Removing Human Bias from Management
Research shows reducing discrimination is harder than we thought.
Can workplace policies prevent bias? | iStock/Yuri_Arcurs
In response to centuries of racial discrimination, American companies spent the last two generations implementing personnel policies designed to limit managerial discretion. Standardized hiring criteria, performance reviews, grievance procedures, and other employment policies aim to remove human bias from management and therefore reduce discrimination and lead to more diverse organizations.
But it’s difficult to tell which policies actually have an effect, says John-Paul Ferguson, a professor of organizational behavior at Stanford Graduate School of Business.
“We don’t actually know what works,” says Ferguson.
In new research, he finds that, surprisingly, even the unionization of companies did not guarantee that diversity would follow.
To better understand the impact of policies that limit manager discretion, Ferguson turned to union elections. Unions, which helped inspire workplace diversity policies in the 1960s and 1970s, typically limit managers’ control over areas like hiring, discipline, promotion, and termination.
Ferguson tracked workforce composition over time, following the onset or failure of union organizing. He focused on establishments with more than 100 employees that were targeted by union-organization drives between 1984 and 2008. These companies are required to file yearly employment surveys with the Equal Employment Opportunity Commission that detail the racial and sexual composition of their workforce in broad occupational categories.
He considered unionization’s impact on the number of women and minorities in management, and occupational segregation within an organization. He also looked at how the composition of the specific workforces compared to the larger labor market. Typically, unionized organizations have more women and minority managers compared to establishments without unions.
Ferguson notes that union organizing was more likely to succeed in establishments that were segregated, relative to their local labor markets (either disproportionately white or disproportionately non-white), and also in firms that already had a larger than average proportion of managers who were female or members of a minority group. This, he says, suggests that the relationship between starting a union and changes in the workforce’s makeup, like greater workforce diversity or more women and minorities in management, might be endogenous. These types of companies were already open to diversity, and therefore any policy changes that followed may have emanated from a certain amount of self-selection.
In other words, workforce diversity may have less to do with the policies and more to do with the characteristics of organizations that adopt these policies.
For example, unions may target workplaces where they observe increasing diversity among managers, assuming that such firms might be more sympathetic to non-white workers, who favor unions at higher rates than their white counterparts. Alternatively, workers in segregated workplaces might vote heavily against a union to retain the status quo. “In each of these examples, you can think of workforce composition causing — or preventing — unionization,” Ferguson says. “The problem is that such cases, where the union has very big wins or losses, are not representative of most union elections.”
Therefore, Ferguson focused on data from establishments where elections on whether to unionize were extremely close. Following these tight races, even when the union won, it had little long-run effect on the eventual composition of the workforce, he found.
“We can’t just assume that putting unions into place in a randomly chosen workplace is going to reduce the level of employment segregation in the workplace,” he says. If managers who want more diversity in their workforce are putting policies into place not so much because of unionization but because of their openness to a more integrated workforce, it would follow that those who don’t want to change the makeup of their workforce are not going to do so just because a policy says they should.
A key takeaway for managers, Ferguson says, is that there is disconnect between compliance and impact in many firms’ diversity policies. “Adopting diversity policies is treated as evidence that a firm is engaged with these issues, and can be a shield against criticism,” he says. “But we lack strong evidence that these policies are effective. We move on when, in fact, we haven’t solved a really important social problem.”
He added that while corporate leaders generally have embraced analytics, this data-driven mentality hasn’t extended to personnel issues. “I’d love to see startup firms that experiment with HR policies,” he says. “I recognize the legal concerns firms have, but we need to recognize that experiments, in and of themselves, are not discriminatory.”
For media inquiries, visit the Newsroom.