Exploiting Neighborhood Interference with Low Order Interactions Under Unit Randomized Design

Watch a 51-minute lecture from Harvard CMSA where Cornell professor Christina Lee Yu explores network interference in causal effect estimation, focusing on how treatment assignments within social networks impact individual outcomes. Learn about estimating total treatment effects under Bernoulli randomized design, utilizing network structures with bounded degree constraints, and understanding low-order interactions among neighbors. Discover a novel framework that balances model flexibility with statistical complexity, particularly useful when analyzing well-connected networks that resist traditional clustering. Follow along as the presentation covers fundamental concepts of causal inference, neighborhood interference challenges, heterogeneous additive network effects, and the implementation of staggered rollout Bernoulli design in real-world applications.