Frontiers retraction controversy

The following is a long post, but on an important issue.

Frontiers is an open source science publisher based in Switzerland. Their aim is to provide an open access, open science platform that empowers researchers in their daily work and where everybody has equal opportunity to seek, share and generate knowledge. They have started up a whole host of “Frontiers in” journals covering a wide range of subjects. They have also been linked with the Nature publishing group who is interested in the open access model Frontiers is developing.

So I jumped at the opportunity to be an associate editor of the newly established area of Interdisciplinary Climate Studies. The Editor in Chief is the Swiss climatologist, Professor Martin Beniston. An associate editor invites a panel of reviewers who review a collection of articles each year. The associate editor establishes their interdisciplinary area with a “challenges” paper to set the ball rolling. Their task is to encourage researchers to submit innovative papers exploring the frontiers of knowledge.

I was partway through writing the challenges paper when the news came that negotiations surrounding a paper withdrawn by Frontiers twelve months ago had broken down. This has set off a chain of events that call Frontiers into question with respect to the boundaries between authors and publishers; in particular, Frontiers’ defence of scholarship and ethics in controversial circumstances.

The paper in question, Recursive fury: conspiracist ideation in the blogosphere in response to research on conspiracist ideation by Stephan Lewandowsky, John Cook, Klaus Oberauer and Michael Marriott was published on 18 March 2013 and retracted on 21 March 2014 (posted 27 March). The retraction notice said the following:

In the light of a small number of complaints received following publication of the original research article cited above, Frontiers carried out a detailed investigation of the academic, ethical, and legal aspects of the work. This investigation did not identify any issues with the academic and ethical aspects of the study. It did, however, determine that the legal context is insufficiently clear and therefore Frontiers wishes to retract the published article. The authors understand this decision, while they stand by their article and regret the limitations on academic freedom which can be caused by legal factors.

The decision was apparently based on the threat of libel, largely to British libel laws that were since changed on January 1 this year. Elaine McKewon, one of the reviewers of the study who was involved in some of the negotiations, has an article describing this sequence of events in The Conversation. Stephan Lewandowsky’s response is here.

The retracted paper undertook a textual analysis of blog comments discussing an earlier paper by some of the authors, NASA Faked the Moon Landing—Therefore, (Climate) Science Is a Hoax: An Anatomy of the Motivated Rejection of Science (LOG12; Lewandowsky, Oberauer and Gignac), who used online internet surveys to link conspiracy theories with the rejection of climate science. Immediately after that paper was published, a number of blog sites were awash with speculation that the data was faked or manipulated in some way because there was no way those responses could be real.

Recursive fury analysed those posts, grouping them under six conspiracist criteria:

• nefarious intent – they mean harm,
• persecution-victimisation – they are persecuting me (while I bravely resist overwhelming forces),
• nihilistic skepticism – if it does not fit my narrative is must not be true,
• not an accident – events cannot be random but are part of the conspiracy,
• must be wrong – the official version is wrong, so I can alter my suspicions if previous evidence no longer fits with no penalty of me being wrong, and
• self-sealing – anyone presenting contrary evidence must be part of the conspiracy.

A number of hypotheses as to how the original survey was manipulated to produce false outcomes were proposed in online statements and blog comments. Statements discussing each hypothesis were examined to see whether the above criteria were used. No judgement was made on whether a particular hypothesis was true unless it could easily be proven as fact.

The criteria of nefarious intent and self-sealing arguments were attributed to all hypotheses by the authors. Most, but not all of the hypotheses were recursive, except for those that crossed into a wider speculation that the authors were involved in a grand climate conspiracy. This would be the conspiracy that all we climate researchers are involved in.

The authors clearly state they are exploring the links between conspiracist ideation and science denial. Their intent is to better understand climate denial, which has a much broader influence than the number of its proponents. This understanding can potentially inform communication and engagement strategies with the general public and decision makers. (Scientists also influence beyond their numbers but the ultimate balance of opinion in public should ideally reflect the view of the scientific community as opposed to non-evidence directed communities.)

The authors identified blogs and originators of hypotheses in the paper and initially included the names of all commenters in supporting data for the paper. These were later removed on request. In LOG12 the authors refer to conspiracist ideation as a personality trait or cognitive style. By taking public comments and analysing them therefore, they are undertaking discourse analysis but are not conducting any psychological diagnoses. After all, anyone can be involved in a discourse on a conspiracy but not be a conspiracist themselves, which would be pretty much anyone who has commented online.

Recursive Fury also documents Freedom of Information requests made of the University of Western Australia by some of the bloggers mentioned who were seeking material to justify their suspicions. These requests are described by journalist Greame Readfearn who broke the Frontiers rejection story.

After a number of subsequent stories and posts, Frontiers posted another comment on April 4. In it, they retreat from the retraction statement:

Frontiers came to the conclusion that it could not continue to carry the paper, which does not sufficiently protect the rights of the studied subjects. Specifically, the article categorizes the behaviour of identifiable individuals within the context of psychopathological characteristics.

This statement raises the tone and volume of the issue by linking conspiracist ideation with psychopathological characteristics, something the authors did not do. A search on Google shows that before April 4 the link between these terms is raised by individuals who are known to have complained to Frontiers. However it gets only 13 hits on Google Scholar, most indirect, all recent, about half in the refereed literature.

After April 4, the links between conspiracist ideation and psychopathology are all over the climate denial blogosphere. By making these incorrect statements, Frontiers have increased recursive justification, amplifying the ‘harm’ that those who feel they have been associated with conspiracist ideation are perceiving.

This misstatement has fuelled the perception that Recursive Fury was a psychological diagnosis of a number of subjects carried out without their permission. It was not – it was a discourse analysis of online comments with an uncomfortable slant for some – that of conspiracist ideation. Recall that in an earlier paper the authors refer to it as being a personality trait or cognitive style.

In the same statement, Frontiers also says:

One of Frontiers’ founding principles is that of authors’ rights. We take this opportunity to reassure our editors, authors and supporters that Frontiers will continue to publish – and stand by – valid research. But we also must uphold the rights and privacy of the subjects included in a study or paper.

So what rights and privacy do the authors of public statements have in social research? How, for instance, is the analysis of political statements in the public domain to be carried out? To what use can social media be put, at least for those comments in the public domain?

Stephan Lewandowsky has most recently posted Frontiers’ expert panel conclusions of the paper, which included a key researcher on web-based psychology studies’ view:

among psychological and linguistic researchers blog posts are regarded as public data and the individuals posting the data are not regarded as participants in the technical sense used by Research Ethics Committees or Institutional Review Boards.   This further entails that no consent is required for the use of such data.

Not all researchers agree, with consent around public statements in social media being somewhat contested. However, most researchers see consistency with print media guidelines (no consent needed) is seen as desirable. This is consistent with my own knowledge of ethical guidelines. For example, with respect to perceived harm, the Canadian Panel on Research Ethics says: Boards

should not prohibit research simply because the research is unpopular or looked upon with disfavour by a community or organization.

and

some research, involving critical assessments of public, political or corporate institutions and associated public figures, for example, may be legitimately critical and/or opposed to the welfare of those individuals in a position of power, and may cause them some harm. There may be a compelling public interest in this research.

In this instance, it appears that Frontiers has reinterpreted offence as harm, without fully understanding the difference between the two.

The above statements suggest if the public interest outweighs disfavour amongst a certain group with respect to public data, then research should not be disadvantaged or prohibited. The University of Western Australia, the Australian Psychological Society and the Union of Concerned Scientists all agree.

Further blog comments by Frontiers Editor-in-Chief Henry Markram state:

For Frontiers, publishing the identities of human subjects without consent cannot be justified in a scientific paper. Some have argued that the subjects and their statements were in the public domain and hence it was acceptable to identify them in a scientific paper, but accepting this will set a dangerous precedent.

And this:

After publication, the community is engaged and a post-publication review naturally follows. Post-publication review is facilitated by the Frontiers’ commenting and social networking platforms. This process may reveal fundamental errors or issues that go against principles of scholarly publishing. Like all other journals, Frontiers seriously investigates any well-founded complaints or allegations, and retraction only happens in cases of absolute necessity and only after extensive analysis. For the paper in question, the issue was clear, the analysis was exhaustive, all efforts were made to work with the authors to find a solution and we even worked on the retraction statement with the authors. But there was no moral dilemma from the start – we do not support scientific publications where human subjects can be identified without their consent.

This potentially disqualifies all psychological analysis of documents in the public domain with searchable content where passages are quoted, statements by public figures identified and analysed, analysis of media reporting and so on and so forth without the express permission of the authors.

How far does this go? Discourse analysis? Political analysis? Framing and structural analysis? If individuals who disagree with any findings complain to Frontiers with a “well-argued and cogent” complaint, by implication Frontiers will withdraw (or proactively reject) any paper that does so even if it has been approved by a research committee of ethics. Reductio ad absurdum, this would lead to any work being withdrawn if it cites a document in a way that aggrieves its author.

The reason for research ethics is to prevent this type of post-publication wrangling and moving of goal posts. In their public statements, Frontiers frequently mentions rights, and infrequently mention ethics. A discourse analysis might have something to say on that score.

Frontiers has, in the course of this process:

• Asked for initial changes to the paper that were addressed.
• Retracted the paper, putting out a statement that given legal doubts following complaints, they wished to retract the paper, having come to a (reluctant) agreement with the authors.
• Have declined to accept a rewritten paper that anonymised the responses without a full explanation as to why (it is hard to see how author’s rights are being represented here).
• After the retraction became public, suggested that ethical concerns were a major reason for retraction, specifically the rights of subjects who made public statements.
• At the same time, inflated the subject of the research from discourse analysis within a cognitive domain to a psychopathological assessment, something the authors never claimed.
• Have posted subsequent comments that make personal consent paramount to any psychologically-related assessment, which seems to include any and all public statements.

Markram continued in comments:

My own personal opinion: The authors of the retracted paper and their followers are doing the climate change crisis a tragic disservice by attacking people personally and saying that it is ethically ok to identify them in a scientific study. They made a monumental mistake, refused to fix it and that rightfully disqualified the study. The planet is headed for a cliff and the scientific evidence for climate change is way past a debate, in my opinion. Why even debate this with contrarians? If scientists think there is a debate, then why not debate this scientifically? Why help the ostriches of society (always are) keep their heads in the sand? Why not focus even more on the science of climate change? Why not develop potential scenarios so that society can get prepared? Is that not what scientists do? Does anyone really believe that a public lynching will help advance anything? Who comes off as the biggest nutter? Activism that abuses science as a weapon is just not helpful at a time of crisis.

This comment is so very unhelpful. The issue is not about winning the debate, it’s about understanding how the negative influences on scientific opinion can be better understood and managed. Conspiracy ideation relating to scientific subjects has become so widespread in public discourse in English-speaking countries that we can see policies to address climate change being wound back in countries like Australia, New Zealand and Canada. The idea that scientists have overstated the risk of global warming has wide currency in industry and government.

The hypothesis that some of the main protagonists within the blogosphere subscribe to conspiracist ideation is interesting. How does such reasoning pass into the mainstream? Can such ideation be seen in policy making? What role does cognitive dissonance play in conspiracy ideation, the maintenance of political and cultural identity and the disconnect between information on climate risk and climate response? These are all valid research questions.

As a potential associate editor, I see this behaviour from Frontiers as counterproductive to science in general and climate science in particular. If I am to be involved in a controversial area such as Interdisciplinary Climate Studies in an editorial capacity, I want to know that the hosting journal is clear on research ethics and is not going to misrepresent my or my colleagues’ research if disputes arise. I want to know if I can reasonably ask a panel of reviewers to associate themselves with Frontiers. Unfortunately, every statement Frontiers has made on the situation has degraded it further. If the statements made by Editor-in-Chief Henry Markram are representative of Frontiers at large, I can’t see how it can be supported by the research community.