By Melissa J. Anderson

Despite decades of effort to recruit, retain, and promote diverse individuals in the workplace, there still seems to be an alarming lack of diversity in senior roles – and a recent study may reveal one reason why.

According to MIT’s Emilio J. Castilla it has to do with the social networks of managers. No performance review really happens in a bubble, he says, as managers talk to one another about their employees. These discussions confirm or introduce biases and notions about individuals they review, and because managers tend to interact with managers who are like themselves in demographic background, these notions often reflect demographic biases as well.

Castilla writes:

“Ultimately, these managerial mechanisms have the potential to operate at many stages of an employee’s career (not just performance appraisals, but also pay, training and career possibilities, and work assignments) and thus can help explain how stratification is sustained in the workplace, as managers’ social context affects their assessments of employees below them.”

These forces contribute to inequality in the workplace, keep wages unequal, and prevent women and minorities from reaching senior positions.

Forces at Work

In his article in the American Sociological Review, Castilla writes, “In particular, my results stress that both managerial network influence and horizontal homophily affect the process of employee performance assessments, over and above the well-studied vertical homophily mechanism.”

Vertical homophily is the tendency for managers to be more agreeable toward employees who are like themselves. Horizontal homophily is the tendency for managers to be more agreeable toward individuals on their same level within the organization – fellow managers. As far as a social network goes, Castilla suggests, the better connected a manager is, the more likely he or she is to agree with other managers on an assessment of a subordinate.

He explains:

“For these theoretical reasons—because assessments are by definition interpersonal and network processes are known to affect them—one manager’s performance evaluation of a given employee can be affected by previous evaluation decisions made by other organizational members connected to her. Performance evaluations can also be amplified toward positive or negative extremes by third-party members of a network (as shown in Burt and Knez 1995).”

Interpersonal relationships outside the manager-employee relationship influence how that employee will be evaluated, and biases relating to race and gender are part of that, Castilla says.

Compounding Effect

Castilla stresses that the three forces at work (manager networks, horizontal homophily, and vertical homophily) often work in concert, and can magnify inequality in performance reviews. He writes:

“My results stress how managers’ evaluations of their employees do not occur in a vacuum but within the social context that surrounds them. My study also underscores that these mechanisms complement each other and may operate simultaneously in many organizational settings and procedures. These managerial network influences are at work (magnified by vertical and horizontal similarity) throughout workers’ careers.”

And because performance reviews are the standard way for companies to assess the skills of their employees, his findings could account for much of the inequality in today’s corporations.  He explains:

“Because these managerial influences in the assessment process in turn affect employees’ final performance ratings, on which rewards are typically based, I show how the managerial construction of performance ratings has implications for workplace disparities in employee rewards, especially inside organizations that use performance-reward procedures to evaluate and pay their employees.”

By being aware of how inequality can be shaped at the mid-level, where diversity efforts often break down, companies can take steps to address issues which may be preventing these efforts from being successful.