Can Job Referrals Improve Employee Diversity?

Image

In new research, scholars examine how valuable referrals are to Black and female employees. | Reuters/Gary Cameron

Sending a job application blind — by email or through some nameless online portal — feels a bit like stuffing your résumé in a bottle and dropping it into the nearest body of water. The crush of candidates makes it unlikely that you’ll even be considered.

That helps to explain why up to 50% of all jobs in the United States are found through social contacts. Referrals give applicants a way to skip the line. They also, theoretically, create a more efficient application process by presenting employers with informally vetted candidates and employees with a workplace that will likely be a good fit.

The downside is that “birds of a feather flock together, and network-based hiring could exclude people who are not as well connected,” says Adina Sterling, an assistant professor of organizational behavior at the Stanford Graduate School of Business. Minorities and women — precisely the kind of people who, historically, are not as well connected —may be at a disadvantage when it comes to finding a job through social networks. “But is that the full story?” she asks. “If network-based hiring is a negative for certain groups at the point of entry, does that effect continue through their tenure within the organization?”

A recent article by Sterling, published in ILR Review, uncovers a surprising result: Referrals can actually boost the rate at which Black employees are promoted and even lift the promotion rates among Black employees high enough to match the rate among their white peers who were not referred under promotions of certain types; however, the effect does not hold for women.

In partnership with Tulane’s Jennifer Merluzzi, Sterling investigated 11 years of personnel records on nearly 16,000 employees hired at a U.S.-based sales organization. They tracked promotions of Black people and women, two groups that are typically promoted at lower rates than whites and males, and differentiated between those who were hired from a referral and those who were not. They found that Black employees hired through a referral were 1.2 times more likely to be promoted than those who were not. However, this did not bring them to parity with white employees. In some rare cases, employees who demonstrated extraordinary ability were given so-called “outstanding promotions” — they were promoted earlier than their colleagues. In these cases, Sterling and Merluzzi found that Black employees with referrals were promoted at the same rate as white employees without referrals.

Women, meanwhile, were largely unaffected by referrals when it came to promotion. “That really surprised me,” says Sterling. To her, this finding helps further add to the scholarly debate about whether race and gender should be studied as a single factor related to minority status. “These results clearly speak to the idea that gender and race are different,” she says. “They suggest that there are distinctive potential biases that occur when women are hired and promoted as opposed to racial minorities.”

To understand why referrals increased promotion among Black employees, Sterling and Merluzzi ran a small vignette in which MBA and executive MBA students, told to role-play as managers, were split into two groups, and each group received a hiring report and a résumé that was identical except for the race of the employee (one white, one Black). Within these two groups, some were told the employee under review was a referred hire while others were told he was brought in through a campus recruiter. They were given a glimpse of his performance — and told that other employees with the same performance were dismissed half the time and promoted half the time — and then they were asked whether to fire or promote him.

Sterling and Merluzzi found evidence that Black employees who were referred were the most likely to be promoted, and they were statistically more likely to be promoted than whites who were not referred. When the MBA and executive MBA students explained their rationale, they were more apt to use hard data to assess employee quality when evaluating Black employees compared with white employees but were less likely to do so for employees who were hired by a referral rather than by formal methods. In other words, having someone to vouch for them and offer greater details “behind the quantitative numbers” was important for Black employees relative to white employees.

For Sterling, this speaks to the possibility that Black candidates face pervasive bias in the hiring process — that, from the outset, they are more heavily screened based on measurable criteria compared with other candidates. Referrals may help dismantle this tendency.

Sterling says women, meanwhile, may face more time-specific discrimination, related to starting a family, for instance. If this is the case, referrals or those with equivalent influence to vouch for individuals — offered at specific times when work-life balance becomes salient — may be more beneficial.

The nuance of these findings, Sterling says, speak to the simple fact that “there is not one silver bullet here that you can walk away with. Hiring and retention with respect to employee diversity are complex problems.” But she offers employers a place to start: At the minimum, track data on new hires, she says. Human resource departments sit atop massive amounts of useful information. Take a look at what each new hire pool looks like when brought in through different methods. If some pools are more diverse than others, or within those pools groups respond or perform differently post-hire, then track that.

“To the degree companies are serious about trying to diversify the labor pool in ways that allow people to be successful — this is not just about hiring, but about retention and getting the most out of employees that you can — HR managers should track these things, look at them, and then be open to surprising aspects of how referrals can help, especially underrepresented groups,” Sterling says.