Player Preference for Local Creators — Metrics and Causal Estimation Without A/B Tests
Context
You are evaluating whether players on Roblox prefer games created by local creators (e.g., creator and player share the same country/region/language). You must:
-
Define clear primary and secondary metrics that capture “preference.”
-
If an A/B test is infeasible, propose a credible observational analysis plan that estimates causal effects while mitigating selection bias.
Assume you can label each game session with whether its creator is “local” to the player, and you have standard engagement and retention telemetry at the session and player level.
Tasks
-
Define primary metric(s) and secondary metric(s) to measure player preference for local creators.
-
When randomized testing is not possible, outline a causal analysis plan that addresses selection bias (e.g., matching, instrumental variables, difference-in-differences), including assumptions and validation.