Verified by 
Recent media coverage of a series of meta-analyses (Greenwald et al., 2009; 2015; Oswald et al., 2013; Blanton et al., 2015) has raised the question of the usefulness of attempting to measure implicit biases (Bartlett, 2017) and the usefulness of using implicit bias models in bias reduction training (French, 2017; Gockowski, 2017). Bartlett’s (2017) article in The Chronicle of Higher Education featured interviews of psychological scientists for their perspectives on the recent findings that the Implicit Association Test (IAT) has very limited power in predicting behavior. Gockowski’s (2017) article in Campus Reform used the findings to argue that the University of Missouri is out of line in requiring the use of an unconscious bias training module for search committee members when hiring job applicants. I have found no evidence that the University of Missouri uses the IAT for this program or how it is used if it is in the program. French’s (2017) criticisms in National Review focused on interpreting these findings as evidence that some progressives are wrong in their ideological stance that implicit bias is “a key explanation for persistent racial disparities in education, housing, employment, and law enforcement—and a justification for cultural retraining.” In my view, this is the problem—the politicization of science.
The Science
The argument for implicit social cognition was initially proposed by Greenwald and Banaji (1995). This idea rested on the assumption that much processing of social stimuli occurs below the threshold of conscious awareness, necessitating the use of indirect measures. Soon after, the Implicit Association Test (IAT) debuted (Greenwald, McGhee, & Schwartz, 1998). In the IAT, different target concepts are paired with different attributes and subjects end up responding faster to more highly associated concepts. Soon researchers showed that the IAT did not measure what we really think of as “attitudes,” but rather measured “environmental associations,” which are connections we have learned through exposure in society (Karpinski & Hilton, 2001; Olson & Fazio, 2001, 2003, 2004). This has always led to the question of “If it doesn’t measure attitude endorsement or predict behavior, what good is this research?”
The implicit associations that the IAT measures are subject to change based on our experiences. Thus, they reveal insights into the biases that are present at a point in time. They do not reveal deep biases of which we strongly hold but wish to not hold. Sometimes implicit associations match explicit attitudes, and sometimes they don’t. Explicit attitudes can predict deliberative behaviors very well (Ajzen, 1991), but implicit associations do not predict behavior all that well. Still, implicit associations can predict behavior and the IAT is not the only measure of implicit biases (e.g., Dovidio, Kawakami, & Gaertner, 2002).
The data show what they show. Researchers who have dedicated their careers to studying implicit social cognition have always sought to uncover applications and useful translational venues for the research. The arguments within the discipline have largely turned away from “does implicit cognition exist” to “what is the extent to which implicit cognition affects behavior?”
Irresponsible Applications of Science
There will always be unsavory characters who take something from science and misuse it to make money in industry. I have not found evidence to make me believe that any diversity training should be done using the IAT with the belief that it will somehow reduce discrimination. Behaviors we can’t anticipate and that occur in highly variable situations are too difficult to predict from a measure that predicts only very specific associations. Under the best of circumstances the IAT only predicts very specific behaviors, not the constellation of events that lead to the broad category of behaviors called discrimination. Scientists often advance agendas, but the data speak for themselves. It is important not to discount informative data.
Science Keeps Moving
Don’t be fooled by the politics that sometimes encroach on science. Greenwald, Banaji, Nosek, Blanton, Tetlock—these are reputable scientists who are doing solid work. Don’t be fooled by some who misrepresent an argument over the magnitude of an effect to look like an area of research has been debunked. Many of us got into this area of research knowing that implicit biases don’t predict behavior well and that it has very small effects. This is not a secret that scientists have been hiding from the public. One day the contributions will be useful. Science works slowly and builds on itself.
Conclusions
The Greenwald group and the Blanton group are in disagreement over the size of the small ability of the IAT to predict behavior. The science behind the IAT is sound but sometimes overstated in popular culture by people who are not trained in science. For a good description of the IAT and its utility, read Blindspot: Hidden Biases of Good People by Banaji and Greenwald (2013) (or read a review by Mather & Hurst, 2014). The meta-analysis research recently in question is only relevant to the IAT and not to other measures of implicit bias that do predict behavior (e.g., Dovidio et al., 1997). Thus any lack of predictive validity of the IAT for behaviors does not invalidate all research on implicit social cognition.
I agree with Gockowski that as a general rule, diversity training programs based on reducing implicit biases are too many steps removed from validation to produce a valid procedure. I agree with David French when he says “I don’t doubt at all the human capacity for prejudice. But the barriers to racial reconciliation in this country don’t lurk in the unconscious. Combine the lingering effects of centuries of explicit discrimination with wildly different concepts of racial justice and you have a recipe for enduring conflict. It’s tempting to believe that there’s a psychological fix, especially when that fix dovetails with your ideology. But it’s wrong.”
I research attitudes and persuasion, particularly implicit social cognition. If you want to develop a marketing campaign, your money is far better spent on consciously perceivable things than chasing after the small yield of targeting the unconscious (). The same applies to reducing discrimination. We should continue to focus our efforts on explicit attitudes. But beware of the implicit associations that are products of what we feed them. Put some careful thought into your next click, because your implicit cognition is watching.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211.
Banaji, M. R., & Greenwald, A. G. (2013). Blindspot: Hidden biases of good people. New York: Delacorte.
Bartlett, T. (2017, January 5). Can we really measure implicit bias? Maybe not. The Chronicle of Higher Education (online).
Blanton, H., Jaccard, J., Strauts, E., Mitchell, G., & Tetlock, P. E. (2015). Toward a meaningful metric of implicit prejudice. Journal of Applied Psychology, 100, 1468-1481.
Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit prejudice and interracial interaction. Journal of Personality and Social Psychology, 82, 62-68.
Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & Howard, A. (1997). On the nature of prejudice: Automatic and controlled processes. Journal of Experimental Social Psychology, 33, 510-540.
French, D. (2017, January 10). Implicit bias gest an explicit debunking. National Review (online).
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.
Greenwald, A. G., Banaji, M. R., & Nosek, B. A. (2015). Statistically small effects of the Implicit Association Test can have societally large effects. Journal of Personality and Social Psychology, 108, 553-562.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464-1480.
Greenwald, A. G., Poehlman, A., Uhlmann, E. L., & Banaji, M. (2009). Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. Journal of Personality and Social Psychology, 97, 17-41.
Karpinski, A., & Hilton, J. L. (2001). Attitudes and the Implicit Association Test. Journal of Personality and Social Psychology, 81, 774-788.
Olson, M. A., & Fazio, R. H. (2001). Implicit attitude formation through classical conditioning. Psychological Science, 12, 413-417.
Mather, R. D., & Hurst, A. C. (2014). The inescapable mental residue of Homo categoricus. [Review of Blindspot: Hidden biases of good people.]. Evolutionary Psychology, 12, 1066-1070.
Olson, M. A., & Fazio, R. H. (2003). Relations between implicit measures of prejudice: What are we measuring? Psychological Science, 14, 636-639.
Olson, M. A., & Fazio, R. H. (2004). Reducing the influence of extrapersonal associations on the Implicit Associations Test: Personalizing the IAT. Journal of Personality and Social Psychology, 86, 653-667.
Oswald, F. L., Mitchell, G., Blanton, H., Jaccard, J., & Tetlock, P. E. (2013). Predicting ethnic and racial discrimination: A meta-analysis of IAT criterion studies. Journal of Personality and Social Psychology, 105, 171-192.
Randolph-Seng, B., & Mather, R. D. (2009). Does subliminal persuasion work? It depends on your motivation and awareness. Skeptical Inquirer, 33, 49-53.
