Getting back to our discussion of the newest paper by Stephan Lewandowsky and John Cook, I'd like to discuss something about the paper I have found troubling since day one. I didn't bring this up before because I wanted to contact the journal about it first. You see, the paper is titled:
The ‘Alice in Wonderland’ mechanics of the rejection of (climate) science: simulating coherence by conspiracism
I immediately recognized this title because it was similar to one I had seen before:
'Alice through the Looking Glass' mechanics: the rejection of (climate) science
This is the title for a media article Lewnandowsky published on October 23rd, 2015. Its text was copied nearly verbatim into the new paper. Today, I'd like to discuss whether or not that qualifies as self-plagiarism.
Before I go on, I feel I should give an example of what I am referring to. The paper's introduction begins:
Over the last 150 years, climate scientists have built an increasingly clear picture of how the greenhouse gas (GHG) emissions that arise from human economic activity are changing the Earth’s climate e.g., IPCC (2013). Current atmospheric CO2 levels are higher than at any time since at least 2.6 million years ago (Masson-Delmotte et al. 2013, Fig. 5.2), and the consensus position that global warming is happening, is human caused, and presents a global problem is shared by more than 95 % of domain experts and more than 95 % of relevant articles in the peer-reviewed literature (Anderegg et al. 2010; Cook et al. 2013, 2016; Doran and Zimmerman 2009; Oreskes 2004; Shwed and Bearman 2010).
This is taken from:
Over the last 50 years, climate scientists have built an increasingly clear picture of how the greenhouse gas (GHG) emissions that arise from human economic activity are changing the Earth’s climate. Current atmospheric CO2 levels are higher than at any time since at least a million years ago, and there is no notable scientific dissent from the consensus position that global warming is happening, is human caused, and presents a global problem.
I chose this example because aside from references being added, there are several changes to the text. "50 years" was changed to "150 years." "at least a million years ago" was changed to "at least 2.6 million years ago." Most importantly, while the "consensus position" was kept constant, an extra bit of text was added to say "95 % of domain experts and more than 95% of relevant articles in the peer-reviewed literature" agree with that consensus, a fabrication not backed up by any of the cited sources.
Interestingly, we can know this fabrication did not exist in a previous version of this paper. John Cook's PhD dissertation includes a copy of this paper which strangely is listed as both "published" and submitted." It is largely the same as the paper we have today, but there are a few changes. One can be seen in this text:
Over the last 150 years, climate scientists have built an increasingly clear picture of how the greenhouse gas (GHG) emissions that arise from human economic activity are changing the Earth's climate (e.g., IPCC, 2013). Current atmospheric CO2 levels are higher than at any time since at least 2.6 millions ago (Masson-Delmotte et al. 2013, Fig. 5.2), and there is no notable scientific dissent from the consensus position that global warming is happening, is human caused, and presents a global problem (Anderegg et al. 2010; Cook et al. 2013, 2016; Doran and Zimmerman 2009; Oreskes 2004; Shwed and Bearman 2010).
The evolution in this text through these three versions is interesting. The first version states what the consensus position (supposedly) is and leaves it at that. The next version repeats the same text but adds the claim "there is no notable scientific dissent from" it. The final version removes that addition and replaces it with the fabricated claim "more than 95 % of domain experts and more than 95% of relevant articles in the peer-reviewed literature" endorse that position.
I won't cover any other examples of copied text in this post. There are plenty more, with most receiving fewer changes than in the example above. You should have no trouble spotting the copied text yourself if you compare the media article to the paper, but to make things easier, I've uploaded a document I made while taking notes which shows the paper with its the copied text side-by-side. What you'll find is well over 1500 words were copied, often verbatim.
I contacted the journal about this to ask what their guidelines on something like this are. I didn't file any complaint, and initially, I didn't even "name names" as I wasn't trying to start anything. I just wanted information. The response I got was somewhat helpful, with the editor telling me:
Synthese is a member of COPE and follows their guidelines: http://publicationethics.org/
When you go to the COPE website, you can find a document listing the standards for authors. Listed in it:
4.1 Authors should adhere to publication requirements that submitted work is
original and has not been published elsewhere in any language. Work should not be
submitted concurrently to more than one publication unless the editors have agreed to
co-publication. If articles are co-published this fact should be made clear to readers.
This would seem dispositive as the guideline says the submitted work must be original and not have "been published elsewhere." This guideline does not restrict itself to publications in scientific journals. It just says the work cannot have "been published elsewhere.
Things are not that simple though. The COPE website offers case studies where practical application and examination of rules can be considered. There are a number of relevant case studies, but one which stood out to me was one where:
n October 2014 it came to our attention via one of the reviewers of a manuscript submitted to our journal that an identical article (100% identical) had been previously published on the website of the author. The submitting author had not made us aware in their submission documentation that the article had been publicly available on their website at the point of submission. Two different but related issues arise from this.
Firstly, as it is the journal’s policy to conduct blind peer reviews of each submission received, it is impossible to uphold this policy where submissions already exist, as does the present one, in an identical form in the public domain. Secondly, there is an issue of self-plagiarism. In academic contexts, it is not permissible to re-use identical copy for multiple submissions, and would in all likelihood be regarded as a case of academic misconduct.
We have consulted the COPE website for advice but there does not appear to be a comparable case whereby the original identical article is in the public domain but not previously published in another journal. We are also aware of the various definitions and types of plagiarism and self-plagiarism which render the details of this case a grey area (COPE Discussion Document: How should editors respond to plagiarism http://publicationethics.org/files/Discussion%20document.pdf), and that copyright and rights of author issues may apply.
In this case, text was not reused. An entire paper had been published online, and then the author submitted it to a journal without informing the journal of the previous publication. That is different from what was done with Lewandowsky and Cook's paper, but it is similar. I should point out I don't know if Lewandowsky informed the journal of his previous media article. It seems unlikely.
In any event, the relevant point is in both cases the original text was published online for the public to see. The previous publication was not in a scientific journal though. This journal was uncertain of whether or not to consider the previous publication to be a "prior publication." The advice given to the journal was:
The Forum advised that it is up to the editor and the journal to decide what they regard as prior publication. Journals should provide guidance on their website, detailing what they do and do not consider prior publication. Many journals provide lists of what they consider prior publication, and these lists vary greatly from journal to journal, and between different disciplines.
It is crucial that every journal discusses this at the editorial level and decide what they consider to be prior publication and then puts this information on their website and on the online submission system. There is no general guidance on what is considered prior publication—it has to be an individual journal decision. In some areas prepublication posting is encouraged, and may be required eg for clinical trials. This is a rapidly changing area and journals should be prepared to modify their policies over time, with the increasing number of prior publication options becoming available (eg, blogs, preprint servers). This does raise issues in relation to blind peer review.
Regarding the present case, if the journal has not been explicit about what it considers prior publication, it may be difficult to accuse the author of self-plagiarism or duplicate publication. The author may reasonably state that he was unaware of the journal policy. Some members of the Forum noted that they would normally allow this form of prior publication but there should be a link to the previous version, and the author should have made the journal aware of the previous publication.
Other members of the Forum stated that they would definitely consider this prior publication, and would reject the paper.
So the editors needs to decide for themselves what they consider to be appropriate for their journal and their discipline.
As I read this, whether what Stephan Lewandowsky did would be considered self-plagiarism is entirely up to the journal. My question is, is that correct? Should authors be allowed to write lengthy media articles then copy them near-verbatim into scientific publications without any mention of the earlier publication?
This case study also brings up a second important issue. The peer-review for this paper was supposed to be "double-blind," where the authors and reviewers could not know who each other were. That's clearly not the case here. Any reviewer familiar with the subject could easily have seen Lewandowsky's media article. Any reviewer who hadn't already seen the article but put the paper's title and/or passages from its text into Google would likely have found Lewandowsky's media article. Is that okay?
Unfortunately, the advice given for this case study does not address that issue. I would like to think everyone could agree the answer to it is, "No." If it is impossible for peer-review to be double blinded, then it should not be accepted through a process which requires it be double-blinded.
I'm obviously a critic of this paper, but I believe these are ethical concerns which should transcend one's view of the paper. As such, I'd like to limit today's discussion to the ethical concerns raised in this post. Given that limitation, do you think this paper should have been published? If you say no, do you think the paper should now be retracted?