Okay, it's time to add another layer to our conversation about gender analytics. I've mentioned the term intersectionality already, but now is our chance to understand it in depth. Let's start with the definition. Intersectionality is the complex, cumulative way in which the effects of multiple forms of discrimination combine, overlap, or intersect, especially in the experiences of marginalized people or groups. The term intersectionality itself was coined by Kimberlé Crenshaw, an American lawyer, activist, legal scholar, and critical race theorist. In the next session, you'll get a chance to watch her explain its origins in greater detail. Intersectionality is a way of understanding how individuals are differently impacted by inequality on the basis of factors such as gender, race, ethnicity, indigeneity, ability, age, immigration status, socioeconomic class, sexual orientation and religion, and other identities. Intersectional inequality affects women and visible minorities differently, depending on their social, cultural, and occupational contexts. Evidence suggests that many of the organizational initiatives promoting diversity and inclusion tend to benefit white women and not other underrepresented groups. In Canada, for example, we can see unequal workplace outcomes on the basis of intersectionality in the gendered and racialized wage gap. Visible minority women, especially first generation immigrants, earn on average $5000 less than nonvisible minority women and $7000 less than visible minority men. Visible minority women are more frequently employed in precarious jobs, characterized by insecurity, low wages, low protection, and limited benefits. Poverty rates for visible minority families are three times higher than for nonvisible minority families. Research has documented many other types of intersectional inequalities in the workplace. For example, white men often experience a glass escalator when working in female dominated occupations, such as nursing and teaching, that enables their promotion through the ranks more quickly. However, the same benefits do not extend to visible minority men. Research finds that black male nurses, for example, are perceived as less skilled than female nurses. Just as organizations and occupations may contribute to inequalities that intersect along lines of race and gender, such inequalities may also be built into the design of policies, products, services, and processes. For example, in the United Kingdom, the UK Women's Budget Group analyzed the cumulative impact of changes in taxes and benefits over the ten-year period 2010 to 2020. They looked at how gender intersected with income level by decile, where the first decile are the poorest people, and the tenth decile are the richest. What they found is that these changes in taxes and benefits were mainly worse for women than for men, and worse for poor people than rich people. And even further the only people who benefited at all where the richest men. On the chart, that's the light pink bar circled in green. They then looked at what it would mean if they added an intersection with race, comparing white, black, and Asian communities. Again, they found that the cuts were worse for women in every community, though dramatically worse for the poorest women in racialized minority groups, the purple bar circled in blue in the chart. Again, the only beneficiaries of the policies were richer man, both white and black, though not Asian men. Without breaking down the data in this way, policymakers exacerbated social inequalities. While the UK Women's Budget Group is an advisory organization that works outside of the UK government, the ideal way to use these kinds of statistics would be in designing policies, so as to avoid these kinds of negative effects. Rather than just evaluating policies after the fact to show that they hurt the most needy. This kind of intersectional analysis can be used to show how biases play out unevenly in products and services as well. The emergence of facial recognition technology has been accompanied by quite a bit of controversy, not least because the algorithms led to biased results. Researcher Joy Buolamwini, founder of the Algorithmic Justice League, discovered in her research at MIT that the machine learning software of leading technology companies was biased. Through her intersectional analysis, she found that IBM's Facial Recognition software, for example, could accurately guess the gender of light skin men 99.7% of the time, nearly perfect. But could only guess the gender of dark skin women 65.3% of the time. Facial recognition software is increasingly being used for security purposes, and such built-in bias has repercussions for people of color. For example, it may be more difficult for dark skinned individuals to use facial recognition software to unlock their phones. More significantly, AI bias and facial recognition may increase surveillance errors made by police,and lead to more frequent misidentification of criminals. Buolamwini's attention to the intersection of race and gender, actually led IBM to address the disparity in their facial recognition software. When she reanalyzed their algorithm after changes had been made, she found that the accuracy for assessing dark skinned individuals had jumped from a rate of 88% to 99.4% for men, and from that 65.3% to 83.5% for women. This is the promise of gender analytics, with real attention to the ways that bias may be built into their products and services, companies can mitigate the bias and create better products. As another example, makeup brands have faced increasing scrutiny for their failure to provide products for a wide range of skin tones. Historically, most products have favored light skinned people and left people of color with few options. While major makeup and fashion houses have tried to expand product offerings in recent years, they frequently come up short. In 2018, for example, Yves St. Laurent launched a new line of concealers that only included one shade of dark brown. Fenty Beauty, a makeup line created by pop star Rihanna, stands out for its emphasis on including a wide range of shades in its product offerings. The company launched with 40 shades of makeup and is perceived as offering the widest and most inclusive range of colors. The products were accompanied by a marketing campaign that prominently featured women of color. The brand has been so successful in providing a wider range of color shades that the media has dubbed its influence, the Fenty Effect. It has led to a chain reaction with other makeup brands launching a wider, more inclusive range in response to the positive reaction to Fenty. And if you follow the hashtag Disabled and Cute on Twitter, you'll know that most fashion lines have overlooked disabled women, even though in the US for example, one in four people have some type of disability. Again, another missed opportunity. In short, a failure to appreciate how gender intersects with other identities could exacerbate inequalities, or conceal insights for improved policies, or innovative new products and services. It can also have other far-reaching social consequences, as you will see in the next session, which presents as I mentioned before, a video of Kimberlé Crenshaw talking about the origin of the idea of intersectionality.
