Strategies for Interrupting Bias in Human Capital Processes
This post is about the specific strategies that leaders can use to identify and address potential biases in their organizational processes, which research suggests is the key lever to achieving diversity and equity goals.
* * *
Recruitment and Hiring Decisions
The short answer: develop clear hiring criteria, interview for evidence of skill, and apply the same tools to every candidate.
POTENTIAL BIAS
Organization disproportionately relies on referrals from current employees to source candidates, limiting itself to the underlying diversity in employees’ social networks
INTERRUPTERS
Post jobs as widely as possible
Encourage recruiters to proactively broaden their networks to increase the reach of job posts
Contract with women and minority headhunters since their networks may be more diverse
HOW THE BIAS SHOWS UP IN DATA
Lack of demographic diversity in job applicant pool
POTENTIAL BIAS
Job descriptions that include language that is more attractive to some groups than to others
INTERRUPTERS
Have someone trained in spotting bias review all job descriptions or use automated tools (e.g., Textio) to do so
Balance the use of potentially biased words (e.g., go back and forth between “build” and “create”)
Go beyond vague language and include the specific skills and experiences you’re looking for in the job description
HOW THE BIAS SHOWS UP IN DATA
Demographic skew in applications, especially by gender
POTENTIAL BIAS
Those screening resumes give preference to people with similar backgrounds and experiences as them and/or discount those with different experiences (affinity bias)
INTERRUPTERS
Use trained resume screeners
Provide diverse examples of good experiences that might be otherwise overlooked (e.g., the names of minority organizations at a college, community service leadership)
Encourage or require screeners to spend more time on each application (implicit bias is more prominent when making snap decisions)
Create “blind” resumes that remove as much information as possible about the candidate’s race, gender, etc.
HOW THE BIAS SHOWS UP IN DATA
Differences in screen pass rates between demographic groups
POTENTIAL BIAS
Technology used to screen applications contains bias (e.g., algorithms are trained on data that contains bias)
INTERRUPTERS
Ask technology providers for evidence that their models reduce the chance of being biased
Scrutinize whether model-driven ratings are correlated with demographic variables or geography
HOW THE BIAS SHOWS UP IN DATA
Differences in screen pass rates between demographic groups
POTENTIAL BIAS
Unstructured interviews and unclear criteria leave room for interviewers and hiring managers to insert biases—unconsciously or consciously—into their decisions
INTERRUPTERS
Develop and use predetermined hiring criteria, including for “culture fit” (i.e., make it explicit what you need and why)
Structure interviews around skill demonstration (e.g., “tell me about a time where you demonstrated…”, “if you were faced with this problem, what would be your approach)
Ask the same questions to every candidate
Require interviewers to submit their ratings of the candidate before discussing with others
Require a diverse candidate slate, ideally with multiple women or people of color. “Research shows that the odds of hiring a woman are 79 times as great if at least two women are in the finalist pool, while the odds of hiring a nonwhite candidate are 194 times as great with at least two finalist minority applicants.” (source)
HOW THE BIAS SHOWS UP IN DATA
Differences in the distribution of interview ratings between demographic groups
Differences in interview pass rates between demographic groups
Presence of irrelevant commentary in interview reports
Assignments and Development Opportunities
The short answer: democratize knowledge of opportunities and formally recognize all contributions.
POTENTIAL BIAS
Unequal distribution of high visibility and high impact projects
Unequal access to senior leaders who decide who gets what opportunities
INTERRUPTERS
Create a norm of reviewing everyone’s potential to take on an assignment rather than relying on the go-to team members
Create a norm of having a diverse team on the highest visibility projects
Review assignment distribution for evidence of bias
Create a norm of considering a diverse slate for promotion opportunities, and don’t rely on self-nomination
Post and communicate new opportunities to democratize knowledge of them
Regularly schedule time with everyone on your team to provide more equal access
Turn mentoring programs into sponsorship programs (aimed at generating promotions) and hold sponsors accountable for the success of their charges
Train mentors/sponsors on the challenges of forming relationships across difference
HOW THE BIAS SHOWS UP IN DATA
Within a given level, women and minorities disproportionately in non-strategic or non-core roles (i.e., those directly affecting profit or the most important outcomes for the organization)
Lower promotion rates and slower promotion timelines for women and minorities
POTENTIAL BIAS
Women and minorities get less frequent, more vague, and less affirming feedback, which hinders their ability to improve and clouds their self-assessment of their readiness for advancement
INTERRUPTERS
Set an expectation that managers have regularly scheduled development conversations
Train managers on giving high-quality feedback
Implement template for performance reviews that includes a forward-looking development plan for each employee
Review performance evaluations before they are sent to ensure quality
Reduce reliance on self-nomination for promotion and stretch assignments
HOW THE BIAS SHOWS UP IN DATA
Qualitative reports by employees
Lower rates of self-nomination for open positions and stretch assignments for women and minorities
Lower promotion rates and slower promotion timelines for women and minorities
POTENTIAL BIASES
Unequal distribution of “office housework”
Partly because they are underrepresented, women and minorities spend disproportionate amount of time on activities like recruiting and mentoring that are outside of their official performance goals
Minorities volunteer in the community at higher rates, but these activities are not always known or credited at work
INTERRUPTERS
“Set up a rotation for office housework, and don’t ask for volunteers.” (source)
Include space in self-assessment and performance review forms for unofficial contributions
Make extracurricular investments a plus factor in ratings
HOW THE BIASES SHOW UP IN DATA
Qualitative reports by employees
Evaluations, Promotion Decisions, and Compensation
The short answer: clear criteria that is communicated broadly and applied uniformly, and after-the-fact review and adjustment for equity.
POTENTIAL BIASES
Unclear performance criteria leaves room for managers to insert biases—unconsciously or consciously—Into their performance ratings and evaluations
Managers are more reluctant to give the highest ratings to women
INTERRUPTERS
Create clear benchmarks for performance ratings
Separate ratings of performance and skill from those of potential
Review performance evaluations for unnecessary commentary and potentially biased language
“Blind” review of performance descriptions against the benchmarks
Review comprehensive results for patterns of bias (and make corrections if necessary)
HOW THE BIASES SHOW UP IN DATA
Differences in distribution of ratings between group demographic
Lower presence of women in the highest rating category
Presence of irrelevant commentary in performance evaluations. “One study found that 66% of women’s reviews contained comments about their personalities, but only 1% of men’s reviews did.” (source)
Qualitative reports by employees or surveys that indicate that employees do not understand ratings criteria
POTENTIAL BIAS
Unspoken criteria for promotion, and those with the strongest relationships to senior leaders get the “real” story
INTERRUPTERS
Clearly communicate evaluation and promotion criteria
Structure group discussions about promotion specifically around those criteria
HOW THE BIAS SHOWS UP IN DATA
Lower promotion rates and slower promotion timelines for women and minorities, especially into the “executive” or “partner” levels of an organization
Discussions of candidates for promotion include statements like, “she’s just not there yet” or “I’m not sure he’s a fit” without a stated rationale
Qualitative reports by employees or surveys that indicate that they do not understand promotion criteria
POTENTIAL BIAS
Using compensation history to determine compensation upon hire. “Even well-intentioned employers, if they peg your salary to your last job, can carry forward discrimination from previous jobs…” (source)
INTERRUPTERS
Do not ask applicants for their previous compensation
Develop clear compensation bands for specific levels and jobs types
Conduct periodic pay equity audits, and make adjustments appropriately
HOW THE BIAS SHOWS UP IN DATA
Pay gaps between demographic groups within the same level and job family at time of hire
POTENTIAL BIAS
Unclear criteria for compensation decisions leaves room for managers to unconsciously or consciously insert biases into decisions
INTERRUPTERS
Reduce manager discretion with clear criteria for how compensation decisions follow from performance ratings
Implement accountability and transparency into the system—e.g., “a performance reward committee was appointed to monitor reward decisions” (source)
Modify initial pay decisions to achieve equity
HOW THE BIAS SHOWS UP IN DATA
Pay gaps between demographic groups within the same level and job family
Final thoughts
While this post focused on how to remove bias from the organizational processes that generate unequal outcomes, leaders must also examine their own personal biases. If leaders can reduce their own biases through education and mitigate the biases through quality decision-making processes, these interventions are more likely to bear fruit.
Second, the research suggests that the impact of any single initiative may be low. Instead, impact likely comes with the cumulative impact of multiple initiatives, done in concert.
Third, while removing bias from existing processes is helpful, many organizations will find that to really drive improvements in diversity, they will need to be proactive and take steps outside of normal practice. For example, it’s unlikely for firms to see greater diversity in applicants if it uses the same channels and networks that it does today.
Finally, efforts to reduce bias require sustained attention. Or as Joan Williams put it in her article “Hacking Tech’s Diversity Problem”:
“Doing anything once will not change organizational culture forever. You need to continually interrupt bias.”
* * *
Additional Resources
Synthesis of diversity and inclusion research
Blog Post: Leading Inclusively
Blog Post: Specific Ideas on Managing for Inclusion
Blog Post: How Business Leaders Can Support Equity...Right Now
Sources
Brett, Jeanne, et al. “Managing Multicultural Teams.” Harvard Business Review, Nov 2006.
Castilla. Emilio J. “Achieving Meritocracy in the Workplace.” MIT Sloan Management Review online, June 13, 2016.
Cohen, Paige and Gavett, Gretchen. “The Day-to-Day Work of Diversity and Inclusion.” Harvard Business Review, Nov 2019.
Dobbin, Frank and Kalev, Alexandra. “Why Diversity Programs Fail.” Harvard Business Review, Jul-Aug 2016.
Eberhardt, Jennifer L. Biased: Uncovering the Hidden Prejudice That Shapes What We See, Think, and Do. 1st Edition, Viking, 2019.
Harbert, Tam. “Compensation bias is bad for business. Here’s how to fix it.” MIT Sloan School of Management website, Apr 17, 2019
Hewlett, Sylvia Ann, et al. “Leadership in Your Midst: Tapping the Hidden Strengths of Minority Executives.” Harvard Business Review, Nov 2005.
Ibarra, Herminia, et al. “Why Men Still Get More Promotions Than Women.” Harvard Business Review, Sept 2010.
Rivera, Lauren and Tilcsik, András. “One Way to Reduce Gender Bias in Performance Reviews.” Harvard Business Review online, Apr 17, 2019.
Society for Human Resource Management. “7 Practical Ways to Reduce Bias in Your Hiring Process.” April 19, 2018.
Thomas, David A. “Race Matters: The Truth About Mentoring Minorities.” Harvard Business Review, Apr 2001.
Williams, Joan C. “Hacking Tech’s Diversity Problem.” Harvard Business Review, Oct 2014.
Williams, Joan C. and Mihaylo, Sky. “How the Best Bosses Interrupt Bias on Their Teams.” Harvard Business Review, Nov-Dec 2019.
Williams, Maxine. “Numbers Take Us Only So Far.” Harvard Business Review, Nov-Dec 2017“