Applying an intersectionality lens in data science
summary
Reframes predictive churn from “women vs men” to predicting employee churn overall, then disaggregate data into subpopulations (race, age, disability, sexual orientation) to spot intersectional effects — e.g., an example finding that Asian men with physical disabilities show higher churn risk; flags confirmation bias and recommends documenting assumptions, rerunning collection if gaps exist, and using identity wheel and unconscious bias training, plus the five whys and external reviewers to evaluate equity before delivery.
AI-written summary · about
chapters (13)
- 0:00Start
- 3:04Applying an Intersectionality Lens 03. Meet Jessica 04. Intersectionality in the Data Science Process
- 3:341. What is intersectionality? A quick introduction
- 4:25Intersectionality MARITAL STATUS female GENDER IDENTITY
- 10:30Meet Jessica!
- 10:48Jessica is a data scientist working in her first role in the human resources department of a Fortune 500 technology company
- 14:52Jessica will collect churn data based on the following factors
- 17:38Analysis and Algorithm Training Jessica trains her prediction model, which tells her that women are more likely to churn than men. She's ready to present her findings!
- 20:55Action Steps How to apply what you've learned to your work
- 21:31Deepen Awareness Self-awareness and unconscious bias training for data teams.
- 22:54Unpack Assumptions Document and challenge assumptions that arise and consult peers representative of your audience.
- 24:44Collect Equitably Disaggregate data for representation of intersectional groups.
- 25:14Analyze with Care Ask appropriate questions when exploring the data.
related talks
12mA New Strategy for Winning
54mHow diversity drives excellence in our data driven tech world
27mDesigning More Inclusive AI with More Inclusive Workflows
10mBeing a force for cultural change in your organization
25mNavigating Unconscious Bias
46mFrom the Unicorn’s Mouth: Stories of Managing Multiple Diverse Identities in Tech
41mDiversity and Inclusion in SRE: A Postmortem
48mKeynote: Data Science Community and You