By Alex Fradera
A pre-registered mass replication attempt published in Perspectives on Psychological Science has raised doubts about another celebrated psychology finding. The collaboration between 40 laboratories found scant evidence for the so-called “Professor Prime”, undermining the famous finding that when people imagined themselves as a professor rather than a football hooligan it led them to perform better on a trivia quiz.
In the original study, published in 1998, the Dutch researchers Ap Dijksterhuis and Daan Van Knippenberg asked participants to spend five minutes writing about themselves as if they were a professor or a football hooligan, thus priming them with one concept or the other. Next, the participants completed an unrelated trivia test of twenty questions, and those primed with the professor concept achieved an average of two and a half more correct answers – a 13 per cent advantage – versus those in the hooligan condition. The authors concluded that “priming a stereotype or trait leads to complex overt behavior in line with this activated stereotype or trait”.
This was part of an exciting new wave of research that moved beyond earlier priming studies that had focused on how “primes” make it easier to recognise related concepts (e.g. encountering the prime word “football” makes us quicker to recognise related words like “goal” or “game”). The field became animated by so-called “social priming” claims, like John Bargh’s classic study that suggested priming the concept of elderliness led participants to walk away from the lab more slowly (a finding that other labs failed to replicate, helping to spark the so-called replication crisis in psychology).
Amidst this crisis, until now the Professor Prime finding – perhaps the kingpin of these effects – was still standing (even if lampooned by, among others, the Dr Primestein Twitter account). The effect became the focus of this latest mass replication attempt (known as a “registered replication report”) when, in a great example of collaborative science, Dijksterhuis and Van Knippenberg volunteered their study as the latest to face scrutiny.
Prior to releasing their original materials to the other participating labs, Dijksterhuis and Van Knippenberg re-ran their own study and found a smaller effect than they had previously, and one that was confined to male participants.
The procedure followed by the other 40 labs was almost identical to the original, except they used a new set of 30 trivia questions pretested to ensure a useful range of difficulty. Unlike the original, they also preregistered their stopping conditions (at what point they would cease looking for further participants; selective stopping is a questionable research practice that can lead to misrepresentation of experimental effects). The labs also aimed for a gender balance between participants, which only 23 of them achieved.
Whereas the original study showed a 13 per cent advantage to participants who got the Professor Prime, the average effect achieved by 23 labs involving over 5,500 participants found an advantage of just 0.014 per cent – almost 100 times weaker – and the variability across studies meant that this tiny effect was not statistically significant. Restricting the analysis to only male participants made no difference, and when the data from all 40 labs was included (i.e. also including those with a gender bias in their sample), the effect was even smaller.
The only hint of an effect came after the researchers excluded anyone who thought the priming task was in any way linked to the trivia task – which was two thirds of the participants – in which case the Professor Prime was associated with a 2 per cent advantage. This is still much smaller than the original and there must be doubts about the meaningfulness of a small effect that is present only in a minority of people.
There is no way to see this as anything other than yet another serious blow to the social priming literature. But it’s also a success for non-adversarial science – an example of psychology cleaning up its issues and coming to a better understanding of the effects that matter enough for us to investigate further.