What are the ways in which we can combat these biases in performance appraisals and narrow the promotion gap? One study we have underway at Harvard looks at whether potential bias will go away when we use data analytics to offer more concrete and more objectively measurable criteria for such traits as analytical skills, emotional intelligence, people skills or client interaction. Generally, the arsenal of evidence-based insights that help address flaws in the promotion process is steadily increasing. Still, fixing the process alone won’t be enough.
We will also need to tackle something called the “thin file,” a term I came across only recently. When explaining to me why a person was not promoted to partner, promotion committee members repeatedly said that a candidate simply didn’t have what it takes, based on a file summarizing his or her work over the past eight years.
The candidate had not been on enough, if any, important deals, and, making matters worse, had received little feedback over the years. Associates with these “thin files” tended to be minorities and women.
Although the promotion process might have some flaws, the flaws of the system had affected these candidates from the time they joined the firm as first-year associates. They were victims of what has become known as performance-support bias, in which some employees receive less support from the start.
Janice Fanning Madden, of the University of Pennsylvania, examined why female stockbrokers in two of the largest brokerage firms in the United States made only about 60 percent of what their male colleagues were paid. A closer look at the data revealed that this was not because female brokers performed worse than their male counterparts but because they were given worse-performing accounts to start with.
At the professional services firms, the partners also tended to look for people like themselves who, ideally, had graduated from the same university. Since the allocation of work in the first year often happens informally, these processes took place below anyone’s radar screen.
Our research shows that when people make such choices more deliberately — by explicitly comparing candidates with one another instead of focusing on just one person, for example — they are more likely to choose based on performance rather than on stereotypical associations. What’s more, variety is more likely to emerge when we make bundle decisions, that is, assign work to or promote more than one person at a time.
The associates with thin files were not only given fewer opportunities to prove themselves but had also received less feedback to help them improve. Typically, giving feedback is discretionary, with little structure provided to the supervisor. And it is hard: We do not like receiving critical feedback, nor do we enjoy giving it.
Research suggests double standards where, for example, the time taken to make a decision is interpreted as a sign of indecisiveness for a female associate but as thoughtfulness in the case of her male counterpart. Given that we hold pre-existing beliefs about the behaviors of people with certain characteristics, we tend to look for evidence that confirms these stereotypes.
Such confirmation bias is well documented and difficult to overcome; we often are unaware of our views or how we bring them to bear in certain situations. In addition, women and minorities have been found to get less and more vague feedback that is less helpful to development and growth. That may be because people are uncomfortable judging or criticizing members of traditionally disadvantaged groups.
My diagnosis of the thin-file problem is based on two observations. First, an informal, decentralized work assignment process where partners and senior associates invited first-year associates to join their teams relatively idiosyncratically could have contributed to the beginning of a vicious cycle in which some associates were able to build a thick file over time and others were left behind.
Second, unstructured, rare feedback based on subjective criteria not calibrated across all employees likely helped the “in-group” or “ideal worker” develop — and grow and accumulate a file filled with feedback, comments and reviews — while the “out-group” file remained half empty.
What are the remedies? More structure, both for the work assignment and the feedback process. Why not have a central work assignment process that monitors how work is allocated and tracks the opportunities given to the various associates over time to make sure equal opportunity indeed is a lived experience?
And why not move to a feedback system where people receive comments regularly, almost instantly, possibly on a weekly basis? And what if feedback were received not only from the supervising partner or associate but also from colleagues and clients?
What if a de-biasing algorithm were used making sure the language in the feedback given was not inadvertently favoring one group over another?
Field experiments by Paola Cecchi-Dimeglio of Harvard Law School suggest that such a process can level the playing field better than traditional, yearly performance appraisals. It makes feedback less prone to distortions in our memory of how people have performed. It also benefits from the wisdom of the crowd and helps overcome the performance-support bias by standardizing and calibrating responses across employees.
Thin files are not hard-wired into the DNA of people — nor of organizations. Indeed, it is the organization that can change and redesign how the files come about, thus giving everyone an equal shot at having a thick file when they are evaluated for promotion.
Continue reading the main story