Last year, Google unveiled a new initiative to combat implicit bias in the workplace. The initiative has garnered widespread interest in the wake of recent disclosures by several leading tech companies, including Google, regarding the diversity (or lack thereof) of their workforces—a problem my colleague Lizzy Gropman discussed last week.
The concept of implicit or unconscious bias – attitudes or stereotypes that unconsciously affect our perceptions and decisions – is hardly new. It has been linked to gender, racial and other disparities in a wide range of contexts, including law enforcement’s disproportionate use of deadly force against black men, professors’ favoring of white male students in university settings, and the disparate medical treatment received by white and non-white patients.
But it is in the employment arena where the concept is really gaining traction. According to one estimate, 20% of large employers with diversity programs now offer some form of unconscious bias training – up from 2% a mere five years ago. This figure is projected to rise to 50% in another five years.
Numerous studies have examined the insidious effects of implicit bias in the workplace – how it subtly colors perceptions of employees’ competence, intelligence, leadership potential and overall worth. Such biases come into play at every step of the employment process, beginning with hiring.
In a classic study entitled “Are Emily and Greg More Employable than Lakisha and Jamal?” Marianne Bertrand and Sendhil Mullainathan found that resumes from candidates with “black” sounding names were 50% less likely to elicit callbacks than those from their “white” sounding counterparts, despite identical qualifications. The same bias has been found for resumes with female names, particularly when it comes to the STEM fields, which are implicitly associated with men. In one experiment, science professors received identical resumes for a lab manager position, but rated the male applicants as more competent, and offered then on average $4,000 more in starting salary. (The results were apparently jarring enough to inspire Google’s unconscious bias initiative.)
Unconscious bias continues to infiltrate the decision-making process well beyond the hiring phase, impacting how employees are evaluated, promoted and paid. In an earlier blog post, I discussed a study showing that women in the tech industry were far more likely to receive personality-driven criticisms than their male counterparts; the same attributes that were deemed assets in men (e.g., assertiveness) were viewed as liabilities for women. In another study released earlier this year, identical writing samples submitted to law firm partners received drastically different scores depending on the writer’s assigned race. Black associates were not only more likely to be penalized for spelling, grammar, factual and formatting errors, but also received very different qualitative feedback. For example, the “Caucasian” version of Thomas Meyer (one of the fictitious associates) was described as having “potential” and “good analytical skills,” whereas “African American” Meyer elicited negative comments such as “can’t believe he went to NYU” and “average at best.”
These and other studies help shed light on why equal employment opportunities continue to elude women and minorities – and why their attrition rates are higher — even at companies that are seemingly committed to diversity. Research shows that even the smallest representation of bias can have cumulative, far-reaching effects; for example, a computer simulation used in Google’s training demonstrates how a 1% bias in performance review scores can, over time, dramatically skew the distribution of women in a company’s leadership ranks.
The subtle, unconscious nature of implicit bias makes it more challenging to address than overt discrimination. But researchers and practitioners are coming up with a number of strategies – not just employee trainings, but broader structural reforms. Companies should scrutinize their employment policies and practices to determine which processes are particularly vulnerable to bias. For example, much has been written about the problems inherent in basing hiring decisions on nebulous criteria like “cultural fit” rather than objective criteria. A more structured interview process – one that limits the discretion of interviewers – can limit the amount of bias that creeps into the decision-making process. In addition, employers should do routine resume and performance review checks (along with reviews of promotion and pay data) to identify any inconsistencies in how particular groups of employees are evaluated. The 2014 “Implicit Bias Review,” published by Ohio State University’s Kirwan Institute for the Study of Race and Ethnicity, contains a number of other strategies for combating implicit bias in the workplace and other arenas.
It remains to be seen how effective Google’s initiative will be. But it represents an important first step in confronting the dismal lack of diversity in the tech industry.
