In one of the first lectures, we discussed the priming study of Bargh, Chen and Burrows (1996). In this study the primed participants with the elderly stereotype. While the participants walked back to the exit through a long corridor, a confederate measured their walking speed. Bargh, Chen and Burrows found that participants primed with the elderly stereotype walked slower than participants who were not primed with the elderly stereotype. Doyen et al. (2012) did a replication of this research and used more advanced measurements. In this experiment they were not able to replicate the effects of the Bargh et al. study. Therefore, they conducted a second experiment in which they told the experimenters the specific hypotheses of the study: half of the experimenters were told the participants would walk slower and the other half that the participants would walk faster. They were also instructed to use a stopwatch for measuring walking speed, because the infrared sensors were not yet calibrated. The results were surprising. The results of Bargh et al. (1996) were replicated for the experimenters in the slow condition. Participants in the prime condition walked slower down the hallway, but only if the experimenter was in the slow condition. This effect was even more prominent for the manually measured walking times. Participant in the prime condition walked slower when the experimenter was in the slow condition and faster when the experimenter was in the fast condition. In this replication the priming effect is thus explained as an experimenter effect.

This and other replications of priming effects and the recent exposure of some fraudulent social psychologist from the priming field made Daniel Kahneman to write an e-mail to his colleagues with a proposal to deal with questions about priming effects (Kahneman, 2012). The most important message in his e-mail is that researchers from the priming field should solve their integrity problem. The best way to do this, according to Kahneman is to examine the replicability of priming results.

The big problem with priming research, is that researchers need specific training to be able to do priming research. So to solve the integrity problem in priming research, researchers from the field itself have to do the replications, because if you are not trained, you don’t find the effect. Do you see the problem here?

However, the solution lies not only in replication and in publishing the replication studies of priming. We need to make the data of all our research available. The availability of data sets should be normality and not an exception. Researchers, who are not willing to share their data, should be approached with a certain amount of suspicion.

Public commitment and pre-commit to publish the results can help solve the integrity problem that, as Kahneman points out to the members of his field, the priming field has to deal with.

An excellent example of data sharing comes from the main author of a recently published article about inhibition of neuroblastoma tumor growth. He explained on national television that all data is available for the public and other researchers in his field (Molenaar et al., 2012). The breakthrough in this area of research is, of course, worth a publication. Not only because the researchers did their jobs well, but more importantly will their research help other children in this world with a specific disease. A clear example of how research is a matter of the public. Science is not about getting on a high position as soon as possible. Science is about exploring the world, wanting to know how it works and sharing that with every human being that is part of that world.

Finally, Let us not forget that science brought to us all the progress in this world, science brought us knowledge, welfare, penicillin, and it is science that extends our life expectancy.

 While you live, tell truth and shame the Devil!

Shakespeare, Henry IV. Part I, 1597


Bargh, J.A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 2, 30-244.

Doyen, S., Klein, O., Pichon, C.L., & Cleeremans, A. (2012). Behavioral priming: It’s all in the mind, but whose mind? PLoS ONE, 7(1): e29081.10.1371/journal.pone.0029081

Kahneman, D. (2012). A proposal to deal with questions about priming effects.

Molenaar et al. (2012). LIN28B induces neuroblastoma and enhances MYCN levels via let-7 suppression. Nature Genetic, in press. doi:10.1038/ng.2436


Methodological Mistakes part 2

Today was part two of the presentations of Methodological Mistakes. Mathias and Anja started with a presentation of Simpson’s Paradox of which you can read more in a previous blog. They explained to us not only what Simpson’s Paradox is, using some fun and understandable examples, but also how we can prevent making this mistake and offering some solutions. As they started out, it occurred to me that it was vaguely familiar and I remembered having seen some of the examples before. However, since this was a long time ago, it was a good thing that they refreshed my memory. Anja and Mathias explained to us during the presentation and in their blog that one of the reasons that this paradox is not recognized by people is “the tendency to interpret correlational data causally”. It was suggested that Simpson’s paradox has a frequency of 1.67%; mistaken correlation for causation is probably an even bigger problem. One harmful example is the mistaken causality between vaccinations and autism, which caused (yes caused!) a sharp drop in children that were vaccinated (for more information on this specific example see:  This shows how dangerous our common sense conclusions can be and how important it is we keep looking for these “causations”.

Next Sara and Rachel presented to us the false discovery rate in behavior genetics research. They explained to us how multiple comparisons inflate the error rate in scientific research. To explain this problem I need more than a blog post, so for more information on this topic see:

A solution for the inflation of the error rate in multiple comparisons in not that simple, nonetheless it is important that researchers think of solutions before they start to analyze their data.  We concluded this day with some discussion on our replications study. And I start to realize that a replication is not that simple at all. Will we be able to finish this in such a short amount of time?