Building a Fairer Workplace—Know Your Processes, Know Your Data, Know Your Bias

How can a company and its leaders ensure that their organizational processes are fair and equitable? And how can organizations use data about their talent management processes to increase diversity and eliminate bias? Many companies have been grappling with these questions recently, while doubling down on their commitment to increasing inclusion on the organizational level. But for Emilio J. Castilla (Nanyang Technological University Professor of Management; Professor of Work and Organization Studies; Co-Director, MIT Institute for Work and Employment Research), such questions have been at the core of his research for over 20 years.

In conducting his research, Castilla partners with companies to systematically and analytically review the processes and routines that drive key employment outcomes, such as recruitment and hiring as well as employee promotion and compensation. While his research projects focus on addressing key challenges for the organizations he works with, there is a common theme: It is paramount that the processes and data used to evaluate applicants, employees, and managers are clearly defined and carefully analyzed.

In a recent article, “The Production of Merit: How Managers Understand and Apply Merit in the Workplace, ” Castilla and co-author Aruna Ranganathan, SM ’13, PhD ’14, interviewed a sample of managers and also worked with a Silicon Valley tech company to study merit-based promotion systems and their limits. The researchers found that managers’ own past evaluation experiences as employees—be they positive or negative—are highly influential in determining the process that the managers use to evaluate employee performance. As a result, managers’ criteria for measuring merit can be highly subjective and vary from one person to another.

More specifically, when evaluating employees, managers tend to include or exclude certain factors based on their perception of how those factors affected their own career trajectory. Castilla and Ranganathan also found that most managers tend to follow one of two approaches when assessing employees for advancement and rewards: focused or diffuse. Managers using a focused approach to evaluation emphasize the individual and quantitative elements of performance, while those using a diffuse approach assess the employee more generally, for not only individual and team contributions but also work actions and personal characteristics. Strikingly, the researchers found that managers who were women and/or nonwhite were more likely to use a focused approach to employee evaluation, in part because these managers generally reported having had more negative evaluation experiences in the past (as employees) than did the white male managers in the study; the female and nonwhite managers saw the focused approach as a way to avoid the unfairness they had experienced.

Castilla and Ranganathan’s findings are an important addition to the conversation around talent management and merit-based evaluation, since they highlight the fact that, even within one organization, merit is a subjective concept. If organizations want to build successful and fair evaluation, promotion, and reward practices for an increasingly diverse workforce, Castilla argues that their leaders, managers, and employees must have a collective understanding of what merit means in the context of their organization. In order to form this definition, Castilla advises leaders to “think about what type of organization you want to build and what type of professionals you want to attract and retain. Then start putting in place methods that will allow you to develop and apply a clear, fair, and valid definition of merit. Only when employees, managers, and executives all share a common understanding of what ‘merit’ actually is (and how to measure it) can we hope that the same standards will be applied consistently to all, regardless of demographic factors. This can help make your workplace more equitable, fair, and diverse.”

This process, Castilla explains, can be achieved successfully through using precise and concrete criteria when refining your organization’s definition of merit. For example, it is not enough to say that an employee should demonstrate “initiative” or “a good attitude”; these terms may be far too subjective in practice and thus are likely to result in bias and misalignment with key organizational goals. Instead, it is up to leadership to precisely define and measure those qualities objectively, so that decision-makers throughout the organization are able to more consistently evaluate employees based on progress made toward organizational goals.

The work does not stop there, however. Defining and identifying the measurements of merit is just the beginning. The next step, Castilla explains, is to determine “whether the measure is really relevant and valid, and that it is not biased.”

This analytical framework aligns with Castilla’s research into talent or people analytics, which he defines as a “data-driven approach to improving people-related decisions for the purpose of advancing the success of not only the organization but also of individual employees.” Castilla explains that not all organizations understand how to do talent analytics effectively. He cited one large organization that pre-pandemic had already collected high-quality and comprehensive data about the skills, capabilities, and knowledge of its employees and managers, encompassing metrics from their entire tenure within the organization.

Through his analysis of the company’s work distribution before and during the COVID-19 pandemic, Castilla found that such a robust database was critical to the company’s being able to successfully redistribute work in a time of crisis, keeping customer satisfaction high while retaining top talent and maintaining employees’ engagement.

However, Castilla notes, employers need to be conscious of what data they are collecting and be careful and responsible when using such data to make people-related decisions. By focusing on too narrow a scope of data, organizations leave a door open for focusing on the wrong metrics or, even worse, activating biases in their workplaces. This is why Castilla enjoys collaborating with organizations and their leaders to help them be strategic and successful in talent management.

“My team and I enjoy working with real companies to identify opportunities and address challenges concerning the management of employees and managers,” he says. “For example, I am currently collaborating with a company to experiment with new approaches to recruiting and hiring talented and diverse candidates. Some of these interventions are surprisingly low cost, with fast returns. My hope is to continue working with many more organizations to help create successful experiences for all their key stakeholders, including applicants, employees, managers, customers, and top executives.”