Applying an intersectionality lens in data science

Data analytics is all about asking good questions and seeking informed answers that address a real need. But how do you decide which questions to ask, and whose needs are prioritized? Intersectionality, or overlapping race, gender, and other identity groupings that can result in discrimination, is a powerful framework to guide the design and execution of data science projects across industries and subject areas. An intersectionality framework challenges us to look more deeply at the structural and systemic underpinnings of the data and unpack assumptions around how the data is collected, analyzed, and presented in its context. At every stage of the data science project lifecycle, an intersectionality lens supports equitable results because it accounts for all identities within the scope of the project. In the definition and design phase, this requires attention to sources of subjectivity around the question or experiment to mitigate possible bias. During data collection, where and how data is collected determines whether the experiment is an accurate reflection of the environment and whether the representation of identities present leaves a population of people behind. Throughout the analysis, structuring and classifying data in traditional ways may not recognize the various groups that do not fall under "traditional" categories, opening the door to preconceived notions that can manifest as bias. Finally, this framework raises critical questions during the presentation and delivery phase: how is data from the experiment being framed? When presenting findings, are you subconsciously omitting negative outcomes?
*What is intersectionality?
*Opportunities for Bias in Data Science Projects
*Action Items