The Stanford political scientist Donald Green has retracted a paper from Science co-authored with a graduate student after the student failed to produce the raw data when concerns emerged over the project.
Columbia’s Chris Blattman – whose blog I highly recommend as a great source of insights on development, conflict, social science, and the occasional cracking gag – is of the opinion that there is not much that social science can learn from this case. I am not so sure.
My first reaction on reading the case was horror that an eminent scholar like Don Green (who has been a hero of mine since I read Pathologies of Rational Choice as a grad student) could have allowed a paper to be submitted to one of the most prestigious journals in the world without, apparently, having seen the data on which it was based for himself.
On reflection, however, I have realized that this was an unreasonable reaction on my part. I have a number of coauthors with whom I regularly write papers. Sometimes, I have done the econometric work; sometimes my coauthor has done the econometric work; and sometimes we have done it together. I have never felt the need to share the datasets or run the Stata code in front of my coauthors when I have taken the number-crunching role, and neither have I demanded to see it when I have not been the lead cruncher.
Collaborative research involves a high level of interpersonal trust, and in this case Donald Green may have misplaced that trust, but we are all allowed one or two such errors of judgment in our lifetimes and it is just unfortunate for Prof Green that he did so in a way that has caused him acknowledged professional embarrassment.
But it does strike me that this case reminds us of the potentially problematic relationship between PhD supervisors and the grad students when it comes to publishing results. In this case, the grad student in question wasn’t one of Prof Green’s own students and, indeed, was at a completely separate institution. But the majority of grad student co-publications are (I would hazard to assume) with their own supervisor.
Coauthorship of papers that emanate from PhD research can, in principle, be of mutually beneficial outcome to everyone involved. PhD students are sharp, hungry, ready to do lots of legwork, but usually inexperienced in researching and writing for journal publication. Supervisors bring experience and guidance in formulating research hypotheses and design and in writing to the particular style required for journal publication.
But it is also true that the supervisor-supervisee relationship in one that is imbued with power relations, and that unscrupulous supervisors can exploit that relationship to gain ‘easy’ publications by simply taking the best chapter of a perhaps poorly supervised thesis, topping and tailing it for a journal, and sending it off with themselves listed as co-author. Indeed, I have heard the process being talked about in more-or-less precisely these terms by established professors.
Moreover, many Universities are seeking quite explicitly to promote such co-publication both for good pedagogical reasons but also, more instrumentally, to increase research output and climb their way up the various international university rankings.
It seems to me that the exploitation of the PhD relationship is particularly liable to be a problem in areas where the PhD co-publication model is not well established, so norms and expectations of the level and nature of each person’s contribution are not clear. In the STEMM subjects, the PhD co-publication model is undoubtedly firmly established. In America, it is better established in the social sciences than in the UK and Australia. And in the more economicsy disciplines, it also seems to be better established.
Hence, for instance, a PhD student in political science in an Ivy League US university would probably enter the programme fully expecting that their first publications coming out of the PhD research would be co-authored with their supervisor, and they would normally have ample shared experience among their peers to identify whether the relationship was tending toward an unusual or exploitative one.
In the UK, however, the norm in political science is still that the PhD gets written up as a monograph with ample gratitude for the supervisor expressed inside the covers, but a single name on the cover itself. In such circumstances, it seems to me, a shift towards the co-publication model needs to be undertaken with extreme care on all sides.
I am not saying the exploitation of PhD students is endemic, and I certainly do not think that this was the case in this particular incident. But what the incident does remind us is that the PhD relationship is one that requires a high level of trust if it is to be of mutual benefit, but that it is hence ripe for exploitation by the unscrupulous.
Chris Blattman suggests that this trust relationship is ‘trickiest for the senior scholars who foster dozens of studies and students’. This may be true in comparison to other less senior scholars, in the sense that they have less time to devote to each individual student, and are hence more likely to be duped by unscrupulous PhD students, in the way that Donald Green appears to have been.
But, it seems to me that given the power dynamics of the PhD relationship, we should be more concerned about honest PhD students being exploited by unscrupulous supervisors than vice versa.
So what do I think we should learn from this case over and above the reasonable points Blattman makes? I think we need to learn to be more critically reflective in our relationship with our grad students and to guard not just against the kind of error of judgment that Prof Green appears to have made, but also against our own potential to exploit that relationship for our personal and institutional advancement.
 Professor Green has noted that because he did not have ethics approval from his institution, he felt constrained not to look at the primary data.