Getting to grips with implicit bias

Post written by guest host Tom Stafford, a psychologist from the University of Sheffield who is a regular contributor to the Mind Hacks blog, for the BPS Research Digest.

Implicit attitudes are one of the hottest topics in social psychology. Now a massive new study directly compares methods for changing them. The results are both good and bad for those who believe that some part of prejudice is our automatic, uncontrollable, reactions to different social groups.

The implicit association test (IAT) is a simple task you can complete online at Project Implicit which records the speed of your responses when sorting targets, such as white and black faces, into different categories, such as good and bad. Even people who disavow any prejudiced beliefs or feelings can have IAT scores which show they find it easier, for example, to associate white faces with goodness and black faces with badness – a so called 'implicit bias'.

The history of implicit bias research is controversial – with arguments over what exactly an implicit bias means, how it should be measured and whether they can be changed [see also this recent Digest item]. Now a new paper in the Journal of Experimental Psychology reports the results of a competition which challenged researchers to design brief interventions aimed at changing people's implicit biases. The interventions had to be completed online, via the Project Implicit website, and take less than five minutes. Samples of 300-400 people were then randomly assigned to take each intervention, allow a high statistical power to estimate the effect of the intervention on IAT scores.

Overall 17 interventions were tested, and nine appeared to work, while eight had estimated effect sizes close to zero. The paper reports that interventions which focused on trying to shift the underlying attitude of the participants fared badly. Interventions such as 'instilling a sense of common humanity', 'training empathetic responding', encouraging taking the perspective of the outgroup or imagining positive interracial contact all seemed not to work.

These failures to shift IAT scores suggest that the IAT measures something which is relative stable – a real thing in our cognitive makeup, and something that can be measured in a way that can't be as easily manipulated as self-report.

The interventions which did work included a some that targeted response strategies, including a straight 'Faking the IAT' intervention, a practicing the IAT intervention and several other priming and training interventions. That these worked is also both good and bad news. That IAT scores can be shifted by faking and training is bad news for the reliability of the measure, but there is some comfort in knowing that the successful interventions all relied on sophisticated knowledge of how the IAT worked – most participants in implicit bias studies wouldn't come up with these strategies on their own.

The big unknown is how long term any of the effects are. It could turn out that sustained change on implicit biases requires longer than five minutes intervention, but with more sustained interventions it really is possible to shift the underlying attitudes, and not just people's response strategies.

_________________________________

Lai CK, Marini M, Lehr SA, Cerruti C, Shin JE, Joy-Gaba JA, Ho AK, Teachman BA, Wojcik SP, Koleva SP, Frazier RS, Heiphetz L, Chen EE, Turner RN, Haidt J, Kesebir S, Hawkins CB, Schaefer HS, Rubichi S, Sartori G, Dial CM, Sriram N, Banaji MR, & Nosek BA (2014). Reducing Implicit Racial Preferences: I. A Comparative Investigation of 17 Interventions. Journal of experimental psychology. General PMID: 24661055