This question evaluates a data scientist's competency in causal inference and observational study design, specifically the ability to define estimands and identification assumptions for self-selected, staggered adoption of a product change.
A mobile app ships a redesigned UI as a new version. Users can choose to upgrade (opt-in). Because adoption is self-selected and staggered, a classic randomized A/B test is not feasible.
How would you measure the causal impact of the redesign when users self-select into the new version?
Describe:
Login required