Mitigating the Effects of Implicit Bias Begins with Data Analysis

By Todd McFall and Megan Regan

Lately it seems everywhere you turn, there is much ado about implicit bias. For good reason, too. Implicit bias, the subconscious altering of reality that impacts humans’ decision making, leads to large organizational costs that keep organizations (and societies) from reaching full potential. There is little doubt implicit bias played a role in fomenting the financial crisis of 2007-8 and is a main factor in instances of bad interactions between police officers and citizens.

Implicit bias is most certainly a problem at workplaces around the world, too. Women and members of protected groups have a litany of stories of mistreatment at workplaces across every conceivable industry. Writer Jessica Nordell qualified and quantified these biases in her book The End of Bias, where she presented evidence from a simulated workplace that showed only a 3% undervaluing of women’s productivity had dire consequences when it came to them advancing through the hierarchy of the firm. Workers’ stories and the evidence shared by writers and researchers like Nordell makes it easy to conclude that HR managers must pay attention to the design of all employee management systems or risk leaving their organization open to problems caused by not mitigating systemic inherent biases.

Figuring out how to mitigate bias is not just for problem avoidance. For instance, a Gallup Workplace article describes some of the benefits of investing in more just HR systems. These workplaces employ better engaged workers who have better relationships with clients, experience lower employee turnover, and more productive employees compared to organizations with less engaged workers.

So, what biases might hinder organizations from evaluating employees in a more accurate and just manner? Consider this short list of possible biases that might be impacting your HR practices right now.

• Overconfidence bias- humans tend to overestimate the chance of success. So, it would not be surprising to hear HR managers state confidently—probably too confidently —that their systems are performing well. If this were the case, then tendency towards overconfidence keeps organizations from even pondering changes that might mitigate outcomes from bias.
• Status quo bias- humans tend not to want to have to make changes to their lives. This hesitancy makes sense because changes are uncertain during the implementation process and payoffs are unknown. Changing HR systems is potentially daunting and a bias toward the current practices makes the likelihood of altering systems to mitigate bias even smaller.
• Confirmation bias- humans like to know their worldview is correct, so we place more weight on evidence that supports our current beliefs. With regards to HR management, this would mean paying too much attention to successful employees who confirm the good aspects of the current system. A more honest accounting of an organization’s systems would give appropriate weight to employees who did not reach such heights and expose flaws in systems that could be changed for the better.
• Recency bias- humans have a limited amount of memory, and remembering past events is difficult compared to recalling recent events. This bias is very important to consider when evaluating employees over a long period. If good work occurred further in the past, it is likely to be discounted for work that occurred more recently, unless an organization has a system that negates this tendency. Creating systems that enable all work to be judged equally across time mitigates problems from this bias.

We cannot go without saying that we advocate for adopting systems that mitigate the effects of bias completely. However, we also recognize that overhauling HR systems for this intent might be incredibly

expensive for many organizations. The good news is that much is known about how to use data analysis to detect the existence of these biases, so organizations that want to push back against bias can use analytics to determine the best place to begin adjustments to their current practices. The recent flourishing of statistical analysis software enables the use of powerful tools like regression analysis at low costs. These software applications allow for a plethora of employee characteristics to be correlated with HR outcomes. In many instances, cause-and-effect relationships can be observed, and accurate measures of how much implicit bias might be impacting an organization can be identified. Investments of these sorts can provide managers of HR systems the opportunity to make data-driven adjustments to help mitigate the effects of implicit bias.

To this end, we view investments in appropriate data analysis as a middle-sized investment for an organization, somewhere between maintaining the status quo and adopting entirely new systems. Good data analysis can offer insights as to what changes your organization can make that will be most valuable, a reality that must be preferred to applying changes that have uncertain returns. The results from good analysis can provide a roadmap towards improvement for your organization to follow instead of taking a leap of faith.

Though it is sometimes a difficult view to embrace, organizations should assume that implicit bias exists in their HR practices and should seek ways to improve these practices. Adopting extreme changes are risky and expensive, but the adoption of half measures has been shown to be unsuccessful in mitigating bias. The use of smart data analysis can help provide a plan for improvement that will make clearer the benefits of mitigation investments and ease the implementation process. Now, all you need is to overcome your status quo bias and start finding ways to improve.