What Works: Gender Equality By Design (Part 2!)

Fresh from the World Economic Forum in Davos, Professor Iris Bohnet presented the second half of her forthcoming book “What Works: Gender Equality by Design” at the first HKS WAPPP Seminar of the spring semester.  Professor Bohnet had presented the first half of her book at the first WAPPP Seminar in the fall, and the group was eager to hear more about her focus on “de-biasing organizations” rather than “de-biasing mindsets.”

Professor Bohnet is the Director of the HKS Women and Public Policy Program and Co-chair of the Behavioral Insights Group at the Center for Public Leadership at HKS. Her research demonstrates a powerful truth: No matter how well-intentioned they are, people can still fall prey to bias. Instead of trying to break people of these often unintentional biases, we should be looking to organizational design to “make it easier for all of us to do the right thing.”

The limits of diversity training

Professor Bohnet began with a survey of existing interventions to overcome bias in the workplace, including diversity training, negotiation training, leadership training, mentorship and sponsorship, and networking. Despite being a billion-dollar industry, the evidence on diversity training is mixed. After hundreds of studies, it is hard to be optimistic about diversity training, if only because it is incredibly difficult to de-bias minds. Beyond stereotypical thinking, humans rely on a number of other cognitive shortcuts that are very difficult to unlearn.

The rest of these interventions are geared toward helping women navigate the workforce more effectively. One critique of this approach is that it places the onus of fixing gender discrimination on women. However, it is important to be pragmatic and to consider how these methods can be helpful. The evidence from mentorship and sponsorship networks is particularly encouraging. Female economists who participated in mentorship training workshops were more productive, more likely to publish in peer-reviewed journals, and more likely to get tenure. This is some of the best causal evidence to date about the impact of mentorship and sponsorship and demonstrates the positive impact of these interventions.

Using behavioral insights to move the needle

Behavioral insights and organizational design present an opportunity to develop novel interventions to decrease bias and improve diversity. For organizations looking to attract the best candidates, Professor Bohnet suggests they look at the language in their job advertisements. She presented a job advertisement for a teacher reading in part, “Looking for a warm and caring teacher with exceptional pedagogical and interpersonal skills to work in a supportive, collaborative work environment.” Gendered language like this resonates with women rather than men and can inhibit great male teachers from applying. And with boys falling about a year behind girls in reading and writing by age 15, partially because of the lack of male role models among their teachers, it makes a difference.

There are numerous behavioral interventions that workplaces can develop to overcome bias. Two simple things organizations can start with: get rid of “potential” rankings in employee evaluation, and stop managers from seeing employees’ self-evaluations before giving their own rankings. There is an enormous amount of bias in potential rankings. In another iteration of “seeing is believing,” a lack of female senior partners at a firm may lead to biased thinking that women don’t have the potential or desire to ascend the career ladder. While it’s not clear how to get rid of this bias, it is simple enough to remove potential rankings altogether in evaluating employees. Similarly, there can be a significant difference in employee rankings. On average, women are inclined to give themselves lower self-evaluation scores than men. If managers can see employees’ self-evaluations before making their own judgments, they may be influenced by these scores and give women lower scores (and men higher) than they otherwise would.  Instead, the manager could have conversations with team members about their performance, but wait to see a numerical self-evaluation until after they have made their own evaluations. These two findings are low-hanging fruit for organizations looking to reduce workplace bias; research increasingly shows that mindsets change once behavior has changed, and these low-impact methods reduce the effect of bias in the workplace.

Gender diversity in teams

In recent years, the “business case for diversity”—evidence that more diverse organizations do better than those that are more homogeneous—has attracted a lot of attention. A meta-analysis of 120 studies finds a small diversity premium for diverse corporate boards. However, these studies are just observational: it could be that diversity really does pay, or that companies that are high-performing are also more inclusive. One study measuring collective intelligence found that gender diverse teams outperformed homogeneous teams on each of the tested tasks. Gender diverse teams demonstrated complementary skill sets and a greater tendency to listen to and build on others’ ideas. Gender diverse teams are much more likely to avoid groupthink, which increases their effectiveness.

However, Professor Bohnet emphasized, we should care about gender equality because it’s the right thing to do, not just because of the business case for diversity. In closing, she discussed several organizational improvements that could encourage equitable behavior, including a “comply or explain” system for corporate diversity. Goal setting, transparency, and accountability all play a crucial role in changing behavior and, by extension, changing mindsets. Professor Bohnet’s book is available on March 4—be on the lookout for more bias-reducing, diversity-maximizing insights!

Gender and Groupthink: Why Don't Smart People Share Their Ideas?

As the cliche goes, we are social animals. We eat in groups; we learn in groups; some of us hunt in groups. And very often, we make lots of big decisions in groups---from corporate boards to faculty committees and elected bodies.

But in those groups, do the best ideas always get adopted? Do they even get contributed to the mix?

What's particularly tough is when good ideas from half of the people in all groups are immediately ruled out. As research by Katherine Phillips and Melissa Thomas-Hunt explored back in 2004, groups across cultures have difficulty recognizing, let alone encouraging female expertise. This is often due to their group dynamics---whether the effects of social networks, group think, or even plain sexism.

But in last week’s WAPPP seminar on “Gender and Group Decisions: Eliciting and Acting Upon Expertise,” Katie Baldiga Coffman, Assistant Professor at Ohio State University, showed how, even in the absence of explicit group dynamics, women are less likely to put their own--often better--ideas forward.

In a controlled experiment at Ohio State, women and men offered their answers in a trivia game--first as individuals, later in blind groups, and finally in groups in which they saw their partners’ faces.

The results? Men contributed their answers more often than women, across categories; women under-contributed even in the subjects in which they were more comfortable about their expertise; and interestingly, even being provided feedback about their strengths did not improve efficiencies in information sharing.

So, even without competition, and with encouragement, talented women were less likely to contribute their better ideas to a group.

How can we fix this to improve efficiencies and make sure that (i) women are more comfortable with contributing their ideas, and that (ii) smart people contribute their good ideas more often? Can we design mechanisms or shape environments that “nudge” people to share and decide according to merit?

Taking a tip from the nudging process from last week's seminar, what would happen if we change the role models that women see--rather than just provide personal feedback--so people get more used to seeing women being successful, and start to associate women with success too? When people start getting used to seeing things a certain way, will they start getting used to doing them--including sharing and adopting the best ideas?

What are some other ways of nudging a system towards the best answers---towards social equity and social efficiency?



Photo Source

Social Norms and Stereotypes: What Happens When Everyone's a Little Bit Sexist?

Upon seeing one of those “reuse your towel” cards in a hotel, have you ever wondered how many people were actually reusing their towels? An experimental study found that when the message was supplemented with information that “nearly 75 percent of hotel guests reuse their towels,” the guests receiving the message became more likely to reuse their towel than the guests who got the standard note. This is an example of the power of social norms – when the desired action is perceived as the norm, compliance increases. The same mechanism is at work in reverse for an undesirable behavior – people are less likely to litter when they perceive that only a few “anti-social types” litter.

Social norms are often more powerful than explicit messages. The signs in the Petrified Forest National Park used to read something like this: “Don’t take petrified wood, 14 tons a year are being stolen in small pieces.” That last part essentially signaled to the visitor: “everybody is doing it.” When that part was removed, compliance improved.

Now, let’s connect this information to racial and gender stereotyping and bias – generally, people view bias as undesirable and actively work against it. However, the 2000’s saw a growing body of research on unconscious biases. The notion that most people have unconscious stereotypes and biases captivated the popular press. The cover of the Washington Post Magazine on January 23, 2005 read: “Many Americans believe they are not prejudiced. Now a new test provides powerful evidence that a majority of us really are.” So, it’s like the song from Broadway musical Avenue Q, “Everyone's a Little Bit Racist.”
 
But what does that mean for social norms? Does knowing that everyone has unconscious biases make it seem permissible to let our own biases slide?

Those are the questions Melissa Thomas-Hunt, Associate Professor of Business Administration at the University of Virginia Darden School of Business, set out to answer with her colleague Michelle Duguid, Assistant Professor of Organizational Behavior at the Washington University in St. Louis Olin Business School. Thomas-Hunt and Duguid designed a series of creative experiments to test how messages of high and low prevalence of stereotyping affect behavior.
 
In one of the experiments, pairs of women and men were negotiating a car sale. All of the men were told to avoid thinking of others in a stereotypical manner. Some of the men were primed with a “high prevalence of stereotyping” message – they were not told what gender biases might be, just that an influential body of research finds that the majority of people do stereotype. Other men were primed with a “low prevalence” message – that very few people hold stereotypical misconceptions. Yet another set of men received a “high prevalence of counter stereotyping” message – that most people actively try to overcome biased misconceptions.

After the negotiation, the women, who did not know anything about these messages, were asked to rank their opponents' assertiveness. The researchers also compared negotiated price outcomes. The men who were told that most people stereotype, despite being cautioned not to do so themselves, consistently claimed more value in the negotiation and were ranked as more assertive by their female counterparts than were the men in both other groups. The division of value was more equitable in the pairs where men received “few people stereotype” and “most people work to avoid stereotyping.” The men were also perceived to be less assertive by their female counterparts. In other words, the men who perceived the social norm to be in line with stereotypes of women as weak negotiators were more likely to treat them as such, while the men who thought the social norm was to counter stereotypes treated their negotiating partners more equitably.

Thomas-Hunt’s and Duguid’s experiments point to the conclusion that it’s not enough to be aware that we have conscious and unconscious biases. We need social norms to counter these biases. We need a culture in which most people work hard not to stereotype. In companies, this could be a consistent internal message from the highest ranks, but in society it will take all of us holding each other accountable and working not to be even a little bit sexist.

Melissa Thomas-Hunt presented at a WAPPP Seminar titled, "Condoning Stereotyping: How Awareness of Stereotyping Prevalence Impacts Stereotype Expression in Negotiations and Beyond," on April 4, 2013.


Anya Malkov is an MPP candidate at the Harvard Kennedy School, a WAPPP Cultural Bridge Fellow, and an alumna of From Harvard Square to the Oval Office.

The Stories We Tell Ourselves

Photo courtesy of Forbes.com

“Most of us think that we’re better drivers than the average person, and we really think we’re more honest and morally upstanding than the average person,” says Michael Norton, Associate Professor of Business Administration at the Harvard Business School. He has been studying how people can lie and yet think that they are perfectly great, honest individuals. You might think he is interested in politicians or car dealers, but actually this is about regular people like you and me. We are quite prone to engaging in questionable behavior, yet we are very good at finding excuses for problematic decisions and even tricking ourselves into believing that we did not cheat.

He described a simple experiment – college men were asked to pick between subscriptions to two sports magazines. The magazine descriptions were identical except for the fact that one offered more articles per issue and the other covered a broader array of sports. The preferences split half and half. However, when one of the magazines also offered a swimsuit edition, that magazine became the overwhelming choice. They were then asked to explain why they picked the magazine. The men in the group whose “more articles” magazine offered the swimsuit edition explained that they valued having more articles, while the men whose swimsuit issue was packaged with the “more sports” magazine were equally emphatic that they liked reading about a wider variety of sports.

Perhaps the guys just did not want the researchers to know their true motivation, but what if they were not even aware of their choice process? And what if their decision was more consequential? In a similar experiment, hypothetical managers of a cement manufacturing company (a stereotypically male field) were asked to select between two applicants – one had more education, the other more experience. Those ‘employers’ whose more experienced candidate was named Lisa, overwhelmingly chose the other candidate named Dan, citing the importance of education credentials. But those whose Lisa was more educated, hired Dan because he had more experience. When told in advance that they would be accountable for their decision, the subjects simply became more vociferous in explaining the importance of whichever non-questionable characteristic they picked, sometimes writing whole essays about their choice and how they never took gender into account.

These experiments have implications for hiring, promotions, political recruiting and college admissions, among other fields.  Are we really guided by objective criteria when we read an application, interview someone or consider them for promotion? All signs point to people making the biased choice and then justifying it with the most conveniently available “objective” information. Perhaps even more troubling is the notion that you and I might be completely unaware of our own questionable behaviors, whether they are gender-biased, race-biased or simply dishonest. After all, we are more morally upstanding and more enlightened than the average person, right?  

Anya Malkov is an MPP candidate at the Harvard Kennedy School, a WAPPP Cultural Bridge Fellow, and an alumna of From Harvard Square to the Oval Office.

Race in Your Face

This will be uncomfortable, so try to relax. Examining our own race and gender stereotypes can be as pleasant as sitting in the dentist’s chair. But for Kerri L. Johnson, Assistant Professor of Communication Studies at UCLA, discovering awkward truths about people’s biases is an occupational hazard. Dr. Johnson investigates how people use physical cues to categorize other people.  Sometimes she even gets hate mail and coverage by Rush Limbaugh.

At the Social Communication Lab at UCLA, Dr. Johnson and her social psychologist colleagues do things like asking students to decide whether a point-light display figure is a man or woman and to categorize faces of different races as male, female, gay or straight. They even have a program that analyzes the relative femininity or masculinity of a person’s face.  Despite the ill-informed hate mail, the researchers persist, because understanding how biases are formed is the first step to combating unfair stereotypes and discrimination. 

In one study, Dr. Johnson demonstrated that race is gendered. For example, in the United States, there is overlap in stereotypes of Asians and women (soft-spoken, shy, family-oriented), while the stereotypes about Blacks overlap with stereotypes about men (dominant, athletic, competitive). The same pattern surfaced when the researchers measured whether respondents were more likely to misidentify the Asian men’s faces as female and Black women’s faces as male. The faces in question were digitally created so that only the race varied (see above). Sure enough, subjects more quickly and more correctly identified the Black men and Asian women than the other way around, and the respondents’ own race or gender did not seem to make a difference.   

The Social Communication Lab runs dozens of similar experiments, finding, for example, that even when information was limited to just photos of faces, subjects were better than chance at categorizing the person as “gay” or “straight.” And, as another angry-letter-provoking study demonstrated, categorizing someone, whether by race, gender or sexual orientation, has implications for how we will interact with that person given our own baggage and biases.

Yet it was a study of politicians and gender that achieved notoriety and misrepresentation by both Samantha Bee at the New York Times and Rush Limbaugh. The conclusions of the study were that Republican women in Congress had more stereotypical feminine facial features than did their Democratic counterparts; and that uninformed observers were more likely to identify the highly feminine women’s faces as Republican and the less feminine ones as Democratic.

You can take a moment to learn about the methods used in this unexpectedly controversial study, and even start another uncomfortable discussion in the comments section here. Please, just go easy on the hate mail.

Anya Malkov is an MPP candidate at the Harvard Kennedy School, a WAPPP Cultural Bridge Fellow, and an alumna of From Harvard Square to the Oval Office.

Professors Are Biased Too

An ordinary e-mail once caught the eye of Modupe Akinola,Assistant Professor of Management at Columbia Business School. It was from a prospective Ph.D. student, an African-American woman, asking for a meeting.

Dr. Akinola wondered: “Would I have reacted differently if this was from a man? Would it make a difference if the request was for today and not next week?” She joined with colleagues Katherine Milkman and DollyChugh to turn these questions into research questions in a field experiment measuring discrimination in academia.

The creative and elaborate experiment started with putting together a representative sample of over 6,500 faculty members from 260 universities. The team used publicly available demographic information on professors to ensure a good sample. The researchers then created a few prospective Ph.D. students. The names of these imaginary students were tested and proved to be easily recognizable as either male or female and as Caucasian, Black, Hispanic, Indian or Chinese.

At 8:00 am on a Monday, e-mails from Brad Andersons, Latoya Browns and 18 others went out to unsuspecting subjects, asking to meet “today” in half the cases and “next Monday” in the other half. The research team was impressed by how quickly most faculty wrote back and how often they agreed to meet, but when and with whom they agreed to meet was more interesting.

One would think academia, with its affirmative action and enlightened world views, might be a post-racial, post-gender sphere. As it turns out – not so much, especially not when there is time to think about it.

In the now condition, Akinola, Milkman and Chugh found no significant difference between number of responses to white males and to others, but when the request was for later, white men received both more responses and more meeting acceptances than any other prospective students. The white men fared better than all other racial categories across almost all disciplines. The effect remained even when the race or gender of faculty and student matched.

According to construal level theory (CLT) immediate events demand concrete reasoning focused on feasibility, whereas decisions about distant events trigger abstract thinking and a focus on desirability of the event. Thus, future events create more room for unconscious stereotypes and biases to enter the decision-making process. In this study, for instance, the responses received to the “now” requests systematically displayed a focus on “how,” whereas the “later” e-mails were more likely to be answered with a request for credentials or more information – a “why” response.

The point of this experiment and other studies that spotlight the influence of stereotypes and biases on consequential decisions is not to call people racist or sexist. The point is to come up with ways to address the biases and to improve minority outcomes. For instance, referring all prospective students to the Ph.D. program coordinator would take the burden off individual professors and ensure that all students get their questions answered. This is just one option that an experiment like this might bring up for discussion.

Unfortunately, some faculty did not take kindly to the experiment. When debriefed and told that they had been used in a study of bias, many expressed outrage and demanded to speak with the Institutional Review Board.

To me, their reaction suggests the thorny subjects of race and gender are not studied enough in academia. Perhaps academia, more so than other spheres, makes it taboo to have biases, but shaming does not make biases go away! Nobody wants to be called a racist or sexist, not even a little bit, but simply feeling guilty is not productive. We can structure decision-making in ways that level the playing field, and we need creative researchers like Akinola, Milkman and Chugh to make that happen.

Anya Malkov is an MPP2 at the Harvard Kennedy School, a WAPPP Cultural Bridge Fellow, and an alumna of From Harvard Square to the Oval Office.

The Gender Pay Gap among Stockbrokers

Wharton professor Janice Fanning Madden proves once again that solid social science research can settle lawsuits and chip away at discrimination. Asked to testify in class-action law-suits of female stockbrokers against two large firms, she used company data to tease out the evidence of gender bias and demonstrate that the firms’ arguments had no basis in fact.

Stockbrokers are the highest paid sales occupation, yet it has the largest gender pay gap. Female stockbrokers earn 54-60% of their male counterparts. Since stockbroker compensation is based almost entirely on commission, the obvious explanation for the pay gap is that women simply generate fewer sales.

The trading firms were convinced that this was a matter of sales capacity – low-performing female stockbrokers stayed while low-performing males left, and women worked less intensely due to household responsibilities and other factors. Professor Madden proved their arguments wrong. Through statistical analysis of account records tied to each stockbroker, she demonstrated that when men and women received equivalent clients and accounts, they generated equivalent sales. Yet she found that on the whole women were being given inferior accounts, which led to lower commissions.

The point is not to accuse male managers of blatant sexism. In most cases, their biases were likely unconscious. Decision-making research demonstrates that people are bad at predicting performance, and that in instances where the workplace is overwhelmingly male, the manager is more likely to have a subjective “hunch” that a man might do better than a woman. So if women start off with inferior accounts, they will generate lower sales and become less and less likely to receive a lucrative account.  

The firms protested this explanation. They claimed that the better accounts went to more experienced brokers, who happened to be male, a disparity that would go away as the female workforce matured. That is where Dr. Fanning Madden really surprised them. The company records demonstrated that when a broker left, the managers redistributed more accounts to the newer, “hungrier” brokers, not those with more experience and an already heavy client base.

Thus both the “experience” and the “sales capacity” arguments of the financial giants fell apart, the lawsuits were settled and the women received substantial damage pay. Yet female stockbrokers are still earning less than men, unconscious biases are still shaping practice, and there is not enough pressure for systemic change. 

The challenge as I see it is to take powerful research about the causes of the pay gap beyond academia and federal court, and put it into the hands of managers, operations professionals and HR staff the world over.    

When Affirmative Action Works. When it doesn’t.

The Woman: Johanna Mollerstrom, Doctoral Candidate, Department of Economics

The Talk: The Downside of Affirmative Action

The Question: Do quotas negatively impact group dynamics?

Discussions around quotas and affirmative action policies can polarize conversations. At their core, we find ourselves asking, “Is this a ‘fair’ way to increase representation?”

The Research from Mollerstrom finds that when groups are created via a quota system, the amount of cooperation within the group decreases. While I won’t go into the details of the study (and more research is still in progress), she essentially hypothesizes three reasons why this may happen:

1. Mood: If a person believes it is unfair for someone to be in the group, then he or she is less likely to cooperate.
2. Entitlement: If a person is in the group for reasons besides credentials or merit, then resentment may occur.
3. Punishment: If a person follows the a type of in-group-out-group bias, then he or she may favor his or her respective in-group.

While few in the United States promote the idea of a quota – the argument being that it tokenizes diversity and discredits qualifications – the current numbers are pretty stark. Only 17 percent of the US Congress is female; women constitute only 3.6 percent of Fortune 500 CEOs; and women hold only 16.1 percent of Fortune 500 board seats.

But globally these numbers are not much better. On average, women hold 19.5 percent of parliamentary seats and represent only 9.8 percent of board members (with this number swinging from less than 1 percent in Japan to 35.6 percent in Norway). Quotas are often discussed as an option to increase representation and parity. In 2006 Norway imposed a quota system that requires companies to have 40 percent of their board to be women. This past January, France joined them.

Is this about diversity, or how this diversity came to be?

While many are doing research in this field, the discussion around why we believe certain groups deserve representation is culturally and socially fascinating. For example, in the US, we consider the athletic quotas that stem from Title IX to be completely acceptable. Why is this? Why are we okay with some quotas, but not others?

Mollerstrom is doing more research around this question. Perhaps it’s rooted in historical legacies. Perhaps it depends on whether the environment is competitive or cooperative. Or perhaps it’s framing and relevance. Regardless, I will be interested to see what she finds next.

Melissa Sandgren is a MPP1 at the Harvard Kennedy School and a participant in WAPPP's From Harvard Square to the Oval Office program. She is also the author of the "For Struggling Boards, the Answer May Be Closer than You Think" in the 2012 Kennedy School Review.