How would you estimate impact without A/B?

Company: Microsoft

Role: Data Scientist

Category: Analytics & Experimentation

Difficulty: medium

Interview Round: Technical Screen

Quick Answer: This question evaluates a data scientist's competency in causal inference, experimental design, metric definition and diagnostics, with specific emphasis on confounding, selection bias, interference, and Simpson's paradox.