Causal Inference in Complex Systems: Network Interference, Strategic Agents, and Beyond

on YouTube ↗49m

" summary

Argues that exact, finite-sample Fisher randomization test methods handle network interference by testing exposure contrasts (e.g., control spillovers vs pure control) instead of unit-level treatments, and introduces the null exposure graph whose biclique structure identifies imputable unit–assignment pairs. Application to a Medellín policing experiment shows focal-unit selection and throwing away non-imputable data are necessary tradeoffs, and the same graph view guides experimental design (maximize biclique size). Also reports minimax designs for temporal habituation where imbalanced randomized assignments perform best.

AI-written summary · about

" chapters (7)

1. 0:00Start
2. 0:50Introduction
3. 2:37Crime spillovers in Medellin, Colombia
4. 4:44Treatment habituation/Learning
5. 9:16Causal Inference
6. 16:43Design-based approach
7. 44:32Spin-off 2: Improving the experimental design

" related talks

• 24mA Friendly Introduction To CausalityAlex Dimakis · MLconf SF 2016
• 33mSimple Yet Efficient Estimators For Network Causal Inference...Christina Yu · Algorithmic Aspects of Causal Inference 2022
• 11mCausal Inference For Socio-Economic & Engineering SystemsAnish Agarwal · Meet the Fellows Welcome Event
• 30mPolicy Evaluation under InterferenceStefan Wager · Reinforcement Learning from Batch Data and Simulation
• 31mLearning And Testing Causal Models: A Property Testing ViewpointArnab Bhattacharyya · Learning from Interventions 2022
• 01h 08mAn Introduction to Potential Outcomes, DAGs and Single-World Intervention GraphsThomas S Richardson · Causality Boot Camp
• 30mCausal Estimands And InterventionsVanessa Didelez · Learning from Interventions 2022
• 01h 02mUnmeasured Confounding and More Recent Developments/Challenges in Causal DiscoveryDaniel Malinsky · Causality Boot Camp