Simple Ways to Take Gender Bias Out of Your Job Ads

New Book: Iris Bohnet's new book, What Works: Gender Equality by Design, discusses how organizations can leverage findings from behavioral science research to fight gender bias in the workplace—starting with job listings.
  • Author Interview

In the Quest for Gender Equality, Job Ads Are Low-hanging Fruit

Interview by Carmen Nobel

Here’s a hint for employers wondering why mostly men (or mostly women) are applying for your company’s open positions. Look at the language in your job listings. Chances are, the wording is more biased toward one gender than you realize.

While the Civil Rights Act of 1964 prohibits employers from overtly soliciting a preferred gender in their job listings, research shows that the language of job descriptions often subtly adheres to gender stereotypes. And that deters members of the opposite gender from applying to those jobs.

For example, a few years ago, social scientists at the University of Waterloo and Duke University coded a long list of adjectives and verbs as masculine or feminine then scanned a popular job site to look for those words. They found that job ads in male-dominated fields (like software programming) tended to use masculine-coded words such as “competitive” and “dominate” much more than job ads in female-dominated fields. Follow-up research confirmed such words made those job listings less appealing to women.

Yet so-called “gendered language” continues to run rampant in online employment listings. Consider the word “ninja,” which increasingly appears in job descriptions in high tech. Among the listings on the employment-related search engine, usage of “ninja” increased nearly 400 percent between January 2012 and October 2016, according to the company’s Job Trends database tool. While the word may make the job sound exciting, it may also dissuade women from applying, as society tends to regard “ninja” as masculine. The word “dominant” rose by 65 percent in the same time period.

It’s unlikely that the world will stop associating certain words with certain genders any time soon. Fortunately for employers looking to narrow the applicant-pool gender gap, there is a simple way to take the gender bias out of job listings: Simply rewrite them.

“Our minds are stubborn beasts that are hard to change, but it’s not hard to de-bias the application process,” says behavioral economist Iris Bohnet, a visiting professor at Harvard Business School, co-chair of Harvard’s Behavioral Insights Group, and director of the Women and Public Policy Program at Harvard Kennedy School. She is the author of the book What Works: Gender Equality by Design, which discusses how organizations can leverage findings of behavioral science research to fight gender bias in the workplace.

“The idea of the book and of my research is to say that it’s easier to de-bias organizations’ practices and procedures than to de-bias mindsets,” she says. “To start with, job ads are super-low-hanging fruit.”

Purging the gendered language

There are two easy key ways to take the gender bias out of job ads, Bohnet says: One, purge the gendered language. Two, limit the number of mandatory qualifications to apply for the job.

In What Works, she cites the example of an elementary school advertising for “a committed teacher with exceptional pedagogical and interpersonal skills to work in a supportive, collaborative work environment.” The potential problem is that “supportive,” “collaborative” and even “committed” are widely associated with femininity, which may detract men from applying.

“Maybe elementary schools want to add still more women to their roughly 80 to 90 percent female faculty,” she writes. “But, I doubt it. Most schools want to benefit from 100 percent of the talent pool and not deter skilled male applicants simply because the gendered adjectives in their advertisements signal to men that they do not belong.”

The easy fix: Nix “supportive” and “collaborative” from the job description. “With a few simple word-choice changes—‘they look for an excellent teacher with exceptional pedagogical skills’—you have expanded the potential talent pool,” she writes.

Bohnet does not mean to imply that men lack the ability to be supportive or collaborative, a point she stresses when discussing the book during a recent interview in her HBS office. “Of course men can be supportive and collaborative, caring and warm,” she says. “But, based on data analytics on the kinds of jobs men and women apply for, research shows that the adjectives matter.”

Limiting the number of qualifications in a job description is another important way to mitigate job-listing gender bias. Bohnet recommends listing only the skills that are absolutely necessary for the role. Often, job descriptions are designed by a committee of opinionated individuals, resulting in a long laundry list of qualifications, some of which are vital, but many of which are just nice-to-haves. Here’s the problem with that list of nice-to-haves: “Many women won’t apply for a job unless they meet almost all of the listed requirements,” Bohnet says. “Men tend to have a lower threshold for applying.”

Lessons from the classroom and tools for the real world

During the past semester, Bohnet taught a graduate-level course called “Behavioral Economics for Organizations,” jointly listed at HBS and the Harvard Kennedy School. Like her book, the course uses social science research and tools, in this case to promote overall organizational health. In one of the lessons, students were tasked with using behavioral design to de-bias talent management; in another, to promote ethics and compliance in organizations.

Guest speakers in her class included Aniela Unguresan, co-founder of the EDGE Certified Foundation, which developed an assessment methodology and business certification standard for gender equality, and Anka Wittenberg, chief diversity officer of technology giant SAP. In addition to becoming one of the first companies to receive EDGE certification, SAP recently announced plans to add de-biasing capabilities to its Success Factors line of human resources software, including the ability to flag job descriptions for potentially biased language.

The course’s guest speakers also included the CEOs of startups Applied Ltd. and Unitive, both of which use behavioral science research to help organizations take the gender bias out of all steps of the employee recruitment process—including writing a job description, evaluating applications, and interviewing applicants.

Bohnet notes that there are also free online tools that automatically scan job descriptions for biased language, such as Gender Decoder for Job Ads. Simply paste the text of a job listing into the decoder, and it scans the text for the list of gender-coded words from the Duke/Waterloo study. In less than a second, the decoder reports whether there are more masculine-coded or feminine-coded words in the ad. It’s not all encompassing—it doesn’t include “ninja,” for example. But, it’s a good start, it’s free, and easy to use.

“There’s a cool story going on,” Bohnet says. “It says, here’s the research, and here are the tools you can use to apply it at work.”

Note to readers: Bohnet and colleagues are currently conducting experimental field research in which they post and track different versions of the same job listing, adjusting the language to gauge how and whether it affects who applies for the job. If your company might be a good fit for this line of research, please reach out to Bohnet at or

  • Book Excerpt

Attracting the right people

from: What Works: Gender Equality by Design
by Iris Bohnet

Attracting the right people instead of managing the wrong ones is one of the most important tasks any organization confronts. This is the mantra Google lives by—or, as Laszlo Bock writes: “Only hire people who are better than you.” In an interview on the company’s hiring and corporate culture, Eric Schmidt, the executive chairman, explained that in addition to judging the technical qualifications of potential hires, a key focus at Google was to determine whether they were passionate and committed to innovation.

Surely, allowing all Google engineers to spend 20 percent of their time developing their own ideas serves as a sorting device. It attracts creative, independent minds who invent Google News, Orkut, or a social networking site. The time is not written in stone nor necessarily utilized, but it matters as an idea: “No one gets a ‘20 percent time’ packet at orientation, or is pushed into distracting themselves with a side project. Twenty percent time has always operated on a somewhat ad hoc basis, providing an outlet for the company’s brightest, most restless, and most persistent employees—for people determined to see an idea through to completion, come hell or high water.”

Not many of those “seeing an idea through to completion, come hell or high water” are women. In the spring of 2015, a gender discrimination trial brought by a former junior partner at a venture capital firm in Silicon Valley drew renewed attention to the low fraction of women in technology. While in the end a jury found against the plaintiff, the low numbers were undeniable: fewer than 20 percent in most tech companies and even fewer in Silicon Valley’s venture capital firms.

Some argue that the “tech bros” mentality of Silicon Valley keeps women out and even discourages female students from focusing on computer science. Perhaps. Surely, the male-dominated environment does not help tech firms attract women. As we know, deviating from behavior that is expected of a social category, either by others or by oneself, can be costly. A woman who acts against the norms by definition doesn’t “belong”; not surprisingly, the fear of not belonging is influential.

Indeed, research by Boris Groysberg, Ashish Nanda, and Nitin Nohria (now dean of Harvard Business School) suggests establishing belonging turns out to be a major concern of female job seekers. They report that women consider more factors than men when screening jobs; in particular, cultural fit, values, and managerial style. There is a surprising silver lining to this research, however: it carries hidden benefits for women and their employers. In follow-up work, Groysberg identifies this scrutiny as one of the key variables explaining why women transition more successfully to new companies than men. Women know better what they are getting themselves into.

The researchers analyzed the performance of more than a thousand “star” analysts working for almost eighty different in- vestment banks over a nine-year period. Analysts were labeled “stars” if they were ranked as one of the best in the industry by Institutional Investor magazine. The team was interested in better understanding whether the analysts’ skills were portable when they switched companies. It turns out most analysts lost their stardom when they changed employers unless they moved to a better firm or brought their whole team along—with the exception of female analysts. Not only had the women studied a potential new employer more carefully before joining, they had also built their expertise differently than their male colleagues.

The top-performing female analysts had “built their franchises on portable, external relationships with clients and the companies they covered, rather than on relationships within their firms.” Or as one female star analyst put it: “For a woman in any business, it’s easier to focus outward, where you can define and deliver the services required to succeed, than to navigate the internal affiliations and power structure within a male-dominant firm.”

People choose organizations based on their preferences and their beliefs about whether or not they could thrive in a given organization. Messages shape those beliefs. Consider the messages sent when Lieutenant General David Morrison stated in a video posted on the Australian army’s official YouTube channel that he was committed to inclusion. “If that does not suit you, then get out,” Morrison flatly declared. “There is no place for you amongst this band of brothers and sisters.”

Acting in response to a 2013 investigation into sexual abuse, Morrison sent a strong message. In 2014, Morrison joined the Australian delegation to the Global Summit to End Sexual Violence in Conflict in London. Speaking again with admirable bluntness, he said that armies that assign more value to men than to women and tolerate sexual violence “do nothing to distinguish the soldier from the brute.”

Will these messages attract and retain soldiers valuing equality and inclusion? Time will tell. And while actions have followed his words, we all know that talk can be cheap. When and how messages affect behavior is a large field of inquiry in itself, but experimental evidence is rare. One example, however, is encouraging.

Robert Jensen and Emily Oster took advantage of the fact that cable television became available at different times in different parts of India, allowing them to trace whether attitudes and behaviors went along with exposure to the new information cable programming provided.

They found that the introduction of cable television was associated with improvements in women’s status in rural areas, including female school enrollment, decreases in fertility, as well as reported increases in autonomy and decreases in the acceptability of beating women and son preference. The information conveyed via cable television, often through somewhat surprising means, such as soap operas, exposed rural viewers to gender attitudes and ways of life, including within the household, more prevalent in urban areas. And it changed behavior.

Sorting mechanisms are powerful and often overlooked. Those charged with attracting the largest, most talented pool of applicants should make sure they scrutinize the messages, overt and biased, conveyed in their advertisements, websites, or other communications.

Excerpted from WHAT WORKS: GENDER EQUALITY BY DESIGN by Iris Bohnet, published by The Belknap Press of Harvard University Press. Copyright © 2016 by Iris Bohnet. Used by permission. All rights reserved.

Post A Comment

In order to be published, comments must be on-topic and civil in tone, with no name calling or personal attacks. Your comment may be edited for clarity and length.