Unconscious Bias Training isn't a Solution

Humans are riddled with biases, to the extent that there is a bias toward the concept of bias (see, the immediate negative reaction many people have at the observation that they are biased).  Absent omniscience and instantaneous perfect judgement biases are an inevitability - we are stuck making judgments with imperfect information under non-ideal situations absolutely constantly and we need some mechanism to arrive at a judgement none-the-less.  That mechanism is bias.  

Sometimes that bias is pretty spot on.  Aversions to spoiled food, revulsions toward unsanitary conditions, and anxiety when walking on raised narrow walkways without handrails are all biases that help keep us safe (Disney, I have your next big D+ Star Wars series - an Imperial Architect secretly working for the rebellion and killing off Imperial Navy personnel through ridiculously dangerous walkway design).  BUT we also have many, many biases that are either the result of evolutionary adaptations optimized for life on the African Savanna 250,000 years ago or are culturally generated through an incredibly flawed process where the flow of time accretes terrible bias and is resistant to shedding them.  These biases serve us very poorly in the conditions of the third decade of the 21st century.

Workplaces have begun to appreciate the implication of these sub-optimal biases when it results in judgement of other people based on their inherent or cultural characteristics, either because the business is operated by decent people who want to do what is Right (fairly rare) or because they now see a financial downside to those biases (most companies).  And thus, the proliferation of "Unconscious Bias Training" - the "we need to do something about this, and this something is simple and cheap" approach to problem solving that business loves.  Unconscious bias training is basically "hey meatbag, you have biases you are often not aware of about other people, and that can skew your judgement.  Try not to let it when the result will be negative for our company".  Reader, it will shock you that doesn't actually work.  

(also, hey companies, look around - you also need to do something about the CONSCIOUS biases that are permeating society.  Do you think the openly racist, misogynistic, transphobe, homophobe employees don't absolutely know they are those things and embrace them?  Though there is actually an easy fix for that, unlike with unconscious bias - you can fire them.  Being biased against intolerance may be a bias, but it's the bias against spoiled food, against something rancid and toxic.  You can only create tolerance by being intolerant of intolerance, and pretending otherwise either marks you as ignorant or disingenuous in your motives.  Also, absolutely do not allow anyone to paint that as an "anti-conservative" bias - they aren't being fired for having an opinion about taxes, the role of government, regulations, public land use, or federalism and it does people who do have those opinions a disservice to suggest they are synonymous with hatred.)

The medical field has been doing robust studies of limiting the impact of certain biases in medical outcomes for decades, since the entire history of medicine is the history of worse outcomes for women, for people ethnically distinct from the politically/economically dominant ethnicity in the region, and in the impoverished, even when given equal access to medicine.  Medicine is taking baby steps to be less atrociously terrible about that, and often trying to guide those steps with robust scientific inquiry these days (unlike basically everyone else, being guided by opportunistic consulting companies with a service to sell).  And what the medical field has found is that Unconscious Bias training in isolation results in WORSE outcomes because of either:

  • the doctor/nurse/lab tech/etc. believes that as the result of the training they are now aware of their biases and won't be impacted by them, resulting in a wrong judgement they are even more confident about than if they hadn't taken the training.
  • they swing far too wide in the other direction, convinced that all of their judgement is impacted by unconscious bias and doubting and second guessing it even in the face of compelling evidence that a judgement is the correct one.

I bolded "isolation" above for a reason, because what scientific studies in medicine have shown is that when Unconscious Bias training is used to raise awareness of the problem, but then areas where that bias are resulting in negative outcomes are identified and DELIBERATE PROCESS is put in place to eliminate the impact of the bias on the outcome, outcomes dramatically improve.  The unconscious bias training becomes the vehicle used to get employees to think critically about where deliberate process might be necessary, and to explain to them why that deliberate process is necessary.  It isn't the solution, its education about the problem.

For example, in Renal Function Tests (evaluation of how healthy the kidneys are), the black population in the US almost always reports lower numbers than the white population.  And for decades the medical profession believed this was from physiological differences between the ethnicities, NOT because any studies provided a mechanistic explanation, but simply from assumptions about the inherent supremacy of white kidneys and how well they worked (yes, really).  And this difference was encoded in the official diagnostics criteria - a low renal function result was simply "natural" if the patient was black.  In the 1980s it was finally pointed out that "holy shit, the black population has been segregated and then exposed to pervasive environmental pollution, and you can see similar low renal function in the impoverished white communities also subjected to environmental injustice.  It's just that those white communities make up a much smaller part of the white population than the black communities do for the black population."  But despite this being known for decades now, when diagnosing kidney illness doctors were still not using a consistent rubric regardless of ethnicity.  The solution was to have the person diagnosing low kidney function be blind to the ethnicity of the patient behind the test results, so that ethnicity couldn't be part of the judgement (or increasingly now, have a single set of thresholds in the medical software, that flags any value below that threshold regardless of ethnicity).

If unconscious bias training can't by itself change judgement when it's a QUANTITATIVE judgement - "is this number below this threshold" - and where an endemic problem for a population is well described - why on earth would anyone think it will change outcomes for the much more frequent qualitative judgments we see commonly in the workplace?  "What do we think about these candidates", "how well did this person perform this year, and what impact did they have", "am I offering the same opportunities evenly to all of my employees", etc.

What we need to be doing, assuming we are either motivated by social justice (it would be really nice if this was enough motivation for businesses), or the mercenary calculus that equality is financially beneficial to the stock price sometimes (at least this is a reliable lever to move businesses), is to instead be saying "We know unconscious bias will impact judgement in this situation in a non-desirable way, how do we build process/methodology/tools  to reduce that negative impact, and how do we measure that the effort is having the desired effect".