Achieving Diversity Through Equity Analytics
- To identify, understand, and address inequities in organizations, students first must become adept at data generating processes.
- Some students must be convinced that sometimes equality generates inequity, and sometimes inequality generates equity.
- Classroom discussions about diversity, equity, and inclusion can be polarizing, but courses like Equity Analytics enable students to ground their beliefs in shared analytical approaches.
In recent years, many business school students have sought timely courses that address their growing interest in fostering a fair and just society. As sociological scholars, we typically have engaged these students by discussing our research on topics like racial disparity and, more broadly, inequality.
Yet, we have continued teaching standard courses on strategy, entrepreneurship, and management without taking these topics into account. But, recently, it occurred to us that students were becoming increasingly interested in what we study. We thought, “Our students could learn how to do what we do (conduct theoretically informed analysis) before assuming positions that enable them to address what we find (disparities in organizations).”
This insight led us to develop Equity Analytics, a course first offered at the University of Michigan’s Ross School of Business in Ann Arbor. This course is a complement to another course we offer called Analytics for Equitable Leadership. Equity Analytics addresses two increasingly salient challenges managers face today: increasingly diverse workforces and growing societal inequality.
Many managers wonder if their business practices are contributing to disparities in opportunities for employees, customers, suppliers, and other stakeholders. Yet, they often lack the analytical skills to answer such questions. Our course provides students with the skills to find the answers they seek and more.
What’s ‘Equal’ Versus What’s ‘Fair’
The course begins with the distinction between equality (equivalent treatment) and equity (fair, but not necessarily equivalent, treatment). The distinction highlights two mechanisms that can generate inequity: differential treatment and disparate impact. For example, a gender-blind algorithm based on gender-independent, individual characteristics satisfies the equivalent treatment condition. The algorithm can, nevertheless, exert disparate impact on women if the included characteristics are correlated with gender in a way that favors people who do not identify as women, which would then result in inequity.
In such cases, treating those characteristics differently for each gender group in the data is one way to make the algorithm more equitable. That is, it is equitable if one is willing to engage in differential treatment and, thus, violate the equality condition.
An important but seemingly paradoxical insight that we hope students will take from our course is this: Sometimes equality generates inequity, and sometimes inequality generates equity. Therefore, students must familiarize themselves with disparate impact and acknowledge that differential treatment can actually be both a cause of inequity and a remedy for it.
After considering accounts related to issues such as bias and merit, we address the (typically unvoiced) possibility that the distribution of income in the U.S. is produced by random luck.
This is a challenging insight to deliver because equal treatment is widely viewed as appropriate—especially in the United States, where the idea of equality is a central part of the U.S. Constitution and anti-discrimination law. This insight also can be especially difficult for many to acknowledge, because it compels them to consider systemic disadvantage and raises questions about the “fairness” of their own achievements.
To get students comfortable with the idea that meritorious achievement and systemic disadvantage are compatible with each other, we conduct an illustrative class exercise. We start with a summary histogram of income distribution in the U.S. and discuss how the observed inequality is generated.
After considering accounts related to issues such as bias and merit, we address the (typically unvoiced) possibility that the distribution is produced by random luck. To underscore the point, we play a coin-flipping game in which all players start with an equal allotment of points and then bet each other on the outcome of the coin flip. After multiple rounds, a few players tend to accumulate a disproportionate number of points. Many are left with none. The resulting distribution resembles the U.S. income distribution.
Students recognize that both distributions emerge from a process of cumulative advantage, with both luck and skill contributing to the observed outcomes. The relative contributions of skill and luck frame the animated discussion that follows, helping students get beyond either/or intuitions.
Such exercises instill in students the idea that they need analytical approaches that elucidate causes of disparity. Only by understanding data generating processes can they address those disparities in ways consistent with their notions of fairness.
Training Skilled Equity Analysts
Our pedagogy in Equity Analytics is informed not only by our research, but also by our institutional service. Ray Reagans is associate dean for diversity, equity, and inclusion at the MIT Sloan School of Management in Cambridge; Chris Rider serves on Michigan Ross’s faculty DEI committee. We have designed the course to teach students simple but insightful analytical frameworks. They learn that, if they are to be skilled equity analysts, they will need to follow three steps:
Use descriptive statistics to document a disparity. We ask our students to calculate group-level means and conduct basic difference-in-means tests to calculate disparities, such as pay inequity among a firm’s employees. Such disparities are typically a necessary but insufficient condition for inferring inequity. For example, students typically consider some pay gaps “fair,” such as those involving sales commissions, while viewing others as “unfair,” such as subjectively determined bonuses.
Consider data-generating processes. For example, we discuss with students how they might discover whether an organization is directing its employees to disparate career advancement prospects either due to differential treatment (assignment to “fast-track” versus “dead-end” jobs, for example) or disparate impact (advancement based on criteria that favor white men).
We also consider valuative processes in which employees can be rewarded differently for equivalent contributions, which some might view as a form of inequality. To adjudicate these possibilities, students engage in everything from multivariate regressions to wage decompositions. These considerations inform not only students’ analyses, but also their personal evaluations of equity.
Based on the results of their equity analytics, students might design an initiative to keep promising students enrolled in college, or they might design a field experiment to test for racial bias on an online platform.
Design interventions to close inequitable gaps in opportunities or outcomes. For example, based on the results of their equity analytics, students might design an initiative to keep promising students facing financial difficulties enrolled in college. Or, they might design a field experiment to test for racial bias on an online platform. Students learn to implement these interventions as randomized controlled trials or A/B tests, so that they produce credible evidence of what is and what is not effective.
Importantly, this pedagogical approach preserves students’ individual notions of fairness while establishing a common basis for analyzing disparities. The analytical emphasis helps students get beyond firsthand experience and intuitive preferences for equality. Students who believe that inequalities in income, wealth, or socioeconomic attainment are generated through meritocratic processes come to recognize that luck and bias also play a role.
Conversely, those who believe that inequality is unfair learn to appreciate the idea that recognitions and rewards can be distributed inequitably, even among people who demonstrate similar capabilities. In short, topics related to diversity, equity, and inclusion can be potentially polarizing in the classroom, but Equity Analytics enables students to ground their beliefs in shared analytical approaches.
We situate these discussions in various contexts, such as Airbnb, higher education, Hollywood, the National Football League (NFL), professional services, and technology. Students begin the Equity Analytics course with the resolve to identify, understand, and address disparities, and they complete the course with the skills to make achieving these goals possible. From an instructor’s perspective, it is the most rewarding course we have ever taught.
Equity Analytics as a DEI Imperative
No one person possesses sufficient firsthand experience with inequity to foster the classroom environment necessary to deliver this course. Addressing disparities related to race, gender, citizenship, age, sexual orientation, and other forms of human difference requires the engagement of professionals who have experienced, analyzed, and addressed inequity.
We have been fortunate to engage experts like Tammy Bennett, partner and chief diversity and inclusion officer of Dinsmore Shohl, LLP; Rachel Brooks, equity lead at Instagram; Ahmmad Brown, co-founder of EBDI; Pam Coukos, co-founder of Working IDEAL; Jeremi Duru, professor of law at American University; Cyrus Mehri, civil rights attorney who helped institute the NFL’s Rooney Rule; Aiwa Shirako, people innovation lab lead at Google; Melissa Thomas-Hunt, former global head of diversity and belonging at Airbnb; Rajkamal Vempati, head of human resources for Axis Bank; and many more.
These guest speakers bring equity analytics to life for students, by sharing stories that illustrate how the use of analytics has supported their organizational and institutional change efforts. As instructors, we greatly appreciate the validation that these speakers provide, because their skills in equity analytics have been central to their career achievements and ongoing efforts. Students appreciate learning more about how to balance conceptual frameworks, rigorous analyses, and practical guidance to close equity gaps.
It is our sincere hope that our initial effort will generate enthusiasm among educators to establish equity analytics courses at other business schools. To that end, we are writing custom cases and teaching notes for other instructors.
The first case, “Cyrus Mehri and the NFL’s Rooney Rule,” won first prize in the WDI Publishing Global DE&I Case Competition and is available from WDI and Harvard Business School Publishing. Forthcoming cases on the Black Lawyers Association of Cincinnati—Cincinnati Bar Association (featuring Tammy Bennet) and Axis Bank (featuring Rajkamal Vempati) will be available soon. And an “Equity Analytics Primer” that details typical analyses performed in class will also be published soon. (Those interested can contact Chris Rider for drafts and other teaching materials.)
Last, we are launching a live virtual cohort on ScholarSite and encourage instructors to join us for a student perspective. Should faculty be interested in launching an Equity Analytics course at their own institutions, we fully encourage them to reach out to us for guidance, materials, and other resources.