Mandatory Implicit Bias Training Is a Bad Idea
It's all the rage. But in the view of some, it's seriously counterproductive.
Posted Dec 02, 2017
“Implicit bias” seems to be everywhere. What is it? “Bias,” to your average layperson, seems to mean something like prejudice or discrimination. “Implicit” is usually taken to mean unconscious or outside of awareness. So “implicit bias” is, supposedly, something like prejudices of which people are not even aware.
However, the research on so-called implicit bias has its serious critics. Almost everything about implicit bias is controversial in scientific circles. It is not clear, for instance, what most implicit bias methods actually measure; their ability to predict discrimination is modest at best; their reliability is low; early claims about their power and immutability have proven unjustified. And yet some colleges and corporations have been rushing to institute "implicit bias trainings" in attempts to reduce discrimination (attempts that are, in my view, misguided and unlikely to be effective)
Over the last few months, I have had several interesting exchanges with Dr. Mahzarin Banaji, one of the most prominent psychological scientists working in the area of implicit methods, beliefs, attitudes, prejudices, and biases. She, along with Dr. Tony Greenwald, created the concept of implicit bias, which has caught on like wildfire, and was even mentioned in a Hillary Clinton election speech.
My own view is that the research framed on implicit bias has been wildly oversold, and its proponents have often leaped to conclusions not justified by the data. And I am not alone. Banaji and I both attended a recent National Science Foundation-supported conference on implicit bias, and one of the other presenters there declared, “I think the sexy story that the public has received is irresponsible."
But if it was just a case of oversold science, the issue would be primarily restricted to the scientific community to resolve what is and is not justified. Instead, however, implicit bias training has been instituted at numerous colleges and universities, as well as some corporations, around the country.
My view is that this is wildly premature—and potentially even dangerous. The overselling of implicit bias has, in my view, along with several other related concepts (microaggressions, stereotype threat, white privilege), contributed to the toxic environment on many campuses and in some corporations in which speech is considered “violence,” and in which, if you say the wrong thing, you can be denounced, ostracized, and even fired. And by “wrong thing,” I am not talking physical threats or sexual harassment. I am talking about making intellectual arguments against affirmative action, acknowledging the evidence that biology contributes to some demographic group differences, or even simply showing a debate regarding Canadian speech laws.
In this context, some of you may find these comments by Dr. Banaji quite interesting. We disagree on many things, most notably the “benefits” from implicit bias educational experiences. As you will see, Dr. Banaji believes it is her responsibility to teach people about implicit bias, and that when she has done so, most people who respond to her consider it a worthwhile experience. She claims that I overestimate the dangers of implicit bias trainings. Despite these and other things about which we disagree, the most interesting, to me, is that even Dr. Banaji, one of the country’s foremost proponents of the concept of implicit bias, comes out here against mandatory implicit bias training. I certainly agree with her on that score.
What appears next is from her.
Yes Lee, you can post it.
In all the verbiage, I hope two points don’t get lost:
- Psychology and sociology data suggest that mandatory training is not as good as voluntary training. Mandatory training has the potential for backlash.
- IAT scores are best considered as medical data or votes. A person may share them, but should not be expected or asked to share them with others at this point.
All the best, mrb
On 10/23/2017 10:31 AM, Banaji, Mahzarin R. wrote:
I don’t remember the comment to which I was responding at the conference. But here are my thoughts, and I don’t mind being cited as long as you do so by providing proper context so that I’m not misunderstood. You’ll understand my being concerned.
I can tell you what I believe and what I do at present.
I said something—it may have been to challenge the use of the term “training." I think “training” is the wrong term to use. I also am not in favor of people teaching about implicit bias without deep knowledge. But that’s not something I can control.
I do offer educational seminars to many places—governmental agencies, nonprofits, for-profits. I find them immensely valuable as a way to learn and to educate. I explicitly do not call them training. I call them seminars—of the sort I would offer to my own students on campus. I start with basic principles of psychology and stick with what I know something about—perception and attention, decision-making of the K&T kind [This refers to Kahneman and Tversky; Kahneman received the Nobel for his work on decision making.] (BTW, that’s where the term “bias” has been used most profligately—although nobody complains when we say “availability bias” or “anchoring bias,” and nobody asks what the anchoring bias actually predicts. We assume it does something important, given the large number of people who veer in one direction, and that we will discover its implications as we progress in our science.)
Towards the end of my seminar, I teach about unconscious cognition, relying on studies that have been widely replicated. I give audiences an IAT showing they have preference for their own institution or group over another—usually an obvious rival. So for example, I show that with a Boston audience, the IAT preference in the room for the Red Sox over the Yankees is massive. As you can imagine, people love this result and are really pleased with their bias. I use this to make an important point: we all have preferences or biases. And that there may be nothing wrong with having a preference or bias in one direction or another. Many of our preferences are very helpful, and we might even want to grow them to be stronger. I would like to grow my preference for vegetables, for example, to be greater than it is. I might prefer my own child over the neighbor’s child—a good preference, I would say, within legal limits. When the audience is comfortable with the idea that we have biases, and that we even should even like that we have these biases, I teach them more.
I tell them about my discovery of my own biases, some of which I personally am not happy to have, but must recognize that they are lodged in my head. I talk about my own race bias, which is hard for me to do, but it’s something I feel I owe those I’m introducing to this idea. You will be impressed with just how relaxing this is for an audience. Then I give them a mild experience with the gender-career test. I say that this test reveals the “thumbprint of culture” on our brain. It reveals that I associate male more with career and female with family compared to the opposite. If they like that they have this association, there’s nothing for them to do. But if they wonder about why it’s there or what it means, they can use their own data to think harder. That’s all.
From thousands of people who have been in these sessions, I have not had a single complaint. From the millions who have visited the website, we have overwhelming responses of interesting questions and suggestions. My experience over the past 15 years of such public teaching is only gratitude and a desire to learn more, irrespective of where in the world the audience comes from. It is a respectful and humble seminar. I make them laugh a lot, mostly by poking fun at myself. What I take pleasure in is that both conservatives and liberals are equally intrigued and educated, because as you can imagine, it’s liberals who are more inconsistent than conservatives are in the disparities they show.
I see it as my duty to teach about this important matter. I treat the public as smart and able to deal with insights that may not be appealing. I derive my belief from the history of science and know that all-important ideas that questioned the primacy and superiority of humans or our planet were met with resistance. I like to call things out as I see them, a quality I hope I can say that I share with you. (FYI: I am about to become the faculty advisor of an undergrad student group called Bullshit Watch. You should start one at Rutgers!)
I am personally opposed to mandatory sessions for anything, but my friends in other disciplines, such as the law, tell me that that’s my libertarian, academic self speaking, and that they disagree. We don’t fight sexual harassment education, even though we know it’s usually lame. I try to tell organizations I work with that even what I do—which is not training, but an educational seminar—should not be mandatory. Many listen. I assume some don’t. I argue that if I do my job of teaching well, people will want to come to it. I try to point out data from Kalev and Dobbin that shows that mandatory training is counterproductive, but that voluntary participation in education has some benefits.
[This next is in response to an example I sent her where a colleague emailed me about participating in an implicit bias training in which they expected to have to discuss their IAT scores.]
I think it is wrong to ask anybody to reveal their IAT score. In the dozens of organizations I’ve worked with, nobody has done this. I suspect there are a few outliers, and given your views, I believe you overestimate them, Lee. If people wish to discuss their implicit cognition data, they should be free to do so, just as we discuss how we voted with our friends and even the public. I consider one’s IAT score to be like a political vote. It’s private. It’s to be used for internal reflection about self, culture, or whatever will help people think about the disparity between two parts of their own mind and how it can be that way. I think given the millions who consume the idea of implicit cognition, a negligible number may do something that isn’t so great with it, such as demand to know the IAT score of anybody.
However, if there’s anything I can do to support the idea that nobody should be asked to reveal their IAT score—in selection contexts especially—I will do more. I would go so far as to say that I will build this topic into some modules I’m producing for general education. You are free to see what I’ve done so far, and hopefully you’ll like them and share them: outsmartinghumanminds.org. I’m not on social media; I assume you are, and that you’ll help me get the word out about these modules that have been newly released. I will consider placing a module on how to understand the IAT—without fear, without guilt, as a way of knowing something you may not know that your daily walk through culture leaves on your brain. That’s what I have done, with no negative repercussions, and will keep on doing. (Quite to the contrary of what you worry about, the number of people who write to say that new knowledge about how their minds work has allowed them to feel freer to think and speak about these matters has created the need for a staff person to simply respond to such email.)
As chair of my department, my time has become even less available. I hope you’ll understand if I’m not able to respond quickly.
This is now Lee Jussim again. First, before commenting, please familiarize yourself with my Rules for Engaging in Controversial Discussions. Short version: no snark, insults, or slurs, keep it short, stay on topic. But please read the whole thing. You are at risk of me taking your comment down if you violate these rules which, as they said in Pirates of the Caribbean, are more like guidelines.
Also, if you are agreeing or disagreeing with someone or something, please be clear as to who. Remember, there are two writers here, with two quite different perspectives: Dr. Banaji, and me. If you write "Your perspective is complete gibberish," we won't really know who you are talking about, unless you are explicit.