Experiment design under network effects
Source: interviewintermediate
WHAT IT TESTS: awareness that SUTVA breaks under interference. OUTLINE: cluster-level randomization, graph or geo clustering to contain spillover, and analysis at the cluster unit.
WHAT IT TESTS: whether you recognize that standard A/B tests assume no interference between units, an assumption a referral feature violates. ANSWER OUTLINE: explain SUTVA and spillover, switch from user-level to cluster-level randomization where clusters are connected components, communities, or geographies so treatment effects stay contained, and analyze at the cluster level with appropriate variance handling.
Read the original → interview
- #network-effects
- #interference
- #cluster-randomization
- #ab-testing
- #causal-inference
Get five bites like this every day.
Tezvyn delivers a daily feed of 60-second tech bites with quizzes to lock in what you learn.