This graph shows a network of policy documents that may have been contamined by a retracted publication after the retraction date. The orange nodes represent policy documents that explicitly cite the retracted paper in question, while the yellow nodes represent policy documents that cite previously contaminated policy documents. Note that only one policy report from the first wave flagged the publication as retracted (square node), while others cited the publication as valid scientific evidence.
The study in question was published in Science in1996 by a group from Tulane University and presented evidence of certain pesticides having estrogenic activity (Arnold et al., 1996). However, the study could not be reproduced, which led its retraction in 1997, resignation of the first author and millions of dollars in lost financial costs due to its quick adoption by legislators (Turrens and Springer, 2019). This paper can be considered a textbook example of how a retracted paper can cause severe efffects due to its use by policy-makers, including after the official retraction.
The graph was created using Altmetric API to extract data on policy mentions of the original Arnold et al. publication and later by using Dimensions API to search additional policy reports where initial policy reports were mentioned. Retraction Watch data were used to extract information on the retraction date and reason for retraction. The network can be expanded further, however, even the extent of contamination of the second wave reports is questionable as they may not directly build on previous policy reports. This assumption can be further tested by using text mining techniques or manual analysis. In this example, text mining was only used to identify first wave policy reports that contained the word retract* when referring to the original publication.
References:
Steven F. Arnold, Diane M. Klotz, Bridgette M. Collins, Peter M. Vonier, Louis J. Guillette Jr., John A. McLachlan. Synergistic Activation of Estrogen Receptor with Combinations of Environmental Chemicals. Science. 1996. DOI: 10.1126/science.272.5267.1489
Turrens, Julio & Michael S Springer. How to Train Undergraduates in Research Integrity and the Responsible Conduct of Research - How-to Series. Blackwell. 2019.
The study in question was published in Science in1996 by a group from Tulane University and presented evidence of certain pesticides having estrogenic activity (Arnold et al., 1996). However, the study could not be reproduced, which led its retraction in 1997, resignation of the first author and millions of dollars in lost financial costs due to its quick adoption by legislators (Turrens and Springer, 2019). This paper can be considered a textbook example of how a retracted paper can cause severe efffects due to its use by policy-makers, including after the official retraction.
The graph was created using Altmetric API to extract data on policy mentions of the original Arnold et al. publication and later by using Dimensions API to search additional policy reports where initial policy reports were mentioned. Retraction Watch data were used to extract information on the retraction date and reason for retraction. The network can be expanded further, however, even the extent of contamination of the second wave reports is questionable as they may not directly build on previous policy reports. This assumption can be further tested by using text mining techniques or manual analysis. In this example, text mining was only used to identify first wave policy reports that contained the word retract* when referring to the original publication.
References:
Steven F. Arnold, Diane M. Klotz, Bridgette M. Collins, Peter M. Vonier, Louis J. Guillette Jr., John A. McLachlan. Synergistic Activation of Estrogen Receptor with Combinations of Environmental Chemicals. Science. 1996. DOI: 10.1126/science.272.5267.1489
Turrens, Julio & Michael S Springer. How to Train Undergraduates in Research Integrity and the Responsible Conduct of Research - How-to Series. Blackwell. 2019.