11 August 2014

Education and scientific misconduct

As an instructor, I never want to say that education or training or teaching is worthless. But I don’t think every problem can be fixed with training.

There is a lot of interest in scientific misconduct. The sad, sad case of STAP cells is only one of the more recent high-profile cases.

A new article in Times Higher Education seems to suggest that training is the way to fix it:

Our own experience as both scientists and teachers suggests to us that the problem begins with, and may be partially solved by, education.

Particularly at the early career stages, it is not surprising to find people do not understand some of the academic norms.

At a workshop that we ran in April this year, involving mainly PhD students and early career postdocs from a wide range of universities, we asked participants to complete a questionnaire on misconduct. Sixty-eight of them answered 51 questions to rate their perception of the severity of different kinds of questionable, unethical and fraudulent research practices, from zero (not really a problem) to three (severe, deserving censure and punishment).
We were relieved to find that most (96 per cent) of them rated “deliberately making up some or all of the data in a manuscript submitted for publication” as three (don’t ask about those 4 per cent who didn’t!). However, we were dismayed that only 54 per cent gave a three to “knowingly selecting only those data that support a hypothesis” and 42 per cent to “deleting some data to make trends clearer”. The naivety is staggering.

From grad students, this might be expected. From postdocs, not so much. I wonder what the scores would be from working professors.

In the United States, ethics training in biomedical research has been the norm for a couple of decades. An article in The Scientist last year was blunt:

But now, 20-plus years later, it is only fair to ask: Does (responsible conduct of research training) work? The simple answer is, “No.”

Education and training can be very effective at combating mistakes due to ignorance. But I doubt ignorance is the major reason for ethical lapses.

I would point to something both the Times Higher Ed and The Scientist articles bring up: competition. Here is something I tweeted during my time as the curator of the Real Scientists Twitter account. The quote is from Jeff Alexander, who was lead designer on Legend of the Five Rings and saw its tournament scene.


In intense competition with few prizes, people will want to quit the game. But what if they can’t quit the game? People who have invested a lot of time and effort into this may conclude that they have no chance playing by the established rules. Is it any surprise people will cheat?

If we want to deal with research misconduct, we need to change the incentive structures in scientific research.

Additional, 12 August 2014: From Neuroskeptic, in reaction to this post:

You can't teach people to be ethical. Not adults, anyway. Ethics is not a skill you get from attending the right Powerpoint presentations.

I disagree a bit here. I think you can teach people to be ethical, but:

  • It’s a long, hard haul to get people to perform their work ethically.
  • External influences can override all that training.

Hat tip to Ed Yong.

2 comments:

alina brate said...

Deluxe Scientific Surgico Pvt Ltd is India's most reputed Laboratory Equipments,Scientific Equipments,Educational Equipments,Exporters,Manufacturers,Suppliers in India involved in Laboratory, Scientific, Educational products. We started as a small enterprise in 1966 in a small town in North of India and finally made our head Office in 2006 in Delhi, the capital city of India and in the same year company got incorporated and is continuing to expand its horizons. The company is exporting quality products to different countries around the world. And is probably the fastest growing company in its field.

Eric Charles said...

Odd question... why are we so worried about "teaching ethics" rather than in getting other people to slow down and do the things that would mostly prevent it. When someone down the hall from you has a cool finding, go check it out. When someone far away has a cool finding, replicate. Sure, you can fake data in any field, but in most fields the findings deemed Very Important, are checked Very Quickly. It should be practically impossible to get away with more than one of these "lapses" in a high impact journal... because the definition of high impact should be something many people will quickly want to do themselves.