Does Blind Screening Make a Difference?
Just how much does inherent bias really influence our decision-making?
In an age when everyone is doing their best to be unbiased, this is a very important question to ask. We would never intentionally favor someone we knew or understood over people who are unfamiliar or favor someone we viewed as stronger or from a better school. We would never knowingly discount someone for traits that suggest their gender or cultural background. But according to the research, we still do.
Decision-Making Bias: The Data Reveals Hidden Influence
Yale Psychologists Reveal “Mainstream” Hiring Decision Bias
Yale University psychologists conducted a landmark study using select differences between otherwise identical resumes to study the inherent bias of hiring decisions. Participants were given identical resumes with only two differences; one had more experience while the other had more education, and one was named “Michael” while the other was named “Michelle.” In both versions of the experiment, Michael was picked either for having “more street smarts” or more education.
Even without the education split, resumes named “Heidi” are picked consistently less than resumes named “Howard.”
At the same time, identical resumes between “Greg” and “Jamal” and other white-sounding vs. black-sounding name tests proved that white-sounding names were called back 50% more often.
These patterns reveal an unfortunate pattern, a bias that we don’t realize and usually rationalize – even though intentional bias would never be exercised.
Hubble Proposals Reveal Gender and Familiarity Bias
Another excellent example of accidental bias is an examination of who is approved to use the Hubble telescope. Thousands of proposals flood the telescope administration board every year, and only the top 20% are approved. Before Ian Neill Reid took a curious look at who was being approved, no one realized that female-led proposals had a much lower chance of being accepted – and the same was true for new researchers who had not used the Hubble before. Often, approvals were pushed through after being questioned by one reviewer personally vouching for the author of the proposal.
In examining the problem, he found that discussions of approval often centered around who the researchers were, not just the academic value of their proposals. A simple change led to startling transformations. They introduced blind evaluation where the identity of the proposing scientist was hidden from reviewers.
In the first year, proposals from women outpaced approved proposals from men. Soon, the number of first-time users of the Hubble increased from about a dozen per year to 50 per year – on the merit of their proposals alone.
This example reveals not only a subtle gender bias but also a familiarity bias. Popular, well-known, and male scientists were all considered more likely to succeed and were more likely to be approved. With a simple blind evaluation, the numbers reveal that bias was successfully removed.
Schema: Human Brains Create Bias to Survive
The good news is that bias, while frustrating, is not anyone’s fault. It’s a natural consequence of trial-and-error survival in the human brain. From the time we are infants, first seeing and hearing the world, we begin to form schema. What is safe? What is familiar? What can we do freely, and what must we do with caution?
The brain automatically tries to recognize patterns and build a safe pattern of behavior in response. We learn how to act in a grocery store vs. a dance club. We learn to recognize a uniform as trustworthy or non-trustworthy. We learn to recognize a person’s mood by their posture. We recognize how cars will act in an intersection by their patterns of motion. These are all schema we build to help make quick, safe decisions. In other words, we can’t help but try to take in all the details and make the best decision based on the big-picture.
But when choosing a resume – or an academic proposal – the schema approach may not be appropriate.
Literal Blinds Remove Gender Bias from Classical Music
We propose that both blind screening and clear identification of what you are screening for can eliminate the natural bias of the human schema-building mind.
In the music industry, they discovered the importance of a literal blind as far back as the 1980s. In the 1970s, it was discovered that only 5% of US orchestra musicians were female. Considering the number of talented female musicians, this was examined. Orchestras seeking the true best candidates began having musicians audition behind a solid white screen so that their identities were hidden from the judges. This led to a rapid 55% increase in women chosen via the quality of their audible performance alone.
Once again, as soon as bias was removed, the results revealed its powerful yet unspoken effect. Because the orchestras knew they wanted musical talent and skill above all else, blind screening allowed them to truly hone their selection results and avoid bias that would select lesser musicians.
How Blind Screening Breaks Schema Bias
Blind screening is the key to building a bias-free candidate selection process — whether you are selecting candidates for a new role or proposals for the Hubble telescope. The two factors that are essential are setting your priorities and preventing schema-based decision-making.
Define Priorities Before Screening Begins
The first step every screening process should start with is defining your goals. What are you screening for? What are the most important features you are looking for when you consider a resume? Do you need more education or more experience? Would you favor candidates with a larger toolkit or dedicated experience to one platform? Do you need someone with a leadership background or a propensity for travel?
The Yale study found that asking reviewers to define their priorities could eliminate the factor of “shifting merits” where the male-named resume was selected when balanced for experience or education. The orchestra assisted their blind screening by defining musical performance as their only important factor. Hubble discovered more first-time and female researchers by ceasing to discuss anything but the academic merit of each proposal.
Remove Identity-Schema Indicators
Next, remove your brain’s ability to build a schema around the resume you are looking at. The most important factor has been the name – suggesting a woman and/or specific cultural background. However, other factors, like the school a person attended or even the cities they have lived in, build a “mental picture” of the candidate – a picture that doesn’t actually help assess their qualifications for the job.
Most programs start by removing specific demographic-revealing information. This is how to create redacted resumes.
- Hide the names of candidates
- Hide school names and locations
- Hide the ages, if mentioned, of candidates
- Remove accidentally revealing comments regarding family or status
Create One-to-One Programmatic Resumes
The best way to create perfectly comparable resumes is to use a program to extract the skills and experiences from each resume to create nearly identical one-to-one comparison documents without any revealing information.
This is because lexical, syntactic, and semantic differences exist in how people of different genders and backgrounds write resumes.
Blind Screening Makes a Difference
Here at TalVista, we are dedicated to using data-driven methods to find the best candidates for each job through bias removal and merit-based selection. Through both published research and our own applied results, we have found that blind screening makes a profound difference in selecting candidates based on their objective merits and skills by eliminating the influence of bias in decision-making.
Explore our technology for easy blind/redacted resume screening and review to optimize your hiring process for the best possible candidate selections. Contact us to build a custom hiring solution, including blind screening for your business.