Academic fraud, data and dishonesty

Academic fraud, data and dishonesty

When I was studying ecology at undergraduate level, I maybe naively believed that academic research and the peer review process was the most robust and rigorous way to discover, evaluate, and build knowledge. 

As work, life and study has shown me, that is maybe not entirely the case. This article in Vox describes how Francesca Gino, a Harvard Business School researcher on (ironically) dishonesty, has been found to have published papers containing manipulated data. Harvard Business School carried out an investigation in 2023 and produced a 1200-page report which found that Gino had “committed research misconduct intentionally, knowingly, or recklessly,” and is available here. 

Gino co-authored some of the papers that we consider canonical in the psychological safety field, including: “Is Yours A Learning Organisation” in HBR in 2008, which is absolutely not in question, and remains an excellent practical resource.

This however highlights a particular problem in the knowledge system: even after a paper is found to be flawed or is retracted, other research that relied and built upon those findings remains; new knowledge is sometimes built upon flawed research. And when an author is found to have falsified research and the rest of their work is thrown into doubt, even high-quality and robust papers co-authored by them can be contaminated by the fog of distrust.

Gino certainly isn’t the only person to have committed academic fraud, malpractice, or research misconduct, and won’t be the last. There are more who have not had such an intense spotlight shone on them, and even more still who have never been identified and probably never will. It’s also been argued, and I can believe it’s true, that Gino was particularly singled out due to her gender. We should think of this as a system problem, not a problem of the odd “bad apple”. Academia suffers from various systemic and structural problems that increase the pressure to manipulate data, exaggerate claims, plagiarise, or commit fraud. There is a huge amount of pressure on academics to publish high volumes of papers, a great deal of institutional and interpersonal competitiveness (including for status as well as funding), an adversarial default stance (i.e. you must “defend” your thesis in a typically very high-pressure, high-consequence environment), and strictly hierarchical structures that can suppress voices of concern. Whilst I don’t endorse what Gino did, I also believe that she was a victim of this system.

There are a few ways in which we could mitigate the problem of research fraud and misconduct. One is to implement a new standard for publication – that any data used in a study must also be made available in its raw format. This would better allow for fraud to be identified, mistakes found, and further valuable data analysis made. Frankly I find it surprising that this isn’t already the standard: it’s a relatively low bar to meet.

Another important improvement to make is to reduce the pressure and incentives for academics to publish ever higher volumes of papers, and to rethink the way that we measure the impact of research. There is a great pressure on academics to “publish or perish“, and to publish papers that (even tenuously or falsely) demonstrate new findings or refute existing papers and knowledge. We’ve seen this effect before in psychological safety. With this level of pressure and incentivisation, it’s hardly surprising that we observe strong Goodhart’s Law effects.

If you’re interested in these ideas, Ben Goldacre wrote about these issues extensively in the excellent book “Bad Science” in 2008 with particular regard to the pharmaceutical industry (where the incentives for hiding “bad” results and manipulating data are much, much, higher). Alongside his other work at Oxford, he now leads the Open Safely project – a healthcare data project aimed at improving transparency in research.