What are our recommendations
Give clear goals and feedback
Similar roles should have similar goals and be evaluated by similar standards. Establish goals and standards as early as possible and communicate them clearly. Be aware that when unmanaged, personal preferences on the importance of contribution areas can become part of structural bias. For new employees, managers can include goal setting as part of the onboarding process.
Furthermore, feedback should be ongoing and part of informal and formal review mechanisms. Feedback and monitoring progress should be part of regular employee one-on-ones. When annual performance reviews are used, schedule more robust conversations on a quarterly or semi-annual basis.
Develop a mechanism that enables employees and managers to easily track feedback. Such a system enables an organization to identify trends in reviewer feedback, potential areas of bias, and roadblocks to advancement. This also helps employees track their progress and set future goals.
Understand how performance reviews can be biased
Bias can come from several different sources. The most common source of bias in reviews is manager or rater bias, which occurs any time one person evaluates the performance of another. Such bias can manifest in considerations of leniency, stereotyping behavior or expectations, preference towards similarity, and the weight or influence negative events are given. Feedback bias can be institutional, influenced by the rating models by which people are evaluated and the organizational culture priorities to which they are held.
Bias can also be encountered when employees rate themselves.1 For self-evaluations, understand that employees can both overrate and underrate. Both can be damaging. If we tend to be judgmental about ourselves, we may tend to underrate. This can be especially true of people in underrepresented groups. Self-rater bias can happen as a result of imposter syndrome, 2 a cultural aversion to self-promotion, or societal expectations and messaging concerning self-promotion.
In practice, anonymity and lack of accountability can bring out bias in the form of subjective, unproductive, and sometimes cruel comments in reviews. 3 A 2014 survey on women in tech found that women were more likely to get negative, less constructive feedback than men. Only 59 percent of the reviews received by men contained critical feedback, compared to 88 percent of the reviews received by women.
The critical feedback men received was heavily geared towards suggestions for additional skills to develop. Feedback for women tended to be more personality based, and called for them to be quieter. Women were advised to “pay attention to your tone” and “let others shine.” This type of negative personality criticism showed up in almost 75 percent of the critical reviews of women compared to roughly two percent of critical reviews of men. 4
A performance review process that mitigates bias and effectively evaluates employees stands to increase the likelihood that a company will not only attract, but also retain and grow, its talent. This must include the early establishment of goals and the use of multiple feedback sources to take advantage of opposing views. Some companies inform reviewers just before giving reviews about the existence and impact of bias; we have not seen any research on the impact of these statements.
Have HR interrogate biases before conducting performance reviews
Reviewers must use a variety of tactics to fully engage with their subjects, setting aside the filters and preconceived notions they take into a performance review. This includes an examination of the biases the reviewer may have faced and how those experiences affected them. Employees may also find that review decisions are influenced by reviewer agendas, rather than the best interests of the employee under review and the company as a whole. It can be important to allow for differences in work style or approach that could contribute to biases, as these may feed into conflict during a review session.
Reviewers should also consider whether a given situation is reminiscent of another event, and if that association is positively, negatively, or neutrally coloring the current review. When discussing the employee’s experiences at the company, the reviewer should consider whether they are projecting their own beliefs about the employee’s career development goals and aspirations onto the conversation. Taking time to explicitly discuss these issues can result in a more equitable and productive outcome.
Consider incorporating joint evaluations into performance reviews. Research indicates that reviewers are less likely to focus on stereotypes and more likely to focus on performance in joint evaluations. 5
Prevent biased language in performance review feedback
Performance review feedback from others and from employees themselves can be subject to bias. Avoid biased language by examining language in performance reviews in the same way you would with job descriptions. Language can reinforce bias. For example, women’s evaluations have been found to be twice as likely to contain language related to their communal or nurturing style. 6
Avoid using language with stereotypical associations to describe performance (for instance, describing a Black employee as “articulate” can have racial overtones). Also be aware that language positively associated with members of one group may have negative connotations when used to describe people from another. For example, the word “aggressive” may be used to positively describe men but is usually used to negatively describe women.
Avoid biases that influence promotion and bonus decisions
In management, the levels of representation are particularly dire. The leadership ranks of technology companies are overwhelmingly white and male. Even minority groups that are overrepresented at lower levels in tech, such as Asian-Americans, face difficulty breaking into the leadership ranks. 7 The numbers of women of color in leadership are so low as to be rendered statistically insignificant. 8
In addition to biased hiring practices and approaches to performance evaluations, low levels of representation at the managerial level can also be attributed to high rates of attrition and biased promotional practices. Women who work in science and tech are 45 percent more likely to leave the industry than their male peers and more likely to report feeling stalled in their careers. 9 Those women who leave do so at a specific moment in their careers, in their mid to late thirties, and do not return to the sector. 10
Increasing representation within leadership is just as important as diversifying at the lower ranks, if not more so. Inclusive leadership teams are more likely to pass on inclusive values to their organizations and serve as role models to people from a variety of backgrounds. Rethink your idea of what a leader looks like. Don’t base it on the current leadership. The idea that you have to be behaving at a certain level to be promoted to that level is inherently biased.
Use standardized systems to measure performance
Have clear and transparent rating areas and weighting processes and identify metrics for team and individual performance. Develop standards of measurement related to core company values and business outcomes. Allow space for additional insights but keep in mind that too much flexibility can allow bias to creep in. Train employees in how to provide constructive feedback on the performance of their colleagues and themselves.
Performance Improvement Plans can be a feedback pitfall
Avoid using Performance Improvement Plans (PIPs) unless there is a chance of the employee pulling out of the situation. In concept, Performance Improvement Plans sound great. They are described as giving an employee a chance to improve on specific areas of failure within a specified time period.
The problems lie in implementation. In many cases, they are designed for (and perceived as) creating a paper trail for terminating an employee. The areas of improvement can be vague, outside the employee’s control, or unlikely to impossible. They can be in areas where the employee has already received feedback for improvement and where they have shown an inability to improve. Ultimately, if you know you want to terminate an employee, and you’ve worked with the employee to try to solve the problem, what is the point of a PIP?
If you want to change behavior, communication is important, but this should be done over time through continuous feedback, and the manager should provide tools or resources to help the employee improve and succeed. When these tools and resources are not available or successful, putting it in writing and setting up a deadline are not going to be meaningfully helpful. If an employee has reached the point of needing a PIP, it reflects a management failure and is usually too late.
Create a clear organizational chart and a consistent promotion ladder
Be consistent about promotion practices. Develop standards for promotion based on performance. Share these standards with all of your employees. Consistently apply these standards when evaluating employees for promotion. This reduces the role of bias in promotion practices. As with regular performance evaluations, considering people collectively rather than individually for a promotion can mitigate bias.
Take the time to carefully consider management levels within your organization and the criteria for reaching them. Be sure that standards associated with each level are neutral or do not unintentionally provide one group with an advantage over another. Publicly share which levels exist within an organization, what the path to advancement looks like, and what the criteria for advancement are.
Finally, be transparent about the compensation ranges associated with each level. Where this information is not transparently shared, access to the information may become privileged, excluding people from underrepresented groups and impeding their ability to advocate on their own behalf or ascertain whether they are appropriately positioned and compensated in the firm. Pay secrecy can contribute to poor performance,11 while pay transparency reduces biases and increases employee confidence.12
Larger startups should review pay across the company periodically to ensure that each employee is being paid fairly. Any salary gap analysis should cover at least gender and race, and ideally all categories. The wage gap analyses announced by companies recently fall short due to their failure to provide enough detail and their failure to extend beyond gender, when we know there is pay disparity due to race.
Install managers who are actively committed to diversity
Moving an individual contributor into a management role is traditionally seen as a promotion. Who you decide you want to make a manager is really important. You should make sure they’re open to diversity, can communicate, understand the role, and treat the role responsibly. Ask them about their views, support them as experiences expand their perspectives, and continuously train them.
Light, J. (May 4 2010). “Self-Assessments Can Be Used Against You.” Wall Street Journal. Retrieved April 2016 from: http://www.wsj.com/articles/SB10001424052748704093204575216740843805662 ↩
Clance, P.R. & Imes, S. (1978). “The Imposter Phenomenon in High-Achieving Women: Dynamics and Therapeutic Intervention.” Psychotherapy Theory, Research, and Practice 15(3). Retrieved December 2016 from: http://www.paulineroseclance.com/pdf/ip_high_achieving_women.pdf ↩
Halverson, M. (February 26 2016). “360 Reviews Often Lead to Cruel, Not Constructive, Criticism.” New York Times. Retrieved April 2016 from: http://www.nytimes.com/2016/02/28/jobs/360-reviews-often-lead-to-cruel-not-constructive-criticism.html ↩
Snyder, K. (August 26 2014). “The abrasiveness trap: High-achieving men and women are described differently in interviews.” Fortune. Retrieved April 2016 from: http://fortune.com/2014/08/26/performance-review-gender-bias/ ↩
Bohnet, I., van Green, A., and Bazerman, M.H. “When Performance Trumps Gender Bias: Joint Versus Separate Evaluation.” Harvard Kennedy School. Retrieved April 2016 from: http://scholar.harvard.edu/files/iris_bohnet/files/when_performance_trumps_gender_bias_final.pdf ↩
Lebowitz, S. (October 2015). “Stanford University researchers analyzed the language in 125 performance reviews from a tech company and found something disturbing.” Business Insider. Retrieved April 2016 from: http://www.businessinsider.com/gendered-language-in-performance-reviews-2015-10 ↩
NPR Staff. (May 7 2015). “Often employees, rarely CEOs: Challenges Asian-Americans Face in Tech.” NPR: All Things Considered. Retrieved April 2016 from: http://www.npr.org/2015/05/17/407478606/often-employees-rarely-ceos-challenges-asian-americans-face-in-tech ↩
Catalyst. (March 2015). “Women in S&P 500 Companies by Race/Ethnicity.” New York: Catalyst. Retrieved April 2016 from: http://www.catalyst.org/knowledge/women-sp-500-companies-raceethnicity ↩
McGregor, J. (January 12 2014). “Keeping women in high-tech fields is big challenge, report finds.” Washington Post. Retrieved April 2016 from: https://www.washingtonpost.com/business/economy/keeping-women-in-high-tech-fields-is-big-challenge-report-finds/2014/02/12/8a53c6ac-93fe-11e3-b46a-5a3d0d2130da_story.html ↩
Alba, D. (November 2014). “Tech’s Gender Gap Wasn’t Always So Bad. Here’s How It Got Worse.” Wired. Retrieved April 2016 from: http://www.wired.com/2014/11/code-documentary-gender-gap/ ↩