116(3), 1771-1783. https://link.springer.com/article/10.1007/s11192-018-2802-y

Temporal characteristics of retracted articles

They accessed the database of ScienceDirect, Elsevier in October of 2014. They extracted a total of 955 articles. Most papers have a note (Retraction Note) that details the reason why a paper was retracted. Three different categories for reason of retraction were established: ethical misconduct, scientific distortion, and administrative error. Year of retraction was determined by the Retraction Note. Altmetric data was collected using Mike Thelwall’s Webometrics Analyst, which has data from both Altmetric.com and Mendeley. Data from Mendeley was extracted four times between 2014 and 2018, and data from Altmetric.com was extracted once in January of 2018. 

Temporal characteristics of retracted articles data collection

Quantitative analysis was done on the atmetric and Mendeley data that was extracted. They compared years of publication to years of eretraction, looked at how many papers were retracted for each of the categories of reasons, and looked at citations trends and social media coverage after retraction.

Temporal characteristics of retracted articles methods

- more than 50% of the articles are retracted within 1 year
- scientific misconduct takes longer to be retracted while administrative errors are corrected within 5 years or less
- 632 (64%) articles retracted because of ethical misconduct 
- 311 (31%) articles retracted because of scientific distortion
- 52 (5%) articles retracted because of administrative reasons or for reasons that did not fit the other two categories 
- citations and reads continue to increase after retraction (in the four years, there were more than 30% increase of citations)
- articles in the scientific distortion category receive more citations than other categories

Temporal characteristics of retracted articles results

This study looked at temporal characteristics of retracted articles such as the year of publication, the time after it was published before being retracted, citation counts after retraction, and social media attention. 

University of California, Berkeley

- "Other than detecting impact in advance, alternative metrics could act as early warning signs of retractions: tentative findings of a study into the papers retracted by PLoS ONE"
- "Allegation of scientific misconduct increases Twitter attention"
- "Temporal characteristics of retracted articles"

Literature on the Retraction of Publications

Researchers studied the paper “In Vitro Modeling of Blood-Brain Barrier with Human iPSC-Derived Endothelial Cells, Pericytes, Neurons, and Astrocytes via Notch Signaling” published in the Stem Cell Reports journal by Yamamizu et al that was retracted due to concerns regarding the authenticity of certain images that seemed to be fabricated. They concluded that tweets are not indicative of research quality because the number of rate of tweets can change drastically due to retractions or research going viral due to reasons unrelated to its content. 

Allegation of scientific misconduct increases Twitter attention

- Researchers wanted to see if Altmetrics can act as warning signs for research that needs to be re-examined because prior studies only found weak or moderate correlations between alternative metrics and citations. They compared 209 papers retracted by PLoS ONE, Public Library of Science - an open access and peer-reviewed scientific journal, and 418 non-retracted papers also from PLoS ONE. 
- Research question: in comparison to non-retracted, contemporary publications, are the retracted articles more likely to be viewed and commented (also tweeted and blogged) before the retraction takes place? 
- Conclusion: deviations from average number of views, downloads, and reads could point to the potential retraction of certain papers 

Other than detecting impact in advance, alternative metrics could act as early warning signs of retractions: tentative findings of a study into the papers retracted by PLoS ONE

Learn Before

Related

Learn After