Also known as unconscious bias training, its recently come into the news after Starbucks announced it will close all stores across the U.S. for one day to have this training. In this article, we will be discussing the science behind implicit bias training and refrain from political discussion.
Before we begin, I’d like to note that this article analyzes the science behind implicit bias and inherently offers an opinion on whether or not it’s valid. This is not an opinion piece, but it does reflect my views. That being said, if you see something wrong with a study I cited, or if you have a piece of evidence that contradicts something in the article, I encourage you to contact us at firstname.lastname@example.org. We want the truth, not a truth. Alright, let’s start.
What is implicit bias training?
Implicit bias is described as an unconscious association of traits to a social group. The argument being made is that this bias affects our prejudice: how you view and feel about another group. It’s reasonable to assume that this bias exists, it’s evolutionarily advantageous to favor members of your own group. This can be seen throughout Kingdom Animalia such as Belding’s ground squirrels whose female members are much more likely to warn other squirrels of prey if the other nearby squirrels are closely related. But before we can test if and how it affects our prejudices we need to know this: can we test and accurately measure it?
The Implicit Association Test (IAC)
What is the difference between implicit and explicit bias? Explicit bias is the bias that an individual can report themselves. This is problematic in measuring bias as an individual may describe themselves as having no bias towards a certain social group because of social repercussions (Being overtly racist is fortunately mostly socially unacceptable). Implicit bias is introduced to account for somebody who discriminates despite being self-described as non-prejudice.
You can take the test here.
I recommend you do to better understand it.
In science, one of the most import aspects of an experiment is its ability to be retested and give similar results. For example, you weigh an unknown mass and it weighs 10 kg, if you measure it again you should measure 10 kg. Otherwise, something went wrong. The IAT’s has a Cronbach’s alpha, which measures this internal consistency, of about 0.6. Here’s a chart to signify what this means:
|Cronbach’s alpha||Internal consistency|
|α ≥ 0.9||Excellent|
|0.9 > α ≥ 0.8||Good|
|0.8 > α ≥ 0.7||Acceptable|
|0.7 > α ≥ 0.6||Questionable|
|0.6 > α ≥ 0.5||Poor|
|0.5 > α||Unacceptable|
The IAT’s internal consistency lies between “questionable” and “poor”. That’s not good; that’s really bad. Why is this?
This study revealed that researchers could alter the results of the IAT by administering a preceding irrelevant task. This is a big deal. You can change the results that are supposedly reflective of one’s bias without actually changing bias. It also calls into question the preceding survey questions the IAT has such as, “I prefer white/black people over black/white people”. If an irrelevant task or survey question can change results, how can these survey questions affect the outcomes? More research on the priming effect of this is needed.
This study also revealed changing and familiarizing the labels of the IAT change outcomes significantly.
Even if the test does show a higher association of African Americans with negative concepts, does this necessarily mean that the association means the individual views that group as negative?
A study from 2006 by Eric Uhlmann of Yale University argued that another explanation for a higher association of negative concepts with blacks could be that the individual associates that group with oppression, or being worse off, opposed to viewing blacks as negative. This means that a high IAT score could mean that one is less racist and/or more racist. Clearly, this can’t be the case indicating that the test contains error.
Conclusion: the IAT does not accurately measure implicit bias. It also does not show what this bias, if detected, indicates.
Even if we can measure implicit bias, does it affect explicit biases?
Short answer? No.
Researchers analyzed data from over 426 studies comprised of a total of 72,063 participants and concluded the following:
“…we found little evidence that changes in implicit bias mediate changes in explicit bias or behavior. Together, these findings suggest that implicit bias is malleable, but that changing implicit bias does not necessarily lead to changes in explicit bias or behavior”
A Meta-Analysis of Change in Implicit Bias. Available from: https://www.researchgate.net/publication/308926636_A_Meta-Analysis_of_Change_in_Implicit_Bias
Another study attempted to link implicit bias with explicit bias; the study showed that a higher preference for whites in the IAT test reflected a higher rate of prescription of heart medicine in whites. But this is only partly true. The data also showed that doctors with a lower score had a much higher tendency to prescribe medicine to blacks and doctors with a higher score actually only slightly prescribed whites more. This actually means that the higher IAT scoring doctors discriminated less on average. (Overall, whites and blacks have an equal chance of being treated). This discredits the strongest argument for implicit bias’ effect on explicit bias.
This meta-analysis also showed that the IAT had very poor predictive skills of explicit bias.
Unconcious Bias Training
What is unconscious bias training and what’s the science behind it? If you google it you’ll find tags such as “eliminate unconscious bias in just 3 minutes”… Can you become “not racist” in the same time it takes to cook ramen noodles? Let’s find out.
For starters, there is no standard that dictates what unconscious bias is. A company can create a seminar of their own, such as this one from Google.
You probably remember that Google’s unconscious bias training made it into the news a while back after James Damor wrote his memo, the Ideological Echo Chamber, which I would encourage you to go through and read. His memo addresses the highly controversial topic of biological and psychological differences between men and women and how their population distribution may affect the number of women in STEM and other jobs. The science behind biological sex and psychological traits is its own article altogether.
The argument for implicit bias training is that if you teach people about their bias, they will consciously adjust their actions to counteract said bias and create a more inclusive environment for the workplace in order to be more productive. This takes the presupposition that a more diverse workforce is more productive. Is it? This study, Workforce Diversity and Productivity: An Analysis of Employer-Employee Match Data, suggests that more productive companies tend to be more diverse, but making an already unproductive company diverse will not increase productivity.
This has businesses wondering what the incentive is to move from a homogenous workforce to a heterogeneous workforce when there is a clear cost in diversifying, such as having to potentially implement implicit bias training. This will be addressed in the op-ed.
So, implicit bias training could essentially be an educational seminar teaching people about biases. This is fine as long as they’re presenting logistically sound science, which so far they appear to have not.
So does making people aware of their biases actually help?
Studies suggest “no”. This study suggests that there is little to no evidence that making people aware of their biases will change their behavior. And this study even suggests that implicit bias training may actually reinforce stereotypes.
Many do not support the idea of mandatory implicit bias training, even the creator of the IAT, Mahzarin Banaji, acknowledges that mandatory bias training is counterproductive and could result in backlash. You could imagine somebody would be upset if they went to work, were labeled a racist, and told that they can’t defend themselves because it’s unconscious. The title “unconsciously biased” is different than “racist”, I agree. But the two are similar enough. When you call somebody racist, they will react adversely despite the evidence. And the fact the scientific evidence is shaky at best? Well, that would just encourage their biases to remain overlooked because the methods that you’re identifying and reducing them with are bad.
That’s not to say any attempt to reduce bias should be overlooked. Bias training could be promising in the future, but as of right now the science is in its infancy. It’s claiming bold things, but it’s most likely doing more damage than good. If the science improves and a standard seminar procedure is put in effect, then bias training could be introduced into the workplace. But until then it seems like band-aid corporations are cover up explicit racism. I’d argue you don’t have to dive into the unconscious to explain the Starbucks incident. Calling the police on two decently presented black men within two minutes of them waiting for an associate isn’t “unconscious bias”; it’s just racist.
As of the present moment, implicit bias training doesn’t help. It feels like it helps. But it could actually make things worse. And right now, worse racial relations is the last thing we need.
Let’s put implicit bias training back in the oven, it’s not ready.
Citations for the nerds.
- Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102(1), http://dx.doi.org/10.1037/0033-295X.102.1.4
- Behavior Evolution and Altruism, Dr. Ben Wagoner, https://faculty.uca.edu/benw/biol4415/presentations/lect7a.pdf
- Han, H. A., Czellar, S., Olson, M. A., & Fazio, R. H. (2010). Malleability of Attitudes or Malleability of the IAT? Journal of Experimental Social Psychology, 46(2), 286–298. http://doi.org/10.1016/j.jesp.2009.11.011
- Uhlmann, E. L., Brescoll, V. L., & Paluck, E. L. (2006). Are members of low status groups perceived as bad, or badly off? Egalitarian negative associations and automatic prejudice. Journal of Experimental Social Psychology, 42(4), 491-499. http://dx.doi.org/10.1016/j.jesp.2004.10.003
- Forscher, Patrick & Lai, Calvin & R. Axt, Jordan & R. Ebersole, Charles & Herman, Michelle & Devine, Patricia & Nosek, Brian. (2016). A Meta-Analysis of Change in Implicit Bias.
- Green, A. R., Carney, D. R., Pallin, D. J., Ngo, L. H., Raymond, K. L., Iezzoni, L. I., & Banaji, M. R. (2007). Implicit Bias among Physicians and its Prediction of Thrombolysis Decisions for Black and White Patients. Journal of General Internal Medicine, 22(9), 1231–1238. http://doi.org/10.1007/s11606-007-0258-5
- Richard Feloni – http://www.businessinsider.com/google-unconscious-bias-training-presentation-2015-12#you-cant-change-everything-all-at-once-begin-with-one-of-the-four-approaches-and-adapt-it-to-your-work-life-40
- Barrington, Linda & Troske, Kenneth. (2001). Workforce Diversity and Productivity: An Analysis of Employer-Employee Match Data.
- Pointless Diversity Training: Unconscious Bias, New Racism and Agency Mike Noon Work, Employment and Society, Vol 32, Issue 1, pp. 198 – 209 , First Published September 1, 2017
- Duguid, M. M., & Thomas-Hunt, M. C. (2015). Condoning stereotyping? How awareness of stereotyping prevalence impacts expression of stereotypes. Journal of Applied Psychology, 100(2), 343-359.
Mandatory Implicit Bias Training Is a Bad Idea https://www.psychologytoday.com/us/blog/rabble-rouser/201712/mandatory-implicit-bias-training-is-bad-idea