Foundations

Routing Summary

This folder covers the conceptual foundations of applied econometrics and causal inference. Contains 4 notes plus the Randomization Inference sub-topic.

Sub-topics

Sub-topicNotesCovers
Randomization Inference5Fisher randomization test & the sharp null, sharp vs weak (Neyman) nulls, studentized randomization tests with dual finite-sample/asymptotic validity, permutation tests & exact inference — Wu & Ding (2021)

Concept Map

ConceptNoteTypeDepends OnKey Result
Four FAQs framework for organizing researchResearch Questions in EconometricsconceptMostly Harmless Econometrics - Overview, The Experimental Ideal, The Selection ProblemGood research answers a causal or descriptive question clearly
Random assignment as causal inference benchmarkThe Experimental IdealconceptThe Selection Problem, Research Questions in Econometrics, Regression and the CEFRandomization eliminates selection bias by design
Potential outcomes, selection bias decompositionThe Selection ProblemconceptResearch Questions in Econometrics, Mostly Harmless Econometrics - OverviewSelection bias = E[Y0i given Di=1] - E[Y0i given Di=0]
DAGs, forks/chains/colliders, backdoor adjustment, d-separationDirected Acyclic GraphsconceptThe Selection Problem, The Experimental IdealConditioning on valid adjustment set (no colliders, all forks/chains blocked) recovers causal effect from observational data

Notes

  • Research Questions in Econometrics — CONTAINS: Four FAQs framework, causal vs descriptive questions, organizing empirical research
  • The Experimental Ideal — CONTAINS: Random assignment, Tennessee STAR experiment, why experiments are the benchmark for causal inference
  • The Selection Problem — CONTAINS: Potential outcomes framework, selection bias decomposition, overview of methods that address it
  • Directed Acyclic Graphs — CONTAINS: DAG definition, forks/chains/colliders, conditioning rules, paths, backdoor paths, backdoor adjustment formula, d-separation, d-connection, valid adjustment sets, worked example with complex DAG, Simpson’s Paradox

Sources

See Also