When Carol Bartz was appointed CEO at Yahoo in January 2009, the internet company was struggling. She was hired on a four-year contract and put forward a strategic plan to turn things around. But in September 2011 – in a phone call with Yahoo's chairman of the board – she was fired, just two years and eight months after she'd joined the company.

"They didn't even let [her plan] come to fruition," says Alison Cook, a professor of management at Utah State University in the US.

Bartz is one of countless female leaders given a precarious leadership position and left standing on the edge of a "glass cliff" with no support.

Research shows that women and people from ethnic minorities are more likely to be chosen to lead a company, sports team, or even country when it is in crisis mode. While those glass cliff positions can provide a way for some leaders to prove themselves, they come with significant downsides – including stress, burnout, and derailed careers.

Writing in The Times in 2003, journalist Elizabeth Judge claimed that women who had broken through the glass ceiling in British firms had "wreaked havoc on companies' performance and share prices", citing data suggesting that FTSE 100 companies with more women on their boards performed worse than those with only men.

There seems to be something about gender that makes us choose women in different types of crisis – Clara Kulich

Researchers at the University of Exeter in the UK set about investigating the apparent correlation. They looked at the share price of FTSE 100 companies immediately before and after the appointment of a board member to figure out which came first: the women or the drop in share price. They found that the relationship actually seemed to run the other way: companies that appointed women to their boards were already performing poorly.

During a crisis, leaders face more barriers to achieving their goals and a higher risk of failure

When they published their results, Michelle Ryan and Alex Haslam described the precariousness of these positions and the subtle discrimination that meant women were over-represented in them as a "glass cliff".

Subsequent research has found similar glass cliffs in politics, sports, and elsewhere. It doesn't just affect women: the glass cliff also extends to people belonging to ethnicities that are typically underrepresented in leadership.

Research by Cook and her colleague Christy Glass found that, in US college men's basketball, coaches belonging to racial minorities were more likely than white coaches to be promoted to losing teams. The researchers also analysed promotion patterns at Fortune 500 companies over a 15-year timespan and saw that – compared to white men – white women and both men and women of colour were more likely to be appointed as CEO in struggling firms.

In the 2005 UK general election, female Conservative party candidates contested harder-to-win seats than their male counterparts. Black and minority ethnic Conservative candidates faced a similar glass cliff across three UK general elections.

The glass cliff has been seen in companies, sports teams and countries

The glass cliff also shows up in experiments. When 80 undergraduate students were asked to choose a fictional candidate to stand in a UK by-election in either a safe seat or one held by the opposition, participants were more likely to choose a man for the safe seat and a woman for the risky one.

But the phenomenon doesn't appear everywhere, and, after failing to find a glass cliff in companies in the UK and Germany, some researchers have gone as far as to suggest it's a "myth".

Ryan, now director of the Global Institute for Women's Leadership at the Australian National University, says the fact the glass cliff is not seen universally reinforces that it is dependent on many factors. "Our research doesn't say that every woman faces the glass cliff, or that non-men occupy risky leadership positions – but rather that women are over-represented in glass cliff positions," she says.

A 2020 meta-analysis by Thekla Morgenroth at Purdue University in Indiana, US, and colleagues took in data from 74 existing studies and found that the evidence for the phenomenon was mixed.

In experimental studies, women were more likely to be selected over men as a leader in times of crisis, and this effect was more pronounced in countries that have more gender inequality. But the part of the analysis looking at real-world data notably didn't find a glass cliff in management – where the phenomenon was first uncovered – though it did find one in politics and the education and non-profit sectors.

"It might be that in the management domain, the glass cliff isn't a thing, or that there's more moderators and more nuances that we just didn't pick up on," says Morgenroth. It could be that it's no longer a problem in some sectors thanks to concerted action, for example, or that it only exists in some countries, and not others, because of cultural differences.

Explanations of the glass cliff fall into three main categories: those hiring a new CEO might think that stereotypically female traits are helpful in a crisis; a political party might want to signal to the outside world that they are changing; or, perhaps, decision-makers are simply prejudiced against certain groups of people.

It's likely that a combination of factors is at work. "Our research suggests that the glass cliff is multiply determined," says Ryan.

Research on gender stereotypes has found that in times of crisis people think that stereotypically "female" traits are more important in a leader, and that stereotypically "male" traits are less desirable. But those studies have tended to consider gendered traits and the actual gender of potential leaders interchangeably.

Those promoted to leadership in difficult times risk higher levels of stress, burnout and derailed careers

In work published last year, Clara Kulich at the University of Geneva, Switzerland, and Leire Gartzia at the University of Deusto, Spain, alongside colleagues, devised a study to look at gender and gender stereotypes separately.

They asked participants to read about a fictional company in three scenarios: internal disharmony among employees, poor financial management, or no crisis. Then they were given brief CVs of candidates for CEO, featuring a short description of their working style peppered with gendered traits, such as that they are considerate and kind, or independent and decisive.

What they found was that candidates described with stereotypically female or communal traits, like being kind, were favoured for the company in a relational crisis, whereas candidates described with stereotypically male traits, like being decisive, were favoured for the financial crisis. Crucially, this was true independent of whether the candidates with these traits presented as male or female.

Analysing their results to separate out the effect of gender, the researchers saw that women were preferred for the relational crisis, then the financial crisis, and lastly the no-crisis situation. In other words, women were preferred in a crisis even when stereotypically feminine traits were not thought desirable. "There seems to be something about gender that makes us choose women in different types of crisis," says Kulich.

She says she thinks stereotyping is one part of the story, but it can't fully explain leadership choices.

If it's not about stereotypes, what is causing the glass cliff? Companies in a crisis might understandably want to show that they're trying to change. To do that, they might choose a new leader who is a marked change from their predecessor.

You're losing these really talented people out of your workforce, because they're struggling through those added barriers – Alison Cook

While there were fewer studies looking at a glass cliff for leaders from ethnic minorities as there are for women, those that do exist suggest it is just as steep. What's more, evidence shows that there is no difference in the size of the glass cliff for people from different minority ethnic backgrounds.

That lends weight to the glass cliff arising because companies want to signal a change. With this explanation "the actual stereotypes or the attributes that you assume these groups have don't really matter. It's really just about, 'Oh, they're different'," says Morgenroth. "It doesn't really matter how they're different as long as they're visibly different."

It could also point to prejudice as the driving force behind the glass cliff. In fact, there can be a fine line between using a new leader to signal change and using them as a scapegoat. "You just put them there until you have a better solution, but you don't really believe in what they can actually do," says Kulich.

In difficult times, people tend to think that stereotypically "female" traits are more important in a leader

Of course, not everyone who wants to see people from underrepresented groups in leadership positions has negative motives. Research in France and Switzerland found that left-wing participants tend to choose minority candidates to run for political office in hard-to-win seats, doing so because they believe the candidate could bring about real change.

The motives of decision-makers, and the support they give, can make a big difference to those in glass cliff positions. Research led by Sarah Robinson at the University of Geneva, Switzerland, shows female Democratic candidates in the US do well in hard-to-win seats, getting more votes and even sometimes winning, whereas Republican women don't fare as well.

"We don't measure the psychology that is going on in the heads of the Republicans and the Democrats, but it gives some reason to believe that in the Democratic Party, there is more support for female candidates," says Kulich.

Of course, there are two sides to every job offer – and the motives of those who accept precarious positions have only recently been looked at in detail.

"One of the early assertions was it might be the only opportunity that they think they're going to be getting, so they're going to go ahead and take it," says Cook. But her research shows that, at least for some minority leaders, this narrative needs flipping. "It's not that it's just happening to them," she says. "They're making this happen."

Focusing on women's disadvantage may also lead us to ignore men's privilege and advantage – Michelle Ryan

Cook interviewed 33 women and people of colour in senior leadership roles across a range of industries in the US and found that nearly all of them recounted multiple risky, make-or-break assignments during the course of their careers. Not only that, they said they actively chose these assignments in order to prove themselves and build a career.

"So many of them have had this agency throughout, and they've set themselves up to prove their leadership worth early on," she says. "They have these situations throughout their entire career. They are known as 'turnaround artists' […] that they're good in these really tough situations, because they've had to be over and over and over."

But going into a glass cliff position willingly doesn't mean there are no downsides. The "risk tax" that members of underrepresented groups pay throughout their careers is high.

For starters, if you don't make it through a risky assignment, you may not get a second chance. If you do make it through, chances are you'll have to keep proving yourself time and time again. "It just gets exhausting," says Cook.

She recalls an interviewee in her early 50s saying she intended to step down from her role soon because she couldn't take it anymore. "It was really hard to hear that," says Cook. "You're losing these really talented people out of your workforce, because they're struggling through those added barriers."

There's also the added pressure of feeling like you're not just representing yourself, but others like you. One of Cook's interviewees, a Latina senior executive, said she thought any failure of hers would disadvantage future Latina job candidates. Similarly, a black male executive said of assessing black candidates for senior roles: "If he fails, then it is an 'I told you so' from others. They may attribute [his or her] lack of success to race."

The idea that a minority leader's failure will provoke a return to the white, male norm is backed up by Cook's research showing that when a firm's performance declines during the tenure of an occupational minority CEO, they are likely to be replaced by a white man. The researchers call it the "saviour effect".

With greater social equality, the glass cliff could become a thing of the past

There's also evidence that some leaders do face more scrutiny than others. Activist investors – who buy shares in a company to shape it's direction – give female CEOs a harder time than their male counterparts. Women can also face shorter tenures as CEO than men, making the difficult task of turning around a failing company even harder.

"With the glass cliff [you're] starting with a deficit," says Cook. "You're not only asking them to help you succeed, but pull you out of the depths and then be successful. It's a tall order."

The consequences of the glass cliff could stretch further than individual careers and companies losing talented employees. "Having women in glass cliff positions can help reinforce gender stereotypes that women aren't good at leadership," says Ryan. "If women are in risky and precarious positions their performance might be evaluated more negatively, and hence the cycle of inequality continues."

As gender equality increases the glass cliff should, in theory, fade away. It already appears to be smaller in studies from countries that rank higher on equality, such as Switzerland and Germany, than in less gender-equal countries like the US and the UK. "These countries are fairly similar culturally," says Morgenroth. "It's surprising and maybe encouraging that even small differences in gender inequality can really make a difference."

There are signs, too, that taking action can stop the glass cliff forming in the first place.

While women standing as Conservative party candidates in recent UK elections have faced a glass cliff, their Labour party counterparts have not. According to one study, strategies used within Labour since 1997, including all-women shortlists in winnable seats, now mean their female candidates are equally likely as men to see election success.

"If you have some form of quotas it is much harder to have these biased decision processes," says Morgenroth. "That's a very practical thing that organisations can implement."

If a concerted effort to promote certain groups of people over others sounds like favouritism, consider the glass cliff from another perspective: in a paper reviewing a decade of work on the glass cliff, Ryan and colleagues suggest that cushy leadership positions given disproportionately to men – a "glass cushion" – could be at the root of the problem. "Focusing on women's disadvantage may also lead us to ignore men's privilege and advantage," she says.

While some leadership assignments will always be trickier than others, affording all kinds of leaders the same support could make the difficult ones a little less perilous.