On having no good options: Why I removed my name from a paper

                Sarah R. Schiavone

I have decided to remove my name from the paper The origins of religious disbelief: A dual inheritance approach, which is posted as a preprint at psyarxiv.com/e29rt. This decision was difficult for me, but I am no longer able to stand behind this manuscript. I am posting this to explain my decision and share my experience as an early career researcher navigating a maze of bad options. These issues are too rarely discussed in academia and too often swept under a rug.

Some notes and disclaimers: 

• This is a summary of the main reasons why I decided to remove myself as a co-author. It is not a complete account of my concerns about this paper and the process behind it. It is also not a summary of the changes that I would have liked to see in the manuscript. I will address the content of the paper very little here.
• The first (and submitting) author of this paper—Dr. Will Gervais—is my former graduate advisor. I left the graduate program in psychology at the University of Kentucky in 2019 and joined Dr. Simine Vazire’s lab as a graduate student at UC Davis.
• The content of this post in no way reflects on or represents the other co-authors of this paper.

Some background on the paper:

The paper is based on a project I began collaborating on in 2017. Will told me that he was ready to submit the manuscript and post it on PsyArXiv in December 2019. I responded with feedback and requested that some edits be made. A few days later, Will sent me the revised manuscript, which had already been submitted and posted as a preprint. This manuscript incorporated some of my feedback, but we did not have collaborative dialogue on resolving the remaining suggestions and edits.^[1]

This was surprising—I did not think Will would submit the paper without my explicit approval or before sending me an updated version. I was also surprised to discover that the submitted and preprinted manuscript stated that “All authors approved the final manuscript.” This statement is false.

Soon after submission, Will wrote that the manuscript had been rejected and that he was planning to submit to a different journal. This was the last he contacted me about the paper (until I reached out a few weeks ago, as described below). I received an automated email from a journal stating that it had been submitted about two hours later. I received several more automated notifications of the paper being submitted to other journals over the next few months. If changes were made to the manuscript between these submissions, I was not informed of them.

At that point, the paper felt out of my hands, and I was frustrated with how it had been handled. However, based on some past interactions, I wanted to avoid conflict with Will. The situation did not

(yet) feel serious enough to confront. Therefore, I did not say anything about these submissions and

was willing to chalk some of these actions up to misinterpretations and differences in expectations

about collaboration. Most importantly and perhaps naively, I expected that Will would at least let me know if and when the manuscript had been reviewed. I assumed that when preparing a revision, we would collaborate to address reviewers’ comments—and resolve any disagreements and differences of opinion among co-authors—about the manuscript before explicitly agreeing upon and submitting a final version.

The primary reasons I requested that my name be removed from this paper:

Will received a revise & resubmit decision, did not inform me, and submitted a revision without my input, knowledge, and approval.

I learned that a revised version of the manuscript had been submitted to Social Psychological and Personality Science (SPPS), on August 4th, 2020, from an automated notification email from the journal. I had not been informed that the paper had been reviewed or that a revision had been invited, despite being listed as an author. I was not given the opportunity to read the handling editor’s decision letter and reviewers’ feedback, nor to contribute to and approve the revised manuscript. From my understanding, submitting a paper without the approval of all authors is inconsistent with COPE guidelines and SAGE’s expectations of authors. This was a significant breach of trust that I could no longer chalk up to miscommunication or different understandings of professional norms. Keeping a co-author in the dark about a submission seemed especially egregious because this was likely the last chance to make changes and approve the final version.

I decided to email Will and the Editor in Chief of SPPS—despite being nervous to do so given Will’s position as an Associate Editor at SPPS. The EiC paused the submission pending the approval of all authors. Only then did Will send me the revised manuscript, along with the cover letter he had submitted with it.

Inaccurate statements about our work were written and submitted—on behalf of all authors—without my knowledge or approval.

After I received a copy of the cover letter, I saw that it contained statements that I consider to be inaccurate and inconsistent with my own values and sense of scientific integrity. The most troubling statements were in response to a reviewer who asked about inconsistencies between the preregistration and what was reported in the paper.^[2] The reviewer asked whether exploratory analyses had been conducted, noting that the preregistration said that we would split the data into two sets and conduct exploratory analyses (see osf.io/9s78c/). In response, Will stated, “We have not (yet) done any of the exploratory work or machine learning stuff.” To the contrary, exploratory analyses were conducted, and we discussed exploratory findings in our correspondence. Will sent me a dataset to explore, titled “go crazy gfk.csv”^[3], on February 2nd, 2018, after he split the sample in half.

I had neither seen nor consented to the claims made in the cover letter, which are inconsistent with my experience on this project. Nevertheless, as a co-author, I would be held accountable for these claims.

Revised versions of this paper were posted to PsyArXiv without my knowledge or approval.

After receiving the notification of the revised submission to SPPS, I learned—by checking PsyArXiv—that Will had posted new versions of the paper on March 4th, 2020 and then again on July 21st, 2020. Both of the revised versions state that “All authors approved the final manuscript.” This is not true. Will had not informed me of these revisions, and I had never seen or approved these versions when he posted them. Curious readers may well have read these updated versions long before I did.

For these reasons, among others, I have lost confidence in this collaboration and could not remain an author on this paper or take responsibility for its content. I wish that this action had not been necessary. Nevertheless, these experiences have caused me to reflect a lot on collaboration, ethical issues in publishing, and the [lack of] systems in place to support authors who find themselves in similarly difficult situations.

Reflections

This experience has been stressful, especially as an early career researcher. It was, and remains, unclear to me how authors should respond in these situations, who they should inform, and what consequences should follow. Having lost trust in the collaboration, there was no obvious path or process I could find to follow. And I tried. I searched the journal’s website, reviewed COPE guidelines, and consulted trusted advisors. Although there was clear agreement that this was wrong and should not have happened, it was unclear what I could do about it and whether doing anything would even be worth the potential career costs. These are difficult and disappointing truths to grapple with that many of us, unfortunately, know too well: that speaking out may negatively affect your career, nothing may come of it,  you may experience retaliation, those in power may assume you are in the wrong, and others may be too hesitant to express support. These realities can leave you feeling powerless.

This situation is not unique. We all know of similar cases and cases that are much worse. We all also know that these issues differentially affect early career researchers in positions with less power and more at stake when situations go awry. This reflects a much larger issue about power and incentive structures in academia that discourage speaking out. If correcting the record has high costs (losing publications, collaborations, jobs, etc.) and few rewards, why would we expect any reasonable (pre-tenure) person to do so?  

Overall, this experience has served as a strong reminder of the importance of transparency and accountability in science, as well as the ways that incentives discourage correcting the scientific record and encourage silence. If we truly want science to be self-correcting, we must create systems that make voicing concerns and addressing problems less costly—systems that level the playing field to support those with less power and hold accountable those with more.