Asymmetries in research misconduct in collaborations

The recent story of the soon-to-be-retracted psychology paper with fake data (see also here) made me think of some more general issues of research misconduct in collaboration.

Just as background: the published article from 2012 had several authors. At least one of the authors of the paper was hawking its findings on NPR years later. The paper from 2020 reporting the failed replications had several authors. In 2021 it came out that some of the data in the 2012 paper were fraudulent. All but one of the authors of that paper had nothing to do with the faked data.

Now for the general questions. Most scientific research projects are collaborative, and they typically involve people at different seniority levels, for example professors and students or a tenured professor and an untenured professor.

I get the occasional email from students or postdocs who feel that their supervisors are conducting unethical research, and they typically want me to keep their names anonymous when writing about these stories. This makes sense: senior scientists who cheat have a lot to lose, and they’re already behaving unethically, so no surprise that they might retaliate against whistleblowers as a way to preserve their reputation. Consider for example the behavior of that disgraced primatologist at Harvard.

One thing that I’ve personally seen in such cases is that the senior researcher can make a coldly calculating, game-theoretic decision, that goes something like this: He (the senior researcher) has a secure job, and he knows the junior researcher’s employment is not so secure, so the senior researcher can win the game of chicken. As a junior researcher, you don’t want to be known as “that guy” who complained about his senior collaborator. You want to be known for your work, not as a victim. The senior researcher knows this, the junior researcher knows the senior researcher knows it, etc., and so nothing needs to be said. That’s the power dynamic. As the junior researcher, you’ve got no voice so you just have to quietly exit. A senior scholar can follow this pattern over and over and get an informal reputation as an abuser (I’m not talking about sexual abuse here, just abuse of the power dynamic), but new students come every year, and it’s hard to distinguish true rumors from fabrications anyway.

But there’s an asymmetry. The junior scholar’s position is more insecure, but the senior scholar has more to lose. So I could imagine a situation in which the junior scholar, if he or she has enough guts, could win the game of chicken by threatening to expose the fraud, plagiarism, misrepresentation of data, etc. The hitch is that it takes A LOT to take down a senior researcher. Lots of big-time professors have clearly committed scientific misconduct but they’re still standing tall. So, yes, the senior guy has a lot to lose, but with rare exceptions he can usually do fine by just brazening it out. And everybody knows this, which allows perpetrators to do it over and over and over again, without losing much sleep over it at all.

13 thoughts on “Asymmetries in research misconduct in collaborations

  1. A side note raised by your comment about academic power dynamic abuse vs other kinds of abuse: I wonder how strong of a correlation there is between the two. I visited a department where I was told over and over again, if Prof A is your advisor, Prof B will need to be on your committee, but Prof B is a serial harasser of women (I’m a man, so doesn’t directly impact me except that it disgusts me) who also has tried to take undue credit for students’ work.

    Years later after I’d chosen another program, I saw this professor try to pick up a master’s student friend of mine at a conference who was literally half his age. She told me he used some really audacious lines, things that neither of us thought real people ever actually said. So clearly this is all an open secret and he’s not afraid of any of this catching up with him. The whole thing was incredibly depressing.

  2. This case is yet another reason why CRediT (Contributor Roles Taxonomy – https://casrai.org/credit/) should be universally implemented on research articles. When it’s clear who did what, there is less risk for individuals of helping out with one part of a project from being blamed for the misconduct of someone working on a different part of the project.

    • Steve:

      I agree. Even if the plagiarist is gonna flat-out lie (in my example, just stating that he did all the work himself and not acknowledging the direct contributions of his collaborator), at least he’s being forced to lie in public, which ultimately could have consequences.

    • Reminds me of when I had undergrad statistics students do projects (planning, carrying out, and analyzing results in groups). Part of my instructions for their project report was that each student in the group had to turn in their own account of who did what. One semester, everyone except one person in one group handed in the same detailed list of who did what. The one person who handed in something different wrote, “We all did everything together.” I gave that person a much lower grade than the other students in her group. No one complained.

  3. This is a very good analysis of the power dynamics of the real world. I was in an awkward situation with a more senior professional and saw the problems that Dr. Gelman described. I thought that one can’t stop the juggernaut by throwing oneself under its wheels and changed jobs and location. I told everyone that my new position would permit more scope for my recreational interests. It was a financial sacrifice for the first three years. I am not saying that I did the right thing, but it has worked out fine for me. One does see young scientists who leave positions to “get back to family”, “go to a better school system for the kids”, and “get into/out of an urban environment.” Some are running from a boss like I did.

  4. I think the comparison to Chicken is off.

    In Chicken, if the other guy pulls over (Cooperate) when you keep going (Defect) so C/D, you get your best outcome (highest payoff). This would correspond to a world in which the senior admits the fraud and the whistleblower is vindicated and goes on to an even better career. In reality, whistleblowers are, no matter how vindicated, severely penalized: directly, because dishonest seniors now want to avoid them and honest seniors will be concerned about potentially harboring a snake in their bosom, and indirectly, simply because being a successful whistleblower doesn’t go on the C.V. or translate into tangible academic status like high-citation-count publications of original work (much like replicating a finding – “you just critiqued some other researcher’s work”). Given how parlous academic prospects are for a grad student or postdoc, even subtle discrimination is enough to torpedo a career. So, the choice is between ‘worse’ and ‘worst’, not ‘best’ and ‘worst’. (I’m not sure there *is* a well-known named game corresponding to whistleblowing, simply because it’s so trivial and just leads to Cooperate/Cooperate in perpetuating the fraud.)

    A would-be whistleblower can benefit greatly from the private information of fraud, by knowing to escape as fast as possible or avoid relying on the claimed results (or letting rivals foolishly build on quicksand…), but there is no return on publicizing it.

    Another factor not mentioned: the junior may have less to lose, true, but their threats are uncreditable simply *because* they have self-selected into being the junior – showing that they still deep down are trying to ‘make it’ and dream of the success of the senior. Perhaps they are one of the few with calibrated expectations and not one of the others… but they have no way of credibly proving this other than by going through with a threat and torching what little chance they had left.

    • Gwern:

      I don’t understand your statement that the junior has “self-selected into being the junior.” When you start out, you’re the junior, that’s just the way it is. I mean, sure, nobody’s holding a gun to anyone’s head and making them go to college and grad school, but if you want to do research, college and grad school make a lot of sense. To put it another way, everyone’s a junior at first.

      Anyway, when this all happened to me, the creepiest thing was seeing in real time the gears turn in the senior guy’s mind, as he seemed to figure out the game theory analysis leading to the conclusion that he could brazen it out. I was just stunned because I’d only worked things out to the stage of realizing that it could blow up his career if he did this; he’d gone one step further and realized that I had no incentive to blow the whistle, which gave him the incentive to hold a hard line.

      From a game-theoretic standpoint, a challenge is that there are two equilibria that involve no whistleblowing and no career damage, one where the senior author takes full credit and the other where the credit is shared, and there’s a logic to ending up at either of these equilibria.

      One interesting thing about the game is how asymmetric it is. In that way it’s much different than classic chicken or prisoner’s dilemma games. The two players have different levers to pull. The senior author can be greedy and take all the credit for himself or be cooperative and share the credit. The junior author can be destructive and blow the whistle or just stay quiet. Also, the game has repeated plays. If at any point the senior author is cooperative or the junior author blows the whistle, the game ends (or, you could say, it enters a new phase, but it’s no longer this particular game of plagiarism and threat). But if the senior author is greedy, the junior author has a choice of what to do, and if the junior author is quiet, the senior author has a choice of what to do.

      The other twist is that offenders are typically repeat offenders. From one perspective, this creates an incentive for the senior author to share credit so that he doesn’t risk exposure as a plagiarist. But, from another perspective, this creates an incentive for the senior author to flex power to deter anyone from looking into his earlier actions. To share credit is a sign of weakness, no? In addition, there’s the selection effect that the kind of person who would consider sharing credit has already demonstrated a level of greed for credit-taking that exceeds his fear of exposure. We’ve talked about this in the context of other plagiarists: we’re talking about people who have built their careers taking credit reframing others’ ideas; it’s just what they do.

      • I think it is the repeated plays and uncertain outcomes that matter the most if you want to insist on using game theory to model this interaction. The junior researcher has a range of outcomes with a lot of uncertainty for blowing the whistle. Everything from she can’t work with some other big shots because they are friends with Prof. Fake Data to maybe the junior researcher is wrong and it turns out she is overstating an innocent mistake. Not to mention that almost all of the upside for whistle blowing in this context is internal, e.g. knowing you have integrity, feeling like you made science incrementally a little better, while all the downside is external, e.g., losing jobs and research opportunities.

        This is whistle blowing statutes give whistle blowers pecuniary incentives to whistle blow and other protections. If you work for a government contractor who has riped off the U.S. government, you can blow the whistle and make a percentage of the money that US was defrauded of. Whistleblowers can makes millions. You need the same type of incentives and protections for whistleblowers in academic research, It is a hard problem because grant agencies don’t have the U.S. Justice Dept. to go around and recoup their research grants, and they probably would rather not spend the money on lawyers to do that work. However, every university could set up a body that would arbitrate all claims of research fraud that all their professors and all research funders would have to agree to if they want to use that University’s researchers or facilities. Then that body could have the power to order research funds be returned and to implement protections for whistleblowers. Whether any university would do it, that is another question, but the asymmetry can be effectively addressed.

        • Steve,

          In my case there was an upside to blowing the whistle in that then I would’ve got credit for some projects that I’d worked very hard on. But I judged the downside—being known as a troublemaker and having a prominent enemy—to be worse. At the time my employment situation was insecure and I had enough enemies as it was.

      • I mean, sure, nobody’s holding a gun to anyone’s head and making them go to college and grad school, but if you want to do research, college and grad school make a lot of sense.

        That’s exactly it. Just like in music, pro sports, game development, opera singing, Broadway plays, Hollywood, streaming, YouTubing, TikTok… There are a lot more people who want to do X than there are slots open to do X. The choice to pursue X then says a lot about you. (It says “I probably have unrealistic aspirations and desperate dreams”.) This is not as bad in STEM as it is in areas like, say, philosophy, where no one bats an eye if a position is announced and they get 500 resumes submitted – but the dynamic is still at work. Do I need to bring up things like ‘the casting couch’ as consequences of that dynamic?

        The fact that every senior was a junior is irrelevant. They were a junior. But they aren’t *now*, because they won the lottery, and that’s all that matters.

        In other areas where the dynamics are different, being a whistleblower is not punished to the point of destroying your career. If I am, say, a deep learning expert or a good web dev, and someone asks me to do something unethical, I can laugh in their face and go get another job, and I’ll even get a pay raise to boot. (One good web dev I know was fired over a matter of principle; he went to a bar upon leaving it for a drink to unwind, and was accosted by a manager at another company who had heard he’d been fired and wanted to headhunt him first. He had a job and a raise before he had returned home from being fired.)

        Steve brings up government whistleblowers who get a bounty. *That* is much closer to Chicken or PD. Actually, there is another situation which is even closer to what you want but is not the case in academia: cartels. In a cartel, they are constantly at risk of collapse because a participant can undercut the others by selling in violation of the agreement, potentially profiting an enormous amount. Further, US law punishes cartels by granting leniency to the first cartelist to defect and assigning *triple* damages to the remaining cartelists. So you have both a carrot and stick for the first cartelist to defect. Cartels struggle to exist under these conditions, when you might make millions of dollars and save billions of dollars by being the first to defect. (People like to bring up the Phoebus lightbulb cartel, but they usually don’t mention that it lasted a few years at most as a real cartel, and it suffered severe problems – looking through the congressional hearing once, I had to laugh when I read about a Scandinavian light bulb manufacturer who got the cartel to loan it money to build its factory, promising to join the cartel… and then promptly began unrestricted exports of light bulbs. Oops.) Needless to say, there is no way to make the analogy of millions of dollars (be granted tenure for uncovering graud?) or avoid losing billions (fraudsters don’t get any additional penalty for denying – quite the opposite, denying & FUDing for all they’re worth will reduce the penalty, perhaps to zero). So it’s unsurprising that the equivalent of academic cartels tend to go on rather than collapse in a race to the authorities as soon as any doubt about the other cartelists sets in.

  5. I find this very interesting and insightful thread to perhaps miss an important aspect of the ‘game’: the context. I’ve now left academia some 6 years ago, but for about 10 years I’ve experienced directly (on myself) or indirectly (on colleagues) to varying degrees — from mild to very serious — bulllish and intimidating behaviour, invariably coming from senior researchers towards less experienced collaborators. In _all_ cases the problem was ‘resolved’ with the small fish eventually leaving the pond. In _all_cases the author of the bullying was not only being shielded by their department — the few reports of harassment or unethical behaviour I have either made myself or supported as a witness landed on deaf ears, and eventually translated into a harsher tratment and marginalisation of the junior researcher(s) — but also went on to being promoted in a short time span to a tenured / more senior position. This adds insult to injury, as IMHO depending on the workplace it indicates that the game may be rigged.

Leave a Reply

Your email address will not be published. Required fields are marked *