Thursday, September 8, 2011

Dear Science, Stop Cheating Already

Last night I was eating dinner at Rubio's (side note: "Baja Fresh," Rubio's slogan is a little confusing considering that their fish tacos are made with Alaskan pollock, just an observation) while surfing the web using my smartphone. What I found was shocking: an article released by Science Magazine, states that dutch social psychologist Diedrik Stapel has "admitted to using faked data" and "will not be asked to return" to Tilburg University where he is a faculty member.

I was, and still am, totally and completely shocked by this news. Stapel is by all accounts "one of Europe's best social psychologists" and I was personally witness to his receipt of a career trajectory award (this award is given to researchers who seem to have a promising upward trajectory in their career) at the Society of Experimental Social Psychology's annual meeting a few years ago. There his colleagues and collaborators gushed about Stapel's prolific writing abilities and provocative findings. At the time I was a young-ish graduate student, and I looked at Stapel with great admiration, as an inspiration to my own academic work. Obviously, I should now re-think my role models.

I could go on and on wondering what made Stapel "fake" his data (according to the report, the data came with a ficticious person who collected and analyzed it). Instead, what I'd like to focus on is the whole issue of data fabrication. Specifically, why faking data is (1) short-sighted and (2) anti-scientific.

Faking data is short-sighted
In the short run, a faked paper might be a big splash and will generate a lot of interest. Such is the case with Stapel's alleged faked data on how eating meat makes people more selfish. This is actually good for you in the short-term because attention that is paid to your paper now is likely to land you a good job, possible media attention, and some respect among your peers.
Apparently not selfish.

The problems come when people try to build the scientific literature. If your paper is indeed a big splash type of finding, people will seek to replicate the results and extend them to other realms. If data is faked, there is no possibility of replication. Read that again. If your findings cannot be replicated, they will in essence slowly become less relevant to the field. That's actually not good for you. It's also possible that your findings will be refuted by another team of researchers. This second possibility will draw questions to your research which could eventually illustrate that you are in fact, faking your data. As Stapel learned, faking data comes with heavy long-term costs and moderate short-term gains.

Faking data is anti-scientific
I imagine that people tend to fake data to confirm their own theories about the world. That's an inherently anti-scientific pursuit. Theories themselves should not be judged on how many things they get right, instead they should be judged on how much of the theory can be systematically tested. Take Freud's psychoanalytic theory for instance, almost everything about it is dead wrong. I mean nearly everything (save perhaps transference and a few other ideas)! No matter, we still talk about his ideas with great enthusiasm because they led to a host of testable hypotheses.

Wrong, but relevant
In more modern research, discreet emotion theory has recently come under fire as researchers react against the notion that emotions may not actually be natural kinds of phenomena with clear and distinguishable neural pathways. Instead, some argue that emotions are cognitive appraisals of good or bad feelings (Barrett, 2006). This debate precisely highlights my point: That we are having a debate in psychology about whether fear, compassion, love, joy, sadness, and shame really are distinct emotions or different cognitive judgments about good and bad is what moves science forward. The discreet emotions perspective has value, not because it is right about discreet emotions [it is ;)], rather, it has value because it provides a theory for which we can generate testable predictions. If scientists were always right about their theories, then we wouldn't need to conduct research.

Faking data is in my view, essentially about ego. I believe my theories are right and valid and I don't need data to prove that, they just are. In science you don't matter, the science matters. I wish we could all keep that in mind.

Are you worried that many researchers fake their data? Let us know in your comments!

Barrett, L. (2006). Are Emotions Natural Kinds? Perspectives on Psychological Science, 1 (1), 28-58 DOI: 10.1111/j.1745-6916.2006.00003.x


  1. My largest concern is the contamination of other research and researchers. How long will social psychologists hear "yes, but can we really trust that finding? That one guy cheated that one time." In addition to being short-sighted an anti-scientific, it's also selfish and damaging to others in the field.

  2. Agreed. Contamination of other research is a third reason why academic cheating is a bad idea. Thanks for reading!

  3. I suspect Stapel's not the only one.

  4. Maybe one should make Stapel the subject of research on short term rewards vs long term negative risk taking. Is this not comparable to those experiments when people chose $1 today instead of $5 next week?

  5. Spotlightmind: it's unknown how many psychologists falsify data, but our field does need to start asking some hard questions about this issue.

    Mihai: nice observation. people often feel that short-term rewards are more valuable because the rewards come right now, versus maybe later.

    Thanks for reading!