Tag: gender
Want To Talk About Abortion On Fox? It Helps To Be A White Man

Want To Talk About Abortion On Fox? It Helps To Be A White Man

In segments discussing the Supreme Court draft opinion revealing the court has decided to repeal Roe v. Wade, Fox News hosted an overwhelming number of men and white people, largely excluding voices most heavily affected by the decision to strike down abortion rights in the United States. In contrast, CNN and MSNBC hosted a majority of women to discuss the news.

Key Findings

In cable news segments that discussed Roe or the draft opinion between May 2, when Politico published the leaked decision, through 5 p.m. EDT on May 5:

  • Men made up nearly two-thirds of all guest appearances on Fox News (64 percent).
  • The majority of guest appearances on CNN (63 percent) and MSNBC (69 percent) were by women.
  • White people made up 87 percent of guest appearances on Fox.
  • CNN and MSNBC featured white people in 70 percent and 62 percent of such guest appearances, respectively.
  • Across the three cable news networks, 21 percent of guest appearances were by women of color — 24 percent of appearances on CNN, 30 percent on MSNBC, and just seven percent on Fox News.
On the evening of May 2, Politico reported a leaked Supreme Court draft opinion by Justice Samuel Alito in the case Dobbs v. Jackson Women’s Health Organization, revealing the court is poised to overturn the landmark 1973 Roe v. Wade decision. The possible repeal would roll back the clock nearly half a century and return abortion law to each state. It would trigger legislation banning abortion in 13 states and would lay the groundwork for an additional 19 states to enforce pre-Roe abortion bans that are still on the books or to institute regressive laws banning abortion extremely early, in the weeks before fetal viability.

The potential ruling would impact people of color the most, and it has already been criticized not only for standing on flimsy and unprecedented legal ground, but also for defying broad and categorical public opinion. In the communities that would be most affected, 63 percent of women, 68 percent of Black adults, and 60 percent of Hispanic adults say abortion should be legal in all or most cases.

Even though this decision impacts women and people of color the most, Fox News predominately hosted men and white guests in segments that discussed the end of Roe: 87 percent of guest appearances in such segments were of white people, 64 percent were by men, and 58 percent were by white men. Only 36 percent were by women.

CNN and MSNBC fared slightly better when it came to featuring a diversity of voices. Both networks featured more guest appearances of women than men in segments on Roe, with CNN featuring women in 63 percent and MSNBC featuring them in 69 percent of such segments.

Across the three cable news networks, 21 percent of guest appearances were by women of color — 24 percent of appearances on CNN, 30 percent on MSNBC, and just seven percent on Fox News.

Black guests also comprised much larger proportions of guest appearances on CNN and MSNBC than on Fox. Approximately 22 percent of guest appearances in segments discussing the decision on CNN and 27 percent on MSNBC were of Black guests. By contrast, only five percent of such appearances on Fox were of Black guests.

All other races and ethnicities – Asian-American/Pacific Islander, Latino/Hispanic, Middle Eastern, and mutiracial guests – were featured in 6% or less of guest appearances on any of the three networks.

It should be noted, however, that MSNBC chose to platform anti-abortion activists in its programming, featuring segments with Kristan Hawkins of Students for Life and Marjorie Dannenfelser of Susan B. Anthony List. In her appearance, Hawkins was allowed to spread anti-abortion misinformation unchecked.

Fox’s decision to host mostly men and white people in segments discussing the opinion might explain why the network’s coverage was largely focused on the leak rather than the damaging impact from the repeal of Roe. Rather than airing segments featuring guests who could describe the genuine impacts of such a decision, Fox News shows like Hannity featured panels of white men to feign outrage over the leak.

When cable media outlets – particularly Fox News – fail to feature guests who can speak to the personal impacts of the story, they fail to adequately inform their viewers about the horrifying ramifications of overturning Roe.

Additional research contributions from Erin Kee

Methodology

Media Matters searched our internal database of all original, weekday programming on CNN, Fox News Channel, and MSNBC (shows airing from 6 a.m. through midnight) from May 2, 2022, when the report of the draft opinion came out, through 5 p.m. EDT May 5, 2022, for guest segments that touched on the leaked Supreme Court opinion overturning Roe v. Wade.

Media Matters searched our internal database of all original, weekday programming on CNN, Fox News Channel, and MSNBC (shows airing from 6 a.m. through midnight) from May 2, 2022, when the report of the draft opinion came out, through 5 p.m. EDT May 5, 2022, for guest segments that touched on the leaked Supreme Court opinion overturning Roe v. Wade.

We reviewed all guest participants in the identified segments for their sex or gender and their race or ethnicity. We classified an individual as “male” or “female” based on their self-identification or publicly available biographical information; no participants in this study publicly identified as nonbinary. We based an individual’s race or ethnicity on their self-identification or publicly available biographical information. If a guest participant’s race or ethnicity could not be determined through such means, we coded them as “unknown.” For guests who identified with multiple races or ethnicities, we coded them as “multiracial.” We used categories as defined by the U.S. Census with the addition of “Middle Eastern” as defined by the U.S. State Department.


We coded guest participants as “white” if they self-identify as white or are of European descent; as “Black” if they self-identify as African American or Black or are of African descent; as “Latinx/Hispanic” if they self-identify as Latino/Latina or Hispanic or are of Spanish/Latin American descent; as “Asian American/Pacific Islander” if they self-identify as Asian American or Pacific Islander or are of Asian descent or Pacific Island descent; or as “Middle Eastern” if they self-identify as Middle Eastern or are of Middle Eastern descent.

We rounded all percentages to the nearest whole.

Reprinted with permission from Media Matters.
Tucker Carlson And The Crisis Of Masculinity

Tucker Carlson And The Crisis Of Masculinity

Tucker Carlson's foray into testicle toasting is only the latest (and possibly most amusing) example of the right wing's masculinity obsession. The manliness theme keeps reappearing. Trump's strutting tough talk was imbibed greedily by fans eager for affirmation of the manly virtues.

Trump and his acolytes didn't invent this; insecure masculinity is an old phenomenon.

In the early years of the 20th century, Europe experienced something of a masculinity crisis. Popular writers, physicians and journalists began to fret that young Englishmen, Frenchmen and Germans had become soft after so many uninterrupted years of peace. In her magisterial history of the period, The War That Ended Peace, Margaret MacMillan traced the currents that coursed through European society in the years before the Great War. Francois Coppee, a French nationalist, worried that "Frenchmen are degenerating ... too absorbed in the race for enjoyment and luxury to retain that grand subordination of self to great causes which has been the historic glory of the French character." In Great Britain, Gen. Robert Baden-Powell founded the Boy Scouts in part because he feared the emasculation of England's youth.

In America, too, many feared that urbanization and industrialization had feminized men. Theodore Roosevelt glorified and personified the "strenuous life."

It's a universal worry. Russian President Vladimir Putin has portrayed himself shirtless on horseback, defeating opponents in hand-to-hand judo combat and shooting tigers (staged, of course). In 2021, the Chinese government banned "effeminate men" from TV and instructed broadcasters to "resolutely put an end to sissy men and other abnormal aesthetics." They were to depict only "revolutionary culture."

It's tempting to dismiss all of this as the pathetic bleats of hollow men who merit only derision. But as anthropologists, psychologists and historians alike can testify, the male need for validation is universal, and when societies fail to offer constructive paths for masculine expression, they court backlash. The negative aspects of masculinity are always lurking just beneath the surface.

In the past 60 years, America and the rest of the developed world have witnessed dramatic and precedent-shattering changes in women's status and in relations between men and women. Not all have been positive. Boys and men have felt neglected in the march toward "girl power" and "woman power." Cutting back on recess denies children not just an outlet for restless limbs but crucial social and emotional learning.

Girls are now outperforming boys at nearly every level of education. They earn 60 percent of bachelor's and master's degrees, and comprise 70 percent of high school valedictorians. Women are also dominating many workplaces. Women today hold a majority of the nation's jobs, including 51.4 percent of managerial and professional jobs — up from 26.1 percent in 1980.

The sexual and feminist revolutions of the 1960s delivered mixed signals to men. At first, the message was: "Women were just as randy as men, and sex was a romp and a frolic." Then it was: "No, wait, failing to get consent for every caress and kiss was assault." Masculinity itself was not a constitutive part of humanity; it was "toxic."

The other great upheaval of the past half-century is the decline of the two-parent family. The great dividing line in American life is not progressive versus conservative, urban versus rural, or black versus white. It's married versus not. For example, African American husbands have higher labor force participation rates than white bachelors. The upper third of the income distribution, who tend to marry and stay together, also tend to raise thriving children. By contrast, the lowest third, who mostly have revolving-door relationships without marriage, tend to have kids who don't. The middle third is more like the bottom than the top. Children in homes with a non-relative adult are 11 times more likely to be the victims of physical, sexual, or emotional abuse than those living with their biological or adoptive parents.

Boys are more disadvantaged than girls when they are raised by single mothers. Two MIT economists studied pairs of siblings in Florida between 1992 and 2002. They found that "Fatherless boys are less ambitious, less hopeful and more likely to get into trouble at school than fatherless girls." Being raised by a single mother significantly decreased the likelihood that a boy would attend college but had no similar effect on girls.

A significant percentage of American men are growing up without models of manliness in the form of fathers. They don't see a man shouldering responsibilities for his wife and children, helping with expenses (or covering them), joking with Mom, taking out the trash, tossing a ball with his kids, helping with homework or preparing a meal. Without a balanced picture of masculinity based upon their life experience, they search for masculinity elsewhere and often find a tawdry version offered up by the Carlsons and Putins of this world.

So, in a sense, we do have a masculinity crisis. We have large numbers of men who never marry, never support their kids and are loosely attached to the community. They are insecure about their masculinity for good reason — and that presents a problem for us all.

Mona Charen is policy editor of The Bulwark and host of the "Beg to Differ" podcast. Her most recent book is Sex Matters: How Modern Feminism Lost Touch with Science, Love, and Common Sense. To read features by other Creators Syndicate writers and cartoonists, visit the Creators Syndicate webpage at www.creators.com.

Printed with permission from Creators.

Misogyny Cannot Build A Better America

Misogyny Cannot Build A Better America

The Elizabeth Warren-Bernie Sanders fight makes good fodder for campaign reporters still trudging through the Democratic primary. The candidates’ exchange during the last debate managed to enliven a dull-as-dishwater two hours during which — apart from the “Can a woman be elected president?” tiff — no one said anything new or remotely newsworthy.

But the passing coverage of the friendly-fire feud manages to gloss over one of the most consequential movements of our time: the changing role of women in the workplace and, importantly, in the home. While the browning of America has animated and frightened voters on the right, so have massive cultural shifts — not only the increasing acceptance of equal rights for gay, lesbian and transgender citizens, but also, fundamentally, the increasing power and status given to women. That has shaken ultraconservatives to the core.

They won’t admit it, of course. While religious fundamentalists such as Franklin Graham regularly denounce same-sex marriage, and white nationalists such as Pat Buchanan frequently lash out against the coming demographic wave, few conservatives since the heyday of Phyllis Schlafly admit their reluctance to allow equal power and influence to women. It’s too small-minded, too misogynistic, too antediluvian.

Schlafly was herself a fascinating study in cynicism. Best known for her opposition to the Equal Rights Amendment, she was an attorney who ran for elected office and held high-profile positions in conservative political organizations while jetting around the country arguing for a return to old-fashioned gender roles.

To see Republican misogyny, you need look no further than the U.S. Senate and House of Representatives, where the vast majority of women who serve are Democrats, not Republicans. Of the 26 women currently serving in the Senate, 17 are Democrats. The 116th Congress ushered in a record number of women in the House — 102, nearly a quarter. Of those, 89 are Democrats. The person third in the line of succession to the presidency is one of those: House Speaker Nancy Pelosi.

Take another look at the fight over abortion rights. If you believe that most of the anti-abortion crowd are sincerely concerned about the lives of children, then you’ll have to explain why the same voters oppose any sort of government assistance — housing, child care, nutrition — for poor children once they emerge from the womb. There’s a reason Margaret Atwood’s The Handmaid’s Tale vaulted to renewed popularity after the election of President Donald J. Trump. Many women understand that the far right is trying to limit women’s choices, not save infants.

Or just examine the far right’s allegiance to Trump, who was elected after dramatic revelations of his abuse of women: He was caught on tape bragging about groping women against their will. Those boasts not only failed to turn off his conservative supporters, but Trump’s behavior also electrified them. Trump fits their ideal of the take-charge alpha male.

It is certainly more than possible for a woman to be elected president of the United States. As Sanders has pointed out, Hillary Clinton won the popular vote, garnering about 3 million more votes than Trump. If Warren were to win the nomination (Amy Klobuchar seems less likely to do so), Democrats would certainly unite behind her, and a strong anti-Trump sentiment could power her into the Oval Office. But that would only further inflame the ultraconservatives who pine for the old days when their values — and straight white men — ruled.

There is no doubt that the women’s movement has shaken up old systems and even created seismic shifts, shifts that a modern society has to reckon with. As just one example, birth rates are in decline across the Western world as women marry later and have fewer children. That has implications for families, for governments, for economies.

But there are no answers in sexism, no reasonable plans or policies in trying to turn back the clock to the time of Father Knows Best. Few thinking women (or men, for that matter) are interested in that. The drive to perfect the American promise of equality for all is ongoing, as it should be. Here’s to the election of a woman or man who will push us forward.

Despite Settlement, Facebook Ads Can Still Discriminate

Despite Settlement, Facebook Ads Can Still Discriminate

Reprinted with permission from ProPublica.

For Dolese Bros. Co. construction and supply company, which has a fleet of 300 trucks, recruiting enough qualified drivers in rural Oklahoma has been a challenge. The company has hung up banners at its plants. It has bought classified ads in newspapers. It has even turned its massive mixer trucks into moving billboards, with bumper stickers telling people how to apply.

But Dolese fills most jobs, according to community relations director Kermit Frank, by advertising on Facebook. On Nov. 4, the company placed a video ad featuring a longtime driver in a hardhat wiping down his truck, talking about all the reasons he appreciates the company: “Here, I’m home every night. And I make really good money. And I get to see my family a lot more.”

The company used Facebook’s new special ads portal, which doesn’t allow targeting by gender, age, race or ethnicity. That was fine with Dolese. While its drivers tend to be men, the company has no gender preference. “The gals we have in our group are fabulous,” Frank said. “We’d take any and all of them we could ever get.”

By the time the ad stopped running ten days later, more than 20,000 people had seen it. Eighty-seven percent of them were men.

In March, Facebook reached a “historic“ settlement of five lawsuits by civil rights groups. Under the terms of the settlement, the social media giant created the special ads portal to prevent discrimination in employment, housing and credit ads against legally protected groups such as women and older workers. The new portal also restricts Facebook’s algorithm from considering gender and age when finding audiences for these ads. “Getting this right,” Facebook CFO Sheryl Sandberg said in a press release, “is deeply important to me and all of us at Facebook because inclusivity is a core value for our company.”

But new research, as well as advertising information available on Facebook, suggest that the social media giant has not gotten this right. As Facebook promised in the settlement, advertisers on the new portal can no longer explicitly target by age or gender. Nevertheless, the composition of audiences can still tilt toward demographic groups such as men or younger workers, according to a study published today by researchers at Northeastern University and Upturn, a nonprofit group that focuses on digital inequities. ProPublica helped design the research with Northeastern and Upturn and placed some additional ads of its own.

“We’ve gone above and beyond others to help prevent discrimination in ads by restricting targeting and adding transparency,” Facebook spokesman Joe Osborne said in an emailed statement. “An advertiser determined to discriminate against people can do so on any online or offline medium today, which is why laws exist…We are the only digital media platform to make such meaningful changes in ads and we’re proud of our progress.” Osborne did not respond to questions about specific ads.

One reason for the persistent bias is that Facebook’s modified algorithm appears to rely on proxy characteristics that correlate with age or gender, said Alan Mislove, a Northeastern University professor of computer science and one of the study’s co-authors. “Our research shows that simply removing a few protected features from an algorithm is unlikely to provide any meaningful protection against discrimination,” Mislove said. If the advertiser provides a sample audience of software engineers, for example, that might be considered a proxy for male profiles.

Facebook’s ad delivery algorithm further skews the audience based on the content of the ad itself, the researchers previously found. As a result, even when advertisers try to reach a diverse audience, they aren’t always able to. Dolese’s ad, for example, could have reached a predominantly male audience because it featured a man, or because an interest in trucking acts as a proxy for maleness, or both. (A Dolese spokeswoman said the ad targeted categories “that would appeal to someone in this line of work.”) The settlement did not resolve the potential bias from proxies and ad content, but said Facebook would study the issue.

ProPublica spotted multiple real-world employment advertisements that favored men or excluded older potential applicants. We found these ads and others in Facebook’s ad library, an archive of advertisements on the platform. Some inactive ads in the library, such as those from Dolese, contain information about how they performed, including a breakdown of age and gender. The analysis does not contain information about the targeting choices the advertiser made on the front end.

In addition, testing by ProPublica found that housing and employment advertisers could circumvent the special ads portal and go through Facebook’s old system, which allowed them to target by age, race and gender. The settlement imposed a Sept. 30 deadline for implementing the new portal, but didn’t set a date for rerouting misplaced ads into it. A Facebook spokesman said the company stepped up policing on Dec. 4 with the launch of a special archive for housing. Employment and credit ads are expected to be added to the archive.

The findings by Northeastern and Upturn are likely to fuel the growing debate over whether algorithms that appear to favor one gender, age group or race over another violate the law even when they don’t explicitly consider such factors — an area of civil rights doctrine known as “disparate impact.” For example, the New York State Department of Financial Services is scrutinizing the criteria used by Apple’s credit card after reports last month that the company was extending smaller lines of credit to women than to men. Goldman Sachs, the bank behind the Apple Card, told ProPublica that it “has not and will never make decisions based on factors like gender, race, age, sexual orientation or any other legally prohibited factors when determining credit worthiness.” New York’s financial services department has also opened an investigation of Facebook over discrimination in ads for housing and other opportunities. The department declined to comment because the investigation is pending.

The Trump administration, which has backed away from Obama-era policies against disparate impact in education, is similarly retreating in its interpretation of housing law. In September, the U.S. Department of Housing and Urban Development introduced a rule that as long as an algorithm involved in housing decisions is not programmed with explicit information about protected classes, like race, or closely related proxies, like zip code, banks and landlords cannot be held liable for its disproportionate effect on people of color. In an apparent contradiction of this stance, HUD also has a pending lawsuit against Facebook, alleging that its advertising system violated fair housing laws. HUD didn’t respond to repeated requests for comment.

Osborne did not respond to emailed questions about the investigations by HUD and the New York financial services department.

Fair housing, employment and credit laws, which prohibit overt discrimination against protected groups such as women, minorities and older people, have been interpreted to apply to digital advertising. In September, the Equal Employment Opportunity Commission found that seven employers broke the law when they excluded women and older workers from seeing job ads on Facebook. The EEOC has said it is considering charges against dozens of more employers.

The settlement talks in the five lawsuits accusing Facebook of enabling advertising discrimination were confidential. But the civil rights groups involved in the case say that while they anticipated the potential for machine bias, the new portal’s potential for disparate impact doesn’t violate the current settlement.

Galen Sherwin, a senior staff attorney at the ACLU, which represented the plaintiffs in the civil rights lawsuits, called the findings “unfortunate but not unexpected. The settlement was a step forward to eliminate the most overt forms of targeting handed to advertisers, but we knew that it wouldn’t solve the continued problem of targeting advertising based on vast troves of user data. The removal of just a few data points is not going to change the targeting outcome. There’s still a lot of work to be done to eradicate the bias that remains on the platform.”

Since targeted advertising is the core of Facebook’s business model, any effort to reduce a marketer’s ability to target by age or gender can hurt the company’s bottom line. Facebook has acknowledged in SEC filings that its anti-discrimination initiatives have had “a small negative impact” on its advertising business.

The company, which controls 22 percent of the U.S. market share for digital ads according to eMarketer, has repeatedly come under fire for allowing advertisers to unlawfully discriminate against black homebuyers, older people and female debtors, among others. In 2016, ProPublica first reported that Facebook’s tools allowed housing marketers to exclude blacks, Hispanics and other groups protected by the Fair Housing Act. Facebook promised to fix the problem, but in 2017, we found that it hadn’t: Advertisers could still target housing ads to whites only and exclude African Americans, Jews and Spanish speakers from seeing them. This reporting spurred the five lawsuits resolved by this year’s settlement.

Prior to the March settlement, housing, credit and employment advertisers had access to a “Lookalike audience” feature, in which they could upload a list of users, known as a “source audience,” and Facebook would reach more people who “look like” them, based on criteria that included age, race and gender. Now, if these advertisers want to upload a source audience, they use a modified version of the “lookalike feature” known as a “special ad audience.” The algorithm for the special ad audiences no longer considers the fields on a Facebook user’s profile identifying “age, gender, relationship status, religious views, school, political views, interested in, or zip code” as inputs.

Nevertheless, the researchers found that Facebook’s algorithm perpetuates existing discrepancies in gender and age in a given source audience through a myriad of related characteristics. The study suggests that a similar bias would appear for race, but it is harder to document; Facebook gives advertisers data on the people they reached by age and gender, but not by race.

In one experiment, the researchers created two source audiences: one composed of more than 11,000 Facebook employees and the other of more than 11,000 random Americans. They uploaded each of these source audiences into Facebook’s new special ad audience system, ran the same generic employment ad — “Find your next career!” with a link to Indeed.com job search site — and compared the results. When the special audience was generated from the source list of Facebook’s employees, the ad was delivered to a new audience that was significantly biased along age and gender lines, reaching 88 percent men, nearly half of whom were aged 25 to 34.

These proportions roughly reflected Facebook’s workforce, which has beencriticized for its lack of diversity. According to its 2019 diversity report, 77 percent of the company’s technical workers are male. Facebook does not disclose ages, but according to the market research firm Statista, the median age of a Facebook employee is 28. In another indication that the algorithm tends to replicate the source demographics, half of this special audience lived in California, where Facebook is headquartered. By comparison, the ad audience generated from the list of random Americans reached 54 percent men, of whom 15 percent were aged 25 to 34, and 2 percent lived in California.

Such examples raise concerns that employers could still tailor their audience to focus on groups that they consider the most appealing job candidates, said Peter Romer-Friedman, a lawyer with Outten & Golden in Washington who represented several plaintiffs in the cases against Facebook. “It was certainly a fear of mine that advertisers in good or bad faith could circumvent anti-discrimination measures by uploading their own unbalanced list of Facebook users,” he said. “We knew that this would be a very real risk and that Facebook had not taken any measures to prevent discriminatory custom audience lists from being used.”

Also potentially contributing to disparate impact is Facebook’s ad delivery algorithm. To make sure that ads are seen by people who Facebook thinks are most likely to click on and engage with them, Facebook skews the audience depending on the content of the ad itself, said Piotr Sapiezynski, an associate research scientist at Northeastern and the lead author of the new study. How many women see a job listing for an open janitorial position, for instance, depends not just on what the advertiser tells Facebook, but on how Facebook interprets the advertisement’s image and text.

This effect persists in the new portal. The researchers uploaded a demographically representative audience and ran ads for artificial intelligence and supermarket jobs without any targeting. An ad for a supermarket job reached an audience that was 72 percent women and mostly aged 35 or older, whereas the audience for AI jobs was 66 percent male and almost exclusively younger than 35. Why the algorithm picked these audiences is not known, though it may have observed that women are more likely to shop for groceries and men are more likely to work in computer science.

Facebook’s system appears to draw similar conclusions about construction, a historically male trade. This assumption, however, can thwart advertisers who are trying to make the future of construction look different than its past.

The Chicago-based chapter of the International Union of Operating Engineers, Local 150, has long used Facebook, along with traditional TV and print advertising, to publicize an apprenticeship program for people interested in a career in construction. Ed Maher, the union’s communications director, said that the goal of its advertising was to attract diverse candidates and show that “there’s room for everyone in Local 150 trades.”

But the union’s goals may have been stymied, as Maher put it, by “nuances of the algorithm that lay entirely outside our control.” Several of the union’s Facebook’s ads, purchased in November, feature videos and photos of female members and women of color; the ads are not targeted by gender, but by location and to those interested in construction. Nevertheless, Maher said, the audience reached by the ads is about two-thirds men.

Paradoxically, because advertisers can no longer target by age or gender, they have little recourse to remedy these disproportions. “It’s an unintended consequence,” Mislove said.

“You can’t say steer it toward men instead. Facebook gives you no way to say, ‘I want this to be balanced.’”

The Muslim Public Affairs Council, a non-partisan advocacy group for Muslim representation, has experienced this constraint firsthand. Each summer, the non-profit runs a 10-week program for college-aged students in Washington D.C. Known as the Congressional Leadership Development Program, the internship offers housing, a stipend and the opportunity to work on the Hill. According to Ann Vallebuona, a digital media manager with the MPAC, the program goes out of its way to reach a diverse group of applicants. One Facebook ad, which started running on Nov. 12, read: “Apply to MPAC’s Congressional Leadership Development Program today. WORK ON CAPITOL HILL! Don’t wait. Apply now. 🏛” The accompanying video features six young Muslim American men and women, discussing the internship’s highlights over peppy techno beats.

When ProPublica told Vallebuona that Facebook’s archive indicated that the council’s ad was reaching 73 percent men, she was shocked. “It really is a quite alarming difference between females and males being reached,” she said, adding that far more women than men actually apply to the program.

Osborne did not respond to questions about whether Facebook’s algorithm considered words like “leadership” or “congressional” more relevant to men. He said that the ad system is designed to show ads to people most likely to take action, based on their behavior and intent.

Similarly, last week, the Christian charity Lifewater International advertised an engineering job to support its missions in Africa and Southeast Asia. Gary Weyel, director of marketing and communications at Lifewater International, said that he used the special ads portal to run his job advertisement, and had no desire to target by age or gender. Yet the ad’s audience was almost 70 percent men, and mostly between the ages of 18 and 34.

“I want to get this in front of qualified applicants, who are aligned with our mission and values. They’re making it difficult to do that, especially if there are behind-the-scenes algorithms like this,” he said.

Most of Facebook’s safeguards against bias live within the new special ads portal. Advertisers who want to run housing, credit or employment ads are supposed to click a button identifying their ads, and Facebook requires them to sign a “self-certifcation” statement that they have agreed to follow anti-discrimination laws.

As part of the settlement, the company has introduced an automated classification system to detect these “special ads” and route them into the portal. It also promised to add automated and human reviewers to make sure that advertisers were using the new system. But, at least until last week, Facebook may not have been catching many of the discriminatory housing, employment or credit ads bought through all of its traditional tools, which allow for targeting by age, gender and race.

Last month, ProPublica bought dozens of housing and employment ads on Facebook, but did not designate them as such. The classification program rejected a few of them, including one that read, “We’re hiring. Are you the right gentleman for the job?”

But it did not stop most of them, even when the ad copy specifically said, “We’re hiring” or, “Rent this home.”

We purchased multiple ads linking to employment sites such as Indeed.com, with copy such as “Come work with us,” targeting these mock opportunities exclusively to men under the age of 40. We also bought housing ads targeted by age, gender and family status. One ad only reached people between the ages of 25 to 45 with the relationship status of “married” and with an interest in parenting school-aged children or pre-teens.

The ads started running within minutes. ProPublica has since removed them.

Facebook’s Osborne said that the company had not been tightly monitoring how employment, housing and credit ads were purchased across its advertising platforms until Dec. 4. Facebook’s protections would have caught the ads had they run after that date, he said.

We may not be the only group to have run employment ads outside of the portal. Some ads in Facebook’s library have been shown to almost 100 percent men or 100 percent women, a breakdown that should be impossible without specific targeting. This includes a Nov. 4 ad from Barker and Sons Plumbing, recruiting plumbers in California into the “Barker and Sons family.” According to the ad, the jobs offered “competitive wages, comprehensive medical insurance, training programs and more.” It was shown only to men.

Several days later, Quantum Health, a medical benefits counseling firm, advertised openings for a “healthcare hero,” a “customer service pro,” or “a recent college graduate driven by a passion for helping people.”

The opportunities were shown almost exclusively to Facebook users under the age of 44.

Barker and Quantum did not respond to repeated requests for comment.

As the settlement requires, Facebook is studying how to protect against potential discrimination in ads, and engaging with academics, civil rights groups, and other experts, Osborne said. The company is not required to take specific actions or make this research public, according to plaintiffs in the case.

Steven Linder, a talent acquisition expert with with the Society for Human Resources Management, a coalition of human resources professionals, said employers concerned about legal liability for recruiting discrimination should take the new findings seriously: “If the algorithm is proving to be discriminatory, the employer should stop using Facebook,” he said. “That’s what I would advise my clients.”