##### Scenario A mobile-app redesign is shipped as a new version; users opt-in by upgrading, so a standard A/B test is not possible. ##### Question How would you measure the causal impact of the redesign when users self-select into the new version? Describe the causal-inference framework you would use, how you would construct comparable treatment/control groups, which features you would match or weight on besides past engagement, and how you would validate your assumptions. ##### Hints Explain propensity-score matching/weighting, covariate selection, balance checks, difference-in-differences or other robustness tests.