Two female and one male social and computer scientists from Abu Dhabi were in for an unpleasant surprise when, three weeks ago, their research article was published by Nature Communications. Scientific Twitter erupted in outrage.
The study showed that having female mentors instead of male mentors correlates with a lower scientific impact in terms of post-mentorship citations, especially for female mentees. You can find the paper and the reviewer reports here.
The greatest offence was caused by the authors’ conclusion: that instead of female scientists seeking out female mentors, they recommended them to seek out male mentors. A request to retract the paper was issued, in part because of this conclusion and in part because the paper’s methodology was fundamentally flawed and the reviewers’ comments had not been addressed.
If you read the reviewer reports, you will see that the reviewers are initially very critical, but as you progress through the 62 page document and see the revision rounds, three of the four reviewers say that their comments have been satisfactorily addressed.
One reviewer remains dissatisfied about one point. Namely, that the proxy they took for mentorship is poor. Senior co-authors (more than seven years since their first publication) were considered mentors if they had the same affiliation as the junior co-authors (within seven years of their first publication).
‘It seems that the authors are simply following the data’
The authors of the study wanted to capture mentorship beyond the formal advisor, hence the inclusion of senior authors from the same institute. The authors included a small survey to establish that this proxy does correlate with mentorship.
Yet, the questions were too general and the scale of the survey too small to confidently establish that it is indeed a good proxy. This begs for a rebuttal. Yet, it doesn’t change the fact that an extensive dataset shows that having more male than female co-authors amongst your more senior co-authors with the same affiliation correlates to a larger number of future citations.
As stated above, the authors concluded that women should hence seek out male mentors (or co-authors). It is this conclusion that is deemed so particularly offensive and damaging. Just looking at the research data, it is not even an illogical conclusion. But is it a wise conclusion? It is not.
You would be silly to judge whom you work with on such a crude metric as the quality of an advisor/collaborator/mentor varies wildly across individuals. You would do yourself a favour by finding out what kind of advisor a person is, where former group members have ended up, and whether that aligns with your ambitions. And morally, you should simply not judge a person based on their gender.
Yet, the finding that having more male than female co-authors amongst your more senior co-authors with the same affiliation, correlates with a larger number of future citations still stands. You may find that an unwelcome finding, but it begs the interesting question of why this is the case. Removing the paper from the scientific record and hence debate, will mean that this question will remain unanswered.
In 2018, another study was published in Nature Communications, mostly by the same authors, which showed that ethnic diversity amongst the team of co-authors correlates with higher impact in terms of the number of citations. Many will consider this a welcome finding, and it did not attract outrage. The research methodology was similar though. It seems that the authors are simply following the data. I keep wondering whether, were the conclusions reversed, would the response to the two papers also have been reversed? Let’s hold the call for retraction for fraudulent and plagiarised papers, shall we, and just have a civil scientific debate.
Monique van der Veen is Associate Professor at the Faculty of Applied Sciences, department of Chemical Engineering. You can read about the work of her research team here and follow her on Twitter at @MAvanderVeen