A new book by Harvard University professor Iris Bohnet, What Works: Gender Equality By Design, argues that tweaking the ways companies identify, develop and promote talent can improve equality and diversity at a “shockingly low cost and high speed.” In the following article, Knowledge at Wharton reviews key takeaways from Bohnet’s book.
When psychologist Daniel Kahneman began exploring cognitive blindspots in the 1960s, he likely had little inkling that his work would later be recognized with a Nobel Prize in Economics. His explorations of human decision-making, along with frequent collaborator Amos Tversky, laid the foundation for the field of behavioral economics. Their essential insight was that the human mind operates in two modes: a slow, deliberate, conscious one; and a fast, intuitive, unconscious one.
We couldn’t get by without the second, faster mode; and the judgments it arrives at are remarkably sound — most of the time. But its use of shortcuts — heuristics, or rules of thumb — leaves it vulnerable to significant error and bias. Behavioral economics, meanwhile, has spawned a growing inquiry into what is known as “choice architecture,” the different ways choices can be designed to impact decision-making.
Now, Iris Bohnet, a public policy professor at Harvard, has turned this lens on the question of gender equality in the workplace. She finds that often, despite the best of intentions, efforts at improving women’s status and opportunities on the job falter and even backfire when they focus on changing mindsets. Unconscious bias proves persistent and elusive.
Simply having musicians audition from behind a screen [increased] by 50% the likelihood that a woman would advance to the next round.
Her new book, What Works: Gender Equality By Design, argues that tweaking the practices and procedures by which companies identify, develop and promote talent, however, can improve equality and diversity at “shockingly low cost and high speed.” Along the way, Bohnet shares her own experiences implementing reforms at Harvard’s Kennedy School of Government, where she has served as a dean, and now as director of the Women and Public Policy Program.
First Impressions
According to one noted neuroscientist, 80% to 90% of the mind works unconsciously. Nowhere is that more evident, and insidious, than in the snap judgments we call first impressions. Within the first few seconds of meeting someone, we take in everything from their attractiveness and body language to their gender and ethnicity, and based on unconscious associations with these qualities, form an assessment of that person. Bohnet cites the work of social psychologists Mahzarin Banaji and Anthony Greenwald, who helped develop the Implicit Association Test, or IAT, to bring this shadow reasoning to light (research they summarize in their book Blindspot: Hidden Biases of Good People).
That early judgment is persistent and stubborn. Once an impression is in place, we tend to use later information to confirm and not challenge it. In simulated hiring experiments, IAT scores were predictive of both gender and racial discrimination, and the stronger that initial bias, the less likely evaluators were to modify their assessments based on subsequent input like individual performance data.
Yet small but powerful design interventions, Bohnet writes, can effectively counter unconscious bias. In 1970, for example, only 5% of the musicians in the nation’s top orchestras were women. Simply having musicians audition from behind a screen changed all that—increasing by 50% the likelihood that a woman would advance to the next round. Today, women account for over a third of orchestra membership.
Likability and Competence: The Double Bind
In a case study used in a number of business schools, students were introduced to Silicon Valley entrepreneur and venture capitalist Howard Roizen. After reviewing his track record, students rated him as highly competent and effective, and indicated they liked him and would be interested in working with him. The case study is based on the real life of former Apple executive Heidi Roizen, and when the identical history was presented to students with the first name changed from Howard to Heidi, students rated her as similarly competent and effective—but did not find her as likable, or as someone they would like to work with.
The findings of this experiment have been replicated in a number of other hiring simulations. In some cases, regardless of the gender and ethnicity of the evaluator, white males were still favored. This bias is particularly pronounced in fields strongly associated with men. In fields seen as stereotypically female, men sometimes suffer a bias in the perception of their competence –but their likability is not affected. “Women, thus, are in a double bind that men are not,” Bohnet writes. “They are perceived as either likable or competent but not both.”
Leaning In: The Cost of Speaking Up
Just as Heidi Roizen’s competence and ambition were seen by many students as a violation of gender norms, women who are assertive in asking for raises are viewed as less likable and less desirable as colleagues. A series of experiments found that both men and women were “significantly less willing to work with demanding female candidates.” Women internalize these gender norms, and are far less likely to negotiate a prospective employer’s initial offer — a dilemma explored by Linda Babock and Sara Laschever in Women Don’t Ask. That reluctance to negotiate has long-term ripple effects on future career advancement.
Simply encouraging women to be more assertive isn’t enough. In a study conducted by Victoria Brescoll of Yale University, a group of professional men and women evaluated the competence of chief executives. Male executives who spoke up the most were rewarded with higher ratings. Yet both male and female evaluators gave lower ratings to female executives who spoke up more than their peers.
Thus, Bohnet asserts, it is “not a matter of timidity, but of backlash.” Women are sometimes encouraged to make more team-oriented appeals in negotiations. Sheryl Sandberg, author of the much-discussed book Lean In, observes that replacing “I” with “we” can be effective. But even a conciliatory approach can backfire, as documented in a New Yorker article entitled “Lean Out: The Dangers for Women Who Negotiate.”
Diversity Training and Leadership Development: Do They Work?
U.S. corporations spend $8 billion annually on diversity training. Yet a meta-review of almost a thousand studies finds a “dearth of evidence” about their efficacy. As Bohnet concludes in the title to the book’s second chapter “De-biasing minds is hard,” attempting to raise awareness about the possibility of bias can be ineffective, or even counter-effective.
U.S. corporations spend $8 billion annually on diversity training. Yet a meta-review of almost a thousand studies finds a ‘dearth of evidence’ about their efficacy….
Conscious suppression of unconscious stereotypes, researchers have found, simply doesn’t work. In one experiment, students shown a diversity training video encouraging them to suppress bias toward the elderly actually evaluated older job applicants more negatively, not less. Moreover, in a phenomenon known as “moral licensing,” when people perceive themselves as engaging in “good” behavior (like participating in a diversity training session), they give themselves subsequent permission to indulge in the opposite. This is especially true among those already biased, “raising the unsettling possibility that diversity programs aimed at influencing the worst offenders might backfire.”
U.S. companies also invest $14 billion a year in leadership training, another traditional approach to broadening the talent pool. A McKinsey report, “Why Leadership Development Programs Fail,” is highly skeptical about their effectiveness and notes a lack of rigorous measurement and evaluation. Among other things, the report finds that such training programs aren’t sufficiently tied to real on-the-job projects and that they “underestimate mind-sets.”
Coupling such training with mentoring has shown promise. But, particularly in the case of programs targeted at women, mentoring is necessary but insufficient.
Catalyst, a nonprofit devoted to greater workplace inclusion, distinguishes between mentoring and sponsorship. In surveys of graduates of top MBA programs from around the world, they found that women were somewhat more likely to have mentors than men — but that their mentors were less senior, and with less organizational clout. Mentors for women tended to coach and advise, while mentors for men became active advocates for their protégés. That advocacy paid off: Two years into the survey, the men in the sample had received 15% more promotions.
Changing Environments Rather Than Mindsets
“Bias is built into our practices and procedures, not just our minds,” Bohnet writes. Because of the limited efficacy of approaches aimed at raising awareness, she encourages a greater focus on how the work environment frames our choices. Dubbed “choice architecture” by Richard Thaler and Cass Sunstein in their book Nudge, this involves simple but critical interventions like curtains in orchestra auditions.
Many of the “nudges” Bohnet argues for involve reducing ambiguity and subjectivity in favor of clarity and well-defined procedure. For example, transparency is an especially fruitful intervention, especially in the tricky matter of negotiation. In a study of MBA students seeking their first job, in fields with high ambiguity as to what could or could not be negotiated about the position, men earned significantly more than women. In fields where applicants were clearly informed about what to negotiate for, the gender gap almost vanished.
Defined rules and procedures also seem effective at weeding out bias. In a study of more than 8,000 employees in a financial-sector firm, the gender gap was much smaller in the area of base salary and much wider when it came to bonuses — one was subject to formal rules, the other was not. In general, Bohnet says, “formulaic approaches measuring individual performance to determine compensation work better for women.”
In … fields with high ambiguity as to what could or could not be negotiated about the position, men earned significantly more than women…. Where applicants were clearly informed about what to negotiate for, the gender gap almost vanished.
Similarly, structured interviews not only prove more equitable, but are better predictors of future performance. Overestimating their own discernment, and underestimating their vulnerability to bias, most managers initially express a preference for unstructured interviews. Yet as Bohnet points out, the case against such interviews is “overwhelming.” Google, a company she singles out for being at the forefront of developing progressive practices, came to this same conclusion after an exhaustive review of its approach to recruitment and hiring.
Finally, how a job is framed has an enormous effect on whether applications will skew more towards men or women. Far fewer women apply when a position’s compensation is advertised as competitive and variable, as opposed to relying on objective performance measures. Jobs with built-in flexibility are also more attractive to women, and a growing number of companies are adopting so-called “flex-time” policies.
Labor economist Claudia Goldin points to pharmacology as a case in point on both counts. It has clear performance criteria and, unlike professions like finance or law, does not place a premium on marathon workdays; in other words, part-time workers aren’t penalized. As a result, it is the third-highest-paying occupation for women, with one of smallest gender pay gaps.
Good Behavioral Design Starts With Data
“What does not get measured cannot be fixed,” Bohnet begins her chapter on the importance of data. Google, a company built on analyzing data, has led the way with its “people analytics.” As Prasad Setty, one of the men behind this initiative, says: “What we try to do is bring the same level of rigor to people decisions that we do to engineering decisions.” Google has taken a data-centered approach not just to its hiring and recruitment practices, but also to a range of issues involving equity and employee happiness.
Bohnet realized the significance of data when, as a dean at Harvard’s Kennedy School, she was confronted by a group of students concerned about the lack of role models for female students. Creating a more diverse faculty was a long and slow process, but Bohnet realized she had never paid attention to the gender breakdown of the many guests visiting the school on any given day as speakers or panelists. She began requiring event sponsors to include gender breakdowns in their annual reports, and found that “the numbers were not pretty.” But they helped inspire change and conversation and provided a baseline by which to measure future progress.
A telling example of the power of deep analytics dates back to the late 1990s, when Janice Fanning Madden of Wharton investigated two large brokerage firms where female stockbrokers earned about 60% of what their male colleagues did. Because compensation was based on commissions, at first glance it might seem that the women simply weren’t as good at selling. But Madden — who as expert witness in a class action lawsuit was given access to detailed records — found that the bottom line was that the women had been given inferior accounts, and that when they were given better accounts, the gender gap nearly disappeared.
There are now a number of analytic tools available to companies that want to gauge their gender pay equity. A leader in the field is EDGE (Economic Dividends for Gender Equality), an initiative launched at the World Economic Forum in 2011. It provides both a rigorous methodology for evaluating gender equity across five different areas, and a certification process for companies that have met a certain global standard. The World Economic Forum now issues an annual Global Gender Gap Report.
Creating Diverse Teams
A fascinating chapter deals with the challenge of creating appropriately diverse groups in the workplace — whether they be work teams, management teams or corporate boards. A growing field of inquiry finds a strong correlation between a group’s “collective intelligence” and its inclusion of women. But Bohnet finds mixed evidence that diversity necessarily leads to better performance. One study of 2,000 management teams in the mutual fund industry over seven years found that homogenous teams significantly outperformed mixed-sex teams.
There are a number of things at play here. The nature of the task at hand seems to matter. But a more fundamental issue is that of “critical mass,” first explored by Bohnet’s Harvard colleague Rosabeth Moss Kanter in a 1977 paper. At low levels of participation, women in a male-dominated group will see themselves as tokens and either overcompensate or identify with the majority. It is only when the percentage of women increases (she suggests 35%) that the benefits of diversity are captured.
Building on this research are groups like the Thirty Percent Coalition, which focuses specifically on diversity among corporate boards. A number of European companies have adopted quotas or targets to ensure boardroom diversity. Yet quotas are in some cases met with resistance.
In Boston, Bohnet’s Women and Public Policy Program has been a key partner in the city’s “100% Talent” initiative — a compact organizations and businesses can sign onto whereby they agree to try out at least three research-based interventions, and to allow independent researchers to evaluate their impact. Bohnet’s program has also set up a free online platform, the Gender Action Portal, compiling research on gender equity interventions.
Men and Counter-stereotypical Careers
Gender bias cuts both ways, but only up to a point. Men can suffer a loss of status in a job traditionally associated with women — yet without, as mentioned before, paying a penalty in likability. And even in a female-dominated industry, they can enjoy superior earnings. A 2015 study published in the Journal of the American Medical Association found that male nurses, despite making up only 7% of the profession, generally had higher salaries.
Both men and women are reluctant to contribute and speak up in areas where their gender is assumed to have less expertise. One of Bohnet’s central points is that, while data is important, diversity is more than a numbers game, and is ultimately about transforming culture. “Often, organizations are failing to hear from their best people.”
Finally, Bohnet notes how, worldwide, boys are falling behind girls in basic literacy. A 2015 study by the Organization for Economic Co-operation and Development found that, in all sixty-four countries surveyed, girls outperformed boys in reading and writing. By age 15, the gender gap was equivalent to a year of schooling. She worries that the “dearth of male teachers and male role models who read” is a major factor behind this decline. And indeed, the World Bank reports that 87% of primary education teachers in the U.S. are female, a percentage not atypical around the world.
