Debunking Social Priming Experiments
Between the 1990s and 2000s, researchers in the field of social psychology published ground-breaking research highlighting how certain stimuli can influence or prime an individual’s perception and behaviour. This came to be known as social or behavioural priming. Ever since, researchers have failed to replicate these crucial findings, raising doubts on the robustness of the results and as a consequence, the field has garnered a controversial reputation. Here, we discuss the past, the present, and the future of social priming.
In the 1980s, the evolution of our understanding and the measurement of associations in our brains led to the foundation of how exposure to a word, picture, or concept evokes, with remarkable ease, related words, pictures, and similar concepts. This came to be known as Priming. Our environment and even our unconscious thoughts prime us with associated ideas. If you are hungry and thinking about food, you’ll be able to identify a related blurry image or understand related whispers faster than usual. But it doesn’t end here. Even our actions and emotions are primed by numerous unconscious experiences (Kahneman, 2011).
The latter claim does raise eyebrows. Can subtle cues in our environment guide our behaviour, and if yes, to what extent? World-famous experiments have been conducted with unbelievable claims. Priming an individual with money makes them more selfish, holding a warm cup of coffee leads to a warmer behaviour are some claims among many others.
The meaning of the term itself is somewhat confusing. In the latest context, social priming is the exposure to any form of an external concept that alters an individual’s behaviour. The origin and evolution of the term have been perfectly explained in this article.
Let’s Look at a Few Replication Failures –
Two experiments were conducted by Harris et al. in 2013, to replicate a highly-cited study by Bargh et al. that claimed to have primed a group of participants to perform better in a particular knowledge quiz than others. In the first experiment, two different puzzles were presented to two different groups of participants from the University of California, San Diego. In the first word-search puzzle, one group was presented with 7 additional high-performance related words: strive, attain, compete, succeed, achieve, master, win to test the priming effects in addition to neutral words in contrast to the other group which received all neutral words. The second puzzle was theme-based where the word-search was based on the theme of each puzzle: bugs, colours, and food. It aimed at showing the enhanced performance of participants who were primed with high-performance words. However, the experimenters found no association between the priming condition and the resulting performance. Additionally, it was found that, on average, participants primed with high-performance words tended towards locating fewer words.
In the second experiment, the same puzzles were used but with a 5-minute delay between the two. In the original study, Bargh et al. claimed that this delay increased the size of the effects of priming by more than double. This time around, the mean score of the two groups was nearly identical and there was no display of the effects of priming.
Another set of experiments were conducted around the same time by Shanks et al. in the hopes of replicating this study by Ap Dijksterhuis and Daan van Knippenberg in 1998. 4 different experiments were conducted with varying combinations of the following:
Displaying a video clip of differing lengths of either a professor or a soccer hooligan.
Listing down the characteristics of either a professor or a soccer hooligan.
Completing questions from Raven's Advanced Progressive Matrices after or both before and after these tasks/task
Solving questions of a general knowledge quiz instead of the Raven’s question
All these combinations were in a close replication of the original experiments.
Yet, the results failed to show any priming effects. When the participants were asked to take the quiz twice, all the participants’ performance increased the second time around, attributing to practice. On many occasions, the mean score of participants primed with soccer hooligans was greater than that of the other group.
Further, another set of experiments was conducted with little modifications trying to replicate similar studies that came after the original. These included priming the participants with an actual professor or a cleaning lady; or using a professor or a soccer player among others. No effects of priming manipulations were found on the quiz performance this time either.
These failed replications are just a few of numerous other publications that later failed to be reproduced. Among other studies of social priming are reports of participants walking slower when primed with words related to old-age; people behaving more selfishly after being primed with visuals of money.
Possible Reasons For The Failure in Replication –
Publication bias, or the ‘file drawer problem’ is a problem where studies failing to produce a statistically significant result are less likely to be published than the contrary. This malpractice raises the average power of the studies on a whole. At the same time, this raises the question of the credibility of such scientific publication journals.
Errors in estimating the size of the effectiveness of the priming intervention while comparing the groups of participants in the study, or effect size. Moreover, it is easier to detect priming effects in the controlled environment of a research laboratory. In reality, however, an individual is exposed to not just one, but several primes at the same time, thus reducing the effect size, if any.
False-positive due to ‘researcher degree of freedom’. This means that the research takes only a subset of the sample taken, resulting in a false statistically significant output.
Failure to recheck the statistical analysis when the results are in the direction of the expectations of the research objective. A researcher will more likely recheck their analysis if the results are contrary to their expectations. This can also point to a confirmation bias in which an individual tends to favour results that are in sync with their beliefs.
What Can Researchers Learn From This Crisis?
There is a need for better statistical methods and their accurate usage. There are positive results but with a lower magnitude.
The sample data mustn't be misused or altered. Moreover, a large enough sample data should be used to reliably detect the small effects of priming.
All findings should be reported, instead of the ones that are surprising or in favour of the objective of the researcher.
It is important to note that the research hypothesis shouldn’t be altered after getting familiar with the data. This problem can be solved by making it public beforehand.
Social priming is the most reliable when we take into account a smaller set of people with a motivational goal already in mind. If an individual wants to fix their sleep routine and wake up early, reading about its possible health benefits can prime them to wake up early. In this case, a relevant goal was already in mind. These interventions lead to short term changes and hence need to be made continuously until a habit is formed. It is also important not to blindly follow and trust every research finding, no matter where or by whom it is published. The ability to pause and think is crucial.